Resolve security warning for macOS users (#249)

* Add support for macOS signing and notarization

Signed-off-by: Dan Luhring <dan.luhring@anchore.com>

* Use Docker to run the changelog generator locally

Signed-off-by: Dan Luhring <dan.luhring@anchore.com>

.github/ISSUE_TEMPLATE/bug_report.md

@@ -16,5 +16,5 @@ assignees: ''
 **Anything else we need to know?**:
 
 **Environment**:
-- syft version (use `syft version -v`):
+- Output of `syft version`:
 - OS (e.g: `cat /etc/os-release` or similar):

.github/scripts/mac-prepare-for-signing.sh

@@ -0,0 +1,31 @@
+#!/usr/bin/env bash
+set -eu
+
+CI_HOME="/Users/runner"
+if [[ "${HOME}" != "${CI_HOME}" ]]; then
+  printf "WARNING! It looks like this isn't the CI environment. This script modifies the macOS Keychain setup in ways you probably wouldn't want for your own machine. It also requires an Apple Developer ID Certificate that you shouldn't have outside of the CI environment.\n\nExiting early to make sure nothing bad happens.\n"
+  exit 1
+fi
+
+# Install gon (see https://github.com/mitchellh/gon for details).
+brew tap mitchellh/gon
+brew install mitchellh/gon/gon
+
+# Write signing certificate to disk from environment variable.
+CERT_FILE="$HOME/developer_id_certificate.p12"
+echo -n "$APPLE_DEVELOPER_ID_CERT" | base64 --decode > "$CERT_FILE"
+
+# In order to have all keychain interactions avoid an interactive user prompt, we need to control the password for the keychain in question, which means we need to create a new keychain into which we'll import the signing certificate and from which we'll later access this certificate during code signing.
+EPHEMERAL_KEYCHAIN="ci-ephemeral-keychain"
+EPHEMERAL_KEYCHAIN_PASSWORD="$(openssl rand -base64 100)"
+security create-keychain -p "${EPHEMERAL_KEYCHAIN_PASSWORD}" "${EPHEMERAL_KEYCHAIN}"
+
+# Import signing certificate into the keychain. (This is a pre-requisite for gon, which is invoked via goreleaser.)
+EPHEMERAL_KEYCHAIN_FULL_PATH="$HOME/Library/Keychains/${EPHEMERAL_KEYCHAIN}-db"
+security import "${CERT_FILE}" -k "${EPHEMERAL_KEYCHAIN_FULL_PATH}" -P "${APPLE_DEVELOPER_ID_CERT_PASS}" -T "$(command -v codesign)"
+
+# Setting the partition list for this certificate's private key to include "apple-tool:" and "apple:" allows the codesign command to access this keychain item without an interactive user prompt. (codesign is invoked by gon.)
+security set-key-partition-list -S "apple-tool:,apple:" -s -k "${EPHEMERAL_KEYCHAIN_PASSWORD}" "${EPHEMERAL_KEYCHAIN_FULL_PATH}"
+
+# Make this new keychain the user's default keychain, so that codesign will be able to find this certificate when we specify it during signing.
+security default-keychain -d "user" -s "${EPHEMERAL_KEYCHAIN_FULL_PATH}"

.github/scripts/mac-sign-and-notarize.sh

@@ -0,0 +1,16 @@
+#!/usr/bin/env bash
+set -eu
+
+IS_SNAPSHOT="$1"                        # e.g. "true", "false"
+
+if [[ "${IS_SNAPSHOT}" == "true" ]]; then
+  # This is a snapshot build —— skipping signing and notarization...
+  exit 0
+fi
+
+GON_CONFIG="$2"                         # e.g. "gon.hcl"
+NEW_DMG_NAME="$3"                       # e.g. "./dist/syft-0.1.0.dmg"
+ORIGINAL_DMG_NAME="./dist/output.dmg"   # This should match dmg output_path in the gon config file.
+
+gon "${GON_CONFIG}"
+mv -v "${ORIGINAL_DMG_NAME}" "${NEW_DMG_NAME}"

.github/scripts/update-version-file.sh

@@ -15,14 +15,9 @@ fi
 echo "creating and publishing version file"
 
 # create a version file for version-update checks
-echo "${VERSION}" | tee ${DISTDIR}/VERSION
+VERSION_FILE="${DISTDIR}/VERSION"
+echo "${VERSION}" | tee "${VERSION_FILE}"
 
 # upload the version file that supports the application version update check
-docker run --rm \
-    -i \
-    -e AWS_DEFAULT_REGION=us-west-2 \
-    -e AWS_ACCESS_KEY_ID=${AWS_ACCESS_KEY_ID} \
-    -e AWS_SECRET_ACCESS_KEY=${AWS_SECRET_ACCESS_KEY} \
-    -v $(pwd)/${DISTDIR}/:/distmount \
-    amazon/aws-cli \
-        s3 cp /distmount/VERSION s3://toolbox-data.anchore.io/${BIN}/releases/latest/VERSION
+export AWS_DEFAULT_REGION=us-west-2
+aws s3 cp "${VERSION_FILE}" s3://toolbox-data.anchore.io/${BIN}/releases/latest/VERSION

.github/workflows/acceptance-test.yaml

@@ -14,7 +14,7 @@ env:
 
 jobs:
   Build-Snapshot-Artifacts:
-    runs-on: ubuntu-latest
+    runs-on: macos-latest # We're creating these snapshot builds on macOS to be consistent with our release workflow's build process, which also takes place on macOS (due to code signing requirements).
     steps:
 
       - uses: actions/setup-go@v2
@@ -35,9 +35,9 @@ jobs:
             ${{ runner.os }}-go-${{ env.GO_VERSION }}-${{ hashFiles('**/go.sum') }}-
             ${{ runner.os }}-go-${{ env.GO_VERSION }}-
 
-      - name: Bootstrap dependencies
-        if: steps.cache.outputs.cache-hit != 'true'
-        run: make ci-bootstrap
+      - name: Bootstrap project dependencies
+        if: steps.bootstrap-cache.outputs.cache-hit != 'true'
+        run: make bootstrap
 
       - name: Import GPG key
         id: import_gpg
@@ -76,7 +76,8 @@ jobs:
   # Note: changing this job name requires making the same update in the .github/workflows/release.yaml pipeline
   Acceptance-Linux:
     needs: [ Build-Snapshot-Artifacts ]
-    runs-on: ubuntu-latest
+    # come Nov 30 2020 ubuntu-latest will be ubuntu-20.04, until then it needs to be explicitly referenced due to python 3.7 specific features being used
+    runs-on: ubuntu-20.04
     steps:
 
       - uses: actions/checkout@v2
@@ -126,7 +127,8 @@ jobs:
   # Note: changing this job name requires making the same update in the .github/workflows/release.yaml pipeline
   Inline-Compare:
     needs: [ Build-Snapshot-Artifacts ]
-    runs-on: ubuntu-latest
+    # come Nov 30 2020 ubuntu-latest will be ubuntu-20.04, until then it needs to be explicitly referenced due to python 3.7 specific features being used
+    runs-on: ubuntu-20.04
     steps:
 
       - uses: actions/checkout@v2
@@ -156,4 +158,4 @@ jobs:
           text: The syft acceptance tests have failed tragically!
         env:
           SLACK_WEBHOOK_URL: ${{ secrets.SLACK_TOOLBOX_WEBHOOK_URL }}
-        if: ${{ failure() }}
+        if: ${{ failure() }}

.github/workflows/codeql-analysis.yml

@@ -0,0 +1,71 @@
+# For most projects, this workflow file will not need changing; you simply need
+# to commit it to your repository.
+#
+# You may wish to alter this file to override the set of languages analyzed,
+# or to provide custom queries or build logic.
+name: "CodeQL Security Scan"
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    # The branches below must be a subset of the branches above
+    branches: [main]
+  schedule:
+    - cron: '0 0 * * 3'
+
+jobs:
+  analyze:
+    name: Analyze
+    runs-on: ubuntu-latest
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # Override automatic language detection by changing the below list
+        # Supported options are ['csharp', 'cpp', 'go', 'java', 'javascript', 'python']
+        language: ['go', 'python']
+        # Learn more...
+        # https://docs.github.com/en/github/finding-security-vulnerabilities-and-errors-in-your-code/configuring-code-scanning#overriding-automatic-language-detection
+
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v2
+      with:
+        # We must fetch at least the immediate parents so that if this is
+        # a pull request then we can checkout the head.
+        fetch-depth: 2
+
+    # If this run was triggered by a pull request event, then checkout
+    # the head of the pull request instead of the merge commit.
+    - run: git checkout HEAD^2
+      if: ${{ github.event_name == 'pull_request' }}
+
+    # Initializes the CodeQL tools for scanning.
+    - name: Initialize CodeQL
+      uses: github/codeql-action/init@v1
+      with:
+        languages: ${{ matrix.language }}
+        # If you wish to specify custom queries, you can do so here or in a config file.
+        # By default, queries listed here will override any specified in a config file. 
+        # Prefix the list here with "+" to use these queries and those in the config file.
+        # queries: ./path/to/local/query, your-org/your-repo/queries@main
+
+    # Autobuild attempts to build any compiled languages  (C/C++, C#, or Java).
+    # If this step fails, then you should remove it and run the build manually (see below)
+    - name: Autobuild
+      uses: github/codeql-action/autobuild@v1
+
+    # ℹ️ Command-line programs to run using the OS shell.
+    # 📚 https://git.io/JvXDl
+
+    # ✏️ If the Autobuild fails above, remove it and uncomment the following three lines
+    #    and modify them (or add more) to build your code if your project
+    #    uses a compiled language
+
+    #- run: |
+    #   make bootstrap
+    #   make release
+
+    - name: Perform CodeQL Analysis
+      uses: github/codeql-action/analyze@v1

.github/workflows/release.yaml

@@ -13,7 +13,7 @@ env:
 
 jobs:
   wait-for-checks:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-latest # This OS choice is arbitrary. None of the steps in this job are specific to either Linux or macOS.
     steps:
 
       - uses: actions/checkout@v2
@@ -82,7 +82,7 @@ jobs:
 
   release:
     needs: [ wait-for-checks ]
-    runs-on: ubuntu-latest
+    runs-on: macos-latest # Due to our code signing process, it's vital that we run our release steps on macOS.
     steps:
 
       - uses: actions/setup-go@v2
@@ -93,6 +93,7 @@ jobs:
         with:
           fetch-depth: 0
 
+      # We are expecting this cache to have been created during the "Build-Snapshot-Artifacts" job in the "Acceptance" workflow.
       - name: Restore bootstrap cache
         id: cache
         uses: actions/cache@v2
@@ -105,10 +106,6 @@ jobs:
             ${{ runner.os }}-go-${{ env.GO_VERSION }}-${{ hashFiles('**/go.sum') }}-
             ${{ runner.os }}-go-${{ env.GO_VERSION }}-
 
-      - name: Bootstrap dependencies
-        if: steps.cache.outputs.cache-hit != 'true'
-        run: make ci-bootstrap
-
       - name: Import GPG key
         id: import_gpg
         uses: crazy-max/ghaction-import-gpg@v2
@@ -126,12 +123,16 @@ jobs:
       - name: Build & publish release artifacts
         run: make release
         env:
-          GITHUB_TOKEN: ${{ secrets.TEMP_RELEASE_TOKEN }}
+          GITHUB_TOKEN: ${{ secrets.ANCHORE_GIT_READ_TOKEN }}
           GPG_PRIVATE_KEY: ${{ secrets.SIGNING_GPG_PRIVATE_KEY }}
           PASSPHRASE: ${{ secrets.SIGNING_GPG_PASSPHRASE }}
           SIGNING_FINGERPRINT: ${{ steps.import_gpg.outputs.fingerprint }}
           AWS_ACCESS_KEY_ID: ${{ secrets.TOOLBOX_AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.TOOLBOX_AWS_SECRET_ACCESS_KEY }}
+          APPLE_DEVELOPER_ID_CERT: ${{ secrets.APPLE_DEVELOPER_ID_CERT }} # Used during macOS code signing.
+          APPLE_DEVELOPER_ID_CERT_PASS: ${{ secrets.APPLE_DEVELOPER_ID_CERT_PASS }} # Used during macOS code signing.
+          AC_USERNAME: ${{ secrets.ENG_CI_APPLE_ID }} # Used during macOS notarization.
+          AC_PASSWORD: ${{ secrets.ENG_CI_APPLE_ID_PASS }} # Used during macOS notarization.
 
       - uses: 8398a7/action-slack@v3
         with:
@@ -145,4 +146,4 @@ jobs:
       - uses: actions/upload-artifact@v2
         with:
           name: artifacts
-          path: dist/**/*
+          path: dist/**/*

.github/workflows/static-unit-integration.yaml

@@ -2,6 +2,7 @@ name: 'Static Analysis + Unit + Integration'
 on:
   workflow_dispatch:
   push:
+  pull_request:
 jobs:
 
   Static-Analysis:

.gitignore

@@ -29,3 +29,6 @@ coverage.txt
 
 # Output of the go coverage tool, specifically when used with LiteIDE
 *.out
+
+# macOS Finder metadata
+.DS_STORE

.goreleaser.yaml

@@ -6,18 +6,37 @@ release:
   # If set to true, will not auto-publish the release. This is done to allow us to review the changelog before publishing.
   draft: true
 
-signs:
-  - artifacts: checksum
-    args: ["--output", "${signature}", "--detach-sign", "${artifact}"]
+  # This ensures any macOS signed artifacts get included with the release.
+  extra_files:
+    - glob: "./dist/*.dmg"
 
 builds:
   - binary: syft
+    id: syft
     env:
       - CGO_ENABLED=0
     goos:
       # windows not supported yet (due to jotframe)
       # - windows
       - linux
+    goarch:
+      - amd64
+    # Set the modified timestamp on the output binary to the git timestamp (to ensure a reproducible build)
+    mod_timestamp: '{{ .CommitTimestamp }}'
+    ldflags: |
+      -w
+      -s
+      -extldflags '-static'
+      -X github.com/anchore/syft/internal/version.version={{.Version}}
+      -X github.com/anchore/syft/internal/version.gitCommit={{.Commit}}
+      -X github.com/anchore/syft/internal/version.buildDate={{.Date}}
+      -X github.com/anchore/syft/internal/version.gitTreeState={{.Env.BUILD_GIT_TREE_STATE}}
+  # For more info on this macOS build, see: https://github.com/mitchellh/gon#usage-with-goreleaser
+  - binary: syft
+    id: syft-macos
+    env:
+      - CGO_ENABLED=0
+    goos:
       - darwin
     goarch:
       - amd64
@@ -31,6 +50,12 @@ builds:
       -X github.com/anchore/syft/internal/version.gitCommit={{.Commit}}
       -X github.com/anchore/syft/internal/version.buildDate={{.Date}}
       -X github.com/anchore/syft/internal/version.gitTreeState={{.Env.BUILD_GIT_TREE_STATE}}
+    hooks:
+      post: ./.github/scripts/mac-sign-and-notarize.sh "{{.IsSnapshot}}" "gon.hcl" "./dist/syft_{{.Tag}}_{{.Target}}.dmg"
+
+signs:
+  - artifacts: checksum
+    args: ["--output", "${signature}", "--detach-sign", "${artifact}"]
 
 nfpms:
   - license: "Apache 2.0"
@@ -50,7 +75,8 @@ brews:
 
 archives:
   - format: tar.gz
+    builds:
+      - syft # i.e. Linux only
     format_overrides:
       - goos: windows
         format: zip
-

Makefile

@@ -74,6 +74,10 @@ help:
 ci-bootstrap:
 	DEBIAN_FRONTEND=noninteractive sudo apt update && sudo -E apt install -y bc jq libxml2-utils
 
+.PHONY:
+ci-bootstrap-mac:
+	github_changelog_generator --version || sudo gem install github_changelog_generator
+
 .PHONY: bootstrap
 bootstrap: ## Download and install all go dependencies (+ prep tooling in the ./tmp dir)
 	$(call title,Bootstrapping dependencies)
@@ -122,7 +126,7 @@ validate-cyclonedx-schema:
 .PHONY: unit
 unit: fixtures ## Run unit tests (with coverage)
 	$(call title,Running unit tests)
-	go test -coverprofile $(COVER_REPORT) ./...
+	go test -coverprofile $(COVER_REPORT) $(shell go list ./... | grep -v anchore/syft/test)
 	@go tool cover -func $(COVER_REPORT) | grep total |  awk '{print substr($$3, 1, length($$3)-1)}' > $(COVER_TOTAL)
 	@echo "Coverage: $$(cat $(COVER_TOTAL))"
 	@if [ $$(echo "$$(cat $(COVER_TOTAL)) >= $(COVERAGE_THRESHOLD)" | bc -l) -ne 1 ]; then echo "$(RED)$(BOLD)Failed coverage quality gate (> $(COVERAGE_THRESHOLD)%)$(RESET)" && false; fi
@@ -130,7 +134,9 @@ unit: fixtures ## Run unit tests (with coverage)
 .PHONY: integration
 integration: ## Run integration tests
 	$(call title,Running integration tests)
-	go test -tags=integration ./test/integration
+
+	go test -v ./test/integration
+
 
 # note: this is used by CI to determine if the integration test fixture cache (docker image tars) should be busted
 integration-fingerprint:
@@ -224,8 +230,7 @@ acceptance-test-rpm-package-install: $(SNAPSHOTDIR)
 changelog-release:
 	@echo "Last tag: $(SECOND_TO_LAST_TAG)"
 	@echo "Current tag: $(VERSION)"
-	@docker run -i --rm  \
-		-v "$(shell pwd)":/usr/local/src/your-app ferrarimarco/github-changelog-generator \
+	@github_changelog_generator \
 		--user anchore \
 		--project $(BIN) \
 		-t ${GITHUB_TOKEN} \
@@ -239,8 +244,9 @@ changelog-release:
 
 .PHONY: changelog-unreleased
 changelog-unreleased: ## show the current changelog that will be produced on the next release (note: requires GITHUB_TOKEN set)
-	@docker run -it --rm  \
-		-v "$(shell pwd)":/usr/local/src/your-app ferrarimarco/github-changelog-generator \
+	@docker run -it --rm \
+		-v "$(shell pwd)":/usr/local/src/your-app \
+		ferrarimarco/github-changelog-generator \
 		--user anchore \
 		--project $(BIN) \
 		-t ${GITHUB_TOKEN} \
@@ -256,8 +262,12 @@ changelog-unreleased: ## show the current changelog that will be produced on the
 			/CHANGELOG.md
 
 .PHONY: release
-release: clean-dist changelog-release ## Build and publish final binaries and packages
+release: clean-dist ci-bootstrap-mac changelog-release ## Build and publish final binaries and packages. Intended to be run only on macOS.
 	$(call title,Publishing release artifacts)
+
+	# Prepare for macOS-specific signing process
+	.github/scripts/mac-prepare-for-signing.sh
+
 	# create a config with the dist dir overridden
 	echo "dist: $(DISTDIR)" > $(TEMPDIR)/goreleaser.yaml
 	cat .goreleaser.yaml >> $(TEMPDIR)/goreleaser.yaml
@@ -288,4 +298,4 @@ clean-dist:
 
 .PHONY: clean-json-schema-examples
 clean-json-schema-examples:
-	rm -f schema/json/examples/*
+	rm -f schema/json/examples/*

README.md

@@ -1,6 +1,7 @@
 # syft
 
-[![CircleCI](https://circleci.com/gh/anchore/syft.svg?style=svg)](https://circleci.com/gh/anchore/syft)
+[![Static Analysis + Unit + Integration](https://github.com/anchore/syft/workflows/Static%20Analysis%20+%20Unit%20+%20Integration/badge.svg)](https://github.com/anchore/syft/actions?query=workflow%3A%22Static+Analysis+%2B+Unit+%2B+Integration%22)
+[![Acceptance](https://github.com/anchore/syft/workflows/Acceptance/badge.svg)](https://github.com/anchore/syft/actions?query=workflow%3AAcceptance)
 [![Go Report Card](https://goreportcard.com/badge/github.com/anchore/syft)](https://goreportcard.com/report/github.com/anchore/syft)
 [![GitHub release](https://img.shields.io/github/release/anchore/syft.svg)](https://github.com/anchore/syft/releases/latest)
 [![License: Apache-2.0](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://github.com/anchore/syft/blob/main/LICENSE)
@@ -15,7 +16,7 @@ A CLI tool and go library for generating a Software Bill of Materials (SBOM) fro
 - Linux distribution identification (supports Alpine, BusyBox, CentOS/RedHat, Debian/Ubuntu flavored distributions)
 - Supports Docker and OCI image formats
 
-> :warning: **This is pre-release software** and it may not work as expected. If you encounter an issue, please [let us know using the issue tracker](https://github.com/anchore/syft/issues).
+If you encounter an issue, please [let us know using the issue tracker](https://github.com/anchore/syft/issues).
 
 ## Getting started
 
@@ -59,7 +60,7 @@ Where the `format`s available are:
 curl -sSfL https://raw.githubusercontent.com/anchore/syft/main/install.sh | sh -s -- -b /usr/local/bin
 
 # install a specific version into a specific dir
-curl -sSfL https://raw.githubusercontent.com/anchore/syft/main/install.sh | sh -s <RELEASE_VERSION> -b <SOME_BIN_PATH>
+curl -sSfL https://raw.githubusercontent.com/anchore/syft/main/install.sh | sh -s -- -b <SOME_BIN_PATH> <RELEASE_VERSION>
 ```
 
 **macOS**
@@ -111,5 +112,4 @@ check-for-app-update: true
 ## Future plans
 
 The following areas of potential development are currently being investigated:
-- Add CycloneDX to list of output formats
 - Establish a stable interchange format w/Grype

go.mod

@@ -8,7 +8,7 @@ require (
 	github.com/anchore/go-testutils v0.0.0-20200925183923-d5f45b0d3c04
 	github.com/anchore/go-version v1.2.2-0.20200701162849-18adb9c92b9b
 	github.com/anchore/stereoscope v0.0.0-20200925184903-c82da54e98fe
-	github.com/bmatcuk/doublestar v1.3.1
+	github.com/bmatcuk/doublestar v1.3.3
 	github.com/docker/docker v17.12.0-ce-rc1.0.20200309214505-aa6a9891b09c+incompatible
 	github.com/dustin/go-humanize v1.0.0
 	github.com/go-test/deep v1.0.6

go.sum

@@ -162,6 +162,8 @@ github.com/blakesmith/ar v0.0.0-20190502131153-809d4375e1fb/go.mod h1:PkYb9DJNAw
 github.com/blang/semver v3.5.0+incompatible/go.mod h1:kRBLl5iJ+tD4TcOOxsy/0fnwebNt5EWlYSAyrTnjyyk=
 github.com/bmatcuk/doublestar v1.3.1 h1:rT8rxDPsavp9G+4ZULzqhhUSaI/OPsTZNG88Z3i0xvY=
 github.com/bmatcuk/doublestar v1.3.1/go.mod h1:wiQtGV+rzVYxB7WIlirSN++5HPtPlXEo9MEoZQC/PmE=
+github.com/bmatcuk/doublestar v1.3.3 h1:pVP1d49CcQQaNOl+PI6sPybIrIOD/6sux31PFdmhTH0=
+github.com/bmatcuk/doublestar v1.3.3/go.mod h1:wiQtGV+rzVYxB7WIlirSN++5HPtPlXEo9MEoZQC/PmE=
 github.com/bombsimon/wsl/v2 v2.0.0/go.mod h1:mf25kr/SqFEPhhcxW1+7pxzGlW+hIl/hYTKY95VwV8U=
 github.com/bombsimon/wsl/v2 v2.2.0/go.mod h1:Azh8c3XGEJl9LyX0/sFC+CKMc7Ssgua0g+6abzXN4Pg=
 github.com/bombsimon/wsl/v3 v3.0.0/go.mod h1:st10JtZYLE4D5sC7b8xV4zTKZwAQjCH/Hy2Pm1FNZIc=

gon.hcl

@@ -0,0 +1,11 @@
+source = ["./dist/syft-macos_darwin_amd64/syft"] # The 'dist' directory path should ideally reference an env var, where the source of truth is the Makefile. I wasn't able to figure out how to solve this.
+bundle_id = "com.anchore.toolbox.syft"
+
+sign {
+  application_identity = "Developer ID Application: ANCHORE, INC. (9MJHKYX5AT)"
+}
+
+dmg {
+  output_path = "./dist/output.dmg"
+  volume_name = "Syft"
+}

internal/file/opener.go

@@ -0,0 +1,14 @@
+package file
+
+import (
+	"io"
+	"os"
+)
+
+type Opener struct {
+	path string
+}
+
+func (o Opener) Open() (io.ReadCloser, error) {
+	return os.Open(o.path)
+}

internal/file/test-fixtures/generate-zip-fixture.sh

@@ -1,4 +1,7 @@
 #!/usr/bin/env bash
 set -eux
 
-zip -r "$1" zip-source
+# $1 —— absolute path to destination file, should end with .zip, ideally
+# $2 —— absolute path to directory from which to add entries to the archive
+
+pushd "$2" && find . -print | zip "$1" -@ && popd

internal/file/zip_file_helpers_test.go

@@ -0,0 +1,131 @@
+package file
+
+import (
+	"fmt"
+	"io/ioutil"
+	"os"
+	"os/exec"
+	"path"
+	"path/filepath"
+	"syscall"
+	"testing"
+)
+
+var expectedZipArchiveEntries = []string{
+	"some-dir" + string(os.PathSeparator),
+	filepath.Join("some-dir", "a-file.txt"),
+	"b-file.txt",
+	"nested.zip",
+}
+
+// fatalIfError calls the supplied function. If the function returns a non-nil error, t.Fatal(err) is called.
+func fatalIfError(t *testing.T, fn func() error) {
+	t.Helper()
+
+	if fn == nil {
+		return
+	}
+
+	err := fn()
+	if err != nil {
+		t.Fatal(err)
+	}
+}
+
+// createZipArchive creates a new ZIP archive file at destinationArchivePath based on the directory found at
+// sourceDirPath.
+func createZipArchive(t *testing.T, sourceDirPath, destinationArchivePath string) error {
+	t.Helper()
+
+	cwd, err := os.Getwd()
+	if err != nil {
+		return fmt.Errorf("unable to get cwd: %+v", err)
+	}
+
+	cmd := exec.Command("./generate-zip-fixture.sh", destinationArchivePath, path.Base(sourceDirPath))
+	cmd.Dir = filepath.Join(cwd, "test-fixtures")
+
+	if err := cmd.Start(); err != nil {
+		return fmt.Errorf("unable to start generate zip fixture script: %+v", err)
+	}
+
+	if err := cmd.Wait(); err != nil {
+		if exiterr, ok := err.(*exec.ExitError); ok {
+			// The program has exited with an exit code != 0
+
+			// This works on both Unix and Windows. Although package
+			// syscall is generally platform dependent, WaitStatus is
+			// defined for both Unix and Windows and in both cases has
+			// an ExitStatus() method with the same signature.
+			if status, ok := exiterr.Sys().(syscall.WaitStatus); ok {
+				if status.ExitStatus() != 0 {
+					return fmt.Errorf("failed to generate fixture: rc=%d", status.ExitStatus())
+				}
+			}
+		} else {
+			return fmt.Errorf("unable to get generate fixture script result: %+v", err)
+		}
+	}
+
+	return nil
+}
+
+// setupZipFileTest encapsulates common test setup work for zip file tests. It returns a cleanup function,
+// which should be called (typically deferred) by the caller, the path of the created zip archive, and an error,
+// which should trigger a fatal test failure in the consuming test. The returned cleanup function will never be nil
+// (even if there's an error), and it should always be called.
+func setupZipFileTest(t *testing.T, sourceDirPath string) (func() error, string, error) {
+	t.Helper()
+
+	// Keep track of any needed cleanup work as we go
+	var cleanupFns []func() error
+	cleanup := func(fns []func() error) func() error {
+		return func() error {
+			for _, fn := range fns {
+				err := fn()
+				if err != nil {
+					return err
+				}
+			}
+
+			return nil
+		}
+	}
+
+	archivePrefix, err := ioutil.TempFile("", "syft-ziputil-archive-TEST-")
+	if err != nil {
+		return cleanup(cleanupFns), "", fmt.Errorf("unable to create tempfile: %+v", err)
+	}
+	cleanupFns = append(cleanupFns, func() error { return os.Remove(archivePrefix.Name()) })
+
+	destinationArchiveFilePath := archivePrefix.Name() + ".zip"
+	t.Logf("archive path: %s", destinationArchiveFilePath)
+	err = createZipArchive(t, sourceDirPath, destinationArchiveFilePath)
+	cleanupFns = append(cleanupFns, func() error { return os.Remove(destinationArchiveFilePath) })
+	if err != nil {
+		return cleanup(cleanupFns), "", err
+	}
+
+	cwd, err := os.Getwd()
+	if err != nil {
+		return cleanup(cleanupFns), "", fmt.Errorf("unable to get cwd: %+v", err)
+	}
+
+	t.Logf("running from: %s", cwd)
+
+	return cleanup(cleanupFns), destinationArchiveFilePath, nil
+}
+
+// TODO: Consider moving any non-git asset generation to a task (e.g. make) that's run ahead of running go tests.
+func ensureNestedZipExists(t *testing.T, sourceDirPath string) error {
+	t.Helper()
+
+	nestedArchiveFilePath := path.Join(sourceDirPath, "nested.zip")
+	err := createZipArchive(t, sourceDirPath, nestedArchiveFilePath)
+
+	if err != nil {
+		return fmt.Errorf("unable to create nested archive for test fixture: %+v", err)
+	}
+
+	return nil
+}

internal/file/zip_file_manifest.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"os"
 	"sort"
+	"strings"
 
 	"github.com/anchore/syft/internal"
 
@@ -26,7 +27,11 @@ func (z ZipFileManifest) GlobMatch(patterns ...string) []string {
 
 	for _, pattern := range patterns {
 		for entry := range z {
-			if GlobMatch(pattern, entry) {
+			// We want to match globs as if entries begin with a leading slash (akin to an absolute path)
+			// so that glob logic is consistent inside and outside of ZIP archives
+			normalizedEntry := normalizeZipEntryName(entry)
+
+			if GlobMatch(pattern, normalizedEntry) {
 				uniqueMatches.Add(entry)
 			}
 		}
@@ -56,3 +61,11 @@ func NewZipFileManifest(archivePath string) (ZipFileManifest, error) {
 	}
 	return manifest, nil
 }
+
+func normalizeZipEntryName(entry string) string {
+	if !strings.HasPrefix(entry, "/") {
+		return "/" + entry
+	}
+
+	return entry
+}

internal/file/zip_file_manifest_test.go

@@ -0,0 +1,109 @@
+package file
+
+import (
+	"encoding/json"
+	"os"
+	"path"
+	"testing"
+)
+
+func TestNewZipFileManifest(t *testing.T) {
+	cwd, err := os.Getwd()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	sourceDirPath := path.Join(cwd, "test-fixtures", "zip-source")
+	err = ensureNestedZipExists(t, sourceDirPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	cleanup, archiveFilePath, err := setupZipFileTest(t, sourceDirPath)
+	defer fatalIfError(t, cleanup)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	actual, err := NewZipFileManifest(archiveFilePath)
+	if err != nil {
+		t.Fatalf("unable to extract from unzip archive: %+v", err)
+	}
+
+	if len(expectedZipArchiveEntries) != len(actual) {
+		t.Fatalf("mismatched manifest: %d != %d", len(actual), len(expectedZipArchiveEntries))
+	}
+
+	for _, e := range expectedZipArchiveEntries {
+		_, ok := actual[e]
+		if !ok {
+			t.Errorf("missing path: %s", e)
+		}
+	}
+
+	if t.Failed() {
+		b, err := json.MarshalIndent(actual, "", "  ")
+		if err != nil {
+			t.Fatalf("can't show results: %+v", err)
+		}
+
+		t.Errorf("full result: %s", string(b))
+	}
+}
+
+func TestZipFileManifest_GlobMatch(t *testing.T) {
+	cwd, err := os.Getwd()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	sourceDirPath := path.Join(cwd, "test-fixtures", "zip-source")
+	err = ensureNestedZipExists(t, sourceDirPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	cleanup, archiveFilePath, err := setupZipFileTest(t, sourceDirPath)
+	//goland:noinspection GoNilness
+	defer fatalIfError(t, cleanup)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	z, err := NewZipFileManifest(archiveFilePath)
+	if err != nil {
+		t.Fatalf("unable to extract from unzip archive: %+v", err)
+	}
+
+	cases := []struct {
+		glob     string
+		expected string
+	}{
+		{
+			"/b*",
+			"b-file.txt",
+		},
+		{
+			"*/a-file.txt",
+			"some-dir/a-file.txt",
+		},
+		{
+			"**/*.zip",
+			"nested.zip",
+		},
+	}
+
+	for _, tc := range cases {
+		t.Run(tc.glob, func(t *testing.T) {
+			glob := tc.glob
+
+			results := z.GlobMatch(glob)
+
+			if len(results) == 1 && results[0] == tc.expected {
+				return
+			}
+
+			t.Errorf("unexpected results for glob '%s': %+v", glob, results)
+		})
+	}
+}

internal/file/zip_file_traversal.go

@@ -63,8 +63,8 @@ func TraverseFilesInZip(archivePath string, visitor func(*zip.File) error, paths
 	return nil
 }
 
-func ExtractFromZipToUniqueTempFile(archivePath, dir string, paths ...string) (map[string]io.Reader, error) {
-	results := make(map[string]io.Reader)
+func ExtractFromZipToUniqueTempFile(archivePath, dir string, paths ...string) (map[string]Opener, error) {
+	results := make(map[string]Opener)
 
 	// don't allow for full traversal, only select traversal from given paths
 	if len(paths) == 0 {
@@ -78,11 +78,21 @@ func ExtractFromZipToUniqueTempFile(archivePath, dir string, paths ...string) (m
 		if err != nil {
 			return fmt.Errorf("unable to create temp file: %w", err)
 		}
+		// we shouldn't try and keep the tempfile open as the returned result may have several files, which takes up
+		// resources (leading to "too many open files"). Instead we'll return a file opener to the caller which
+		// provides a ReadCloser. It is up to the caller to handle closing the file explicitly.
+		defer tempFile.Close()
 
 		zippedFile, err := file.Open()
 		if err != nil {
 			return fmt.Errorf("unable to read file=%q from zip=%q: %w", file.Name, archivePath, err)
 		}
+		defer func() {
+			err := zippedFile.Close()
+			if err != nil {
+				log.Errorf("unable to close source file=%q from zip=%q: %+v", file.Name, archivePath, err)
+			}
+		}()
 
 		if file.FileInfo().IsDir() {
 			return fmt.Errorf("unable to extract directories, only files: %s", file.Name)
@@ -103,12 +113,8 @@ func ExtractFromZipToUniqueTempFile(archivePath, dir string, paths ...string) (m
 			return fmt.Errorf("unable to reset file pointer (%s): %w", tempFile.Name(), err)
 		}
 
-		results[file.Name] = tempFile
+		results[file.Name] = Opener{path: tempFile.Name()}
 
-		err = zippedFile.Close()
-		if err != nil {
-			return fmt.Errorf("unable to close source file=%q from zip=%q: %w", file.Name, archivePath, err)
-		}
 		return nil
 	}
 

internal/file/zip_file_traversal_test.go

@@ -6,47 +6,14 @@ import (
 	"io"
 	"io/ioutil"
 	"os"
-	"os/exec"
+	"path"
 	"path/filepath"
 	"strings"
-	"syscall"
 	"testing"
 
 	"github.com/go-test/deep"
 )
 
-func generateFixture(t *testing.T, archivePath string) {
-	cwd, err := os.Getwd()
-	if err != nil {
-		t.Errorf("unable to get cwd: %+v", err)
-	}
-
-	cmd := exec.Command("./generate-zip-fixture.sh", archivePath)
-	cmd.Dir = filepath.Join(cwd, "test-fixtures")
-
-	if err := cmd.Start(); err != nil {
-		t.Fatalf("unable to start generate zip fixture script: %+v", err)
-	}
-
-	if err := cmd.Wait(); err != nil {
-		if exiterr, ok := err.(*exec.ExitError); ok {
-			// The program has exited with an exit code != 0
-
-			// This works on both Unix and Windows. Although package
-			// syscall is generally platform dependent, WaitStatus is
-			// defined for both Unix and Windows and in both cases has
-			// an ExitStatus() method with the same signature.
-			if status, ok := exiterr.Sys().(syscall.WaitStatus); ok {
-				if status.ExitStatus() != 0 {
-					t.Fatalf("failed to generate fixture: rc=%d", status.ExitStatus())
-				}
-			}
-		} else {
-			t.Fatalf("unable to get generate fixture script result: %+v", err)
-		}
-	}
-}
-
 func equal(r1, r2 io.Reader) (bool, error) {
 	w1 := sha256.New()
 	w2 := sha256.New()
@@ -67,54 +34,50 @@ func equal(r1, r2 io.Reader) (bool, error) {
 }
 
 func TestUnzipToDir(t *testing.T) {
-	archivePrefix, err := ioutil.TempFile("", "syft-ziputil-archive-TEST-")
+	cwd, err := os.Getwd()
 	if err != nil {
-		t.Fatalf("unable to create tempfile: %+v", err)
+		t.Fatal(err)
 	}
-	defer os.Remove(archivePrefix.Name())
-	// the zip utility will add ".zip" to the end of the given name
-	archivePath := archivePrefix.Name() + ".zip"
-	defer os.Remove(archivePath)
-	t.Logf("archive path: %s", archivePath)
 
-	generateFixture(t, archivePrefix.Name())
+	goldenRootDir := filepath.Join(cwd, "test-fixtures")
+	sourceDirPath := path.Join(goldenRootDir, "zip-source")
+	cleanup, archiveFilePath, err := setupZipFileTest(t, sourceDirPath)
+	defer fatalIfError(t, cleanup)
+	if err != nil {
+		t.Fatal(err)
+	}
 
-	contentsDir, err := ioutil.TempDir("", "syft-ziputil-contents-TEST-")
+	unzipDestinationDir, err := ioutil.TempDir("", "syft-ziputil-contents-TEST-")
+	defer os.RemoveAll(unzipDestinationDir)
 	if err != nil {
 		t.Fatalf("unable to create tempdir: %+v", err)
 	}
-	defer os.RemoveAll(contentsDir)
 
-	t.Logf("content path: %s", contentsDir)
+	t.Logf("content path: %s", unzipDestinationDir)
 
-	cwd, err := os.Getwd()
-	if err != nil {
-		t.Errorf("unable to get cwd: %+v", err)
-	}
-
-	t.Logf("running from: %s", cwd)
-
-	// note: zip utility already includes "zip-source" as a parent dir for all contained files
-	goldenRootDir := filepath.Join(cwd, "test-fixtures")
-	expectedPaths := 4
+	expectedPaths := len(expectedZipArchiveEntries)
 	observedPaths := 0
 
-	err = UnzipToDir(archivePath, contentsDir)
+	err = UnzipToDir(archiveFilePath, unzipDestinationDir)
 	if err != nil {
 		t.Fatalf("unable to unzip archive: %+v", err)
 	}
 
 	// compare the source dir tree and the unzipped tree
-	err = filepath.Walk(filepath.Join(contentsDir, "zip-source"),
+	err = filepath.Walk(unzipDestinationDir,
 		func(path string, info os.FileInfo, err error) error {
-			t.Logf("unzipped path: %s", path)
-			observedPaths++
+			// We don't unzip the root archive dir, since there's no archive entry for it
+			if path != unzipDestinationDir {
+				t.Logf("unzipped path: %s", path)
+				observedPaths++
+			}
+
 			if err != nil {
 				t.Fatalf("this should not happen")
 				return err
 			}
 
-			goldenPath := filepath.Join(goldenRootDir, strings.TrimPrefix(path, contentsDir))
+			goldenPath := filepath.Join(sourceDirPath, strings.TrimPrefix(path, unzipDestinationDir))
 
 			if info.IsDir() {
 				i, err := os.Stat(goldenPath)
@@ -156,12 +119,11 @@ func TestUnzipToDir(t *testing.T) {
 	}
 
 	if observedPaths != expectedPaths {
-		t.Errorf("missed test paths: %d!=%d", observedPaths, expectedPaths)
+		t.Errorf("missed test paths: %d != %d", observedPaths, expectedPaths)
 	}
-
 }
 
-func TestExtractFilesFromZipFile(t *testing.T) {
+func TestContentsFromZip(t *testing.T) {
 	archivePrefix, err := ioutil.TempFile("", "syft-ziputil-archive-TEST-")
 	if err != nil {
 		t.Fatalf("unable to create tempfile: %+v", err)
@@ -172,7 +134,10 @@ func TestExtractFilesFromZipFile(t *testing.T) {
 	defer os.Remove(archivePath)
 	t.Logf("archive path: %s", archivePath)
 
-	generateFixture(t, archivePrefix.Name())
+	err = createZipArchive(t, "zip-source", archivePrefix.Name())
+	if err != nil {
+		t.Fatal(err)
+	}
 
 	cwd, err := os.Getwd()
 	if err != nil {
@@ -181,8 +146,8 @@ func TestExtractFilesFromZipFile(t *testing.T) {
 
 	t.Logf("running from: %s", cwd)
 
-	aFilePath := filepath.Join("zip-source", "some-dir", "a-file.txt")
-	bFilePath := filepath.Join("zip-source", "b-file.txt")
+	aFilePath := filepath.Join("some-dir", "a-file.txt")
+	bFilePath := filepath.Join("b-file.txt")
 
 	expected := map[string]string{
 		aFilePath: "A file! nice!",
@@ -207,60 +172,4 @@ func TestExtractFilesFromZipFile(t *testing.T) {
 
 		t.Errorf("full result: %s", string(b))
 	}
-
-}
-
-func TestZipFileManifest(t *testing.T) {
-	archivePrefix, err := ioutil.TempFile("", "syft-ziputil-archive-TEST-")
-	if err != nil {
-		t.Fatalf("unable to create tempfile: %+v", err)
-	}
-	defer os.Remove(archivePrefix.Name())
-	// the zip utility will add ".zip" to the end of the given name
-	archivePath := archivePrefix.Name() + ".zip"
-	defer os.Remove(archivePath)
-	t.Logf("archive path: %s", archivePath)
-
-	generateFixture(t, archivePrefix.Name())
-
-	cwd, err := os.Getwd()
-	if err != nil {
-		t.Errorf("unable to get cwd: %+v", err)
-	}
-
-	t.Logf("running from: %s", cwd)
-
-	expected := []string{
-		filepath.Join("zip-source") + string(os.PathSeparator),
-		filepath.Join("zip-source", "some-dir") + string(os.PathSeparator),
-		filepath.Join("zip-source", "some-dir", "a-file.txt"),
-		filepath.Join("zip-source", "b-file.txt"),
-	}
-
-	actual, err := NewZipFileManifest(archivePath)
-	if err != nil {
-		t.Fatalf("unable to extract from unzip archive: %+v", err)
-	}
-
-	if len(expected) != len(actual) {
-		t.Fatalf("mismatched manifest: %d != %d", len(actual), len(expected))
-	}
-
-	for _, e := range expected {
-		_, ok := actual[e]
-		if !ok {
-			t.Errorf("missing path: %s", e)
-		}
-	}
-
-	if t.Failed() {
-
-		b, err := json.MarshalIndent(actual, "", "  ")
-		if err != nil {
-			t.Fatalf("can't show results: %+v", err)
-		}
-
-		t.Errorf("full result: %s", string(b))
-	}
-
 }

internal/parse.go

@@ -0,0 +1,15 @@
+package internal
+
+import "regexp"
+
+// MatchCaptureGroups takes a regular expression and string and returns all of the named capture group results in a map.
+func MatchCaptureGroups(regEx *regexp.Regexp, str string) map[string]string {
+	match := regEx.FindStringSubmatch(str)
+	results := make(map[string]string)
+	for i, name := range regEx.SubexpNames() {
+		if i > 0 && i <= len(match) {
+			results[name] = match[i]
+		}
+	}
+	return results
+}

schema/README.md

@@ -0,0 +1,9 @@
+## Updating the JSON schema
+Today the JSON schema is generated from integration test data. Specifically, when integration tests are run, the `/schema/json/examples` directory is populated with syft JSON output data. This examples directory is used to drive automatically generating the JSON schema.
+The caveats with this approach is:
+1) the JSON schema is only as good as the examples provided
+2) there is an integration test that ensures that the JSON schema is valid relative to what the code currently generates.
+This means to update the JSON schema you need to
+1) Open up `test/integration/json_schema_test.go` and comment out invocations of the `validateAgainstV1Schema` function.
+2) From the root of the repo run `generate-json-schema`. Now there should be a new schema generated at `/schema/json/schema.json`
+3) Uncomment the `validateAgainstV1Schema` function.

schema/json/schema.json

@@ -40,8 +40,11 @@
                             "architecture": {
                                 "type": "string"
                             },
-                            "authors": {
-                                "type": "null"
+                            "author": {
+                                "type": "string"
+                            },
+                            "authorEmail": {
+                                "type": "string"
                             },
                             "description": {
                                 "type": "string"
@@ -60,6 +63,21 @@
                                                 "checksum": {
                                                     "type": "string"
                                                 },
+                                                "digest": {
+                                                    "properties": {
+                                                        "algorithm": {
+                                                            "type": "string"
+                                                        },
+                                                        "value": {
+                                                            "type": "string"
+                                                        }
+                                                    },
+                                                    "required": [
+                                                        "algorithm",
+                                                        "value"
+                                                    ],
+                                                    "type": "object"
+                                                },
                                                 "ownerGid": {
                                                     "type": "string"
                                                 },
@@ -71,14 +89,13 @@
                                                 },
                                                 "permissions": {
                                                     "type": "string"
+                                                },
+                                                "size": {
+                                                    "type": "string"
                                                 }
                                             },
                                             "required": [
-                                                "checksum",
-                                                "ownerGid",
-                                                "ownerUid",
-                                                "path",
-                                                "permissions"
+                                                "path"
                                             ],
                                             "type": "object"
                                         }
@@ -89,6 +106,9 @@
                             "gitCommitOfApkPort": {
                                 "type": "string"
                             },
+                            "homepage": {
+                                "type": "string"
+                            },
                             "installedSize": {
                                 "type": "integer"
                             },
@@ -105,116 +125,90 @@
                                 "type": "string"
                             },
                             "manifest": {
-                                "anyOf": [
-                                    {
-                                        "type": "null"
-                                    },
-                                    {
+                                "properties": {
+                                    "main": {
                                         "properties": {
-                                            "extraFields": {
-                                                "properties": {
-                                                    "Archiver-Version": {
-                                                        "type": "string"
-                                                    },
-                                                    "Build-Jdk": {
-                                                        "type": "string"
-                                                    },
-                                                    "Built-By": {
-                                                        "type": "string"
-                                                    },
-                                                    "Created-By": {
-                                                        "type": "string"
-                                                    },
-                                                    "Extension-Name": {
-                                                        "type": "string"
-                                                    },
-                                                    "Group-Id": {
-                                                        "type": "string"
-                                                    },
-                                                    "Hudson-Version": {
-                                                        "type": "string"
-                                                    },
-                                                    "Jenkins-Version": {
-                                                        "type": "string"
-                                                    },
-                                                    "Long-Name": {
-                                                        "type": "string"
-                                                    },
-                                                    "Main-Class": {
-                                                        "type": "string"
-                                                    },
-                                                    "Minimum-Java-Version": {
-                                                        "type": "string"
-                                                    },
-                                                    "Plugin-Dependencies": {
-                                                        "type": "string"
-                                                    },
-                                                    "Plugin-Developers": {
-                                                        "type": "string"
-                                                    },
-                                                    "Plugin-License-Name": {
-                                                        "type": "string"
-                                                    },
-                                                    "Plugin-License-Url": {
-                                                        "type": "string"
-                                                    },
-                                                    "Plugin-ScmUrl": {
-                                                        "type": "string"
-                                                    },
-                                                    "Plugin-Version": {
-                                                        "type": "string"
-                                                    },
-                                                    "Short-Name": {
-                                                        "type": "string"
-                                                    }
-                                                },
-                                                "required": [
-                                                    "Archiver-Version",
-                                                    "Build-Jdk",
-                                                    "Built-By",
-                                                    "Created-By"
-                                                ],
-                                                "type": "object"
-                                            },
-                                            "implementationTitle": {
+                                            "Archiver-Version": {
                                                 "type": "string"
                                             },
-                                            "implementationVendor": {
+                                            "Build-Jdk": {
                                                 "type": "string"
                                             },
-                                            "implementationVersion": {
+                                            "Built-By": {
                                                 "type": "string"
                                             },
-                                            "manifestVersion": {
+                                            "Created-By": {
                                                 "type": "string"
                                             },
-                                            "name": {
+                                            "Extension-Name": {
                                                 "type": "string"
                                             },
-                                            "specificationTitle": {
+                                            "Group-Id": {
                                                 "type": "string"
                                             },
-                                            "specificationVendor": {
+                                            "Hudson-Version": {
                                                 "type": "string"
                                             },
-                                            "specificationVersion": {
+                                            "Implementation-Title": {
+                                                "type": "string"
+                                            },
+                                            "Implementation-Version": {
+                                                "type": "string"
+                                            },
+                                            "Jenkins-Version": {
+                                                "type": "string"
+                                            },
+                                            "Long-Name": {
+                                                "type": "string"
+                                            },
+                                            "Main-Class": {
+                                                "type": "string"
+                                            },
+                                            "Manifest-Version": {
+                                                "type": "string"
+                                            },
+                                            "Minimum-Java-Version": {
+                                                "type": "string"
+                                            },
+                                            "Plugin-Dependencies": {
+                                                "type": "string"
+                                            },
+                                            "Plugin-Developers": {
+                                                "type": "string"
+                                            },
+                                            "Plugin-License-Name": {
+                                                "type": "string"
+                                            },
+                                            "Plugin-License-Url": {
+                                                "type": "string"
+                                            },
+                                            "Plugin-ScmUrl": {
+                                                "type": "string"
+                                            },
+                                            "Plugin-Version": {
+                                                "type": "string"
+                                            },
+                                            "Short-Name": {
+                                                "type": "string"
+                                            },
+                                            "Specification-Title": {
                                                 "type": "string"
                                             }
                                         },
                                         "required": [
-                                            "extraFields",
-                                            "implementationTitle",
-                                            "implementationVendor",
-                                            "implementationVersion",
-                                            "manifestVersion",
-                                            "name",
-                                            "specificationTitle",
-                                            "specificationVendor",
-                                            "specificationVersion"
+                                            "Archiver-Version",
+                                            "Build-Jdk",
+                                            "Built-By",
+                                            "Created-By",
+                                            "Manifest-Version"
                                         ],
                                         "type": "object"
                                     }
-                                ]
+                                },
+                                "required": [
+                                    "main"
+                                ],
+                                "type": "object"
                             },
                             "name": {
                                 "type": "string"
@@ -225,203 +219,11 @@
                             "package": {
                                 "type": "string"
                             },
-                            "parentPackage": {
-                                "anyOf": [
-                                    {
-                                        "type": "null"
-                                    },
-                                    {
-                                        "properties": {
-                                            "foundBy": {
-                                                "type": "string"
-                                            },
-                                            "language": {
-                                                "type": "integer"
-                                            },
-                                            "licenses": {
-                                                "type": "null"
-                                            },
-                                            "manifest": {
-                                                "type": "string"
-                                            },
-                                            "metadata": {
-                                                "properties": {
-                                                    "manifest": {
-                                                        "properties": {
-                                                            "extraFields": {
-                                                                "properties": {
-                                                                    "Archiver-Version": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Build-Jdk": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Built-By": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Created-By": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Extension-Name": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Group-Id": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Hudson-Version": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Jenkins-Version": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Long-Name": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Main-Class": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Minimum-Java-Version": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Plugin-Dependencies": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Plugin-Developers": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Plugin-License-Name": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Plugin-License-Url": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Plugin-ScmUrl": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Plugin-Version": {
-                                                                        "type": "string"
-                                                                    },
-                                                                    "Short-Name": {
-                                                                        "type": "string"
-                                                                    }
-                                                                },
-                                                                "required": [
-                                                                    "Archiver-Version",
-                                                                    "Build-Jdk",
-                                                                    "Built-By",
-                                                                    "Created-By"
-                                                                ],
-                                                                "type": "object"
-                                                            },
-                                                            "implementationTitle": {
-                                                                "type": "string"
-                                                            },
-                                                            "implementationVendor": {
-                                                                "type": "string"
-                                                            },
-                                                            "implementationVersion": {
-                                                                "type": "string"
-                                                            },
-                                                            "manifestVersion": {
-                                                                "type": "string"
-                                                            },
-                                                            "name": {
-                                                                "type": "string"
-                                                            },
-                                                            "specificationTitle": {
-                                                                "type": "string"
-                                                            },
-                                                            "specificationVendor": {
-                                                                "type": "string"
-                                                            },
-                                                            "specificationVersion": {
-                                                                "type": "string"
-                                                            }
-                                                        },
-                                                        "required": [
-                                                            "extraFields",
-                                                            "implementationTitle",
-                                                            "implementationVendor",
-                                                            "implementationVersion",
-                                                            "manifestVersion",
-                                                            "name",
-                                                            "specificationTitle",
-                                                            "specificationVendor",
-                                                            "specificationVersion"
-                                                        ],
-                                                        "type": "object"
-                                                    },
-                                                    "parentPackage": {
-                                                        "type": "null"
-                                                    },
-                                                    "pomProperties": {
-                                                        "properties": {
-                                                            "Path": {
-                                                                "type": "string"
-                                                            },
-                                                            "artifactId": {
-                                                                "type": "string"
-                                                            },
-                                                            "extraFields": {
-                                                                "type": "null"
-                                                            },
-                                                            "groupId": {
-                                                                "type": "string"
-                                                            },
-                                                            "name": {
-                                                                "type": "string"
-                                                            },
-                                                            "version": {
-                                                                "type": "string"
-                                                            }
-                                                        },
-                                                        "required": [
-                                                            "Path",
-                                                            "artifactId",
-                                                            "extraFields",
-                                                            "groupId",
-                                                            "name",
-                                                            "version"
-                                                        ],
-                                                        "type": "object"
-                                                    }
-                                                },
-                                                "required": [
-                                                    "manifest",
-                                                    "parentPackage",
-                                                    "pomProperties"
-                                                ],
-                                                "type": "object"
-                                            },
-                                            "sources": {
-                                                "type": "null"
-                                            },
-                                            "type": {
-                                                "type": "string"
-                                            },
-                                            "version": {
-                                                "type": "string"
-                                            }
-                                        },
-                                        "required": [
-                                            "foundBy",
-                                            "language",
-                                            "licenses",
-                                            "manifest",
-                                            "metadata",
-                                            "sources",
-                                            "type",
-                                            "version"
-                                        ],
-                                        "type": "object"
-                                    }
-                                ]
+                            "platform": {
+                                "type": "string"
                             },
                             "pomProperties": {
                                 "properties": {
-                                    "Path": {
-                                        "type": "string"
-                                    },
                                     "artifactId": {
                                         "type": "string"
                                     },
@@ -434,16 +236,19 @@
                                     "name": {
                                         "type": "string"
                                     },
+                                    "path": {
+                                        "type": "string"
+                                    },
                                     "version": {
                                         "type": "string"
                                     }
                                 },
                                 "required": [
-                                    "Path",
                                     "artifactId",
                                     "extraFields",
                                     "groupId",
                                     "name",
+                                    "path",
                                     "version"
                                 ],
                                 "type": "object"
@@ -457,6 +262,9 @@
                             "release": {
                                 "type": "string"
                             },
+                            "sitePackagesRootPath": {
+                                "type": "string"
+                            },
                             "size": {
                                 "type": "integer"
                             },
@@ -466,6 +274,12 @@
                             "sourceRpm": {
                                 "type": "string"
                             },
+                            "topLevelPackages": {
+                                "items": {
+                                    "type": "string"
+                                },
+                                "type": "array"
+                            },
                             "url": {
                                 "type": "string"
                             },
@@ -474,6 +288,9 @@
                             },
                             "version": {
                                 "type": "string"
+                            },
+                            "virtualPath": {
+                                "type": "string"
                             }
                         },
                         "type": "object"

syft/cataloger/apkdb/parse_apk_db.go

@@ -20,9 +20,14 @@ var _ common.ParserFn = parseApkDB
 // parseApkDb parses individual packages from a given Alpine DB file. For more information on specific fields
 // see https://wiki.alpinelinux.org/wiki/Apk_spec .
 func parseApkDB(_ string, reader io.Reader) ([]pkg.Package, error) {
+	// larger capacity for the scanner.
+	const maxScannerCapacity = 1024 * 1024
+	// a new larger buffer for the scanner
+	bufScan := make([]byte, maxScannerCapacity)
 	packages := make([]pkg.Package, 0)
 
 	scanner := bufio.NewScanner(reader)
+	scanner.Buffer(bufScan, maxScannerCapacity)
 	onDoubleLF := func(data []byte, atEOF bool) (advance int, token []byte, err error) {
 		for i := 0; i < len(data); i++ {
 			if i > 0 && data[i-1] == '\n' && data[i] == '\n' {
@@ -44,11 +49,12 @@ func parseApkDB(_ string, reader io.Reader) ([]pkg.Package, error) {
 		}
 		if metadata != nil {
 			packages = append(packages, pkg.Package{
-				Name:     metadata.Package,
-				Version:  metadata.Version,
-				Licenses: strings.Split(metadata.License, " "),
-				Type:     pkg.ApkPkg,
-				Metadata: *metadata,
+				Name:         metadata.Package,
+				Version:      metadata.Version,
+				Licenses:     strings.Split(metadata.License, " "),
+				Type:         pkg.ApkPkg,
+				MetadataType: pkg.ApkMetadataType,
+				Metadata:     *metadata,
 			})
 		}
 	}
@@ -83,7 +89,16 @@ func parseApkDBEntry(reader io.Reader) (*pkg.ApkMetadata, error) {
 
 		switch key {
 		case "F":
-			lastFile = "/" + value
+			currentFile := "/" + value
+
+			newFileRecord := pkg.ApkFileRecord{
+				Path: currentFile,
+			}
+			files = append(files, newFileRecord)
+			fileRecord = &files[len(files)-1]
+
+			// future aux references are relative to previous "F" records
+			lastFile = currentFile
 			continue
 		case "R":
 			newFileRecord := pkg.ApkFileRecord{
@@ -91,7 +106,7 @@ func parseApkDBEntry(reader io.Reader) (*pkg.ApkMetadata, error) {
 			}
 			files = append(files, newFileRecord)
 			fileRecord = &files[len(files)-1]
-		case "a":
+		case "a", "M":
 			ownershipFields := strings.Split(value, ":")
 			if len(ownershipFields) < 3 {
 				log.Errorf("unexpected APK ownership field: %q", value)
@@ -102,7 +117,7 @@ func parseApkDBEntry(reader io.Reader) (*pkg.ApkMetadata, error) {
 				continue
 			}
 			fileRecord.OwnerUID = ownershipFields[0]
-			fileRecord.OwnerGUI = ownershipFields[1]
+			fileRecord.OwnerGID = ownershipFields[1]
 			fileRecord.Permissions = ownershipFields[2]
 			// note: there are more optional fields available that we are not capturing, e.g.:
 			// "0:0:755:Q1JaDEHQHBbizhEzoWK1YxuraNU/4="

syft/cataloger/apkdb/parse_apk_db_test.go

@@ -19,10 +19,25 @@ func TestExtraFileAttributes(t *testing.T) {
 			name: "test extra file attributes (checksum) are ignored",
 			expected: pkg.ApkMetadata{
 				Files: []pkg.ApkFileRecord{
+					{
+						Path: "/usr",
+					},
+					{
+						Path: "/usr/lib",
+					},
+					{
+						Path: "/usr/lib/jvm",
+					},
+					{
+						Path: "/usr/lib/jvm/java-1.8-openjdk",
+					},
+					{
+						Path: "/usr/lib/jvm/java-1.8-openjdk/bin",
+					},
 					{
 						Path:        "/usr/lib/jvm/java-1.8-openjdk/bin/policytool",
 						OwnerUID:    "0",
-						OwnerGUI:    "0",
+						OwnerGID:    "0",
 						Permissions: "755",
 						Checksum:    "Q1M0C9qfC/+kdRiOodeihG2GMRtkE=",
 					},
@@ -60,13 +75,13 @@ func TestExtraFileAttributes(t *testing.T) {
 	}
 }
 
-func TestSinglePackage(t *testing.T) {
+func TestSinglePackageDetails(t *testing.T) {
 	tests := []struct {
-		name     string
+		fixture  string
 		expected pkg.ApkMetadata
 	}{
 		{
-			name: "Test Single Package",
+			fixture: "test-fixtures/single",
 			expected: pkg.ApkMetadata{
 				Package:          "musl-utils",
 				OriginPackage:    "musl",
@@ -82,51 +97,428 @@ func TestSinglePackage(t *testing.T) {
 				PullChecksum:     "Q1bTtF5526tETKfL+lnigzIDvm+2o=",
 				GitCommitOfAport: "4024cc3b29ad4c65544ad068b8f59172b5494306",
 				Files: []pkg.ApkFileRecord{
+					{
+						Path: "/sbin",
+					},
 					{
 						Path:        "/sbin/ldconfig",
 						OwnerUID:    "0",
-						OwnerGUI:    "0",
+						OwnerGID:    "0",
 						Permissions: "755",
 						Checksum:    "Q1Kja2+POZKxEkUOZqwSjC6kmaED4=",
 					},
+					{
+						Path: "/usr",
+					},
+					{
+						Path: "/usr/bin",
+					},
 					{
 						Path:        "/usr/bin/iconv",
 						OwnerUID:    "0",
-						OwnerGUI:    "0",
+						OwnerGID:    "0",
 						Permissions: "755",
 						Checksum:    "Q1CVmFbdY+Hv6/jAHl1gec2Kbx1EY=",
 					},
 					{
 						Path:        "/usr/bin/ldd",
 						OwnerUID:    "0",
-						OwnerGUI:    "0",
+						OwnerGID:    "0",
 						Permissions: "755",
 						Checksum:    "Q1yFAhGggmL7ERgbIA7KQxyTzf3ks=",
 					},
 					{
 						Path:        "/usr/bin/getconf",
 						OwnerUID:    "0",
-						OwnerGUI:    "0",
+						OwnerGID:    "0",
 						Permissions: "755",
 						Checksum:    "Q1dAdYK8M/INibRQF5B3Rw7cmNDDA=",
 					},
 					{
 						Path:        "/usr/bin/getent",
 						OwnerUID:    "0",
-						OwnerGUI:    "0",
+						OwnerGID:    "0",
 						Permissions: "755",
 						Checksum:    "Q1eR2Dz/WylabgbWMTkd2+hGmEya4=",
 					},
 				},
 			},
 		},
+		{
+			fixture: "test-fixtures/base",
+			expected: pkg.ApkMetadata{
+				Package:          "alpine-baselayout",
+				OriginPackage:    "alpine-baselayout",
+				Version:          "3.2.0-r6",
+				Description:      "Alpine base dir structure and init scripts",
+				Maintainer:       "Natanael Copa <ncopa@alpinelinux.org>",
+				License:          "GPL-2.0-only",
+				Architecture:     "x86_64",
+				URL:              "https://git.alpinelinux.org/cgit/aports/tree/main/alpine-baselayout",
+				Size:             19917,
+				InstalledSize:    409600,
+				PullDependencies: "/bin/sh so:libc.musl-x86_64.so.1",
+				PullChecksum:     "Q1myMNfd7u5v5UTgNHeq1e31qTjZU=",
+				GitCommitOfAport: "e1c51734fa96fa4bac92e9f14a474324c67916fc",
+				Files: []pkg.ApkFileRecord{
+					{
+						Path: "/dev",
+					},
+					{
+						Path: "/dev/pts",
+					},
+					{
+						Path: "/dev/shm",
+					},
+					{
+						Path: "/etc",
+					},
+					{
+						Path:     "/etc/fstab",
+						Checksum: "Q11Q7hNe8QpDS531guqCdrXBzoA/o=",
+					},
+					{
+						Path:     "/etc/group",
+						Checksum: "Q1oJ16xWudgKOrXIEquEDzlF2Lsm4=",
+					},
+					{
+						Path:     "/etc/hostname",
+						Checksum: "Q16nVwYVXP/tChvUPdukVD2ifXOmc=",
+					},
+					{
+						Path:     "/etc/hosts",
+						Checksum: "Q1BD6zJKZTRWyqGnPi4tSfd3krsMU=",
+					},
+					{
+						Path:     "/etc/inittab",
+						Checksum: "Q1TsthbhW7QzWRe1E/NKwTOuD4pHc=",
+					},
+					{
+						Path:     "/etc/modules",
+						Checksum: "Q1toogjUipHGcMgECgPJX64SwUT1M=",
+					},
+					{
+						Path:     "/etc/motd",
+						Checksum: "Q1XmduVVNURHQ27TvYp1Lr5TMtFcA=",
+					},
+					{
+						Path:        "/etc/mtab",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "777",
+						Checksum:    "Q1kiljhXXH1LlQroHsEJIkPZg2eiw=",
+					},
+					{
+						Path:     "/etc/passwd",
+						Checksum: "Q1TchuuLUfur0izvfZQZxgN/LJhB8=",
+					},
+					{
+						Path:     "/etc/profile",
+						Checksum: "Q1KpFb8kl5LvwXWlY3e58FNsjrI34=",
+					},
+					{
+						Path:     "/etc/protocols",
+						Checksum: "Q13FqXUnvuOpMDrH/6rehxuYAEE34=",
+					},
+					{
+						Path:     "/etc/services",
+						Checksum: "Q1C6HJNgQvLWqt5VY+n7MZJ1rsDuY=",
+					},
+					{
+						Path:        "/etc/shadow",
+						OwnerUID:    "0",
+						OwnerGID:    "42",
+						Permissions: "640",
+						Checksum:    "Q1ltrPIAW2zHeDiajsex2Bdmq3uqA=",
+					},
+					{
+						Path:     "/etc/shells",
+						Checksum: "Q1ojm2YdpCJ6B/apGDaZ/Sdb2xJkA=",
+					},
+					{
+						Path:     "/etc/sysctl.conf",
+						Checksum: "Q14upz3tfnNxZkIEsUhWn7Xoiw96g=",
+					},
+					{
+						Path: "/etc/apk",
+					},
+					{
+						Path: "/etc/conf.d",
+					},
+					{
+						Path: "/etc/crontabs",
+					},
+					{
+						Path:        "/etc/crontabs/root",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "600",
+						Checksum:    "Q1vfk1apUWI4yLJGhhNRd0kJixfvY=",
+					},
+					{
+						Path: "/etc/init.d",
+					},
+					{
+						Path: "/etc/modprobe.d",
+					},
+					{
+						Path:     "/etc/modprobe.d/aliases.conf",
+						Checksum: "Q1WUbh6TBYNVK7e4Y+uUvLs/7viqk=",
+					},
+					{
+						Path:     "/etc/modprobe.d/blacklist.conf",
+						Checksum: "Q1xxYGU6S6TLQvb7ervPrWWwAWqMg=",
+					},
+					{
+						Path:     "/etc/modprobe.d/i386.conf",
+						Checksum: "Q1pnay/njn6ol9cCssL7KiZZ8etlc=",
+					},
+					{
+						Path:     "/etc/modprobe.d/kms.conf",
+						Checksum: "Q1ynbLn3GYDpvajba/ldp1niayeog=",
+					},
+					{
+						Path: "/etc/modules-load.d",
+					},
+					{
+						Path: "/etc/network",
+					},
+					{
+						Path: "/etc/network/if-down.d",
+					},
+					{
+						Path: "/etc/network/if-post-down.d",
+					},
+					{
+						Path: "/etc/network/if-pre-up.d",
+					},
+					{
+						Path: "/etc/network/if-up.d",
+					},
+					{
+						Path: "/etc/opt",
+					},
+					{
+						Path: "/etc/periodic",
+					},
+					{
+						Path: "/etc/periodic/15min",
+					},
+					{
+						Path: "/etc/periodic/daily",
+					},
+					{
+						Path: "/etc/periodic/hourly",
+					},
+					{
+						Path: "/etc/periodic/monthly",
+					},
+					{
+						Path: "/etc/periodic/weekly",
+					},
+					{
+						Path: "/etc/profile.d",
+					},
+					{
+						Path:     "/etc/profile.d/color_prompt",
+						Checksum: "Q10wL23GuSCVfumMRgakabUI6EsSk=",
+					},
+					{
+						Path:     "/etc/profile.d/locale",
+						Checksum: "Q1R4bIEpnKxxOSrlnZy9AoawqZ5DU=",
+					},
+					{
+						Path: "/etc/sysctl.d",
+					},
+					{
+						Path: "/home",
+					},
+					{
+						Path: "/lib",
+					},
+					{
+						Path: "/lib/firmware",
+					},
+					{
+						Path: "/lib/mdev",
+					},
+					{
+						Path: "/lib/modules-load.d",
+					},
+					{
+						Path: "/lib/sysctl.d",
+					},
+					{
+						Path:     "/lib/sysctl.d/00-alpine.conf",
+						Checksum: "Q1HpElzW1xEgmKfERtTy7oommnq6c=",
+					},
+					{
+						Path: "/media",
+					},
+					{
+						Path: "/media/cdrom",
+					},
+					{
+						Path: "/media/floppy",
+					},
+					{
+						Path: "/media/usb",
+					},
+					{
+						Path: "/mnt",
+					},
+					{
+						Path: "/opt",
+					},
+					{
+						Path: "/proc",
+					},
+					{
+						Path:        "/root",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "700",
+					},
+					{
+						Path: "/run",
+					},
+					{
+						Path: "/sbin",
+					},
+					{
+						Path:        "/sbin/mkmntdirs",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "755",
+						Checksum:    "Q1YeuSmC7iDbEWrusPzA/zUQF6YSg=",
+					},
+					{
+						Path: "/srv",
+					},
+					{
+						Path: "/sys",
+					},
+					{
+						Path:        "/tmp",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "1777",
+					},
+					{
+						Path: "/usr",
+					},
+					{
+						Path: "/usr/lib",
+					},
+					{
+						Path: "/usr/lib/modules-load.d",
+					},
+					{
+						Path: "/usr/local",
+					},
+					{
+						Path: "/usr/local/bin",
+					},
+					{
+						Path: "/usr/local/lib",
+					},
+					{
+						Path: "/usr/local/share",
+					},
+					{
+						Path: "/usr/sbin",
+					},
+					{
+						Path: "/usr/share",
+					},
+					{
+						Path: "/usr/share/man",
+					},
+					{
+						Path: "/usr/share/misc",
+					},
+					{
+						Path: "/var",
+					},
+					{
+						Path:        "/var/run",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "777",
+						Checksum:    "Q11/SNZz/8cK2dSKK+cJpVrZIuF4Q=",
+					},
+					{
+						Path: "/var/cache",
+					},
+					{
+						Path: "/var/cache/misc",
+					},
+					{
+						Path:        "/var/empty",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "555",
+					},
+					{
+						Path: "/var/lib",
+					},
+					{
+						Path: "/var/lib/misc",
+					},
+					{
+						Path: "/var/local",
+					},
+					{
+						Path: "/var/lock",
+					},
+					{
+						Path: "/var/lock/subsys",
+					},
+					{
+						Path: "/var/log",
+					},
+					{
+						Path: "/var/mail",
+					},
+					{
+						Path: "/var/opt",
+					},
+					{
+						Path: "/var/spool",
+					},
+					{
+						Path:        "/var/spool/mail",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "777",
+						Checksum:    "Q1dzbdazYZA2nTzSIG3YyNw7d4Juc=",
+					},
+					{
+						Path: "/var/spool/cron",
+					},
+					{
+						Path:        "/var/spool/cron/crontabs",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "777",
+						Checksum:    "Q1OFZt+ZMp7j0Gny0rqSKuWJyqYmA=",
+					},
+					{
+						Path:        "/var/tmp",
+						OwnerUID:    "0",
+						OwnerGID:    "0",
+						Permissions: "1777",
+					},
+				},
+			},
+		},
 	}
 
 	for _, test := range tests {
-		t.Run(test.name, func(t *testing.T) {
-			file, err := os.Open("test-fixtures/single")
+		t.Run(test.fixture, func(t *testing.T) {
+			file, err := os.Open(test.fixture)
 			if err != nil {
-				t.Fatal("Unable to read test_fixtures/single: ", err)
+				t.Fatal("Unable to read fixture: ", err)
 			}
 			defer func() {
 				err := file.Close()
@@ -160,10 +552,11 @@ func TestMultiplePackages(t *testing.T) {
 			fixture: "test-fixtures/multiple",
 			expected: []pkg.Package{
 				{
-					Name:     "libc-utils",
-					Version:  "0.7.2-r0",
-					Licenses: []string{"BSD"},
-					Type:     pkg.ApkPkg,
+					Name:         "libc-utils",
+					Version:      "0.7.2-r0",
+					Licenses:     []string{"BSD"},
+					Type:         pkg.ApkPkg,
+					MetadataType: pkg.ApkMetadataType,
 					Metadata: pkg.ApkMetadata{
 						Package:          "libc-utils",
 						OriginPackage:    "libc-dev",
@@ -182,10 +575,11 @@ func TestMultiplePackages(t *testing.T) {
 					},
 				},
 				{
-					Name:     "musl-utils",
-					Version:  "1.1.24-r2",
-					Licenses: []string{"MIT", "BSD", "GPL2+"},
-					Type:     pkg.ApkPkg,
+					Name:         "musl-utils",
+					Version:      "1.1.24-r2",
+					Licenses:     []string{"MIT", "BSD", "GPL2+"},
+					Type:         pkg.ApkPkg,
+					MetadataType: pkg.ApkMetadataType,
 					Metadata: pkg.ApkMetadata{
 						Package:          "musl-utils",
 						OriginPackage:    "musl",
@@ -201,38 +595,47 @@ func TestMultiplePackages(t *testing.T) {
 						PullChecksum:     "Q1bTtF5526tETKfL+lnigzIDvm+2o=",
 						GitCommitOfAport: "4024cc3b29ad4c65544ad068b8f59172b5494306",
 						Files: []pkg.ApkFileRecord{
+							{
+								Path: "/sbin",
+							},
 							{
 								Path:        "/sbin/ldconfig",
 								OwnerUID:    "0",
-								OwnerGUI:    "0",
+								OwnerGID:    "0",
 								Permissions: "755",
 								Checksum:    "Q1Kja2+POZKxEkUOZqwSjC6kmaED4=",
 							},
+							{
+								Path: "/usr",
+							},
+							{
+								Path: "/usr/bin",
+							},
 							{
 								Path:        "/usr/bin/iconv",
 								OwnerUID:    "0",
-								OwnerGUI:    "0",
+								OwnerGID:    "0",
 								Permissions: "755",
 								Checksum:    "Q1CVmFbdY+Hv6/jAHl1gec2Kbx1EY=",
 							},
 							{
 								Path:        "/usr/bin/ldd",
 								OwnerUID:    "0",
-								OwnerGUI:    "0",
+								OwnerGID:    "0",
 								Permissions: "755",
 								Checksum:    "Q1yFAhGggmL7ERgbIA7KQxyTzf3ks=",
 							},
 							{
 								Path:        "/usr/bin/getconf",
 								OwnerUID:    "0",
-								OwnerGUI:    "0",
+								OwnerGID:    "0",
 								Permissions: "755",
 								Checksum:    "Q1dAdYK8M/INibRQF5B3Rw7cmNDDA=",
 							},
 							{
 								Path:        "/usr/bin/getent",
 								OwnerUID:    "0",
-								OwnerGUI:    "0",
+								OwnerGID:    "0",
 								Permissions: "755",
 								Checksum:    "Q1eR2Dz/WylabgbWMTkd2+hGmEya4=",
 							},

syft/cataloger/apkdb/test-fixtures/base

@@ -0,0 +1,149 @@
+C:Q1myMNfd7u5v5UTgNHeq1e31qTjZU=
+P:alpine-baselayout
+V:3.2.0-r6
+A:x86_64
+S:19917
+I:409600
+T:Alpine base dir structure and init scripts
+U:https://git.alpinelinux.org/cgit/aports/tree/main/alpine-baselayout
+L:GPL-2.0-only
+o:alpine-baselayout
+m:Natanael Copa <ncopa@alpinelinux.org>
+t:1590679557
+c:e1c51734fa96fa4bac92e9f14a474324c67916fc
+D:/bin/sh so:libc.musl-x86_64.so.1
+p:cmd:mkmntdirs
+F:dev
+F:dev/pts
+F:dev/shm
+F:etc
+R:fstab
+Z:Q11Q7hNe8QpDS531guqCdrXBzoA/o=
+R:group
+Z:Q1oJ16xWudgKOrXIEquEDzlF2Lsm4=
+R:hostname
+Z:Q16nVwYVXP/tChvUPdukVD2ifXOmc=
+R:hosts
+Z:Q1BD6zJKZTRWyqGnPi4tSfd3krsMU=
+R:inittab
+Z:Q1TsthbhW7QzWRe1E/NKwTOuD4pHc=
+R:modules
+Z:Q1toogjUipHGcMgECgPJX64SwUT1M=
+R:motd
+Z:Q1XmduVVNURHQ27TvYp1Lr5TMtFcA=
+R:mtab
+a:0:0:777
+Z:Q1kiljhXXH1LlQroHsEJIkPZg2eiw=
+R:passwd
+Z:Q1TchuuLUfur0izvfZQZxgN/LJhB8=
+R:profile
+Z:Q1KpFb8kl5LvwXWlY3e58FNsjrI34=
+R:protocols
+Z:Q13FqXUnvuOpMDrH/6rehxuYAEE34=
+R:services
+Z:Q1C6HJNgQvLWqt5VY+n7MZJ1rsDuY=
+R:shadow
+a:0:42:640
+Z:Q1ltrPIAW2zHeDiajsex2Bdmq3uqA=
+R:shells
+Z:Q1ojm2YdpCJ6B/apGDaZ/Sdb2xJkA=
+R:sysctl.conf
+Z:Q14upz3tfnNxZkIEsUhWn7Xoiw96g=
+F:etc/apk
+F:etc/conf.d
+F:etc/crontabs
+R:root
+a:0:0:600
+Z:Q1vfk1apUWI4yLJGhhNRd0kJixfvY=
+F:etc/init.d
+F:etc/modprobe.d
+R:aliases.conf
+Z:Q1WUbh6TBYNVK7e4Y+uUvLs/7viqk=
+R:blacklist.conf
+Z:Q1xxYGU6S6TLQvb7ervPrWWwAWqMg=
+R:i386.conf
+Z:Q1pnay/njn6ol9cCssL7KiZZ8etlc=
+R:kms.conf
+Z:Q1ynbLn3GYDpvajba/ldp1niayeog=
+F:etc/modules-load.d
+F:etc/network
+F:etc/network/if-down.d
+F:etc/network/if-post-down.d
+F:etc/network/if-pre-up.d
+F:etc/network/if-up.d
+F:etc/opt
+F:etc/periodic
+F:etc/periodic/15min
+F:etc/periodic/daily
+F:etc/periodic/hourly
+F:etc/periodic/monthly
+F:etc/periodic/weekly
+F:etc/profile.d
+R:color_prompt
+Z:Q10wL23GuSCVfumMRgakabUI6EsSk=
+R:locale
+Z:Q1R4bIEpnKxxOSrlnZy9AoawqZ5DU=
+F:etc/sysctl.d
+F:home
+F:lib
+F:lib/firmware
+F:lib/mdev
+F:lib/modules-load.d
+F:lib/sysctl.d
+R:00-alpine.conf
+Z:Q1HpElzW1xEgmKfERtTy7oommnq6c=
+F:media
+F:media/cdrom
+F:media/floppy
+F:media/usb
+F:mnt
+F:opt
+F:proc
+F:root
+M:0:0:700
+F:run
+F:sbin
+R:mkmntdirs
+a:0:0:755
+Z:Q1YeuSmC7iDbEWrusPzA/zUQF6YSg=
+F:srv
+F:sys
+F:tmp
+M:0:0:1777
+F:usr
+F:usr/lib
+F:usr/lib/modules-load.d
+F:usr/local
+F:usr/local/bin
+F:usr/local/lib
+F:usr/local/share
+F:usr/sbin
+F:usr/share
+F:usr/share/man
+F:usr/share/misc
+F:var
+R:run
+a:0:0:777
+Z:Q11/SNZz/8cK2dSKK+cJpVrZIuF4Q=
+F:var/cache
+F:var/cache/misc
+F:var/empty
+M:0:0:555
+F:var/lib
+F:var/lib/misc
+F:var/local
+F:var/lock
+F:var/lock/subsys
+F:var/log
+F:var/mail
+F:var/opt
+F:var/spool
+R:mail
+a:0:0:777
+Z:Q1dzbdazYZA2nTzSIG3YyNw7d4Juc=
+F:var/spool/cron
+R:crontabs
+a:0:0:777
+Z:Q1OFZt+ZMp7j0Gny0rqSKuWJyqYmA=
+F:var/tmp
+M:0:0:1777

syft/cataloger/catalog.go

@@ -1,7 +1,6 @@
 package cataloger
 
 import (
-	"github.com/anchore/stereoscope/pkg/file"
 	"github.com/anchore/syft/internal/bus"
 	"github.com/anchore/syft/internal/log"
 	"github.com/anchore/syft/syft/event"
@@ -39,36 +38,21 @@ func newMonitor() (*progress.Manual, *progress.Manual) {
 // request.
 func Catalog(resolver scope.Resolver, catalogers ...Cataloger) (*pkg.Catalog, error) {
 	catalog := pkg.NewCatalog()
-	fileSelection := make([]file.Reference, 0)
-
 	filesProcessed, packagesDiscovered := newMonitor()
 
-	// ask catalogers for files to extract from the image tar
-	for _, a := range catalogers {
-		fileSelection = append(fileSelection, a.SelectFiles(resolver)...)
-		log.Debugf("cataloger '%s' selected '%d' files", a.Name(), len(fileSelection))
-		filesProcessed.N += int64(len(fileSelection))
-	}
-
-	// fetch contents for requested selection by catalogers
-	// TODO: we should consider refactoring to return a set of io.Readers instead of the full contents themselves (allow for optional buffering).
-	contents, err := resolver.MultipleFileContentsByRef(fileSelection...)
-	if err != nil {
-		return nil, err
-	}
-
 	// perform analysis, accumulating errors for each failed analysis
 	var errs error
-	for _, a := range catalogers {
-		// TODO: check for multiple rounds of analyses by Iterate error
-		packages, err := a.Catalog(contents)
+	for _, theCataloger := range catalogers {
+		packages, err := theCataloger.Catalog(resolver)
 		if err != nil {
 			errs = multierror.Append(errs, err)
 			continue
 		}
 
-		log.Debugf("cataloger '%s' discovered '%d' packages", a.Name(), len(packages))
-		packagesDiscovered.N += int64(len(packages))
+		catalogedPackages := len(packages)
+
+		log.Debugf("cataloger '%s' discovered '%d' packages", theCataloger.Name(), catalogedPackages)
+		packagesDiscovered.N += int64(catalogedPackages)
 
 		for _, p := range packages {
 			catalog.Add(p)

syft/cataloger/cataloger.go

@@ -6,7 +6,6 @@ catalogers defined in child packages as well as the interface definition to impl
 package cataloger
 
 import (
-	"github.com/anchore/stereoscope/pkg/file"
 	"github.com/anchore/syft/syft/cataloger/apkdb"
 	"github.com/anchore/syft/syft/cataloger/deb"
 	"github.com/anchore/syft/syft/cataloger/golang"
@@ -25,20 +24,16 @@ import (
 type Cataloger interface {
 	// Name returns a string that uniquely describes a cataloger
 	Name() string
-	// SelectFiles discovers and returns specific files that the cataloger would like to inspect the contents of.
-	SelectFiles(scope.FileResolver) []file.Reference
-	// Catalog is given the file contents and should return any discovered Packages after analyzing the contents.
-	Catalog(map[file.Reference]string) ([]pkg.Package, error)
-	// TODO: add "IterationNeeded" error to indicate to the driver to continue with another Select/Catalog pass
-	// TODO: we should consider refactoring to return a set of io.Readers instead of the full contents themselves (allow for optional buffering).
+	// Catalog is given an object to resolve file references and content, this function returns any discovered Packages after analyzing the catalog source.
+	Catalog(resolver scope.Resolver) ([]pkg.Package, error)
 }
 
 // ImageCatalogers returns a slice of locally implemented catalogers that are fit for detecting installations of packages.
 func ImageCatalogers() []Cataloger {
 	return []Cataloger{
 		ruby.NewGemSpecCataloger(),
-		python.NewPythonCataloger(),         // TODO: split and replace me
-		javascript.NewJavascriptCataloger(), // TODO: split and replace me
+		python.NewPythonPackageCataloger(),
+		javascript.NewJavascriptPackageCataloger(),
 		deb.NewDpkgdbCataloger(),
 		rpmdb.NewRpmdbCataloger(),
 		java.NewJavaCataloger(),
@@ -51,8 +46,9 @@ func ImageCatalogers() []Cataloger {
 func DirectoryCatalogers() []Cataloger {
 	return []Cataloger{
 		ruby.NewGemFileLockCataloger(),
-		python.NewPythonCataloger(),         // TODO: split and replace me
-		javascript.NewJavascriptCataloger(), // TODO: split and replace me
+		python.NewPythonIndexCataloger(),
+		python.NewPythonPackageCataloger(),
+		javascript.NewJavascriptLockCataloger(),
 		deb.NewDpkgdbCataloger(),
 		rpmdb.NewRpmdbCataloger(),
 		java.NewJavaCataloger(),

syft/cataloger/common/generic_cataloger.go

@@ -34,73 +34,83 @@ func NewGenericCataloger(pathParsers map[string]ParserFn, globParsers map[string
 }
 
 // Name returns a string that uniquely describes the upstream cataloger that this Generic Cataloger represents.
-func (a *GenericCataloger) Name() string {
-	return a.upstreamCataloger
+func (c *GenericCataloger) Name() string {
+	return c.upstreamCataloger
 }
 
 // register pairs a set of file references with a parser function for future cataloging (when the file contents are resolved)
-func (a *GenericCataloger) register(files []file.Reference, parser ParserFn) {
-	a.selectedFiles = append(a.selectedFiles, files...)
+func (c *GenericCataloger) register(files []file.Reference, parser ParserFn) {
+	c.selectedFiles = append(c.selectedFiles, files...)
 	for _, f := range files {
-		a.parsers[f] = parser
+		c.parsers[f] = parser
 	}
 }
 
 // clear deletes all registered file-reference-to-parser-function pairings from former SelectFiles() and register() calls
-func (a *GenericCataloger) clear() {
-	a.selectedFiles = make([]file.Reference, 0)
-	a.parsers = make(map[file.Reference]ParserFn)
+func (c *GenericCataloger) clear() {
+	c.selectedFiles = make([]file.Reference, 0)
+	c.parsers = make(map[file.Reference]ParserFn)
+}
+
+// Catalog is given an object to resolve file references and content, this function returns any discovered Packages after analyzing the catalog source.
+func (c *GenericCataloger) Catalog(resolver scope.Resolver) ([]pkg.Package, error) {
+	fileSelection := c.selectFiles(resolver)
+	contents, err := resolver.MultipleFileContentsByRef(fileSelection...)
+	if err != nil {
+		return nil, err
+	}
+	return c.catalog(contents)
 }
 
 // SelectFiles takes a set of file trees and resolves and file references of interest for future cataloging
-func (a *GenericCataloger) SelectFiles(resolver scope.FileResolver) []file.Reference {
+func (c *GenericCataloger) selectFiles(resolver scope.FileResolver) []file.Reference {
 	// select by exact path
-	for path, parser := range a.pathParsers {
+	for path, parser := range c.pathParsers {
 		files, err := resolver.FilesByPath(file.Path(path))
 		if err != nil {
 			log.Errorf("cataloger failed to select files by path: %+v", err)
 		}
 		if files != nil {
-			a.register(files, parser)
+			c.register(files, parser)
 		}
 	}
 
 	// select by glob pattern
-	for globPattern, parser := range a.globParsers {
+	for globPattern, parser := range c.globParsers {
 		fileMatches, err := resolver.FilesByGlob(globPattern)
 		if err != nil {
 			log.Errorf("failed to find files by glob: %s", globPattern)
 		}
 		if fileMatches != nil {
-			a.register(fileMatches, parser)
+			c.register(fileMatches, parser)
 		}
 	}
 
-	return a.selectedFiles
+	return c.selectedFiles
 }
 
-// Catalog takes a set of file contents and uses any configured parser functions to resolve and return discovered packages
-func (a *GenericCataloger) Catalog(contents map[file.Reference]string) ([]pkg.Package, error) {
-	defer a.clear()
+// catalog takes a set of file contents and uses any configured parser functions to resolve and return discovered packages
+func (c *GenericCataloger) catalog(contents map[file.Reference]string) ([]pkg.Package, error) {
+	defer c.clear()
 
 	packages := make([]pkg.Package, 0)
 
-	for reference, parser := range a.parsers {
+	for reference, parser := range c.parsers {
 		content, ok := contents[reference]
 		if !ok {
-			log.Errorf("cataloger '%s' missing file content: %+v", a.upstreamCataloger, reference)
+			log.Errorf("cataloger '%s' missing file content: %+v", c.upstreamCataloger, reference)
 			continue
 		}
 
 		entries, err := parser(string(reference.Path), strings.NewReader(content))
 		if err != nil {
 			// TODO: should we fail? or only log?
-			log.Errorf("cataloger '%s' failed to parse entries (reference=%+v): %+v", a.upstreamCataloger, reference, err)
+			log.Errorf("cataloger '%s' failed to parse entries (reference=%+v): %+v", c.upstreamCataloger, reference, err)
 			continue
 		}
 
 		for _, entry := range entries {
-			entry.FoundBy = a.upstreamCataloger
+			entry.FoundBy = c.upstreamCataloger
 			entry.Source = []file.Reference{reference}
 
 			packages = append(packages, entry)

syft/cataloger/common/generic_cataloger_test.go

@@ -7,21 +7,28 @@ import (
 	"testing"
 
 	"github.com/anchore/stereoscope/pkg/file"
-	"github.com/anchore/syft/internal"
 	"github.com/anchore/syft/syft/pkg"
 )
 
-type testResolver struct {
+type testResolverMock struct {
 	contents map[file.Reference]string
 }
 
-func newTestResolver() *testResolver {
-	return &testResolver{
+func newTestResolver() *testResolverMock {
+	return &testResolverMock{
 		contents: make(map[file.Reference]string),
 	}
 }
 
-func (r *testResolver) FilesByPath(paths ...file.Path) ([]file.Reference, error) {
+func (r *testResolverMock) FileContentsByRef(_ file.Reference) (string, error) {
+	return "", fmt.Errorf("not implemented")
+}
+
+func (r *testResolverMock) MultipleFileContentsByRef(_ ...file.Reference) (map[file.Reference]string, error) {
+	return r.contents, nil
+}
+
+func (r *testResolverMock) FilesByPath(paths ...file.Path) ([]file.Reference, error) {
 	results := make([]file.Reference, len(paths))
 
 	for idx, p := range paths {
@@ -32,13 +39,17 @@ func (r *testResolver) FilesByPath(paths ...file.Path) ([]file.Reference, error)
 	return results, nil
 }
 
-func (r *testResolver) FilesByGlob(patterns ...string) ([]file.Reference, error) {
+func (r *testResolverMock) FilesByGlob(_ ...string) ([]file.Reference, error) {
 	path := "/a-path.txt"
 	ref := file.NewFileReference(file.Path(path))
 	r.contents[ref] = fmt.Sprintf("%s file contents!", path)
 	return []file.Reference{ref}, nil
 }
 
+func (r *testResolverMock) RelativeFileByPath(_ file.Reference, _ string) (*file.Reference, error) {
+	return nil, fmt.Errorf("not implemented")
+}
+
 func parser(_ string, reader io.Reader) ([]pkg.Package, error) {
 	contents, err := ioutil.ReadAll(reader)
 	if err != nil {
@@ -64,31 +75,16 @@ func TestGenericCataloger(t *testing.T) {
 	resolver := newTestResolver()
 	cataloger := NewGenericCataloger(pathParsers, globParsers, upstream)
 
-	selected := cataloger.SelectFiles(resolver)
-
-	if len(selected) != 3 {
-		t.Fatalf("unexpected selection length: %d", len(selected))
-	}
-
-	expectedSelection := internal.NewStringSetFromSlice([]string{"/last/path.txt", "/another-path.txt", "/a-path.txt"})
-	selectionByPath := make(map[string]file.Reference)
-	for _, s := range selected {
-		if !expectedSelection.Contains(string(s.Path)) {
-			t.Errorf("unexpected selection path: %+v", s.Path)
-		}
-		selectionByPath[string(s.Path)] = s
-	}
-
-	expectedPkgs := make(map[file.Reference]pkg.Package)
-	for path, ref := range selectionByPath {
-		expectedPkgs[ref] = pkg.Package{
+	expectedSelection := []string{"/last/path.txt", "/another-path.txt", "/a-path.txt"}
+	expectedPkgs := make(map[string]pkg.Package)
+	for _, path := range expectedSelection {
+		expectedPkgs[path] = pkg.Package{
 			FoundBy: upstream,
-			Source:  []file.Reference{ref},
 			Name:    fmt.Sprintf("%s file contents!", path),
 		}
 	}
 
-	actualPkgs, err := cataloger.Catalog(resolver.contents)
+	actualPkgs, err := cataloger.Catalog(resolver)
 	if err != nil {
 		t.Fatalf("cataloger catalog action failed: %+v", err)
 	}
@@ -99,7 +95,7 @@ func TestGenericCataloger(t *testing.T) {
 
 	for _, p := range actualPkgs {
 		ref := p.Source[0]
-		exP, ok := expectedPkgs[ref]
+		exP, ok := expectedPkgs[string(ref.Path)]
 		if !ok {
 			t.Errorf("missing expected pkg: ref=%+v", ref)
 			continue

syft/cataloger/deb/parse_dpkg_status.go

@@ -30,10 +30,11 @@ func parseDpkgStatus(_ string, reader io.Reader) ([]pkg.Package, error) {
 			return nil, err
 		}
 		packages = append(packages, pkg.Package{
-			Name:     entry.Package,
-			Version:  entry.Version,
-			Type:     pkg.DebPkg,
-			Metadata: entry,
+			Name:         entry.Package,
+			Version:      entry.Version,
+			Type:         pkg.DebPkg,
+			MetadataType: pkg.DpkgMetadataType,
+			Metadata:     entry,
 		})
 	}
 

syft/cataloger/java/archive_filename.go

@@ -10,11 +10,20 @@ import (
 	"github.com/anchore/syft/syft/pkg"
 )
 
+// match on versions and anything after the version. This is used to isolate the name from the version.
+// match examples:
+// wagon-webdav-1.0.2-rc1-hudson.jar      --->     -1.0.2-rc1-hudson.jar
+// windows-remote-command-1.0.jar         --->     -1.0.jar
+// wstx-asl-1-2.jar                       --->     -1-2.jar
+// guava-rc0.jar                          --->     -rc0.jar
+var versionAreaPattern = regexp.MustCompile(`-(?P<version>(\d+\.)?(\d+\.)?(r?c?\d+)(-[a-zA-Z0-9\-.]+)*)(?P<remaining>.*)$`)
+
+// match on explicit versions. This is used for extracting version information.
 // match examples:
 //	pkg-extra-field-4.3.2-rc1 --> match(name=pkg-extra-field version=4.3.2-rc1)
 //	pkg-extra-field-4.3-rc1   --> match(name=pkg-extra-field version=4.3-rc1)
 //	pkg-extra-field-4.3       --> match(name=pkg-extra-field version=4.3)
-var versionPattern = regexp.MustCompile(`(?P<name>.+)-(?P<version>(\d+\.)?(\d+\.)?(\*|\d+)(-[a-zA-Z0-9\-\.]+)*)`)
+var versionPattern = regexp.MustCompile(`-(?P<version>(\d+\.)?(\d+\.)?(r?c?\d+)(-[a-zA-Z0-9\-.]+)*)`)
 
 type archiveFilename struct {
 	raw    string
@@ -70,10 +79,8 @@ func (a archiveFilename) version() string {
 }
 
 func (a archiveFilename) name() string {
-	// there should be only one name, if there is more or less then something is wrong
-	if len(a.fields) != 1 {
-		return ""
-	}
-
-	return a.fields[0]["name"]
+	// derive the name from the archive name (no path or extension) and remove any versions found
+	basename := filepath.Base(a.raw)
+	cleaned := strings.TrimSuffix(basename, filepath.Ext(basename))
+	return versionAreaPattern.ReplaceAllString(cleaned, "")
 }

syft/cataloger/java/archive_filename_test.go

@@ -1,9 +1,10 @@
 package java
 
 import (
+	"testing"
+
 	"github.com/anchore/syft/syft/pkg"
 	"github.com/sergi/go-diff/diffmatchpatch"
-	"testing"
 )
 
 func TestExtractInfoFromJavaArchiveFilename(t *testing.T) {
@@ -56,12 +57,78 @@ func TestExtractInfoFromJavaArchiveFilename(t *testing.T) {
 			name:      "pkg-extra-field-maven",
 			ty:        pkg.JenkinsPluginPkg,
 		},
+		{
+			filename:  "/some/path-with-version-5.4.3/wagon-webdav-1.0.2-beta-2.2.3a-hudson.jar",
+			version:   "1.0.2-beta-2.2.3a-hudson",
+			extension: "jar",
+			name:      "wagon-webdav",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename:  "/some/path-with-version-5.4.3/wagon-webdav-1.0.2-beta-2.2.3-hudson.jar",
+			version:   "1.0.2-beta-2.2.3-hudson",
+			extension: "jar",
+			name:      "wagon-webdav",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename:  "/some/path-with-version-5.4.3/windows-remote-command-1.0.jar",
+			version:   "1.0",
+			extension: "jar",
+			name:      "windows-remote-command",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename:  "/some/path-with-version-5.4.3/wagon-http-lightweight-1.0.5-beta-2.jar",
+			version:   "1.0.5-beta-2",
+			extension: "jar",
+			name:      "wagon-http-lightweight",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename:  "/hudson.war:WEB-INF/lib/commons-jelly-1.1-hudson-20100305.jar",
+			version:   "1.1-hudson-20100305",
+			extension: "jar",
+			name:      "commons-jelly",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename: "/hudson.war:WEB-INF/lib/jtidy-4aug2000r7-dev-hudson-1.jar",
+			// I don't see how we can reliably account for this case
+			//version:   "4aug2000r7-dev-hudson-1",
+			version:   "",
+			extension: "jar",
+			name:      "jtidy",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename: "/hudson.war:WEB-INF/lib/trilead-ssh2-build212-hudson-5.jar",
+			// I don't see how we can reliably account for this case
+			//version:   "build212-hudson-5",
+			version:   "5",
+			extension: "jar",
+			// name: "trilead-ssh2",
+			name: "trilead-ssh2-build212-hudson",
+			ty:   pkg.JavaPkg,
+		},
+		{
+			filename:  "/hudson.war:WEB-INF/lib/guava-r06.jar",
+			version:   "r06",
+			extension: "jar",
+			name:      "guava",
+			ty:        pkg.JavaPkg,
+		},
 	}
 
 	for _, test := range tests {
 		t.Run(test.filename, func(t *testing.T) {
 			obj := newJavaArchiveFilename(test.filename)
 
+			ty := obj.pkgType()
+			if ty != test.ty {
+				t.Errorf("mismatched type: %+v != %v", ty, test.ty)
+			}
+
 			version := obj.version()
 			if version != test.version {
 				dmp := diffmatchpatch.New()

syft/cataloger/java/archive_parser.go

@@ -5,6 +5,8 @@ import (
 	"io"
 	"strings"
 
+	"github.com/anchore/syft/internal/log"
+
 	"github.com/anchore/syft/internal"
 	"github.com/anchore/syft/internal/file"
 	"github.com/anchore/syft/syft/cataloger/common"
@@ -64,13 +66,17 @@ func newJavaArchiveParser(virtualPath string, reader io.Reader, detectNested boo
 		return nil, cleanupFn, fmt.Errorf("unable to read files from java archive: %w", err)
 	}
 
+	// fetch the last element of the virtual path
+	virtualElements := strings.Split(virtualPath, ":")
+	currentFilepath := virtualElements[len(virtualElements)-1]
+
 	return &archiveParser{
 		discoveredPkgs: internal.NewStringSet(),
 		fileManifest:   fileManifest,
 		virtualPath:    virtualPath,
 		archivePath:    archivePath,
 		contentPath:    contentPath,
-		fileInfo:       newJavaArchiveFilename(virtualPath),
+		fileInfo:       newJavaArchiveFilename(currentFilepath),
 		detectNested:   detectNested,
 	}, cleanupFn, nil
 }
@@ -118,7 +124,7 @@ func (j *archiveParser) parse() ([]pkg.Package, error) {
 // discoverMainPackage parses the root Java manifest used as the parent package to all discovered nested packages.
 func (j *archiveParser) discoverMainPackage() (*pkg.Package, error) {
 	// search and parse java manifest files
-	manifestMatches := j.fileManifest.GlobMatch(manifestPath)
+	manifestMatches := j.fileManifest.GlobMatch(manifestGlob)
 	if len(manifestMatches) > 1 {
 		return nil, fmt.Errorf("found multiple manifests in the jar: %+v", manifestMatches)
 	} else if len(manifestMatches) == 0 {
@@ -134,24 +140,27 @@ func (j *archiveParser) discoverMainPackage() (*pkg.Package, error) {
 
 	// parse the manifest file into a rich object
 	manifestContents := contents[manifestMatches[0]]
-	manifest, err := parseJavaManifest(strings.NewReader(manifestContents))
+	manifest, err := parseJavaManifest(j.archivePath, strings.NewReader(manifestContents))
 	if err != nil {
 		return nil, fmt.Errorf("failed to parse java manifest (%s): %w", j.virtualPath, err)
 	}
 
 	return &pkg.Package{
-		Name:     selectName(manifest, j.fileInfo),
-		Version:  selectVersion(manifest, j.fileInfo),
-		Language: pkg.Java,
-		Type:     pkg.JavaPkg,
+		Name:         selectName(manifest, j.fileInfo),
+		Version:      selectVersion(manifest, j.fileInfo),
+		Language:     pkg.Java,
+		Type:         pkg.JavaPkg,
+		MetadataType: pkg.JavaMetadataType,
 		Metadata: pkg.JavaMetadata{
-			Manifest: manifest,
+			VirtualPath: j.virtualPath,
+			Manifest:    manifest,
 		},
 	}, nil
 }
 
 // discoverPkgsFromPomProperties parses Maven POM properties for a given parent package, returning all listed Java packages found and
 // associating each discovered package to the given parent package.
+// nolint:funlen,gocognit
 func (j *archiveParser) discoverPkgsFromPomProperties(parentPkg *pkg.Package) ([]pkg.Package, error) {
 	var pkgs = make([]pkg.Package, 0)
 	parentKey := uniquePkgKey(parentPkg)
@@ -173,13 +182,22 @@ func (j *archiveParser) discoverPkgsFromPomProperties(parentPkg *pkg.Package) ([
 			if propsObj.Version != "" && propsObj.ArtifactID != "" {
 				// TODO: if there is no parentPkg (no java manifest) one of these poms could be the parent. We should discover the right parent and attach the correct info accordingly to each discovered package
 
+				// keep the artifact name within the virtual path if this package does not match the parent package
+				vPathSuffix := ""
+				if !strings.HasPrefix(propsObj.ArtifactID, parentPkg.Name) {
+					vPathSuffix += ":" + propsObj.ArtifactID
+				}
+				virtualPath := j.virtualPath + vPathSuffix
+
 				// discovered props = new package
 				p := pkg.Package{
-					Name:     propsObj.ArtifactID,
-					Version:  propsObj.Version,
-					Language: pkg.Java,
-					Type:     pkg.JavaPkg,
+					Name:         propsObj.ArtifactID,
+					Version:      propsObj.Version,
+					Language:     pkg.Java,
+					Type:         pkg.JavaPkg,
+					MetadataType: pkg.JavaMetadataType,
 					Metadata: pkg.JavaMetadata{
+						VirtualPath:   virtualPath,
 						PomProperties: propsObj,
 						Parent:        parentPkg,
 					},
@@ -187,16 +205,35 @@ func (j *archiveParser) discoverPkgsFromPomProperties(parentPkg *pkg.Package) ([
 
 				pkgKey := uniquePkgKey(&p)
 
-				if !j.discoveredPkgs.Contains(pkgKey) {
-					// only keep packages we haven't seen yet
-					pkgs = append(pkgs, p)
-				} else if pkgKey == parentKey {
+				// the name/version pair matches...
+				matchesParentPkg := pkgKey == parentKey
+
+				// the virtual path matches...
+				matchesParentPkg = matchesParentPkg || parentPkg.Metadata.(pkg.JavaMetadata).VirtualPath == virtualPath
+
+				// the pom artifactId has the parent name or vice versa
+				if propsObj.ArtifactID != "" {
+					matchesParentPkg = matchesParentPkg || strings.Contains(parentPkg.Name, propsObj.ArtifactID) || strings.Contains(propsObj.ArtifactID, parentPkg.Name)
+				}
+
+				if matchesParentPkg {
 					// we've run across more information about our parent package, add this info to the parent package metadata
+					// the pom properties is typically a better source of information for name and version than the manifest
+					if p.Name != parentPkg.Name {
+						parentPkg.Name = p.Name
+					}
+					if p.Version != parentPkg.Version {
+						parentPkg.Version = p.Version
+					}
+
 					parentMetadata, ok := parentPkg.Metadata.(pkg.JavaMetadata)
 					if ok {
 						parentMetadata.PomProperties = propsObj
 						parentPkg.Metadata = parentMetadata
 					}
+				} else if !j.discoveredPkgs.Contains(pkgKey) {
+					// only keep packages we haven't seen yet (and are not related to the parent package)
+					pkgs = append(pkgs, p)
 				}
 			}
 		}
@@ -214,18 +251,28 @@ func (j *archiveParser) discoverPkgsFromNestedArchives(parentPkg *pkg.Package) (
 	}
 
 	// search and parse pom.properties files & fetch the contents
-	readers, err := file.ExtractFromZipToUniqueTempFile(j.archivePath, j.contentPath, j.fileManifest.GlobMatch(archiveFormatGlobs...)...)
+	openers, err := file.ExtractFromZipToUniqueTempFile(j.archivePath, j.contentPath, j.fileManifest.GlobMatch(archiveFormatGlobs...)...)
 	if err != nil {
 		return nil, fmt.Errorf("unable to extract files from zip: %w", err)
 	}
 
 	// discover nested artifacts
-	for archivePath, archiveReader := range readers {
-		nestedPath := fmt.Sprintf("%s:%s", j.virtualPath, archivePath)
-		nestedPkgs, err := parseJavaArchive(nestedPath, archiveReader)
+	for archivePath, archiveOpener := range openers {
+		archiveReadCloser, err := archiveOpener.Open()
 		if err != nil {
+			return nil, fmt.Errorf("unable to open archived file from tempdir: %w", err)
+		}
+		nestedPath := fmt.Sprintf("%s:%s", j.virtualPath, archivePath)
+		nestedPkgs, err := parseJavaArchive(nestedPath, archiveReadCloser)
+		if err != nil {
+			if closeErr := archiveReadCloser.Close(); closeErr != nil {
+				log.Errorf("unable to close archived file from tempdir: %+v", closeErr)
+			}
 			return nil, fmt.Errorf("unable to process nested java archive (%s): %w", archivePath, err)
 		}
+		if err = archiveReadCloser.Close(); err != nil {
+			return nil, fmt.Errorf("unable to close archived file from tempdir: %w", err)
+		}
 
 		// attach the parent package to all discovered packages that are not already associated with a java archive
 		for _, p := range nestedPkgs {

syft/cataloger/java/archive_parser_test.go

@@ -78,51 +78,6 @@ func generateJavaBuildFixture(t *testing.T, fixturePath string) {
 	}
 }
 
-func TestSelectName(t *testing.T) {
-	tests := []struct {
-		desc     string
-		manifest pkg.JavaManifest
-		archive  archiveFilename
-		expected string
-	}{
-		{
-			desc:    "name from Implementation-Title",
-			archive: archiveFilename{},
-			manifest: pkg.JavaManifest{
-				Name:      "",
-				SpecTitle: "",
-				ImplTitle: "maven-wrapper",
-			},
-			expected: "maven-wrapper",
-		},
-		{
-			desc: "Implementation-Title does not override",
-			manifest: pkg.JavaManifest{
-				Name:      "Foo",
-				SpecTitle: "",
-				ImplTitle: "maven-wrapper",
-			},
-			archive: archiveFilename{
-				fields: []map[string]string{
-					{"name": "omg"},
-				},
-			},
-			expected: "omg",
-		},
-	}
-
-	for _, test := range tests {
-		t.Run(test.desc, func(t *testing.T) {
-			result := selectName(&test.manifest, test.archive)
-
-			if result != test.expected {
-				t.Errorf("mismatch in names: '%s' != '%s'", result, test.expected)
-			}
-		})
-	}
-
-}
-
 func TestParseJar(t *testing.T) {
 	tests := []struct {
 		fixture      string
@@ -137,17 +92,20 @@ func TestParseJar(t *testing.T) {
 			},
 			expected: map[string]pkg.Package{
 				"example-jenkins-plugin": {
-					Name:     "example-jenkins-plugin",
-					Version:  "1.0-SNAPSHOT",
-					Language: pkg.Java,
-					Type:     pkg.JenkinsPluginPkg,
+					Name:         "example-jenkins-plugin",
+					Version:      "1.0-SNAPSHOT",
+					Language:     pkg.Java,
+					Type:         pkg.JenkinsPluginPkg,
+					MetadataType: pkg.JavaMetadataType,
 					Metadata: pkg.JavaMetadata{
+						VirtualPath: "test-fixtures/java-builds/packages/example-jenkins-plugin.hpi",
 						Manifest: &pkg.JavaManifest{
-							ManifestVersion: "1.0",
-							SpecTitle:       "The Jenkins Plugins Parent POM Project",
-							ImplTitle:       "example-jenkins-plugin",
-							ImplVersion:     "1.0-SNAPSHOT",
-							Extra: map[string]string{
+							Main: map[string]string{
+								"Manifest-Version":       "1.0",
+								"Specification-Title":    "The Jenkins Plugins Parent POM Project",
+								"Implementation-Title":   "example-jenkins-plugin",
+								"Implementation-Version": "1.0-SNAPSHOT",
+								// extra fields...
 								"Archiver-Version":    "Plexus Archiver",
 								"Plugin-License-Url":  "https://opensource.org/licenses/MIT",
 								"Plugin-License-Name": "MIT License",
@@ -181,13 +139,17 @@ func TestParseJar(t *testing.T) {
 			fixture: "test-fixtures/java-builds/packages/example-java-app-gradle-0.1.0.jar",
 			expected: map[string]pkg.Package{
 				"example-java-app-gradle": {
-					Name:     "example-java-app-gradle",
-					Version:  "0.1.0",
-					Language: pkg.Java,
-					Type:     pkg.JavaPkg,
+					Name:         "example-java-app-gradle",
+					Version:      "0.1.0",
+					Language:     pkg.Java,
+					Type:         pkg.JavaPkg,
+					MetadataType: pkg.JavaMetadataType,
 					Metadata: pkg.JavaMetadata{
+						VirtualPath: "test-fixtures/java-builds/packages/example-java-app-gradle-0.1.0.jar",
 						Manifest: &pkg.JavaManifest{
-							ManifestVersion: "1.0",
+							Main: map[string]string{
+								"Manifest-Version": "1.0",
+							},
 						},
 					},
 				},
@@ -200,14 +162,17 @@ func TestParseJar(t *testing.T) {
 			},
 			expected: map[string]pkg.Package{
 				"example-java-app-maven": {
-					Name:     "example-java-app-maven",
-					Version:  "0.1.0",
-					Language: pkg.Java,
-					Type:     pkg.JavaPkg,
+					Name:         "example-java-app-maven",
+					Version:      "0.1.0",
+					Language:     pkg.Java,
+					Type:         pkg.JavaPkg,
+					MetadataType: pkg.JavaMetadataType,
 					Metadata: pkg.JavaMetadata{
+						VirtualPath: "test-fixtures/java-builds/packages/example-java-app-maven-0.1.0.jar",
 						Manifest: &pkg.JavaManifest{
-							ManifestVersion: "1.0",
-							Extra: map[string]string{
+							Main: map[string]string{
+								"Manifest-Version": "1.0",
+								// extra fields...
 								"Archiver-Version": "Plexus Archiver",
 								"Created-By":       "Apache Maven 3.6.3",
 								"Built-By":         "?",
@@ -224,11 +189,15 @@ func TestParseJar(t *testing.T) {
 					},
 				},
 				"joda-time": {
-					Name:     "joda-time",
-					Version:  "2.9.2",
-					Language: pkg.Java,
-					Type:     pkg.JavaPkg,
+					Name:         "joda-time",
+					Version:      "2.9.2",
+					Language:     pkg.Java,
+					Type:         pkg.JavaPkg,
+					MetadataType: pkg.JavaMetadataType,
 					Metadata: pkg.JavaMetadata{
+						// ensure that nested packages with different names than that of the parent are appended as
+						// a suffix on the virtual path
+						VirtualPath: "test-fixtures/java-builds/packages/example-java-app-maven-0.1.0.jar:joda-time",
 						PomProperties: &pkg.PomProperties{
 							Path:       "META-INF/maven/joda-time/joda-time/pom.properties",
 							GroupID:    "joda-time",
@@ -295,11 +264,11 @@ func TestParseJar(t *testing.T) {
 				metadata := a.Metadata.(pkg.JavaMetadata)
 				metadata.Parent = nil
 
-				// ignore select fields
+				// ignore select fields (only works for the main section)
 				for _, field := range test.ignoreExtras {
-					if metadata.Manifest != nil && metadata.Manifest.Extra != nil {
-						if _, ok := metadata.Manifest.Extra[field]; ok {
-							delete(metadata.Manifest.Extra, field)
+					if metadata.Manifest != nil && metadata.Manifest.Main != nil {
+						if _, ok := metadata.Manifest.Main[field]; ok {
+							delete(metadata.Manifest.Main, field)
 						}
 					}
 				}
@@ -334,7 +303,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-boot-starter",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "jul-to-slf4j",
@@ -346,7 +315,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-boot-starter-validation",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "hibernate-validator",
@@ -358,7 +327,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-expression",
-					Version: "5.2.2.RELEASE",
+					Version: "5.2.2",
 				},
 				{
 					Name:    "jakarta.validation-api",
@@ -366,11 +335,11 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-web",
-					Version: "5.2.2.RELEASE",
+					Version: "5.2.2",
 				},
 				{
 					Name:    "spring-boot-starter-actuator",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "log4j-api",
@@ -390,23 +359,23 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-aop",
-					Version: "5.2.2.RELEASE",
+					Version: "5.2.2",
 				},
 				{
 					Name:    "spring-boot-actuator-autoconfigure",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "spring-jcl",
-					Version: "5.2.2.RELEASE",
+					Version: "5.2.2",
 				},
 				{
 					Name:    "spring-boot",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "spring-boot-starter-logging",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "jakarta.annotation-api",
@@ -414,7 +383,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-webmvc",
-					Version: "5.2.2.RELEASE",
+					Version: "5.2.2",
 				},
 				{
 					Name:    "HdrHistogram",
@@ -422,7 +391,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-boot-starter-web",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "logback-classic",
@@ -434,7 +403,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-boot-starter-json",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "jackson-databind",
@@ -450,7 +419,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-boot-autoconfigure",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "jackson-datatype-jdk8",
@@ -466,11 +435,11 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-beans",
-					Version: "5.2.2.RELEASE",
+					Version: "5.2.2",
 				},
 				{
 					Name:    "spring-boot-actuator",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "slf4j-api",
@@ -478,7 +447,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-core",
-					Version: "5.2.2.RELEASE",
+					Version: "5.2.2",
 				},
 				{
 					Name:    "logback-core",
@@ -498,7 +467,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-boot-starter-tomcat",
-					Version: "2.2.2.RELEASE",
+					Version: "2.2.2",
 				},
 				{
 					Name:    "classmate",
@@ -506,7 +475,7 @@ func TestParseNestedJar(t *testing.T) {
 				},
 				{
 					Name:    "spring-context",
-					Version: "5.2.2.RELEASE",
+					Version: "5.2.2",
 				},
 			},
 		},
@@ -527,7 +496,7 @@ func TestParseNestedJar(t *testing.T) {
 				t.Fatalf("failed to parse java archive: %+v", err)
 			}
 
-			nameVersionPairSet := internal.NewStringSet()
+			expectedNameVersionPairSet := internal.NewStringSet()
 
 			makeKey := func(p *pkg.Package) string {
 				if p == nil {
@@ -537,20 +506,32 @@ func TestParseNestedJar(t *testing.T) {
 			}
 
 			for _, e := range test.expected {
-				nameVersionPairSet.Add(makeKey(&e))
+				expectedNameVersionPairSet.Add(makeKey(&e))
 			}
 
-			if len(actual) != len(nameVersionPairSet) {
+			if len(actual) != len(expectedNameVersionPairSet) {
+				actualNameVersionPairSet := internal.NewStringSet()
 				for _, a := range actual {
-					t.Log("   ", a)
+					key := makeKey(&a)
+					actualNameVersionPairSet.Add(key)
+					if !expectedNameVersionPairSet.Contains(key) {
+						t.Logf("extra package: %s", a)
+					}
 				}
-				t.Fatalf("unexpected package count: %d!=%d", len(actual), len(nameVersionPairSet))
+
+				for _, key := range expectedNameVersionPairSet.ToSlice() {
+					if !actualNameVersionPairSet.Contains(key) {
+						t.Logf("missing package: %s", key)
+					}
+				}
+
+				t.Fatalf("unexpected package count: %d!=%d", len(actual), len(expectedNameVersionPairSet))
 			}
 
 			for _, a := range actual {
 				actualKey := makeKey(&a)
 
-				if !nameVersionPairSet.Contains(actualKey) {
+				if !expectedNameVersionPairSet.Contains(actualKey) {
 					t.Errorf("unexpected pkg: %q", actualKey)
 				}
 
@@ -563,7 +544,7 @@ func TestParseNestedJar(t *testing.T) {
 					if metadata.Parent == nil {
 						t.Errorf("unassigned error for pkg=%q", actualKey)
 					} else if makeKey(metadata.Parent) != "spring-boot|0.0.1-SNAPSHOT" {
-						// NB: this is a hard-coded condition to simplify the test harness
+						// NB: this is a hard-coded condition to simplify the test harness to account for https://github.com/micrometer-metrics/micrometer/issues/1785
 						if a.Name == "pcollections" {
 							if metadata.Parent.Name != "micrometer-core" {
 								t.Errorf("nested 'pcollections' pkg has wrong parent: %q", metadata.Parent.Name)

syft/cataloger/java/java_manifest.go

@@ -4,25 +4,41 @@ import (
 	"bufio"
 	"fmt"
 	"io"
+	"strconv"
 	"strings"
 
+	"github.com/anchore/syft/internal/log"
+
 	"github.com/anchore/syft/syft/pkg"
-	"github.com/mitchellh/mapstructure"
 )
 
-const manifestPath = "META-INF/MANIFEST.MF"
+const manifestGlob = "/META-INF/MANIFEST.MF"
 
-func parseJavaManifest(reader io.Reader) (*pkg.JavaManifest, error) {
+// nolint:funlen
+// parseJavaManifest takes MANIFEST.MF file content and returns sections of parsed key/value pairs.
+// For more information: https://docs.oracle.com/en/java/javase/11/docs/specs/jar/jar.html#jar-manifest
+func parseJavaManifest(path string, reader io.Reader) (*pkg.JavaManifest, error) {
 	var manifest pkg.JavaManifest
-	manifestMap := make(map[string]string)
+	sections := []map[string]string{
+		make(map[string]string),
+	}
+	currentSection := 0
 	scanner := bufio.NewScanner(reader)
 	var lastKey string
 	for scanner.Scan() {
 		line := scanner.Text()
 
-		// ignore empty lines
+		// empty lines denote section separators
 		if strings.TrimSpace(line) == "" {
+			currentSection++
+			// we don't want to allocate a new section map that wont necessarily be used, do that once there is
+			// a non-empty line to process
+
+			// do not process line continuations after this
+			lastKey = ""
 			continue
+		} else if currentSection >= len(sections) {
+			sections = append(sections, make(map[string]string))
 		}
 
 		if line[0] == ' ' {
@@ -30,7 +46,7 @@ func parseJavaManifest(reader io.Reader) (*pkg.JavaManifest, error) {
 			if lastKey == "" {
 				return nil, fmt.Errorf("found continuation with no previous key (%s)", line)
 			}
-			manifestMap[lastKey] += strings.TrimSpace(line)
+			sections[currentSection][lastKey] += strings.TrimSpace(line)
 		} else {
 			// this is a new key-value pair
 			idx := strings.Index(line, ":")
@@ -40,9 +56,9 @@ func parseJavaManifest(reader io.Reader) (*pkg.JavaManifest, error) {
 
 			key := strings.TrimSpace(line[0:idx])
 			value := strings.TrimSpace(line[idx+1:])
-			manifestMap[key] = value
+			sections[currentSection][key] = value
 
-			// keep track of key for future continuations
+			// keep track of key for potential future continuations
 			lastKey = key
 		}
 	}
@@ -51,14 +67,24 @@ func parseJavaManifest(reader io.Reader) (*pkg.JavaManifest, error) {
 		return nil, fmt.Errorf("unable to read java manifest: %w", err)
 	}
 
-	if err := mapstructure.Decode(manifestMap, &manifest); err != nil {
-		return nil, fmt.Errorf("unable to parse java manifest: %w", err)
-	}
-
-	// clean select fields
-	if strings.Trim(manifest.ImplVersion, " ") != "" {
-		// transform versions with dates attached to just versions (e.g. "1.3 2244 October 5 2008" --> "1.3")
-		manifest.ImplVersion = strings.Split(manifest.ImplVersion, " ")[0]
+	if len(sections) > 0 {
+		manifest.Main = sections[0]
+		if len(sections) > 1 {
+			manifest.NamedSections = make(map[string]map[string]string)
+			for i, s := range sections[1:] {
+				name, ok := s["Name"]
+				if !ok {
+					// per the manifest spec (https://docs.oracle.com/en/java/javase/11/docs/specs/jar/jar.html#jar-manifest)
+					// this should never happen. If it does we want to know about it, but not necessarily stop
+					// cataloging entirely... for this reason we only log.
+					log.Errorf("java manifest section found without a name: %s", path)
+					name = strconv.Itoa(i)
+				} else {
+					delete(s, "Name")
+				}
+				manifest.NamedSections[name] = s
+			}
+		}
 	}
 
 	return &manifest, nil
@@ -69,24 +95,21 @@ func selectName(manifest *pkg.JavaManifest, filenameObj archiveFilename) string
 	switch {
 	case filenameObj.name() != "":
 		name = filenameObj.name()
-	case manifest.Name != "":
+	case manifest.Main["Name"] != "":
 		// Manifest original spec...
-		name = manifest.Name
-	case manifest.Extra["Bundle-Name"] != "":
+		name = manifest.Main["Name"]
+	case manifest.Main["Bundle-Name"] != "":
 		// BND tooling...
-		name = manifest.Extra["Bundle-Name"]
-	case manifest.Extra["Short-Name"] != "":
+		name = manifest.Main["Bundle-Name"]
+	case manifest.Main["Short-Name"] != "":
 		// Jenkins...
-		name = manifest.Extra["Short-Name"]
-	case manifest.Extra["Extension-Name"] != "":
+		name = manifest.Main["Short-Name"]
+	case manifest.Main["Extension-Name"] != "":
 		// Jenkins...
-		name = manifest.Extra["Extension-Name"]
-	}
-
-	// in situations where we hit this point and no name was
-	// determined, look at the Implementation-Title
-	if name == "" && manifest.ImplTitle != "" {
-		name = manifest.ImplTitle
+		name = manifest.Main["Extension-Name"]
+	case manifest.Main["Implementation-Title"] != "":
+		// last ditch effort...
+		name = manifest.Main["Implementation-Title"]
 	}
 	return name
 }
@@ -94,14 +117,14 @@ func selectName(manifest *pkg.JavaManifest, filenameObj archiveFilename) string
 func selectVersion(manifest *pkg.JavaManifest, filenameObj archiveFilename) string {
 	var version string
 	switch {
-	case manifest.ImplVersion != "":
-		version = manifest.ImplVersion
 	case filenameObj.version() != "":
 		version = filenameObj.version()
-	case manifest.SpecVersion != "":
-		version = manifest.SpecVersion
-	case manifest.Extra["Plugin-Version"] != "":
-		version = manifest.Extra["Plugin-Version"]
+	case manifest.Main["Implementation-Version"] != "":
+		version = manifest.Main["Implementation-Version"]
+	case manifest.Main["Specification-Version"] != "":
+		version = manifest.Main["Specification-Version"]
+	case manifest.Main["Plugin-Version"] != "":
+		version = manifest.Main["Plugin-Version"]
 	}
 	return version
 }

syft/cataloger/java/java_manifest_test.go

@@ -2,10 +2,11 @@ package java
 
 import (
 	"encoding/json"
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/go-test/deep"
 	"os"
 	"testing"
+
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/go-test/deep"
 )
 
 func TestParseJavaManifest(t *testing.T) {
@@ -16,49 +17,62 @@ func TestParseJavaManifest(t *testing.T) {
 		{
 			fixture: "test-fixtures/manifest/small",
 			expected: pkg.JavaManifest{
-				ManifestVersion: "1.0",
+				Main: map[string]string{
+					"Manifest-Version": "1.0",
+				},
 			},
 		},
 		{
 			fixture: "test-fixtures/manifest/standard-info",
 			expected: pkg.JavaManifest{
-				ManifestVersion: "1.0",
-				Name:            "the-best-name",
-				SpecTitle:       "the-spec-title",
-				SpecVersion:     "the-spec-version",
-				SpecVendor:      "the-spec-vendor",
-				ImplTitle:       "the-impl-title",
-				ImplVersion:     "the-impl-version",
-				ImplVendor:      "the-impl-vendor",
+				Main: map[string]string{
+					"Name":                   "the-best-name",
+					"Manifest-Version":       "1.0",
+					"Specification-Title":    "the-spec-title",
+					"Specification-Version":  "the-spec-version",
+					"Specification-Vendor":   "the-spec-vendor",
+					"Implementation-Title":   "the-impl-title",
+					"Implementation-Version": "the-impl-version",
+					"Implementation-Vendor":  "the-impl-vendor",
+				},
 			},
 		},
 		{
 			fixture: "test-fixtures/manifest/extra-info",
 			expected: pkg.JavaManifest{
-				ManifestVersion: "1.0",
-				Extra: map[string]string{
+				Main: map[string]string{
+					"Manifest-Version": "1.0",
 					"Archiver-Version": "Plexus Archiver",
-					"Build-Jdk":        "14.0.1",
-					"Built-By":         "?",
 					"Created-By":       "Apache Maven 3.6.3",
-					"Main-Class":       "hello.HelloWorld",
+				},
+				NamedSections: map[string]map[string]string{
+					"thing-1": {
+						"Built-By": "?",
+					},
+					"1": {
+						"Build-Jdk":  "14.0.1",
+						"Main-Class": "hello.HelloWorld",
+					},
 				},
 			},
 		},
 		{
 			fixture: "test-fixtures/manifest/continuation",
 			expected: pkg.JavaManifest{
-				ManifestVersion: "1.0",
-				Extra: map[string]string{
-					"Plugin-ScmUrl": "https://github.com/jenkinsci/plugin-pom/example-jenkins-plugin",
+				Main: map[string]string{
+					"Manifest-Version": "1.0",
+					"Plugin-ScmUrl":    "https://github.com/jenkinsci/plugin-pom/example-jenkins-plugin",
 				},
 			},
 		},
 		{
+			// regression test, we should always keep the full version
 			fixture: "test-fixtures/manifest/version-with-date",
 			expected: pkg.JavaManifest{
-				ManifestVersion: "1.0",
-				ImplVersion:     "1.3", // ensure the date is stripped off during processing
+				Main: map[string]string{
+					"Manifest-Version":       "1.0",
+					"Implementation-Version": "1.3 2244 October 5 2005",
+				},
 			},
 		},
 	}
@@ -70,7 +84,7 @@ func TestParseJavaManifest(t *testing.T) {
 				t.Fatalf("could not open fixture: %+v", err)
 			}
 
-			actual, err := parseJavaManifest(fixture)
+			actual, err := parseJavaManifest(test.fixture, fixture)
 			if err != nil {
 				t.Fatalf("failed to parse manifest: %+v", err)
 			}
@@ -91,3 +105,45 @@ func TestParseJavaManifest(t *testing.T) {
 		})
 	}
 }
+
+func TestSelectName(t *testing.T) {
+	tests := []struct {
+		desc     string
+		manifest pkg.JavaManifest
+		archive  archiveFilename
+		expected string
+	}{
+		{
+			desc:    "Get name from Implementation-Title",
+			archive: archiveFilename{},
+			manifest: pkg.JavaManifest{
+				Main: map[string]string{
+					"Implementation-Title": "maven-wrapper",
+				},
+			},
+			expected: "maven-wrapper",
+		},
+		{
+			desc: "Implementation-Title does not override name from filename",
+			manifest: pkg.JavaManifest{
+				Main: map[string]string{
+					"Name":                 "foo",
+					"Implementation-Title": "maven-wrapper",
+				},
+			},
+			archive:  newJavaArchiveFilename("/something/omg.jar"),
+			expected: "omg",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.desc, func(t *testing.T) {
+			result := selectName(&test.manifest, test.archive)
+
+			if result != test.expected {
+				t.Errorf("mismatch in names: '%s' != '%s'", result, test.expected)
+			}
+		})
+	}
+
+}

syft/cataloger/java/save_archive_to_tmp.go

@@ -35,6 +35,7 @@ func saveArchiveToTmp(reader io.Reader) (string, string, func(), error) {
 	if err != nil {
 		return contentDir, "", cleanupFn, fmt.Errorf("unable to create archive: %w", err)
 	}
+	defer archiveFile.Close()
 
 	_, err = io.Copy(archiveFile, reader)
 	if err != nil {

syft/cataloger/java/test-fixtures/manifest/extra-info

@@ -2,6 +2,7 @@ Manifest-Version: 1.0
 Archiver-Version: Plexus Archiver
 Created-By: Apache Maven 3.6.3
 
+Name: thing-1
 Built-By: ?
 
 Build-Jdk: 14.0.1

syft/cataloger/javascript/cataloger.go

@@ -7,12 +7,21 @@ import (
 	"github.com/anchore/syft/syft/cataloger/common"
 )
 
-// NewJavascriptCataloger returns a new JavaScript cataloger object.
-func NewJavascriptCataloger() *common.GenericCataloger {
+// NewJavascriptPackageCataloger returns a new JavaScript cataloger object based on detection of npm based packages.
+func NewJavascriptPackageCataloger() *common.GenericCataloger {
+	globParsers := map[string]common.ParserFn{
+		"**/package.json": parsePackageJSON,
+	}
+
+	return common.NewGenericCataloger(nil, globParsers, "javascript-package-cataloger")
+}
+
+// NewJavascriptLockCataloger returns a new Javascript cataloger object base on package lock files.
+func NewJavascriptLockCataloger() *common.GenericCataloger {
 	globParsers := map[string]common.ParserFn{
 		"**/package-lock.json": parsePackageLock,
 		"**/yarn.lock":         parseYarnLock,
 	}
 
-	return common.NewGenericCataloger(nil, globParsers, "javascript-cataloger")
+	return common.NewGenericCataloger(nil, globParsers, "javascript-lock-cataloger")
 }

syft/cataloger/javascript/parse_package_json.go

@@ -0,0 +1,140 @@
+package javascript
+
+import (
+	"encoding/json"
+	"fmt"
+	"io"
+	"regexp"
+
+	"github.com/anchore/syft/internal"
+
+	"github.com/mitchellh/mapstructure"
+
+	"github.com/anchore/syft/syft/cataloger/common"
+	"github.com/anchore/syft/syft/pkg"
+)
+
+// integrity check
+var _ common.ParserFn = parsePackageLock
+
+// PackageJSON represents a JavaScript package.json file
+type PackageJSON struct {
+	Version      string            `json:"version"`
+	Latest       []string          `json:"latest"`
+	Author       Author            `json:"author"`
+	License      string            `json:"license"`
+	Name         string            `json:"name"`
+	Homepage     string            `json:"homepage"`
+	Description  string            `json:"description"`
+	Dependencies map[string]string `json:"dependencies"`
+	Repository   Repository        `json:"repository"`
+}
+
+type Author struct {
+	Name  string `json:"name" mapstruct:"name"`
+	Email string `json:"email" mapstruct:"email"`
+	URL   string `json:"url" mapstruct:"url"`
+}
+
+type Repository struct {
+	Type string `json:"type" mapstructure:"type"`
+	URL  string `json:"url" mapstructure:"url"`
+}
+
+// match example: "author": "Isaac Z. Schlueter <i@izs.me> (http://blog.izs.me)"
+// ---> name: "Isaac Z. Schlueter" email: "i@izs.me" url: "http://blog.izs.me"
+var authorPattern = regexp.MustCompile(`^\s*(?P<name>[^<(]*)(\s+<(?P<email>.*)>)?(\s\((?P<url>.*)\))?\s*$`)
+
+// Exports Author.UnmarshalJSON interface to help normalize the json structure.
+func (a *Author) UnmarshalJSON(b []byte) error {
+	var authorStr string
+	var fields map[string]string
+	var author Author
+
+	if err := json.Unmarshal(b, &authorStr); err != nil {
+		// string parsing did not work, assume a map was given
+		// for more information: https://docs.npmjs.com/files/package.json#people-fields-author-contributors
+		if err := json.Unmarshal(b, &fields); err != nil {
+			return fmt.Errorf("unable to parse package.json author: %w", err)
+		}
+	} else {
+		// parse out "name <email> (url)" into an Author struct
+		fields = internal.MatchCaptureGroups(authorPattern, authorStr)
+	}
+
+	// translate the map into a structure
+	if err := mapstructure.Decode(fields, &author); err != nil {
+		return fmt.Errorf("unable to decode package.json author: %w", err)
+	}
+
+	*a = author
+
+	return nil
+}
+
+func (a *Author) AuthorString() string {
+	result := a.Name
+	if a.Email != "" {
+		result += fmt.Sprintf(" <%s>", a.Email)
+	}
+	if a.URL != "" {
+		result += fmt.Sprintf(" (%s)", a.URL)
+	}
+	return result
+}
+
+func (r *Repository) UnmarshalJSON(b []byte) error {
+	var repositoryStr string
+	var fields map[string]string
+	var repository Repository
+
+	if err := json.Unmarshal(b, &repositoryStr); err != nil {
+		// string parsing did not work, assume a map was given
+		// for more information: https://docs.npmjs.com/files/package.json#people-fields-author-contributors
+		if err := json.Unmarshal(b, &fields); err != nil {
+			return fmt.Errorf("unable to parse package.json author: %w", err)
+		}
+		// translate the map into a structure
+		if err := mapstructure.Decode(fields, &repository); err != nil {
+			return fmt.Errorf("unable to decode package.json author: %w", err)
+		}
+
+		*r = repository
+	} else {
+		r.URL = repositoryStr
+	}
+
+	return nil
+}
+
+// parsePackageJson parses a package.json and returns the discovered JavaScript packages.
+func parsePackageJSON(_ string, reader io.Reader) ([]pkg.Package, error) {
+	packages := make([]pkg.Package, 0)
+	dec := json.NewDecoder(reader)
+
+	for {
+		var p PackageJSON
+		if err := dec.Decode(&p); err == io.EOF {
+			break
+		} else if err != nil {
+			return nil, fmt.Errorf("failed to parse package.json file: %w", err)
+		}
+
+		packages = append(packages, pkg.Package{
+			Name:         p.Name,
+			Version:      p.Version,
+			Licenses:     []string{p.License},
+			Language:     pkg.JavaScript,
+			Type:         pkg.NpmPkg,
+			MetadataType: pkg.NpmPackageJSONMetadataType,
+			Metadata: pkg.NpmPackageJSONMetadata{
+				Author:   p.Author.AuthorString(),
+				Homepage: p.Homepage,
+				URL:      p.Repository.URL,
+				Licenses: []string{p.License},
+			},
+		})
+	}
+
+	return packages, nil
+}

syft/cataloger/javascript/parse_package_json_test.go

@@ -0,0 +1,93 @@
+package javascript
+
+import (
+	"os"
+	"testing"
+
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/go-test/deep"
+)
+
+func TestParsePackageJSON(t *testing.T) {
+	tests := []struct {
+		Fixture     string
+		ExpectedPkg pkg.Package
+	}{
+		{
+			Fixture: "test-fixtures/pkg-json/package.json",
+			ExpectedPkg: pkg.Package{
+				Name:         "npm",
+				Version:      "6.14.6",
+				Type:         pkg.NpmPkg,
+				Licenses:     []string{"Artistic-2.0"},
+				Language:     pkg.JavaScript,
+				MetadataType: pkg.NpmPackageJSONMetadataType,
+				Metadata: pkg.NpmPackageJSONMetadata{
+					Author:   "Isaac Z. Schlueter <i@izs.me> (http://blog.izs.me)",
+					Homepage: "https://docs.npmjs.com/",
+					URL:      "https://github.com/npm/cli",
+					Licenses: []string{"Artistic-2.0"},
+				},
+			},
+		},
+		{
+			Fixture: "test-fixtures/pkg-json/package-nested-author.json",
+			ExpectedPkg: pkg.Package{
+				Name:         "npm",
+				Version:      "6.14.6",
+				Type:         pkg.NpmPkg,
+				Licenses:     []string{"Artistic-2.0"},
+				Language:     pkg.JavaScript,
+				MetadataType: pkg.NpmPackageJSONMetadataType,
+				Metadata: pkg.NpmPackageJSONMetadata{
+					Author:   "Isaac Z. Schlueter <i@izs.me> (http://blog.izs.me)",
+					Homepage: "https://docs.npmjs.com/",
+					URL:      "https://github.com/npm/cli",
+					Licenses: []string{"Artistic-2.0"},
+				},
+			},
+		},
+		{
+			Fixture: "test-fixtures/pkg-json/package-repo-string.json",
+			ExpectedPkg: pkg.Package{
+				Name:         "function-bind",
+				Version:      "1.1.1",
+				Type:         pkg.NpmPkg,
+				Licenses:     []string{"MIT"},
+				Language:     pkg.JavaScript,
+				MetadataType: pkg.NpmPackageJSONMetadataType,
+				Metadata: pkg.NpmPackageJSONMetadata{
+					Author:   "Raynos <raynos2@gmail.com>",
+					Homepage: "https://github.com/Raynos/function-bind",
+					URL:      "git://github.com/Raynos/function-bind.git",
+					Licenses: []string{"MIT"},
+				},
+			},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.Fixture, func(t *testing.T) {
+			fixture, err := os.Open(test.Fixture)
+			if err != nil {
+				t.Fatalf("failed to open fixture: %+v", err)
+			}
+
+			actual, err := parsePackageJSON(fixture.Name(), fixture)
+			if err != nil {
+				t.Fatalf("failed to parse package-lock.json: %+v", err)
+			}
+			if len(actual) != 1 {
+				for _, a := range actual {
+					t.Log("   ", a)
+				}
+				t.Fatalf("unexpected package count: %d!=1", len(actual))
+			}
+
+			for _, d := range deep.Equal(actual[0], test.ExpectedPkg) {
+
+				t.Errorf("diff: %+v", d)
+			}
+		})
+	}
+}

syft/cataloger/javascript/parse_yarn_lock.go

@@ -66,7 +66,7 @@ func parseYarnLock(_ string, reader io.Reader) ([]pkg.Package, error) {
 				Name:     currentName,
 				Version:  strings.Trim(version, "\""),
 				Language: pkg.JavaScript,
-				Type:     pkg.YarnPkg,
+				Type:     pkg.NpmPkg,
 			})
 		}
 	}

syft/cataloger/javascript/parse_yarn_lock_test.go

@@ -13,43 +13,43 @@ func TestParseYarnLock(t *testing.T) {
 			Name:     "@babel/code-frame",
 			Version:  "7.10.4",
 			Language: pkg.JavaScript,
-			Type:     pkg.YarnPkg,
+			Type:     pkg.NpmPkg,
 		},
 		"@types/minimatch": {
 			Name:     "@types/minimatch",
 			Version:  "3.0.3",
 			Language: pkg.JavaScript,
-			Type:     pkg.YarnPkg,
+			Type:     pkg.NpmPkg,
 		},
 		"@types/qs": {
 			Name:     "@types/qs",
 			Version:  "6.9.4",
 			Language: pkg.JavaScript,
-			Type:     pkg.YarnPkg,
+			Type:     pkg.NpmPkg,
 		},
 		"ajv": {
 			Name:     "ajv",
 			Version:  "6.12.3",
 			Language: pkg.JavaScript,
-			Type:     pkg.YarnPkg,
+			Type:     pkg.NpmPkg,
 		},
 		"atob": {
 			Name:     "atob",
 			Version:  "2.1.2",
 			Language: pkg.JavaScript,
-			Type:     pkg.YarnPkg,
+			Type:     pkg.NpmPkg,
 		},
 		"aws-sdk": {
 			Name:     "aws-sdk",
 			Version:  "2.706.0",
 			Language: pkg.JavaScript,
-			Type:     pkg.YarnPkg,
+			Type:     pkg.NpmPkg,
 		},
 		"jhipster-core": {
 			Name:     "jhipster-core",
 			Version:  "7.3.4",
 			Language: pkg.JavaScript,
-			Type:     pkg.YarnPkg,
+			Type:     pkg.NpmPkg,
 		},
 	}
 	fixture, err := os.Open("test-fixtures/yarn/yarn.lock")

syft/cataloger/javascript/test-fixtures/pkg-json/package-nested-author.json

@@ -0,0 +1,16 @@
+{
+  "version": "6.14.6",
+  "name": "npm",
+  "description": "a package manager for JavaScript",
+  "homepage": "https://docs.npmjs.com/",
+  "author": {
+    "name": "Isaac Z. Schlueter",
+    "email": "i@izs.me",
+    "url": "http://blog.izs.me"
+  },
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/npm/cli"
+  },
+  "license": "Artistic-2.0"
+}

syft/cataloger/javascript/test-fixtures/pkg-json/package-repo-string.json

@@ -0,0 +1,66 @@
+{
+  "name": "function-bind",
+  "version": "1.1.1",
+  "description": "Implementation of Function.prototype.bind",
+  "keywords": [
+    "function",
+    "bind",
+    "shim",
+    "es5"
+  ],
+  "author": "Raynos <raynos2@gmail.com>",
+  "repository": "git://github.com/Raynos/function-bind.git",
+  "main": "index",
+  "homepage": "https://github.com/Raynos/function-bind",
+  "contributors": [
+    {
+      "name": "Raynos"
+    },
+    {
+      "name": "Jordan Harband",
+      "url": "https://github.com/ljharb"
+    }
+  ],
+  "bugs": {
+    "url": "https://github.com/Raynos/function-bind/issues",
+    "email": "raynos2@gmail.com"
+  },
+  "dependencies": {},
+  "devDependencies": {
+    "@ljharb/eslint-config": "^12.2.1",
+    "covert": "^1.1.0",
+    "eslint": "^4.5.0",
+    "jscs": "^3.0.7",
+    "tape": "^4.8.0"
+  },
+  "license": "MIT",
+  "scripts": {
+    "pretest": "npm run lint",
+    "test": "npm run tests-only",
+    "posttest": "npm run coverage -- --quiet",
+    "tests-only": "node test",
+    "coverage": "covert test/*.js",
+    "lint": "npm run jscs && npm run eslint",
+    "jscs": "jscs *.js */*.js",
+    "eslint": "eslint *.js */*.js"
+  },
+  "testling": {
+    "files": "test/index.js",
+    "browsers": [
+      "ie/8..latest",
+      "firefox/16..latest",
+      "firefox/nightly",
+      "chrome/22..latest",
+      "chrome/canary",
+      "opera/12..latest",
+      "opera/next",
+      "safari/5.1..latest",
+      "ipad/6.0..latest",
+      "iphone/6.0..latest",
+      "android-browser/4.2..latest"
+    ]
+  }
+,"_resolved": "https://registry.npmjs.org/function-bind/-/function-bind-1.1.1.tgz"
+,"_integrity": "sha512-yIovAzMX49sF8Yl58fSCWJ5svSLuaibPxXQJFLmBObTuCr0Mf1KiPopGM9NiFjiYBCbfaa2Fh6breQ6ANVTI0A=="
+,"_from": "function-bind@1.1.1"
+}

syft/cataloger/javascript/test-fixtures/pkg-json/package.json

@@ -0,0 +1,314 @@
+{
+  "version": "6.14.6",
+  "name": "npm",
+  "description": "a package manager for JavaScript",
+  "keywords": [
+    "install",
+    "modules",
+    "package manager",
+    "package.json"
+  ],
+  "preferGlobal": true,
+  "config": {
+    "publishtest": false
+  },
+  "homepage": "https://docs.npmjs.com/",
+  "author": "Isaac Z. Schlueter <i@izs.me> (http://blog.izs.me)",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/npm/cli"
+  },
+  "bugs": {
+    "url": "https://npm.community/c/bugs"
+  },
+  "directories": {
+    "bin": "./bin",
+    "doc": "./doc",
+    "lib": "./lib",
+    "man": "./man"
+  },
+  "main": "./lib/npm.js",
+  "bin": {
+    "npm": "./bin/npm-cli.js",
+    "npx": "./bin/npx-cli.js"
+  },
+  "dependencies": {
+    "JSONStream": "^1.3.5",
+    "abbrev": "~1.1.1",
+    "ansicolors": "~0.3.2",
+    "ansistyles": "~0.1.3",
+    "aproba": "^2.0.0",
+    "archy": "~1.0.0",
+    "bin-links": "^1.1.7",
+    "bluebird": "^3.5.5",
+    "byte-size": "^5.0.1",
+    "cacache": "^12.0.3",
+    "call-limit": "^1.1.1",
+    "chownr": "^1.1.4",
+    "ci-info": "^2.0.0",
+    "cli-columns": "^3.1.2",
+    "cli-table3": "^0.5.1",
+    "cmd-shim": "^3.0.3",
+    "columnify": "~1.5.4",
+    "config-chain": "^1.1.12",
+    "detect-indent": "~5.0.0",
+    "detect-newline": "^2.1.0",
+    "dezalgo": "~1.0.3",
+    "editor": "~1.0.0",
+    "figgy-pudding": "^3.5.1",
+    "find-npm-prefix": "^1.0.2",
+    "fs-vacuum": "~1.2.10",
+    "fs-write-stream-atomic": "~1.0.10",
+    "gentle-fs": "^2.3.0",
+    "glob": "^7.1.6",
+    "graceful-fs": "^4.2.4",
+    "has-unicode": "~2.0.1",
+    "hosted-git-info": "^2.8.8",
+    "iferr": "^1.0.2",
+    "infer-owner": "^1.0.4",
+    "inflight": "~1.0.6",
+    "inherits": "^2.0.4",
+    "ini": "^1.3.5",
+    "init-package-json": "^1.10.3",
+    "is-cidr": "^3.0.0",
+    "json-parse-better-errors": "^1.0.2",
+    "lazy-property": "~1.0.0",
+    "libcipm": "^4.0.7",
+    "libnpm": "^3.0.1",
+    "libnpmaccess": "^3.0.2",
+    "libnpmhook": "^5.0.3",
+    "libnpmorg": "^1.0.1",
+    "libnpmsearch": "^2.0.2",
+    "libnpmteam": "^1.0.2",
+    "libnpx": "^10.2.2",
+    "lock-verify": "^2.1.0",
+    "lockfile": "^1.0.4",
+    "lodash._baseuniq": "~4.6.0",
+    "lodash.clonedeep": "~4.5.0",
+    "lodash.union": "~4.6.0",
+    "lodash.uniq": "~4.5.0",
+    "lodash.without": "~4.4.0",
+    "lru-cache": "^5.1.1",
+    "meant": "~1.0.1",
+    "mississippi": "^3.0.0",
+    "mkdirp": "^0.5.5",
+    "move-concurrently": "^1.0.1",
+    "node-gyp": "^5.1.0",
+    "nopt": "^4.0.3",
+    "normalize-package-data": "^2.5.0",
+    "npm-audit-report": "^1.3.2",
+    "npm-cache-filename": "~1.0.2",
+    "npm-install-checks": "^3.0.2",
+    "npm-lifecycle": "^3.1.4",
+    "npm-package-arg": "^6.1.1",
+    "npm-packlist": "^1.4.8",
+    "npm-pick-manifest": "^3.0.2",
+    "npm-profile": "^4.0.4",
+    "npm-registry-fetch": "^4.0.5",
+    "npm-user-validate": "~1.0.0",
+    "npmlog": "~4.1.2",
+    "once": "~1.4.0",
+    "opener": "^1.5.1",
+    "osenv": "^0.1.5",
+    "pacote": "^9.5.12",
+    "path-is-inside": "~1.0.2",
+    "promise-inflight": "~1.0.1",
+    "qrcode-terminal": "^0.12.0",
+    "query-string": "^6.8.2",
+    "qw": "~1.0.1",
+    "read": "~1.0.7",
+    "read-cmd-shim": "^1.0.5",
+    "read-installed": "~4.0.3",
+    "read-package-json": "^2.1.1",
+    "read-package-tree": "^5.3.1",
+    "readable-stream": "^3.6.0",
+    "readdir-scoped-modules": "^1.1.0",
+    "request": "^2.88.0",
+    "retry": "^0.12.0",
+    "rimraf": "^2.7.1",
+    "safe-buffer": "^5.1.2",
+    "semver": "^5.7.1",
+    "sha": "^3.0.0",
+    "slide": "~1.1.6",
+    "sorted-object": "~2.0.1",
+    "sorted-union-stream": "~2.1.3",
+    "ssri": "^6.0.1",
+    "stringify-package": "^1.0.1",
+    "tar": "^4.4.13",
+    "text-table": "~0.2.0",
+    "tiny-relative-date": "^1.3.0",
+    "uid-number": "0.0.6",
+    "umask": "~1.1.0",
+    "unique-filename": "^1.1.1",
+    "unpipe": "~1.0.0",
+    "update-notifier": "^2.5.0",
+    "uuid": "^3.3.3",
+    "validate-npm-package-license": "^3.0.4",
+    "validate-npm-package-name": "~3.0.0",
+    "which": "^1.3.1",
+    "worker-farm": "^1.7.0",
+    "write-file-atomic": "^2.4.3"
+  },
+  "bundleDependencies": [
+    "abbrev",
+    "ansicolors",
+    "ansistyles",
+    "aproba",
+    "archy",
+    "bin-links",
+    "bluebird",
+    "byte-size",
+    "cacache",
+    "call-limit",
+    "chownr",
+    "ci-info",
+    "cli-columns",
+    "cli-table3",
+    "cmd-shim",
+    "columnify",
+    "config-chain",
+    "debuglog",
+    "detect-indent",
+    "detect-newline",
+    "dezalgo",
+    "editor",
+    "figgy-pudding",
+    "find-npm-prefix",
+    "fs-vacuum",
+    "fs-write-stream-atomic",
+    "gentle-fs",
+    "glob",
+    "graceful-fs",
+    "has-unicode",
+    "hosted-git-info",
+    "iferr",
+    "imurmurhash",
+    "infer-owner",
+    "inflight",
+    "inherits",
+    "ini",
+    "init-package-json",
+    "is-cidr",
+    "json-parse-better-errors",
+    "JSONStream",
+    "lazy-property",
+    "libcipm",
+    "libnpm",
+    "libnpmaccess",
+    "libnpmhook",
+    "libnpmorg",
+    "libnpmsearch",
+    "libnpmteam",
+    "libnpx",
+    "lock-verify",
+    "lockfile",
+    "lodash._baseindexof",
+    "lodash._baseuniq",
+    "lodash._bindcallback",
+    "lodash._cacheindexof",
+    "lodash._createcache",
+    "lodash._getnative",
+    "lodash.clonedeep",
+    "lodash.restparam",
+    "lodash.union",
+    "lodash.uniq",
+    "lodash.without",
+    "lru-cache",
+    "meant",
+    "mississippi",
+    "mkdirp",
+    "move-concurrently",
+    "node-gyp",
+    "nopt",
+    "normalize-package-data",
+    "npm-audit-report",
+    "npm-cache-filename",
+    "npm-install-checks",
+    "npm-lifecycle",
+    "npm-package-arg",
+    "npm-packlist",
+    "npm-pick-manifest",
+    "npm-profile",
+    "npm-registry-fetch",
+    "npm-user-validate",
+    "npmlog",
+    "once",
+    "opener",
+    "osenv",
+    "pacote",
+    "path-is-inside",
+    "promise-inflight",
+    "qrcode-terminal",
+    "query-string",
+    "qw",
+    "read-cmd-shim",
+    "read-installed",
+    "read-package-json",
+    "read-package-tree",
+    "read",
+    "readable-stream",
+    "readdir-scoped-modules",
+    "request",
+    "retry",
+    "rimraf",
+    "safe-buffer",
+    "semver",
+    "sha",
+    "slide",
+    "sorted-object",
+    "sorted-union-stream",
+    "ssri",
+    "stringify-package",
+    "tar",
+    "text-table",
+    "tiny-relative-date",
+    "uid-number",
+    "umask",
+    "unique-filename",
+    "unpipe",
+    "update-notifier",
+    "uuid",
+    "validate-npm-package-license",
+    "validate-npm-package-name",
+    "which",
+    "worker-farm",
+    "write-file-atomic"
+  ],
+  "devDependencies": {
+    "deep-equal": "^1.0.1",
+    "get-stream": "^4.1.0",
+    "licensee": "^7.0.3",
+    "marked": "^0.6.3",
+    "marked-man": "^0.6.0",
+    "npm-registry-couchapp": "^2.7.4",
+    "npm-registry-mock": "^1.3.1",
+    "require-inject": "^1.4.4",
+    "sprintf-js": "^1.1.2",
+    "standard": "^11.0.1",
+    "tacks": "^1.3.0",
+    "tap": "^12.7.0",
+    "tar-stream": "^2.1.0"
+  },
+  "scripts": {
+    "dumpconf": "env | grep npm | sort | uniq",
+    "prepare": "node bin/npm-cli.js rebuild && node bin/npm-cli.js --no-audit --no-timing prune --prefix=. --no-global && rimraf test/*/*/node_modules && make -j4 mandocs",
+    "preversion": "bash scripts/update-authors.sh && git add AUTHORS && git commit -m \"update AUTHORS\" || true",
+    "licenses": "licensee --production --errors-only",
+    "tap": "tap -J --timeout 300 --no-esm",
+    "tap-cover": "tap -J --nyc-arg=--cache --coverage --timeout 600 --no-esm",
+    "lint": "standard",
+    "pretest": "npm run lint",
+    "test": "npm run test-tap --",
+    "test:nocleanup": "NO_TEST_CLEANUP=1 npm run test --",
+    "sudotest": "sudo npm run tap -- \"test/tap/*.js\"",
+    "sudotest:nocleanup": "sudo NO_TEST_CLEANUP=1 npm run tap -- \"test/tap/*.js\"",
+    "posttest": "rimraf test/npm_cache*",
+    "test-coverage": "npm run tap-cover -- \"test/tap/*.js\" \"test/network/*.js\"",
+    "test-tap": "npm run tap -- \"test/tap/*.js\" \"test/network/*.js\"",
+    "test-node": "tap --timeout 240 \"test/tap/*.js\" \"test/network/*.js\""
+  },
+  "license": "Artistic-2.0",
+  "engines": {
+    "node": "6 >=6.2.0 || 8 || >=9.3.0"
+  }
+}

syft/cataloger/python/cataloger.go

@@ -1,21 +0,0 @@
-/*
-Package python provides a concrete Cataloger implementation for Python ecosystem files (egg, wheel, requirements.txt).
-*/
-package python
-
-import (
-	"github.com/anchore/syft/syft/cataloger/common"
-)
-
-// NewPythonCataloger returns a new Python cataloger object.
-func NewPythonCataloger() *common.GenericCataloger {
-	globParsers := map[string]common.ParserFn{
-		"**/*egg-info/PKG-INFO":  parseEggMetadata,
-		"**/*dist-info/METADATA": parseWheelMetadata,
-		"**/*requirements*.txt":  parseRequirementsTxt,
-		"**/poetry.lock":         parsePoetryLock,
-		"**/setup.py":            parseSetup,
-	}
-
-	return common.NewGenericCataloger(nil, globParsers, "python-cataloger")
-}

syft/cataloger/python/index_cataloger.go

@@ -0,0 +1,19 @@
+/*
+Package python provides a concrete Cataloger implementation for Python ecosystem files (egg, wheel, requirements.txt).
+*/
+package python
+
+import (
+	"github.com/anchore/syft/syft/cataloger/common"
+)
+
+// NewPythonIndexCataloger returns a new cataloger for python packages referenced from poetry lock files, requirements.txt files, and setup.py files.
+func NewPythonIndexCataloger() *common.GenericCataloger {
+	globParsers := map[string]common.ParserFn{
+		"**/*requirements*.txt": parseRequirementsTxt,
+		"**/poetry.lock":        parsePoetryLock,
+		"**/setup.py":           parseSetup,
+	}
+
+	return common.NewGenericCataloger(nil, globParsers, "python-index-cataloger")
+}

syft/cataloger/python/package_cataloger.go

@@ -0,0 +1,179 @@
+package python
+
+import (
+	"bufio"
+	"fmt"
+	"path/filepath"
+	"strings"
+
+	"github.com/anchore/stereoscope/pkg/file"
+
+	"github.com/anchore/syft/syft/pkg"
+
+	"github.com/anchore/syft/syft/scope"
+)
+
+const (
+	eggMetadataGlob   = "**/*egg-info/PKG-INFO"
+	wheelMetadataGlob = "**/*dist-info/METADATA"
+)
+
+type PackageCataloger struct{}
+
+// NewPythonPackageCataloger returns a new cataloger for python packages within egg or wheel installation directories.
+func NewPythonPackageCataloger() *PackageCataloger {
+	return &PackageCataloger{}
+}
+
+// Name returns a string that uniquely describes a cataloger
+func (c *PackageCataloger) Name() string {
+	return "python-package-cataloger"
+}
+
+// Catalog is given an object to resolve file references and content, this function returns any discovered Packages after analyzing python egg and wheel installations.
+func (c *PackageCataloger) Catalog(resolver scope.Resolver) ([]pkg.Package, error) {
+	// nolint:prealloc
+	var fileMatches []file.Reference
+
+	for _, glob := range []string{eggMetadataGlob, wheelMetadataGlob} {
+		matches, err := resolver.FilesByGlob(glob)
+		if err != nil {
+			return nil, fmt.Errorf("failed to find files by glob: %s", glob)
+		}
+		fileMatches = append(fileMatches, matches...)
+	}
+
+	var pkgs []pkg.Package
+	for _, ref := range fileMatches {
+		p, err := c.catalogEggOrWheel(resolver, ref)
+		if err != nil {
+			return nil, fmt.Errorf("unable to catalog python package=%+v: %w", ref.Path, err)
+		}
+		if p != nil {
+			pkgs = append(pkgs, *p)
+		}
+	}
+	return pkgs, nil
+}
+
+// catalogEggOrWheel takes the primary metadata file reference and returns the python package it represents.
+func (c *PackageCataloger) catalogEggOrWheel(resolver scope.Resolver, metadataRef file.Reference) (*pkg.Package, error) {
+	metadata, sources, err := c.assembleEggOrWheelMetadata(resolver, metadataRef)
+	if err != nil {
+		return nil, err
+	}
+
+	var licenses []string
+	if metadata.License != "" {
+		licenses = []string{metadata.License}
+	}
+
+	return &pkg.Package{
+		Name:         metadata.Name,
+		Version:      metadata.Version,
+		FoundBy:      c.Name(),
+		Source:       sources,
+		Licenses:     licenses,
+		Language:     pkg.Python,
+		Type:         pkg.PythonPkg,
+		MetadataType: pkg.PythonPackageMetadataType,
+		Metadata:     *metadata,
+	}, nil
+}
+
+// fetchRecordFiles finds a corresponding RECORD file for the given python package metadata file and returns the set of file records contained.
+func (c *PackageCataloger) fetchRecordFiles(resolver scope.Resolver, metadataRef file.Reference) (files []pkg.PythonFileRecord, sources []file.Reference, err error) {
+	// we've been given a file reference to a specific wheel METADATA file. note: this may be for a directory
+	// or for an image... for an image the METADATA file may be present within multiple layers, so it is important
+	// to reconcile the RECORD path to the same layer (or the next adjacent lower layer).
+
+	// lets find the RECORD file relative to the directory where the METADATA file resides (in path AND layer structure)
+	recordPath := filepath.Join(filepath.Dir(string(metadataRef.Path)), "RECORD")
+	recordRef, err := resolver.RelativeFileByPath(metadataRef, recordPath)
+	if err != nil {
+		return nil, nil, err
+	}
+
+	if recordRef != nil {
+		sources = append(sources, *recordRef)
+
+		recordContents, err := resolver.FileContentsByRef(*recordRef)
+		if err != nil {
+			return nil, nil, err
+		}
+
+		// parse the record contents
+		records, err := parseWheelOrEggRecord(strings.NewReader(recordContents))
+		if err != nil {
+			return nil, nil, err
+		}
+
+		files = append(files, records...)
+	}
+	return files, sources, nil
+}
+
+// fetchTopLevelPackages finds a corresponding top_level.txt file for the given python package metadata file and returns the set of package names contained.
+func (c *PackageCataloger) fetchTopLevelPackages(resolver scope.Resolver, metadataRef file.Reference) (pkgs []string, sources []file.Reference, err error) {
+	// a top_level.txt file specifies the python top-level packages (provided by this python package) installed into site-packages
+	parentDir := filepath.Dir(string(metadataRef.Path))
+	topLevelPath := filepath.Join(parentDir, "top_level.txt")
+	topLevelRef, err := resolver.RelativeFileByPath(metadataRef, topLevelPath)
+	if err != nil {
+		return nil, nil, err
+	}
+	if topLevelRef == nil {
+		return nil, nil, fmt.Errorf("missing python package top_level.txt (package=%q)", string(metadataRef.Path))
+	}
+
+	sources = append(sources, *topLevelRef)
+
+	topLevelContents, err := resolver.FileContentsByRef(*topLevelRef)
+	if err != nil {
+		return nil, nil, err
+	}
+
+	scanner := bufio.NewScanner(strings.NewReader(topLevelContents))
+	for scanner.Scan() {
+		pkgs = append(pkgs, scanner.Text())
+	}
+
+	if err := scanner.Err(); err != nil {
+		return nil, nil, fmt.Errorf("could not read python package top_level.txt: %w", err)
+	}
+
+	return pkgs, sources, nil
+}
+
+// assembleEggOrWheelMetadata discovers and accumulates python package metadata from multiple file sources and returns a single metadata object as well as a list of files where the metadata was derived from.
+func (c *PackageCataloger) assembleEggOrWheelMetadata(resolver scope.Resolver, metadataRef file.Reference) (*pkg.PythonPackageMetadata, []file.Reference, error) {
+	var sources = []file.Reference{metadataRef}
+
+	metadataContents, err := resolver.FileContentsByRef(metadataRef)
+	if err != nil {
+		return nil, nil, err
+	}
+
+	metadata, err := parseWheelOrEggMetadata(metadataRef.Path, strings.NewReader(metadataContents))
+	if err != nil {
+		return nil, nil, err
+	}
+
+	// attach any python files found for the given wheel/egg installation
+	r, s, err := c.fetchRecordFiles(resolver, metadataRef)
+	if err != nil {
+		return nil, nil, err
+	}
+	sources = append(sources, s...)
+	metadata.Files = r
+
+	// attach any top-level package names found for the given wheel/egg installation
+	p, s, err := c.fetchTopLevelPackages(resolver, metadataRef)
+	if err != nil {
+		return nil, nil, err
+	}
+	sources = append(sources, s...)
+	metadata.TopLevelPackages = p
+
+	return &metadata, sources, nil
+}

syft/cataloger/python/package_cataloger_test.go

@@ -0,0 +1,241 @@
+package python
+
+import (
+	"fmt"
+	"io"
+	"io/ioutil"
+	"os"
+	"strings"
+	"testing"
+
+	"github.com/anchore/stereoscope/pkg/file"
+
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/go-test/deep"
+)
+
+type pythonTestResolverMock struct {
+	metadataReader io.Reader
+	recordReader   io.Reader
+	topLevelReader io.Reader
+	metadataRef    *file.Reference
+	recordRef      *file.Reference
+	topLevelRef    *file.Reference
+	contents       map[file.Reference]string
+}
+
+func newTestResolver(metaPath, recordPath, topPath string) *pythonTestResolverMock {
+	metadataReader, err := os.Open(metaPath)
+	if err != nil {
+		panic(fmt.Errorf("failed to open metadata: %+v", err))
+	}
+
+	var recordReader io.Reader
+	if recordPath != "" {
+		recordReader, err = os.Open(recordPath)
+		if err != nil {
+			panic(fmt.Errorf("failed to open record: %+v", err))
+		}
+	}
+
+	var topLevelReader io.Reader
+	if topPath != "" {
+		topLevelReader, err = os.Open(topPath)
+		if err != nil {
+			panic(fmt.Errorf("failed to open top level: %+v", err))
+		}
+	}
+
+	var recordRef *file.Reference
+	if recordReader != nil {
+		ref := file.NewFileReference("test-fixtures/dist-info/RECORD")
+		recordRef = &ref
+	}
+	var topLevelRef *file.Reference
+	if topLevelReader != nil {
+		ref := file.NewFileReference("test-fixtures/dist-info/top_level.txt")
+		topLevelRef = &ref
+	}
+	metadataRef := file.NewFileReference("test-fixtures/dist-info/METADATA")
+	return &pythonTestResolverMock{
+		recordReader:   recordReader,
+		metadataReader: metadataReader,
+		topLevelReader: topLevelReader,
+		metadataRef:    &metadataRef,
+		recordRef:      recordRef,
+		topLevelRef:    topLevelRef,
+		contents:       make(map[file.Reference]string),
+	}
+}
+
+func (r *pythonTestResolverMock) FileContentsByRef(ref file.Reference) (string, error) {
+	switch ref.Path {
+	case r.topLevelRef.Path:
+		b, err := ioutil.ReadAll(r.topLevelReader)
+		if err != nil {
+			return "", err
+		}
+		return string(b), nil
+	case r.metadataRef.Path:
+		b, err := ioutil.ReadAll(r.metadataReader)
+		if err != nil {
+			return "", err
+		}
+		return string(b), nil
+	case r.recordRef.Path:
+		b, err := ioutil.ReadAll(r.recordReader)
+		if err != nil {
+			return "", err
+		}
+		return string(b), nil
+	}
+	return "", fmt.Errorf("invalid value given")
+}
+
+func (r *pythonTestResolverMock) MultipleFileContentsByRef(_ ...file.Reference) (map[file.Reference]string, error) {
+	return nil, fmt.Errorf("not implemented")
+}
+
+func (r *pythonTestResolverMock) FilesByPath(_ ...file.Path) ([]file.Reference, error) {
+	return nil, fmt.Errorf("not implemented")
+}
+
+func (r *pythonTestResolverMock) FilesByGlob(_ ...string) ([]file.Reference, error) {
+	return nil, fmt.Errorf("not implemented")
+}
+func (r *pythonTestResolverMock) RelativeFileByPath(_ file.Reference, path string) (*file.Reference, error) {
+	switch {
+	case strings.Contains(path, "RECORD"):
+		return r.recordRef, nil
+	case strings.Contains(path, "top_level.txt"):
+		return r.topLevelRef, nil
+	default:
+		return nil, fmt.Errorf("invalid RelativeFileByPath value given: %q", path)
+	}
+}
+
+func TestPythonPackageWheelCataloger(t *testing.T) {
+	tests := []struct {
+		MetadataFixture string
+		RecordFixture   string
+		TopLevelFixture string
+		ExpectedPackage pkg.Package
+	}{
+		{
+			MetadataFixture: "test-fixtures/egg-info/PKG-INFO",
+			RecordFixture:   "test-fixtures/egg-info/RECORD",
+			TopLevelFixture: "test-fixtures/egg-info/top_level.txt",
+			ExpectedPackage: pkg.Package{
+				Name:         "requests",
+				Version:      "2.22.0",
+				Type:         pkg.PythonPkg,
+				Language:     pkg.Python,
+				Licenses:     []string{"Apache 2.0"},
+				FoundBy:      "python-package-cataloger",
+				MetadataType: pkg.PythonPackageMetadataType,
+				Metadata: pkg.PythonPackageMetadata{
+					Name:                 "requests",
+					Version:              "2.22.0",
+					License:              "Apache 2.0",
+					Platform:             "UNKNOWN",
+					Author:               "Kenneth Reitz",
+					AuthorEmail:          "me@kennethreitz.org",
+					SitePackagesRootPath: "test-fixtures",
+					Files: []pkg.PythonFileRecord{
+						{Path: "requests-2.22.0.dist-info/INSTALLER", Digest: &pkg.Digest{"sha256", "zuuue4knoyJ-UwPPXg8fezS7VCrXJQrAP7zeNuwvFQg"}, Size: "4"},
+						{Path: "requests/__init__.py", Digest: &pkg.Digest{"sha256", "PnKCgjcTq44LaAMzB-7--B2FdewRrE8F_vjZeaG9NhA"}, Size: "3921"},
+						{Path: "requests/__pycache__/__version__.cpython-38.pyc"},
+						{Path: "requests/__pycache__/utils.cpython-38.pyc"},
+						{Path: "requests/__version__.py", Digest: &pkg.Digest{"sha256", "Bm-GFstQaFezsFlnmEMrJDe8JNROz9n2XXYtODdvjjc"}, Size: "436"},
+						{Path: "requests/utils.py", Digest: &pkg.Digest{"sha256", "LtPJ1db6mJff2TJSJWKi7rBpzjPS3mSOrjC9zRhoD3A"}, Size: "30049"},
+					},
+					TopLevelPackages: []string{"requests"},
+				},
+			},
+		},
+		{
+			MetadataFixture: "test-fixtures/dist-info/METADATA",
+			RecordFixture:   "test-fixtures/dist-info/RECORD",
+			TopLevelFixture: "test-fixtures/dist-info/top_level.txt",
+			ExpectedPackage: pkg.Package{
+				Name:         "Pygments",
+				Version:      "2.6.1",
+				Type:         pkg.PythonPkg,
+				Language:     pkg.Python,
+				Licenses:     []string{"BSD License"},
+				FoundBy:      "python-package-cataloger",
+				MetadataType: pkg.PythonPackageMetadataType,
+				Metadata: pkg.PythonPackageMetadata{
+					Name:                 "Pygments",
+					Version:              "2.6.1",
+					License:              "BSD License",
+					Platform:             "any",
+					Author:               "Georg Brandl",
+					AuthorEmail:          "georg@python.org",
+					SitePackagesRootPath: "test-fixtures",
+					Files: []pkg.PythonFileRecord{
+						{Path: "../../../bin/pygmentize", Digest: &pkg.Digest{"sha256", "dDhv_U2jiCpmFQwIRHpFRLAHUO4R1jIJPEvT_QYTFp8"}, Size: "220"},
+						{Path: "Pygments-2.6.1.dist-info/AUTHORS", Digest: &pkg.Digest{"sha256", "PVpa2_Oku6BGuiUvutvuPnWGpzxqFy2I8-NIrqCvqUY"}, Size: "8449"},
+						{Path: "Pygments-2.6.1.dist-info/RECORD"},
+						{Path: "pygments/__pycache__/__init__.cpython-38.pyc"},
+						{Path: "pygments/util.py", Digest: &pkg.Digest{"sha256", "586xXHiJGGZxqk5PMBu3vBhE68DLuAe5MBARWrSPGxA"}, Size: "10778"},
+					},
+					TopLevelPackages: []string{"pygments", "something_else"},
+				},
+			},
+		},
+		{
+			// in cases where the metadata file is available and the record is not we should still record there is a package
+			// additionally empty top_level.txt files should not result in an error
+			MetadataFixture: "test-fixtures/partial.dist-info/METADATA",
+			TopLevelFixture: "test-fixtures/partial.dist-info/top_level.txt",
+			ExpectedPackage: pkg.Package{
+				Name:         "Pygments",
+				Version:      "2.6.1",
+				Type:         pkg.PythonPkg,
+				Language:     pkg.Python,
+				Licenses:     []string{"BSD License"},
+				FoundBy:      "python-package-cataloger",
+				MetadataType: pkg.PythonPackageMetadataType,
+				Metadata: pkg.PythonPackageMetadata{
+					Name:                 "Pygments",
+					Version:              "2.6.1",
+					License:              "BSD License",
+					Platform:             "any",
+					Author:               "Georg Brandl",
+					AuthorEmail:          "georg@python.org",
+					SitePackagesRootPath: "test-fixtures",
+				},
+			},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.MetadataFixture, func(t *testing.T) {
+			resolver := newTestResolver(test.MetadataFixture, test.RecordFixture, test.TopLevelFixture)
+
+			// note that the source is the record ref created by the resolver mock... attach the expected values
+			test.ExpectedPackage.Source = []file.Reference{*resolver.metadataRef}
+			if resolver.recordRef != nil {
+				test.ExpectedPackage.Source = append(test.ExpectedPackage.Source, *resolver.recordRef)
+			}
+
+			if resolver.topLevelRef != nil {
+				test.ExpectedPackage.Source = append(test.ExpectedPackage.Source, *resolver.topLevelRef)
+			}
+			// end patching expected values with runtime data...
+
+			pyPkgCataloger := NewPythonPackageCataloger()
+
+			actual, err := pyPkgCataloger.catalogEggOrWheel(resolver, *resolver.metadataRef)
+			if err != nil {
+				t.Fatalf("failed to catalog python package: %+v", err)
+			}
+
+			for _, d := range deep.Equal(actual, &test.ExpectedPackage) {
+				t.Errorf("diff: %+v", d)
+			}
+		})
+	}
+
+}

syft/cataloger/python/parse_poetry_lock_test.go

@@ -1,10 +1,11 @@
 package python
 
 import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/go-test/deep"
 	"os"
 	"testing"
+
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/go-test/deep"
 )
 
 func TestParsePoetryLock(t *testing.T) {
@@ -13,28 +14,28 @@ func TestParsePoetryLock(t *testing.T) {
 			Name:     "added-value",
 			Version:  "0.14.2",
 			Language: pkg.Python,
-			Type:     pkg.PoetryPkg,
+			Type:     pkg.PythonPkg,
 			Licenses: nil,
 		},
 		{
 			Name:     "alabaster",
 			Version:  "0.7.12",
 			Language: pkg.Python,
-			Type:     pkg.PoetryPkg,
+			Type:     pkg.PythonPkg,
 			Licenses: nil,
 		},
 		{
 			Name:     "appnope",
 			Version:  "0.1.0",
 			Language: pkg.Python,
-			Type:     pkg.PoetryPkg,
+			Type:     pkg.PythonPkg,
 			Licenses: nil,
 		},
 		{
 			Name:     "asciitree",
 			Version:  "0.3.3",
 			Language: pkg.Python,
-			Type:     pkg.PoetryPkg,
+			Type:     pkg.PythonPkg,
 			Licenses: nil,
 		},
 	}

syft/cataloger/python/parse_requirements.go

@@ -47,7 +47,7 @@ func parseRequirementsTxt(_ string, reader io.Reader) ([]pkg.Package, error) {
 				Name:     name,
 				Version:  version,
 				Language: pkg.Python,
-				Type:     pkg.PythonRequirementsPkg,
+				Type:     pkg.PythonPkg,
 			})
 		default:
 			continue

syft/cataloger/python/parse_requirements_test.go

@@ -4,24 +4,45 @@ import (
 	"os"
 	"testing"
 
+	"github.com/go-test/deep"
+
 	"github.com/anchore/syft/syft/pkg"
 )
 
+func assertPackagesEqual(t *testing.T, actual []pkg.Package, expected map[string]pkg.Package) {
+	t.Helper()
+	if len(actual) != len(expected) {
+		for _, a := range actual {
+			t.Log("   ", a)
+		}
+		t.Fatalf("unexpected package count: %d!=%d", len(actual), len(expected))
+	}
+
+	for _, a := range actual {
+		expectedPkg, ok := expected[a.Name]
+		if !ok {
+			t.Errorf("unexpected package found: '%s'", a.Name)
+		}
+
+		for _, d := range deep.Equal(a, expectedPkg) {
+			t.Errorf("diff: %+v", d)
+		}
+	}
+}
+
 func TestParseRequirementsTxt(t *testing.T) {
 	expected := map[string]pkg.Package{
 		"foo": {
 			Name:     "foo",
 			Version:  "1.0.0",
 			Language: pkg.Python,
-			Type:     pkg.PythonRequirementsPkg,
-			Licenses: []string{},
+			Type:     pkg.PythonPkg,
 		},
 		"flask": {
 			Name:     "flask",
 			Version:  "4.0.0",
 			Language: pkg.Python,
-			Type:     pkg.PythonRequirementsPkg,
-			Licenses: []string{},
+			Type:     pkg.PythonPkg,
 		},
 	}
 	fixture, err := os.Open("test-fixtures/requires/requirements.txt")
@@ -34,6 +55,6 @@ func TestParseRequirementsTxt(t *testing.T) {
 		t.Fatalf("failed to parse requirements: %+v", err)
 	}
 
-	assertPkgsEqual(t, actual, expected)
+	assertPackagesEqual(t, actual, expected)
 
 }

syft/cataloger/python/parse_setup.go

@@ -41,7 +41,7 @@ func parseSetup(_ string, reader io.Reader) ([]pkg.Package, error) {
 				Name:     strings.Trim(name, "'\""),
 				Version:  strings.Trim(version, "'\""),
 				Language: pkg.Python,
-				Type:     pkg.PythonSetupPkg,
+				Type:     pkg.PythonPkg,
 			})
 		}
 	}

syft/cataloger/python/parse_setup_test.go

@@ -13,36 +13,31 @@ func TestParseSetup(t *testing.T) {
 			Name:     "pathlib3",
 			Version:  "2.2.0",
 			Language: pkg.Python,
-			Type:     pkg.PythonSetupPkg,
-			Licenses: []string{},
+			Type:     pkg.PythonPkg,
 		},
 		"mypy": {
 			Name:     "mypy",
 			Version:  "v0.770",
 			Language: pkg.Python,
-			Type:     pkg.PythonSetupPkg,
-			Licenses: []string{},
+			Type:     pkg.PythonPkg,
 		},
 		"mypy1": {
 			Name:     "mypy1",
 			Version:  "v0.770",
 			Language: pkg.Python,
-			Type:     pkg.PythonSetupPkg,
-			Licenses: []string{},
+			Type:     pkg.PythonPkg,
 		},
 		"mypy2": {
 			Name:     "mypy2",
 			Version:  "v0.770",
 			Language: pkg.Python,
-			Type:     pkg.PythonSetupPkg,
-			Licenses: []string{},
+			Type:     pkg.PythonPkg,
 		},
 		"mypy3": {
 			Name:     "mypy3",
 			Version:  "v0.770",
 			Language: pkg.Python,
-			Type:     pkg.PythonSetupPkg,
-			Licenses: []string{},
+			Type:     pkg.PythonPkg,
 		},
 	}
 	fixture, err := os.Open("test-fixtures/setup/setup.py")
@@ -55,6 +50,6 @@ func TestParseSetup(t *testing.T) {
 		t.Fatalf("failed to parse requirements: %+v", err)
 	}
 
-	assertPkgsEqual(t, actual, expected)
+	assertPackagesEqual(t, actual, expected)
 
 }

syft/cataloger/python/parse_wheel_egg.go

@@ -1,100 +0,0 @@
-package python
-
-import (
-	"bufio"
-	"fmt"
-	"io"
-	"strings"
-
-	"github.com/anchore/syft/syft/cataloger/common"
-	"github.com/anchore/syft/syft/pkg"
-)
-
-// integrity check
-var _ common.ParserFn = parseWheelMetadata
-var _ common.ParserFn = parseEggMetadata
-
-// parseWheelMetadata is a parser function for individual Python Wheel metadata file contents, returning all Python
-// packages listed.
-func parseWheelMetadata(_ string, reader io.Reader) ([]pkg.Package, error) {
-	packages, err := parseWheelOrEggMetadata(reader)
-	for idx := range packages {
-		packages[idx].Type = pkg.WheelPkg
-	}
-	return packages, err
-}
-
-// parseEggMetadata is a parser function for individual Python Egg metadata file contents, returning all Python
-// packages listed.
-func parseEggMetadata(_ string, reader io.Reader) ([]pkg.Package, error) {
-	packages, err := parseWheelOrEggMetadata(reader)
-	for idx := range packages {
-		packages[idx].Type = pkg.EggPkg
-	}
-	return packages, err
-}
-
-// parseWheelOrEggMetadata takes a Python Egg or Wheel (which share the same format and values for our purposes),
-// returning all Python packages listed.
-func parseWheelOrEggMetadata(reader io.Reader) ([]pkg.Package, error) {
-	fields := make(map[string]string)
-	var key string
-
-	scanner := bufio.NewScanner(reader)
-	for scanner.Scan() {
-		line := scanner.Text()
-
-		line = strings.TrimRight(line, "\n")
-
-		// empty line indicates end of entry
-		if len(line) == 0 {
-			// if the entry has not started, keep parsing lines
-			if len(fields) == 0 {
-				continue
-			}
-			break
-		}
-
-		switch {
-		case strings.HasPrefix(line, " "):
-			// a field-body continuation
-			if len(key) == 0 {
-				return nil, fmt.Errorf("no match for continuation: line: '%s'", line)
-			}
-
-			val, ok := fields[key]
-			if !ok {
-				return nil, fmt.Errorf("no previous key exists, expecting: %s", key)
-			}
-			// concatenate onto previous value
-			val = fmt.Sprintf("%s\n %s", val, strings.TrimSpace(line))
-			fields[key] = val
-		default:
-			// parse a new key (note, duplicate keys are overridden)
-			if i := strings.Index(line, ":"); i > 0 {
-				key = strings.TrimSpace(line[0:i])
-				val := strings.TrimSpace(line[i+1:])
-
-				fields[key] = val
-			} else {
-				return nil, fmt.Errorf("cannot parse field from line: '%s'", line)
-			}
-		}
-	}
-
-	if err := scanner.Err(); err != nil {
-		return nil, fmt.Errorf("failed to parse python wheel/egg: %w", err)
-	}
-
-	p := pkg.Package{
-		Name:     fields["Name"],
-		Version:  fields["Version"],
-		Language: pkg.Python,
-	}
-
-	if license, ok := fields["License"]; ok && license != "" {
-		p.Licenses = []string{license}
-	}
-
-	return []pkg.Package{p}, nil
-}

syft/cataloger/python/parse_wheel_egg_metadata.go

@@ -0,0 +1,80 @@
+package python
+
+import (
+	"bufio"
+	"fmt"
+	"io"
+	"path/filepath"
+	"strings"
+
+	"github.com/anchore/stereoscope/pkg/file"
+
+	"github.com/mitchellh/mapstructure"
+
+	"github.com/anchore/syft/syft/pkg"
+)
+
+// parseWheelOrEggMetadata takes a Python Egg or Wheel (which share the same format and values for our purposes),
+// returning all Python packages listed.
+func parseWheelOrEggMetadata(path file.Path, reader io.Reader) (pkg.PythonPackageMetadata, error) {
+	fields := make(map[string]string)
+	var key string
+
+	scanner := bufio.NewScanner(reader)
+	for scanner.Scan() {
+		line := scanner.Text()
+		line = strings.TrimRight(line, "\n")
+
+		// empty line indicates end of entry
+		if len(line) == 0 {
+			// if the entry has not started, keep parsing lines
+			if len(fields) == 0 {
+				continue
+			}
+			break
+		}
+
+		switch {
+		case strings.HasPrefix(line, " "):
+			// a field-body continuation
+			if len(key) == 0 {
+				return pkg.PythonPackageMetadata{}, fmt.Errorf("no match for continuation: line: '%s'", line)
+			}
+
+			val, ok := fields[key]
+			if !ok {
+				return pkg.PythonPackageMetadata{}, fmt.Errorf("no previous key exists, expecting: %s", key)
+			}
+			// concatenate onto previous value
+			val = fmt.Sprintf("%s\n %s", val, strings.TrimSpace(line))
+			fields[key] = val
+		default:
+			// parse a new key (note, duplicate keys are overridden)
+			if i := strings.Index(line, ":"); i > 0 {
+				// mapstruct cannot map keys with dashes, and we are expected to persist the "Author-email" field
+				key = strings.ReplaceAll(strings.TrimSpace(line[0:i]), "-", "")
+				val := strings.TrimSpace(line[i+1:])
+
+				fields[key] = val
+			} else {
+				return pkg.PythonPackageMetadata{}, fmt.Errorf("cannot parse field from line: '%s'", line)
+			}
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		return pkg.PythonPackageMetadata{}, fmt.Errorf("failed to parse python wheel/egg: %w", err)
+	}
+
+	var metadata pkg.PythonPackageMetadata
+	if err := mapstructure.Decode(fields, &metadata); err != nil {
+		return pkg.PythonPackageMetadata{}, fmt.Errorf("unable to parse APK metadata: %w", err)
+	}
+
+	// add additional metadata not stored in the egg/wheel metadata file
+
+	sitePackagesRoot := filepath.Clean(filepath.Join(filepath.Dir(string(path)), ".."))
+	metadata.SitePackagesRootPath = sitePackagesRoot
+
+	return metadata, nil
+}

syft/cataloger/python/parse_wheel_egg_metadata_test.go

@@ -0,0 +1,62 @@
+package python
+
+import (
+	"os"
+	"testing"
+
+	"github.com/anchore/stereoscope/pkg/file"
+
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/go-test/deep"
+)
+
+func TestParseWheelEggMetadata(t *testing.T) {
+	tests := []struct {
+		Fixture          string
+		ExpectedMetadata pkg.PythonPackageMetadata
+	}{
+		{
+			Fixture: "test-fixtures/egg-info/PKG-INFO",
+			ExpectedMetadata: pkg.PythonPackageMetadata{
+				Name:                 "requests",
+				Version:              "2.22.0",
+				License:              "Apache 2.0",
+				Platform:             "UNKNOWN",
+				Author:               "Kenneth Reitz",
+				AuthorEmail:          "me@kennethreitz.org",
+				SitePackagesRootPath: "test-fixtures",
+			},
+		},
+		{
+			Fixture: "test-fixtures/dist-info/METADATA",
+			ExpectedMetadata: pkg.PythonPackageMetadata{
+				Name:                 "Pygments",
+				Version:              "2.6.1",
+				License:              "BSD License",
+				Platform:             "any",
+				Author:               "Georg Brandl",
+				AuthorEmail:          "georg@python.org",
+				SitePackagesRootPath: "test-fixtures",
+			},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.Fixture, func(t *testing.T) {
+			fixture, err := os.Open(test.Fixture)
+			if err != nil {
+				t.Fatalf("failed to open fixture: %+v", err)
+			}
+
+			actual, err := parseWheelOrEggMetadata(file.Path(test.Fixture), fixture)
+			if err != nil {
+				t.Fatalf("failed to parse: %+v", err)
+			}
+
+			for _, d := range deep.Equal(actual, test.ExpectedMetadata) {
+				t.Errorf("diff: %+v", d)
+			}
+		})
+	}
+
+}

syft/cataloger/python/parse_wheel_egg_record.go

@@ -0,0 +1,60 @@
+package python
+
+import (
+	"encoding/csv"
+	"fmt"
+	"io"
+	"strings"
+
+	"github.com/anchore/syft/syft/pkg"
+)
+
+// parseWheelOrEggMetadata takes a Python Egg or Wheel (which share the same format and values for our purposes),
+// returning all Python packages listed.
+func parseWheelOrEggRecord(reader io.Reader) ([]pkg.PythonFileRecord, error) {
+	var records []pkg.PythonFileRecord
+	r := csv.NewReader(reader)
+
+	for {
+		recordList, err := r.Read()
+		if err == io.EOF {
+			break
+		}
+		if err != nil {
+			return nil, fmt.Errorf("unable to read python record file: %w", err)
+		}
+
+		if len(recordList) != 3 {
+			return nil, fmt.Errorf("python record an unexpected length=%d: %q", len(recordList), recordList)
+		}
+
+		var record pkg.PythonFileRecord
+
+		for idx, item := range recordList {
+			switch idx {
+			case 0:
+				record.Path = item
+			case 1:
+				if item == "" {
+					continue
+				}
+				fields := strings.Split(item, "=")
+
+				if len(fields) != 2 {
+					return nil, fmt.Errorf("unexpected python record digest: %q", item)
+				}
+
+				record.Digest = &pkg.Digest{
+					Algorithm: fields[0],
+					Value:     fields[1],
+				}
+			case 2:
+				record.Size = item
+			}
+		}
+
+		records = append(records, record)
+	}
+
+	return records, nil
+}

syft/cataloger/python/parse_wheel_egg_record_test.go

@@ -0,0 +1,57 @@
+package python
+
+import (
+	"os"
+	"testing"
+
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/go-test/deep"
+)
+
+func TestParseWheelEggRecord(t *testing.T) {
+	tests := []struct {
+		Fixture          string
+		ExpectedMetadata []pkg.PythonFileRecord
+	}{
+		{
+			Fixture: "test-fixtures/egg-info/RECORD",
+			ExpectedMetadata: []pkg.PythonFileRecord{
+				{Path: "requests-2.22.0.dist-info/INSTALLER", Digest: &pkg.Digest{"sha256", "zuuue4knoyJ-UwPPXg8fezS7VCrXJQrAP7zeNuwvFQg"}, Size: "4"},
+				{Path: "requests/__init__.py", Digest: &pkg.Digest{"sha256", "PnKCgjcTq44LaAMzB-7--B2FdewRrE8F_vjZeaG9NhA"}, Size: "3921"},
+				{Path: "requests/__pycache__/__version__.cpython-38.pyc"},
+				{Path: "requests/__pycache__/utils.cpython-38.pyc"},
+				{Path: "requests/__version__.py", Digest: &pkg.Digest{"sha256", "Bm-GFstQaFezsFlnmEMrJDe8JNROz9n2XXYtODdvjjc"}, Size: "436"},
+				{Path: "requests/utils.py", Digest: &pkg.Digest{"sha256", "LtPJ1db6mJff2TJSJWKi7rBpzjPS3mSOrjC9zRhoD3A"}, Size: "30049"},
+			},
+		},
+		{
+			Fixture: "test-fixtures/dist-info/RECORD",
+			ExpectedMetadata: []pkg.PythonFileRecord{
+				{Path: "../../../bin/pygmentize", Digest: &pkg.Digest{"sha256", "dDhv_U2jiCpmFQwIRHpFRLAHUO4R1jIJPEvT_QYTFp8"}, Size: "220"},
+				{Path: "Pygments-2.6.1.dist-info/AUTHORS", Digest: &pkg.Digest{"sha256", "PVpa2_Oku6BGuiUvutvuPnWGpzxqFy2I8-NIrqCvqUY"}, Size: "8449"},
+				{Path: "Pygments-2.6.1.dist-info/RECORD"},
+				{Path: "pygments/__pycache__/__init__.cpython-38.pyc"},
+				{Path: "pygments/util.py", Digest: &pkg.Digest{"sha256", "586xXHiJGGZxqk5PMBu3vBhE68DLuAe5MBARWrSPGxA"}, Size: "10778"},
+			},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.Fixture, func(t *testing.T) {
+			fixture, err := os.Open(test.Fixture)
+			if err != nil {
+				t.Fatalf("failed to open fixture: %+v", err)
+			}
+
+			actual, err := parseWheelOrEggRecord(fixture)
+			if err != nil {
+				t.Fatalf("failed to parse: %+v", err)
+			}
+
+			for _, d := range deep.Equal(actual, test.ExpectedMetadata) {
+				t.Errorf("diff: %+v", d)
+			}
+		})
+	}
+
+}

syft/cataloger/python/parse_wheel_egg_test.go

@@ -1,93 +0,0 @@
-package python
-
-import (
-	"os"
-	"testing"
-
-	"github.com/anchore/syft/syft/pkg"
-)
-
-func assertPkgsEqual(t *testing.T, actual []pkg.Package, expected map[string]pkg.Package) {
-	t.Helper()
-	if len(actual) != len(expected) {
-		for _, a := range actual {
-			t.Log("   ", a)
-		}
-		t.Fatalf("unexpected package count: %d!=%d", len(actual), len(expected))
-	}
-
-	for _, a := range actual {
-		expectedPkg, ok := expected[a.Name]
-		if !ok {
-			t.Errorf("unexpected package found: '%s'", a.Name)
-		}
-
-		if expectedPkg.Version != a.Version {
-			t.Errorf("unexpected package version: '%s'", a.Version)
-		}
-
-		if a.Language != expectedPkg.Language {
-			t.Errorf("bad language: '%+v'", a.Language)
-		}
-
-		if a.Type != expectedPkg.Type {
-			t.Errorf("bad package type: %+v", a.Type)
-		}
-
-		if len(a.Licenses) < len(expectedPkg.Licenses) {
-			t.Errorf("bad package licenses count: '%+v'", a.Licenses)
-		}
-		if len(a.Licenses) > 0 {
-			if a.Licenses[0] != expectedPkg.Licenses[0] {
-				t.Errorf("bad package licenses: '%+v'", a.Licenses)
-			}
-		}
-
-	}
-}
-
-func TestParseEggMetadata(t *testing.T) {
-	expected := map[string]pkg.Package{
-		"requests": {
-			Name:     "requests",
-			Version:  "2.22.0",
-			Language: pkg.Python,
-			Type:     pkg.EggPkg,
-			Licenses: []string{"Apache 2.0"},
-		},
-	}
-	fixture, err := os.Open("test-fixtures/egg-info/PKG-INFO")
-	if err != nil {
-		t.Fatalf("failed to open fixture: %+v", err)
-	}
-
-	actual, err := parseEggMetadata(fixture.Name(), fixture)
-	if err != nil {
-		t.Fatalf("failed to parse egg-info: %+v", err)
-	}
-
-	assertPkgsEqual(t, actual, expected)
-}
-
-func TestParseWheelMetadata(t *testing.T) {
-	expected := map[string]pkg.Package{
-		"Pygments": {
-			Name:     "Pygments",
-			Version:  "2.6.1",
-			Language: pkg.Python,
-			Type:     pkg.WheelPkg,
-			Licenses: []string{"BSD License"},
-		},
-	}
-	fixture, err := os.Open("test-fixtures/dist-info/METADATA")
-	if err != nil {
-		t.Fatalf("failed to open fixture: %+v", err)
-	}
-
-	actual, err := parseWheelMetadata(fixture.Name(), fixture)
-	if err != nil {
-		t.Fatalf("failed to parse dist-info: %+v", err)
-	}
-
-	assertPkgsEqual(t, actual, expected)
-}

syft/cataloger/python/poetry_metadata_package.go

@@ -16,6 +16,6 @@ func (p PoetryMetadataPackage) Pkg() pkg.Package {
 		Name:     p.Name,
 		Version:  p.Version,
 		Language: pkg.Python,
-		Type:     pkg.PoetryPkg,
+		Type:     pkg.PythonPkg,
 	}
 }

syft/cataloger/python/test-fixtures/dist-info/RECORD

@@ -0,0 +1,5 @@
+../../../bin/pygmentize,sha256=dDhv_U2jiCpmFQwIRHpFRLAHUO4R1jIJPEvT_QYTFp8,220
+Pygments-2.6.1.dist-info/AUTHORS,sha256=PVpa2_Oku6BGuiUvutvuPnWGpzxqFy2I8-NIrqCvqUY,8449
+Pygments-2.6.1.dist-info/RECORD,,
+pygments/__pycache__/__init__.cpython-38.pyc,,
+pygments/util.py,sha256=586xXHiJGGZxqk5PMBu3vBhE68DLuAe5MBARWrSPGxA,10778

syft/cataloger/python/test-fixtures/dist-info/top_level.txt

@@ -0,0 +1,2 @@
+pygments
+something_else

syft/cataloger/python/test-fixtures/egg-info/RECORD

@@ -0,0 +1,6 @@
+requests-2.22.0.dist-info/INSTALLER,sha256=zuuue4knoyJ-UwPPXg8fezS7VCrXJQrAP7zeNuwvFQg,4
+requests/__init__.py,sha256=PnKCgjcTq44LaAMzB-7--B2FdewRrE8F_vjZeaG9NhA,3921
+requests/__pycache__/__version__.cpython-38.pyc,,
+requests/__pycache__/utils.cpython-38.pyc,,
+requests/__version__.py,sha256=Bm-GFstQaFezsFlnmEMrJDe8JNROz9n2XXYtODdvjjc,436
+requests/utils.py,sha256=LtPJ1db6mJff2TJSJWKi7rBpzjPS3mSOrjC9zRhoD3A,30049

syft/cataloger/python/test-fixtures/egg-info/top_level.txt

@@ -0,0 +1 @@
+requests

syft/cataloger/python/test-fixtures/partial.dist-info/METADATA

@@ -0,0 +1,47 @@
+Metadata-Version: 2.1
+Name: Pygments
+Version: 2.6.1
+Summary: Pygments is a syntax highlighting package written in Python.
+Home-page: https://pygments.org/
+Author: Georg Brandl
+Author-email: georg@python.org
+License: BSD License
+Keywords: syntax highlighting
+Platform: any
+Classifier: License :: OSI Approved :: BSD License
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: End Users/Desktop
+Classifier: Intended Audience :: System Administrators
+Classifier: Development Status :: 6 - Mature
+Classifier: Programming Language :: Python
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.5
+Classifier: Programming Language :: Python :: 3.6
+Classifier: Programming Language :: Python :: 3.7
+Classifier: Programming Language :: Python :: 3.8
+Classifier: Programming Language :: Python :: Implementation :: CPython
+Classifier: Programming Language :: Python :: Implementation :: PyPy
+Classifier: Operating System :: OS Independent
+Classifier: Topic :: Text Processing :: Filters
+Classifier: Topic :: Utilities
+Requires-Python: >=3.5
+
+
+Pygments
+~~~~~~~~
+
+Pygments is a syntax highlighting package written in Python.
+
+It is a generic syntax highlighter suitable for use in code hosting, forums,
+wikis or other applications that need to prettify source code.  Highlights
+are:
+
+* a wide range of over 500 languages and other text formats is supported
+* special attention is paid to details, increasing quality by a fair amount
+* support for new languages and formats are added easily
+* a number of output formats, presently HTML, LaTeX, RTF, SVG, all image     formats that PIL supports and ANSI sequences
+* it is usable as a command-line tool and as a library
+
+:copyright: Copyright 2006-2019 by the Pygments team, see AUTHORS.
+:license: BSD, see LICENSE for details.
+

syft/cataloger/rpmdb/parse_rpmdb.go

@@ -52,8 +52,9 @@ func parseRpmDB(_ string, reader io.Reader) ([]pkg.Package, error) {
 			Name:    entry.Name,
 			Version: fmt.Sprintf("%s-%s", entry.Version, entry.Release), // this is what engine does
 			//Version: fmt.Sprintf("%d:%s-%s.%s", entry.Epoch, entry.Version, entry.Release, entry.Arch),
-			Type: pkg.RpmPkg,
-			Metadata: pkg.RpmMetadata{
+			Type:         pkg.RpmPkg,
+			MetadataType: pkg.RpmdbMetadataType,
+			Metadata: pkg.RpmdbMetadata{
 				Name:      entry.Name,
 				Version:   entry.Version,
 				Epoch:     entry.Epoch,

syft/cataloger/rpmdb/parse_rpmdb_test.go

@@ -1,19 +1,21 @@
 package rpmdb
 
 import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/go-test/deep"
 	"os"
 	"testing"
+
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/go-test/deep"
 )
 
 func TestParseRpmDB(t *testing.T) {
 	expected := map[string]pkg.Package{
 		"dive": {
-			Name:    "dive",
-			Version: "0.9.2-1",
-			Type:    pkg.RpmPkg,
-			Metadata: pkg.RpmMetadata{
+			Name:         "dive",
+			Version:      "0.9.2-1",
+			Type:         pkg.RpmPkg,
+			MetadataType: pkg.RpmdbMetadataType,
+			Metadata: pkg.RpmdbMetadata{
 				Name:      "dive",
 				Epoch:     0,
 				Arch:      "x86_64",

syft/cataloger/ruby/parse_gemspec.go

@@ -8,6 +8,8 @@ import (
 	"regexp"
 	"strings"
 
+	"github.com/anchore/syft/internal"
+
 	"github.com/mitchellh/mapstructure"
 
 	"github.com/anchore/syft/syft/cataloger/common"
@@ -76,7 +78,7 @@ func parseGemSpecEntries(_ string, reader io.Reader) ([]pkg.Package, error) {
 		}
 
 		for field, pattern := range patterns {
-			matchMap := matchCaptureGroups(pattern, sanitizedLine)
+			matchMap := internal.MatchCaptureGroups(pattern, sanitizedLine)
 			if value := matchMap[field]; value != "" {
 				if postProcessor := postProcessors[field]; postProcessor != nil {
 					fields[field] = postProcessor(value)
@@ -96,12 +98,13 @@ func parseGemSpecEntries(_ string, reader io.Reader) ([]pkg.Package, error) {
 		}
 
 		pkgs = append(pkgs, pkg.Package{
-			Name:     metadata.Name,
-			Version:  metadata.Version,
-			Licenses: metadata.Licenses,
-			Language: pkg.Ruby,
-			Type:     pkg.GemPkg,
-			Metadata: metadata,
+			Name:         metadata.Name,
+			Version:      metadata.Version,
+			Licenses:     metadata.Licenses,
+			Language:     pkg.Ruby,
+			Type:         pkg.GemPkg,
+			MetadataType: pkg.GemMetadataType,
+			Metadata:     metadata,
 		})
 	}
 
@@ -121,15 +124,3 @@ func renderUtf8(s string) string {
 	})
 	return fullReplacement
 }
-
-// matchCaptureGroups takes a regular expression and string and returns all of the named capture group results in a map.
-func matchCaptureGroups(regEx *regexp.Regexp, str string) map[string]string {
-	match := regEx.FindStringSubmatch(str)
-	results := make(map[string]string)
-	for i, name := range regEx.SubexpNames() {
-		if i > 0 && i <= len(match) {
-			results[name] = match[i]
-		}
-	}
-	return results
-}

syft/cataloger/ruby/parse_gemspec_test.go

@@ -10,17 +10,19 @@ import (
 
 func TestParseGemspec(t *testing.T) {
 	var expectedPkg = pkg.Package{
-		Name:     "bundler",
-		Version:  "2.1.4",
-		Type:     pkg.GemPkg,
-		Licenses: []string{"MIT"},
-		Language: pkg.Ruby,
+		Name:         "bundler",
+		Version:      "2.1.4",
+		Type:         pkg.GemPkg,
+		Licenses:     []string{"MIT"},
+		Language:     pkg.Ruby,
+		MetadataType: pkg.GemMetadataType,
 		Metadata: pkg.GemMetadata{
 			Name:     "bundler",
 			Version:  "2.1.4",
 			Files:    []string{"exe/bundle", "exe/bundler"},
 			Authors:  []string{"André Arko", "Samuel Giddins", "Colby Swandale", "Hiroshi Shibata", "David Rodríguez", "Grey Baker", "Stephanie Morillo", "Chris Morris", "James Wen", "Tim Moore", "André Medeiros", "Jessica Lynn Suttles", "Terence Lee", "Carl Lerche", "Yehuda Katz"},
 			Licenses: []string{"MIT"},
+			Homepage: "https://bundler.io",
 		},
 	}
 

syft/pkg/apk_metadata.go

@@ -29,10 +29,10 @@ type ApkMetadata struct {
 // ApkFileRecord represents a single file listing and metadata from a APK DB entry (which may have many of these file records).
 type ApkFileRecord struct {
 	Path        string `json:"path"`
-	OwnerUID    string `json:"ownerUid"`
-	OwnerGUI    string `json:"ownerGid"`
-	Permissions string `json:"permissions"`
-	Checksum    string `json:"checksum"`
+	OwnerUID    string `json:"ownerUid,omitempty"`
+	OwnerGID    string `json:"ownerGid,omitempty"`
+	Permissions string `json:"permissions,omitempty"`
+	Checksum    string `json:"checksum,omitempty"`
 }
 
 func (m ApkMetadata) PackageURL() string {

syft/pkg/gem_metadata.go

@@ -3,7 +3,8 @@ package pkg
 type GemMetadata struct {
 	Name     string   `mapstructure:"name" json:"name"`
 	Version  string   `mapstructure:"version" json:"version"`
-	Files    []string `mapstructure:"files" json:"files"`
-	Authors  []string `mapstructure:"authors" json:"authors"`
-	Licenses []string `mapstructure:"licenses" json:"licenses"`
+	Files    []string `mapstructure:"files" json:"files,omitempty"`
+	Authors  []string `mapstructure:"authors" json:"authors,omitempty"`
+	Licenses []string `mapstructure:"licenses" json:"licenses,omitempty"`
+	Homepage string   `mapstructure:"homepage" json:"homepage,omitempty"`
 }

syft/pkg/java_metadata.go

@@ -4,14 +4,15 @@ import "github.com/package-url/packageurl-go"
 
 // JavaMetadata encapsulates all Java ecosystem metadata for a package as well as an (optional) parent relationship.
 type JavaMetadata struct {
-	Manifest      *JavaManifest  `mapstructure:"Manifest" json:"manifest"`
-	PomProperties *PomProperties `mapstructure:"PomProperties" json:"pomProperties"`
-	Parent        *Package       `json:"parentPackage"` // TODO: should this be included in the json output?
+	VirtualPath   string         `json:"virtualPath"`
+	Manifest      *JavaManifest  `mapstructure:"Manifest" json:"manifest,omitempty"`
+	PomProperties *PomProperties `mapstructure:"PomProperties" json:"pomProperties,omitempty"`
+	Parent        *Package       `json:"-"`
 }
 
 // PomProperties represents the fields of interest extracted from a Java archive's pom.xml file.
 type PomProperties struct {
-	Path       string
+	Path       string            `mapstructure:"path" json:"path"`
 	Name       string            `mapstructure:"name" json:"name"`
 	GroupID    string            `mapstructure:"groupId" json:"groupId"`
 	ArtifactID string            `mapstructure:"artifactId" json:"artifactId"`
@@ -21,15 +22,8 @@ type PomProperties struct {
 
 // JavaManifest represents the fields of interest extracted from a Java archive's META-INF/MANIFEST.MF file.
 type JavaManifest struct {
-	Name            string            `mapstructure:"Name" json:"name"`
-	ManifestVersion string            `mapstructure:"Manifest-Version" json:"manifestVersion"`
-	SpecTitle       string            `mapstructure:"Specification-Title" json:"specificationTitle"`
-	SpecVersion     string            `mapstructure:"Specification-Version" json:"specificationVersion"`
-	SpecVendor      string            `mapstructure:"Specification-Vendor" json:"specificationVendor"`
-	ImplTitle       string            `mapstructure:"Implementation-Title" json:"implementationTitle"`
-	ImplVersion     string            `mapstructure:"Implementation-Version" json:"implementationVersion"`
-	ImplVendor      string            `mapstructure:"Implementation-Vendor" json:"implementationVendor"`
-	Extra           map[string]string `mapstructure:",remain" json:"extraFields"`
+	Main          map[string]string            `json:"main,omitempty"`
+	NamedSections map[string]map[string]string `json:"namedSections,omitempty"`
 }
 
 func (m JavaMetadata) PackageURL() string {

syft/pkg/metadata.go

@@ -0,0 +1,14 @@
+package pkg
+
+type MetadataType string
+
+const (
+	UnknownMetadataType        MetadataType = "UnknownMetadata"
+	ApkMetadataType            MetadataType = "apk-metadata"
+	DpkgMetadataType           MetadataType = "dpkg-metadata"
+	GemMetadataType            MetadataType = "gem-metadata"
+	JavaMetadataType           MetadataType = "java-metadata"
+	NpmPackageJSONMetadataType MetadataType = "npm-package-json-metadata"
+	RpmdbMetadataType          MetadataType = "rpmdb-metadata"
+	PythonPackageMetadataType  MetadataType = "python-package-metadata"
+)

syft/pkg/npm_metadata.go

@@ -0,0 +1,11 @@
+package pkg
+
+// NpmPackageJSONMetadata holds extra information that is used in pkg.Package
+type NpmPackageJSONMetadata struct {
+	Files       []string `mapstructure:"files" json:"files,omitempty"`
+	Author      string   `mapstructure:"author" json:"author"`
+	Licenses    []string `mapstructure:"licenses" json:"licenses"`
+	Homepage    string   `mapstructure:"homepage" json:"homepage"`
+	Description string   `mapstructure:"description" json:"description"`
+	URL         string   `mapstructure:"url" json:"url"`
+}

syft/pkg/package.go

@@ -23,10 +23,11 @@ type Package struct {
 	FoundBy string           `json:"foundBy"`  // the specific cataloger that discovered this package
 	Source  []file.Reference `json:"sources"`  // the locations that lead to the discovery of this package (note: this is not necessarily the locations that make up this package)
 	// TODO: should we move licenses into metadata?
-	Licenses []string    `json:"licenses"`           // licenses discovered with the package metadata
-	Language Language    `json:"language"`           // the language ecosystem this package belongs to (e.g. JavaScript, Python, etc)
-	Type     Type        `json:"type"`               // the package type (e.g. Npm, Yarn, Egg, Wheel, Rpm, Deb, etc)
-	Metadata interface{} `json:"metadata,omitempty"` // additional data found while parsing the package source
+	Licenses     []string     `json:"licenses"`           // licenses discovered with the package metadata
+	Language     Language     `json:"language"`           // the language ecosystem this package belongs to (e.g. JavaScript, Python, etc)
+	Type         Type         `json:"type"`               // the package type (e.g. Npm, Yarn, Python, Rpm, Deb, etc)
+	MetadataType MetadataType `json:"metadataType"`       // the shape of the additional data in the "metadata" field
+	Metadata     interface{}  `json:"metadata,omitempty"` // additional data found while parsing the package source
 }
 
 // ID returns the package ID, which is unique relative to a package catalog.

syft/pkg/package_test.go

@@ -25,7 +25,7 @@ func TestPackage_pURL(t *testing.T) {
 			pkg: Package{
 				Name:    "name",
 				Version: "v0.1.0",
-				Type:    WheelPkg,
+				Type:    PythonPkg,
 			},
 			expected: "pkg:pypi/name@v0.1.0",
 		},
@@ -33,7 +33,7 @@ func TestPackage_pURL(t *testing.T) {
 			pkg: Package{
 				Name:    "name",
 				Version: "v0.1.0",
-				Type:    EggPkg,
+				Type:    PythonPkg,
 			},
 			expected: "pkg:pypi/name@v0.1.0",
 		},
@@ -41,7 +41,7 @@ func TestPackage_pURL(t *testing.T) {
 			pkg: Package{
 				Name:    "name",
 				Version: "v0.1.0",
-				Type:    PythonSetupPkg,
+				Type:    PythonPkg,
 			},
 			expected: "pkg:pypi/name@v0.1.0",
 		},
@@ -49,7 +49,7 @@ func TestPackage_pURL(t *testing.T) {
 			pkg: Package{
 				Name:    "name",
 				Version: "v0.1.0",
-				Type:    PythonRequirementsPkg,
+				Type:    PythonPkg,
 			},
 			expected: "pkg:pypi/name@v0.1.0",
 		},
@@ -69,14 +69,6 @@ func TestPackage_pURL(t *testing.T) {
 			},
 			expected: "pkg:npm/name@v0.1.0",
 		},
-		{
-			pkg: Package{
-				Name:    "name",
-				Version: "v0.1.0",
-				Type:    YarnPkg,
-			},
-			expected: "pkg:npm/name@v0.1.0",
-		},
 		{
 			distro: distro.Distro{
 				Type: distro.Ubuntu,
@@ -101,7 +93,7 @@ func TestPackage_pURL(t *testing.T) {
 				Name:    "bad-name",
 				Version: "bad-v0.1.0",
 				Type:    RpmPkg,
-				Metadata: RpmMetadata{
+				Metadata: RpmdbMetadata{
 					Name:    "name",
 					Version: "v0.1.0",
 					Epoch:   2,

syft/pkg/python_package_metadata.go

@@ -0,0 +1,26 @@
+package pkg
+
+type Digest struct {
+	Algorithm string `json:"algorithm"`
+	Value     string `json:"value"`
+}
+
+// PythonFileRecord represents a single entry within a RECORD file for a python wheel or egg package
+type PythonFileRecord struct {
+	Path   string  `json:"path"`
+	Digest *Digest `json:"digest,omitempty"`
+	Size   string  `json:"size,omitempty"`
+}
+
+// PythonPackageMetadata represents all captured data for a python egg or wheel package.
+type PythonPackageMetadata struct {
+	Name                 string             `json:"name" mapstruct:"Name"`
+	Version              string             `json:"version" mapstruct:"Version"`
+	License              string             `json:"license" mapstruct:"License"`
+	Author               string             `json:"author" mapstruct:"Author"`
+	AuthorEmail          string             `json:"authorEmail" mapstruct:"Authoremail"`
+	Platform             string             `json:"platform" mapstruct:"Platform"`
+	Files                []PythonFileRecord `json:"files,omitempty"`
+	SitePackagesRootPath string             `json:"sitePackagesRootPath"`
+	TopLevelPackages     []string           `json:"topLevelPackages,omitempty"`
+}

syft/pkg/rpmdb_metadata.go

@@ -7,8 +7,8 @@ import (
 	"github.com/package-url/packageurl-go"
 )
 
-// RpmMetadata represents all captured data for a RPM DB package entry.
-type RpmMetadata struct {
+// RpmdbMetadata represents all captured data for a RPM DB package entry.
+type RpmdbMetadata struct {
 	Name      string `json:"name"`
 	Version   string `json:"version"`
 	Epoch     int    `json:"epoch"`
@@ -20,7 +20,7 @@ type RpmMetadata struct {
 	Vendor    string `json:"vendor"`
 }
 
-func (m RpmMetadata) PackageURL(d distro.Distro) string {
+func (m RpmdbMetadata) PackageURL(d distro.Distro) string {
 	pURL := packageurl.NewPackageURL(
 		packageurl.TypeRPM,
 		d.Type.String(),

syft/pkg/rpmdb_metadata_test.go

@@ -1,22 +1,23 @@
 package pkg
 
 import (
+	"testing"
+
 	"github.com/anchore/syft/syft/distro"
 	"github.com/sergi/go-diff/diffmatchpatch"
-	"testing"
 )
 
 func TestRpmMetadata_pURL(t *testing.T) {
 	tests := []struct {
 		distro   distro.Distro
-		metadata RpmMetadata
+		metadata RpmdbMetadata
 		expected string
 	}{
 		{
 			distro: distro.Distro{
 				Type: distro.CentOS,
 			},
-			metadata: RpmMetadata{
+			metadata: RpmdbMetadata{
 				Name:    "p",
 				Version: "v",
 				Arch:    "a",
@@ -29,7 +30,7 @@ func TestRpmMetadata_pURL(t *testing.T) {
 			distro: distro.Distro{
 				Type: distro.RedHat,
 			},
-			metadata: RpmMetadata{
+			metadata: RpmdbMetadata{
 				Name:    "p",
 				Version: "v",
 				Arch:    "a",

syft/pkg/type.go

@@ -6,36 +6,25 @@ import "github.com/package-url/packageurl-go"
 type Type string
 
 const (
-	UnknownPkg Type = "UnknownPackage"
-	ApkPkg     Type = "apk"
-	GemPkg     Type = "gem"
-	DebPkg     Type = "deb"
-	EggPkg     Type = "egg"
-	// PacmanPkg Type = "pacman"
-	RpmPkg                Type = "rpm"
-	WheelPkg              Type = "wheel"
-	PoetryPkg             Type = "poetry"
-	NpmPkg                Type = "npm"
-	YarnPkg               Type = "yarn"
-	PythonRequirementsPkg Type = "python-requirements"
-	PythonSetupPkg        Type = "python-setup"
-	JavaPkg               Type = "java-archive"
-	JenkinsPluginPkg      Type = "jenkins-plugin"
-	GoModulePkg           Type = "go-module"
+	UnknownPkg       Type = "UnknownPackage"
+	ApkPkg           Type = "apk"
+	GemPkg           Type = "gem"
+	DebPkg           Type = "deb"
+	RpmPkg           Type = "rpm"
+	NpmPkg           Type = "npm"
+	PythonPkg        Type = "python"
+	JavaPkg          Type = "java-archive"
+	JenkinsPluginPkg Type = "jenkins-plugin"
+	GoModulePkg      Type = "go-module"
 )
 
 var AllPkgs = []Type{
 	ApkPkg,
 	GemPkg,
 	DebPkg,
-	EggPkg,
-	// PacmanPkg,
 	RpmPkg,
-	WheelPkg,
 	NpmPkg,
-	YarnPkg,
-	PythonRequirementsPkg,
-	PythonSetupPkg,
+	PythonPkg,
 	JavaPkg,
 	JenkinsPluginPkg,
 	GoModulePkg,
@@ -49,9 +38,9 @@ func (t Type) PackageURLType() string {
 		return packageurl.TypeGem
 	case DebPkg:
 		return "deb"
-	case EggPkg, WheelPkg, PythonRequirementsPkg, PythonSetupPkg:
+	case PythonPkg:
 		return packageurl.TypePyPi
-	case NpmPkg, YarnPkg:
+	case NpmPkg:
 		return packageurl.TypeNPM
 	case JavaPkg, JenkinsPluginPkg:
 		return packageurl.TypeMaven

syft/presenter/cyclonedx/presenter_test.go

@@ -3,11 +3,12 @@ package cyclonedx
 import (
 	"bytes"
 	"flag"
-	"github.com/anchore/stereoscope/pkg/imagetest"
-	"github.com/anchore/syft/syft/distro"
 	"regexp"
 	"testing"
 
+	"github.com/anchore/stereoscope/pkg/imagetest"
+	"github.com/anchore/syft/syft/distro"
+
 	"github.com/anchore/go-testutils"
 	"github.com/anchore/stereoscope/pkg/file"
 	"github.com/anchore/syft/syft/pkg"
@@ -109,7 +110,7 @@ func TestCycloneDxImgsPresenter(t *testing.T) {
 		},
 		Type:    pkg.RpmPkg,
 		FoundBy: "the-cataloger-1",
-		Metadata: pkg.RpmMetadata{
+		Metadata: pkg.RpmdbMetadata{
 			Name:      "package1",
 			Epoch:     0,
 			Arch:      "x86_64",
@@ -133,7 +134,7 @@ func TestCycloneDxImgsPresenter(t *testing.T) {
 			"MIT",
 			"Apache-v2",
 		},
-		Metadata: pkg.RpmMetadata{
+		Metadata: pkg.RpmdbMetadata{
 			Name:      "package2",
 			Epoch:     0,
 			Arch:      "x86_64",

syft/scope/resolver.go

@@ -16,13 +16,20 @@ type Resolver interface {
 
 // ContentResolver knows how to get file content for given file.References
 type ContentResolver interface {
+	FileContentsByRef(ref file.Reference) (string, error)
 	MultipleFileContentsByRef(f ...file.Reference) (map[file.Reference]string, error)
+	// TODO: we should consider refactoring to return a set of io.Readers or file.Openers instead of the full contents themselves (allow for optional buffering).
 }
 
 //  FileResolver knows how to get file.References for given string paths and globs
 type FileResolver interface {
+	// FilesByPath fetches a set of file references which have the given path (for an image, there may be multiple matches)
 	FilesByPath(paths ...file.Path) ([]file.Reference, error)
+	// FilesByGlob fetches a set of file references which the given glob matches
 	FilesByGlob(patterns ...string) ([]file.Reference, error)
+	// RelativeFileByPath fetches a single file at the given path relative to the layer squash of the given reference.
+	// This is helpful when attempting to find a file that is in the same layer or lower as another file.
+	RelativeFileByPath(reference file.Reference, path string) (*file.Reference, error)
 }
 
 // getImageResolver returns the appropriate resolve for a container image given the scope option

syft/scope/resolvers/all_layers_resolver.go

@@ -109,8 +109,23 @@ func (r *AllLayersResolver) FilesByGlob(patterns ...string) ([]file.Reference, e
 	return uniqueFiles, nil
 }
 
+func (r *AllLayersResolver) RelativeFileByPath(reference file.Reference, path string) (*file.Reference, error) {
+	entry, err := r.img.FileCatalog.Get(reference)
+	if err != nil {
+		return nil, err
+	}
+
+	return entry.Source.SquashedTree.File(file.Path(path)), nil
+}
+
 // MultipleFileContentsByRef returns the file contents for all file.References relative to the image. Note that a
 // file.Reference is a path relative to a particular layer.
 func (r *AllLayersResolver) MultipleFileContentsByRef(f ...file.Reference) (map[file.Reference]string, error) {
 	return r.img.MultipleFileContentsByRef(f...)
 }
+
+// FileContentsByRef fetches file contents for a single file reference, irregardless of the source layer.
+// If the path does not exist an error is returned.
+func (r *AllLayersResolver) FileContentsByRef(ref file.Reference) (string, error) {
+	return r.img.FileContentsByRef(ref)
+}

syft/scope/resolvers/directory_resolver.go

@@ -5,6 +5,7 @@ import (
 	"io/ioutil"
 	"os"
 	"path"
+	"path/filepath"
 
 	"github.com/anchore/stereoscope/pkg/file"
 	"github.com/anchore/syft/internal/log"
@@ -18,7 +19,7 @@ type DirectoryResolver struct {
 
 // Stringer to represent a directory path data source
 func (s DirectoryResolver) String() string {
-	return fmt.Sprintf("dir://%s", s.Path)
+	return fmt.Sprintf("dir:%s", s.Path)
 }
 
 // FilesByPath returns all file.References that match the given paths from the directory.
@@ -26,15 +27,19 @@ func (s DirectoryResolver) FilesByPath(userPaths ...file.Path) ([]file.Reference
 	var references = make([]file.Reference, 0)
 
 	for _, userPath := range userPaths {
-		resolvedPath := path.Join(s.Path, string(userPath))
-		_, err := os.Stat(resolvedPath)
+		userStrPath := string(userPath)
+
+		if filepath.IsAbs(userStrPath) {
+			// a path relative to root should be prefixed with the resolvers directory path, otherwise it should be left as is
+			userStrPath = path.Join(s.Path, userStrPath)
+		}
+		_, err := os.Stat(userStrPath)
 		if os.IsNotExist(err) {
 			continue
 		} else if err != nil {
-			log.Errorf("path (%s) is not valid: %v", resolvedPath, err)
+			log.Errorf("path (%s) is not valid: %v", userStrPath, err)
 		}
-		filePath := file.Path(resolvedPath)
-		references = append(references, file.NewFileReference(filePath))
+		references = append(references, file.NewFileReference(file.Path(userStrPath)))
 	}
 
 	return references, nil
@@ -75,6 +80,18 @@ func (s DirectoryResolver) FilesByGlob(patterns ...string) ([]file.Reference, er
 	return result, nil
 }
 
+func (s *DirectoryResolver) RelativeFileByPath(_ file.Reference, path string) (*file.Reference, error) {
+	paths, err := s.FilesByPath(file.Path(path))
+	if err != nil {
+		return nil, err
+	}
+	if len(paths) == 0 {
+		return nil, nil
+	}
+
+	return &paths[0], nil
+}
+
 // MultipleFileContentsByRef returns the file contents for all file.References relative a directory.
 func (s DirectoryResolver) MultipleFileContentsByRef(f ...file.Reference) (map[file.Reference]string, error) {
 	refContents := make(map[file.Reference]string)
@@ -82,9 +99,20 @@ func (s DirectoryResolver) MultipleFileContentsByRef(f ...file.Reference) (map[f
 		contents, err := fileContents(fileRef.Path)
 
 		if err != nil {
-			return refContents, fmt.Errorf("could not read contents of file: %s", fileRef.Path)
+			return nil, fmt.Errorf("could not read contents of file: %s", fileRef.Path)
 		}
 		refContents[fileRef] = string(contents)
 	}
 	return refContents, nil
 }
+
+// FileContentsByRef fetches file contents for a single file reference relative to a directory.
+// If the path does not exist an error is returned.
+func (s DirectoryResolver) FileContentsByRef(reference file.Reference) (string, error) {
+	contents, err := fileContents(reference.Path)
+	if err != nil {
+		return "", fmt.Errorf("could not read contents of file: %s", reference.Path)
+	}
+
+	return string(contents), nil
+}

syft/scope/resolvers/directory_resolver_test.go

@@ -1,7 +1,6 @@
 package resolvers
 
 import (
-	"path"
 	"testing"
 
 	"github.com/anchore/stereoscope/pkg/file"
@@ -10,24 +9,49 @@ import (
 func TestDirectoryResolver_FilesByPath(t *testing.T) {
 	cases := []struct {
 		name     string
+		root     string
 		input    string
+		expected string
 		refCount int
 	}{
 		{
-			name:     "finds a file",
-			input:    "image-symlinks/file-1.txt",
+			name:     "finds a file (relative)",
+			root:     "./test-fixtures/",
+			input:    "test-fixtures/image-symlinks/file-1.txt",
+			expected: "test-fixtures/image-symlinks/file-1.txt",
 			refCount: 1,
 		},
 		{
-			name:     "managed non-existing files",
-			input:    "image-symlinks/bogus.txt",
+			name:     "finds a file with relative indirection",
+			root:     "./test-fixtures/../test-fixtures",
+			input:    "test-fixtures/image-symlinks/file-1.txt",
+			expected: "test-fixtures/image-symlinks/file-1.txt",
+			refCount: 1,
+		},
+		{
+			// note: this is asserting the old behavior is not supported
+			name:     "relative lookup with wrong path fails",
+			root:     "./test-fixtures/",
+			input:    "image-symlinks/file-1.txt",
 			refCount: 0,
 		},
+		{
+			name:     "managed non-existing files (relative)",
+			root:     "./test-fixtures/",
+			input:    "test-fixtures/image-symlinks/bogus.txt",
+			refCount: 0,
+		},
+		{
+			name:     "finds a file (absolute)",
+			root:     "./test-fixtures/",
+			input:    "/image-symlinks/file-1.txt",
+			expected: "test-fixtures/image-symlinks/file-1.txt",
+			refCount: 1,
+		},
 	}
 	for _, c := range cases {
 		t.Run(c.name, func(t *testing.T) {
-			resolver := DirectoryResolver{"test-fixtures"}
-			expected := path.Join("test-fixtures", c.input)
+			resolver := DirectoryResolver{c.root}
 			refs, err := resolver.FilesByPath(file.Path(c.input))
 			if err != nil {
 				t.Fatalf("could not use resolver: %+v, %+v", err, refs)
@@ -38,8 +62,8 @@ func TestDirectoryResolver_FilesByPath(t *testing.T) {
 			}
 
 			for _, actual := range refs {
-				if actual.Path != file.Path(expected) {
-					t.Errorf("bad resolve path: '%s'!='%s'", actual.Path, c.input)
+				if actual.Path != file.Path(c.expected) {
+					t.Errorf("bad resolve path: '%s'!='%s'", actual.Path, c.expected)
 				}
 			}
 		})
@@ -54,17 +78,17 @@ func TestDirectoryResolver_MultipleFilesByPath(t *testing.T) {
 	}{
 		{
 			name:     "finds multiple files",
-			input:    []file.Path{file.Path("image-symlinks/file-1.txt"), file.Path("image-symlinks/file-2.txt")},
+			input:    []file.Path{file.Path("test-fixtures/image-symlinks/file-1.txt"), file.Path("test-fixtures/image-symlinks/file-2.txt")},
 			refCount: 2,
 		},
 		{
 			name:     "skips non-existing files",
-			input:    []file.Path{file.Path("image-symlinks/bogus.txt"), file.Path("image-symlinks/file-1.txt")},
+			input:    []file.Path{file.Path("test-fixtures/image-symlinks/bogus.txt"), file.Path("test-fixtures/image-symlinks/file-1.txt")},
 			refCount: 1,
 		},
 		{
 			name:     "does not return anything for non-existing directories",
-			input:    []file.Path{file.Path("non-existing/bogus.txt"), file.Path("non-existing/file-1.txt")},
+			input:    []file.Path{file.Path("test-fixtures/non-existing/bogus.txt"), file.Path("test-fixtures/non-existing/file-1.txt")},
 			refCount: 0,
 		},
 	}
@@ -93,17 +117,17 @@ func TestDirectoryResolver_MultipleFileContentsByRef(t *testing.T) {
 	}{
 		{
 			name:     "gets multiple file contents",
-			input:    []file.Path{file.Path("image-symlinks/file-1.txt"), file.Path("image-symlinks/file-2.txt")},
+			input:    []file.Path{file.Path("test-fixtures/image-symlinks/file-1.txt"), file.Path("test-fixtures/image-symlinks/file-2.txt")},
 			refCount: 2,
 		},
 		{
 			name:     "skips non-existing files",
-			input:    []file.Path{file.Path("image-symlinks/bogus.txt"), file.Path("image-symlinks/file-1.txt")},
+			input:    []file.Path{file.Path("test-fixtures/image-symlinks/bogus.txt"), file.Path("test-fixtures/image-symlinks/file-1.txt")},
 			refCount: 1,
 		},
 		{
 			name:     "does not return anything for non-existing directories",
-			input:    []file.Path{file.Path("non-existing/bogus.txt"), file.Path("non-existing/file-1.txt")},
+			input:    []file.Path{file.Path("test-fixtures/non-existing/bogus.txt"), file.Path("test-fixtures/non-existing/file-1.txt")},
 			refCount: 0,
 		},
 	}

syft/scope/resolvers/image_squash_resolver.go

@@ -73,8 +73,26 @@ func (r *ImageSquashResolver) FilesByGlob(patterns ...string) ([]file.Reference,
 	return uniqueFiles, nil
 }
 
+func (r *ImageSquashResolver) RelativeFileByPath(reference file.Reference, path string) (*file.Reference, error) {
+	paths, err := r.FilesByPath(file.Path(path))
+	if err != nil {
+		return nil, err
+	}
+	if len(paths) == 0 {
+		return nil, nil
+	}
+
+	return &paths[0], nil
+}
+
 // MultipleFileContentsByRef returns the file contents for all file.References relative to the image. Note that a
 // file.Reference is a path relative to a particular layer, in this case only from the squashed representation.
 func (r *ImageSquashResolver) MultipleFileContentsByRef(f ...file.Reference) (map[file.Reference]string, error) {
 	return r.img.MultipleFileContentsByRef(f...)
 }
+
+// FileContentsByRef fetches file contents for a single file reference, irregardless of the source layer.
+// If the path does not exist an error is returned.
+func (r *ImageSquashResolver) FileContentsByRef(ref file.Reference) (string, error) {
+	return r.img.FileContentsByRef(ref)
+}

syft/scope/scope_test.go

@@ -61,13 +61,13 @@ func TestDirectoryScope(t *testing.T) {
 		{
 			desc:       "path detected",
 			input:      "test-fixtures",
-			inputPaths: []file.Path{file.Path("path-detected")},
+			inputPaths: []file.Path{file.Path("test-fixtures/path-detected")},
 			expRefs:    1,
 		},
 		{
 			desc:       "no files-by-path detected",
 			input:      "test-fixtures",
-			inputPaths: []file.Path{file.Path("no-path-detected")},
+			inputPaths: []file.Path{file.Path("test-fixtures/no-path-detected")},
 			expRefs:    0,
 		},
 	}
@@ -105,13 +105,13 @@ func TestMultipleFileContentsByRefContents(t *testing.T) {
 		{
 			input:    "test-fixtures/path-detected",
 			desc:     "empty file",
-			path:     "empty",
+			path:     "test-fixtures/path-detected/empty",
 			expected: "",
 		},
 		{
 			input:    "test-fixtures/path-detected",
 			desc:     "file has contents",
-			path:     ".vimrc",
+			path:     "test-fixtures/path-detected/.vimrc",
 			expected: "\" A .vimrc file\n",
 		},
 	}
@@ -127,7 +127,7 @@ func TestMultipleFileContentsByRefContents(t *testing.T) {
 			}
 
 			if len(refs) != 1 {
-				t.Errorf("expected a single ref to be generated but got: %d", len(refs))
+				t.Fatalf("expected a single ref to be generated but got: %d", len(refs))
 			}
 			ref := refs[0]
 

test/acceptance/mac.sh

@@ -37,9 +37,10 @@ skopeo --override-os linux copy docker://docker.io/${TEST_IMAGE} docker-archive:
 ls -alh ${TEST_IMAGE_TAR}
 
 # run syft
-chmod 755 ${DISTDIR}/syft_darwin_amd64/syft
-${DISTDIR}/syft_darwin_amd64/syft version
-SYFT_CHECK_FOR_APP_UPDATE=0 ${DISTDIR}/syft_darwin_amd64/syft docker-archive://${TEST_IMAGE_TAR} -vv -o json > ${REPORT}
+SYFT_PATH="${DISTDIR}/syft-macos_darwin_amd64/syft"
+chmod 755 "${SYFT_PATH}"
+"${SYFT_PATH}" version
+SYFT_CHECK_FOR_APP_UPDATE=0 "${SYFT_PATH}" docker-archive://${TEST_IMAGE_TAR} -vv -o json > "${REPORT}"
 
 # keep the generated report around
 mkdir -p ${RESULTSDIR}

test/inline-compare/.gitignore

@@ -1,2 +1,3 @@
 *.json
+*.pyc
 inline-reports

test/inline-compare/Makefile

@@ -2,7 +2,7 @@ ifndef SYFT_CMD
 	SYFT_CMD = go run ../../main.go
 endif
 
-IMAGE_CLEAN = $(shell echo $(COMPARE_IMAGE) | tr ":" "_")
+IMAGE_CLEAN = $(shell basename $(COMPARE_IMAGE) | tr ":" "_" )
 SYFT_DIR = syft-reports
 SYFT_REPORT = $(SYFT_DIR)/$(IMAGE_CLEAN).json
 INLINE_DIR = inline-reports

test/inline-compare/compare-all.sh

@@ -1,8 +1,7 @@
 #!/usr/bin/env bash
 set -eu
 
-# TODO: Add "alpine:3.12.0" back in when we've figured out how to handle the apk version field w/ and w/o release information (see issue: https://github.com/anchore/syft/pull/195)
-images=("debian:10.5" "centos:8.2.2004" "rails:5.0.1")
+images=("debian:10.5" "centos:8.2.2004" "rails:5.0.1" "alpine:3.12.0" "anchore/test_images:java" "anchore/test_images:py38" "anchore/anchore-engine:v0.8.2" "jenkins/jenkins:2.249.2-lts-jdk11" )
 
 # gather all image analyses
 for img in "${images[@]}"; do

test/inline-compare/compare.py

@@ -1,13 +1,22 @@
 #!/usr/bin/env python3
 import os
 import sys
-import json
+import difflib
 import collections
 
-QUALITY_GATE_THRESHOLD = 0.95
-INDENT = "    "
-IMAGE_QUALITY_GATE = collections.defaultdict(lambda: QUALITY_GATE_THRESHOLD, **{
+import utils.package
+from utils.format import Colors, print_rows
+from utils.inline import InlineScan
+from utils.syft import Syft
 
+DEFAULT_QUALITY_GATE_THRESHOLD = 0.95
+INDENT = "    "
+
+PACKAGE_QUALITY_GATE = collections.defaultdict(lambda: DEFAULT_QUALITY_GATE_THRESHOLD, **{})
+METADATA_QUALITY_GATE = collections.defaultdict(lambda: DEFAULT_QUALITY_GATE_THRESHOLD, **{
+    # syft is better at detecting package versions in specific cases, leading to a drop in matching metadata
+    "anchore/test_images:java": 0.61,
+    "jenkins/jenkins:2.249.2-lts-jdk11": 0.85,
 })
 
 # We additionally fail if an image is above a particular threshold. Why? We expect the lower threshold to be 90%,
@@ -16,271 +25,207 @@ IMAGE_QUALITY_GATE = collections.defaultdict(lambda: QUALITY_GATE_THRESHOLD, **{
 # issues/enhancements are done we want to ensure that the lower threshold is bumped up to catch regression. The only way
 # to do this is to select an upper threshold for images with known threshold values, so we have a failure that
 # loudly indicates the lower threshold should be bumped.
-IMAGE_UPPER_THRESHOLD = collections.defaultdict(lambda: 1, **{
-
+PACKAGE_UPPER_THRESHOLD = collections.defaultdict(lambda: 1, **{})
+METADATA_UPPER_THRESHOLD = collections.defaultdict(lambda: 1, **{
+    # syft is better at detecting package versions in specific cases, leading to a drop in matching metadata
+    "anchore/test_images:java": 0.65,
+    "jenkins/jenkins:2.249.2-lts-jdk11": 0.9,
 })
-Metadata = collections.namedtuple("Metadata", "version")
-Package = collections.namedtuple("Package", "name type")
 
 
-class InlineScan:
+def report(image, analysis):
+    if analysis.extra_packages:
+        rows = []
+        print(
+            Colors.bold + "Syft found extra packages:",
+            Colors.reset,
+            "Syft discovered packages that Inline did not",
+        )
+        for package in sorted(list(analysis.extra_packages)):
+            rows.append([INDENT, repr(package)])
+        print_rows(rows)
+        print()
 
-    report_tmpl = "{image}-{report}.json"
+    if analysis.missing_packages:
+        rows = []
+        print(
+            Colors.bold + "Syft missed packages:",
+            Colors.reset,
+            "Inline discovered packages that Syft did not",
+        )
+        for package in sorted(list(analysis.missing_packages)):
+            rows.append([INDENT, repr(package)])
+        print_rows(rows)
+        print()
 
-    def __init__(self, image, report_dir="./"):
-        self.report_dir = report_dir
-        self.image = image
+    if analysis.missing_metadata:
+        print(
+            Colors.bold + "Syft mismatched metadata:",
+            Colors.reset,
+            "the packages between Syft and Inline are the same, the metadata is not",
+        )
+        for inline_metadata_pair in sorted(list(analysis.missing_metadata)):
+            pkg, metadata = inline_metadata_pair
+            if pkg not in analysis.syft_data.metadata[pkg.type]:
+                continue
+            syft_metadata_item = analysis.syft_data.metadata[pkg.type][pkg]
 
-    def packages(self):
-        python_packages, python_metadata = self._python_packages()
-        gem_packages, gem_metadata = self._gem_packages()
-        os_packages, os_metadata = self._os_packages()
-        return python_packages | os_packages | gem_packages , {**python_metadata, **os_metadata, **gem_metadata}
+            diffs = difflib.ndiff([repr(syft_metadata_item)], [repr(metadata)])
 
-    def _report_path(self, report):
-        return os.path.join(
-            self.report_dir,
-            self.report_tmpl.format(image=self.image.replace(":", "_"), report=report),
+            print(INDENT + "for: " + repr(pkg), "(top is syft, bottom is inline)")
+            print(INDENT+INDENT+("\n"+INDENT+INDENT).join(list(diffs)))
+
+        if not analysis.missing_metadata:
+            print(
+                INDENT,
+                "There are mismatches, but only due to packages Syft did not find (but inline did).\n",
+            )
+
+    if analysis.similar_missing_packages:
+        rows = []
+        print(
+            Colors.bold + "Probably pairings of missing/extra packages:",
+            Colors.reset,
+            "to aid in troubleshooting missed/extra packages",
+        )
+        for similar_packages in analysis.similar_missing_packages:
+            rows.append(
+                [
+                    INDENT,
+                    repr(similar_packages.pkg),
+                    "--->",
+                    repr(similar_packages.missed),
+                ]
+            )
+        print_rows(rows)
+        print()
+
+    show_probable_mismatches = analysis.unmatched_missing_packages and analysis.extra_packages and len(analysis.unmatched_missing_packages) != len(analysis.missing_packages)
+
+    if show_probable_mismatches:
+        rows = []
+        print(
+            Colors.bold + "Probably missed packages:",
+            Colors.reset,
+            "a probable pair was not found",
+        )
+        for p in analysis.unmatched_missing_packages:
+            rows.append([INDENT, repr(p)])
+        print_rows(rows)
+        print()
+
+    print(Colors.bold + "Summary:", Colors.reset, image)
+    print("   Inline Packages : %d" % len(analysis.inline_data.packages))
+    print("   Syft Packages   : %d" % len(analysis.syft_data.packages))
+    print(
+        "         (extra)   : %d (note: this is ignored by the quality gate!)"
+        % len(analysis.extra_packages)
+    )
+    print("       (missing)   : %d" % len(analysis.missing_packages))
+    print()
+
+    if show_probable_mismatches:
+        print(
+            "   Probable Package Matches  : %d (matches not made, but were probably found by both Inline and Syft)"
+            % len(analysis.similar_missing_packages)
+        )
+        print(
+            "   Probable Packages Matched : %2.3f %% (%d/%d packages)"
+            % (
+                analysis.percent_probable_overlapping_packages,
+                len(analysis.overlapping_packages)
+                + len(analysis.similar_missing_packages),
+                len(analysis.inline_data.packages),
+            )
+        )
+        print(
+            "   Probable Packages Missing : %d "
+            % len(analysis.unmatched_missing_packages)
+        )
+        print()
+    print(
+        "   Baseline Packages Matched : %2.3f %% (%d/%d packages)"
+        % (
+            analysis.percent_overlapping_packages,
+            len(analysis.overlapping_packages),
+            len(analysis.inline_data.packages),
+        )
+    )
+    print(
+        "   Baseline Metadata Matched : %2.3f %% (%d/%d metadata)"
+        % (
+            analysis.percent_overlapping_metadata,
+            len(analysis.overlapping_metadata),
+            len(analysis.inline_metadata),
+        )
+    )
+
+
+def enforce_quality_gate(title, actual_value, lower_gate_value, upper_gate_value):
+
+    if actual_value < lower_gate_value:
+        print(
+            Colors.bold
+            + "   %s Quality Gate:\t" % title
+            + Colors.FG.red
+            + "FAIL (is not >= %d %%)" % lower_gate_value,
+            Colors.reset,
+            )
+        return False
+    elif actual_value > upper_gate_value:
+        print(
+            Colors.bold
+            + "   %s Quality Gate:\t" % title
+            + Colors.FG.orange
+            + "FAIL (lower threshold is artificially low and should be updated)",
+            Colors.reset,
+            )
+        return False
+
+    print(
+        Colors.bold
+        + "   %s Quality Gate:\t" % title
+        + Colors.FG.green
+        + "Pass (>= %d %%)" % lower_gate_value,
+        Colors.reset,
         )
 
-    def _enumerate_section(self, report, section):
-        report_path = self._report_path(report=report)
-        os_report_path = self._report_path(report="content-os")
-
-        if os.path.exists(os_report_path) and not os.path.exists(report_path):
-            # if the OS report is there but the target report is not, that is engine's way of saying "no findings"
-            return
-
-        with open(report_path) as json_file:
-            data = json.load(json_file)
-            for entry in data[section]:
-                yield entry
-
-    def _python_packages(self):
-        packages = set()
-        metadata = collections.defaultdict(dict)
-        for entry in self._enumerate_section(
-            report="content-python", section="content"
-        ):
-            package = Package(name=entry["package"], type=entry["type"].lower(),)
-            packages.add(package)
-            metadata[package.type][package] = Metadata(version=entry["version"])
-
-        return packages, metadata
-
-    def _gem_packages(self):
-        packages = set()
-        metadata = collections.defaultdict(dict)
-        for entry in self._enumerate_section(
-                report="content-gem", section="content"
-        ):
-            package = Package(name=entry["package"], type=entry["type"].lower(),)
-            packages.add(package)
-            metadata[package.type][package] = Metadata(version=entry["version"])
-
-        return packages, metadata
-
-    def _os_packages(self):
-        packages = set()
-        metadata = collections.defaultdict(dict)
-        for entry in self._enumerate_section(report="content-os", section="content"):
-            package = Package(name=entry["package"], type=entry["type"].lower())
-            packages.add(package)
-            metadata[package.type][package] = Metadata(version=entry["version"])
-
-        return packages, metadata
-
-
-class Syft:
-
-    report_tmpl = "{image}.json"
-
-    def __init__(self, image, report_dir="./"):
-        self.report_path = os.path.join(
-            report_dir, self.report_tmpl.format(image=image.replace(":", "_"))
-        )
-
-    def _enumerate_section(self, section):
-        with open(self.report_path) as json_file:
-            data = json.load(json_file)
-            for entry in data[section]:
-                yield entry
-
-    def packages(self):
-        packages = set()
-        metadata = collections.defaultdict(dict)
-        for entry in self._enumerate_section(section="artifacts"):
-
-            # normalize to inline
-            pkg_type = entry["type"].lower()
-            if pkg_type in ("wheel", "egg"):
-                pkg_type = "python"
-            elif pkg_type in ("deb",):
-                pkg_type = "dpkg"
-            elif pkg_type in ("java-archive",):
-                pkg_type = "java"
-            elif pkg_type in ("apk",):
-                pkg_type = "apkg"
-
-            package = Package(name=entry["name"], type=pkg_type,)
-
-            packages.add(package)
-            metadata[package.type][package] = Metadata(version=entry["version"])
-        return packages, metadata
-
-
-def print_rows(rows):
-    if not rows:
-        return
-    widths = []
-    for col, _ in enumerate(rows[0]):
-        width = max(len(row[col]) for row in rows) + 2  # padding
-        widths.append(width)
-    for row in rows:
-        print("".join(word.ljust(widths[col_idx]) for col_idx, word in enumerate(row)))
-
+    return True
 
 def main(image):
-    print(colors.bold+"Image:", image, colors.reset)
+    cwd = os.path.dirname(os.path.abspath(__file__))
 
-    inline = InlineScan(image=image, report_dir="inline-reports")
-    inline_packages, inline_metadata = inline.packages()
+    # parse the inline-scan and syft reports on disk
+    inline = InlineScan(image=image, report_dir=os.path.join(cwd, "inline-reports"))
+    syft = Syft(image=image, report_dir=os.path.join(cwd, "syft-reports"))
 
-    syft = Syft(image=image, report_dir="syft-reports")
-    syft_packages, syft_metadata = syft.packages()
-
-    if len(inline_packages) == 0:
-        # we are purposefully selecting test images that are guaranteed to have packages, so this should never happen
-        print(colors.bold + colors.fg.red + "inline found no packages!", colors.reset)
-        return 1
-
-    if len(syft_packages) == 0 and len(inline_packages) == 0:
-        print("nobody found any packages")
-        return 0
-
-    same_packages = syft_packages & inline_packages
-    percent_overlap_packages = (
-        float(len(same_packages)) / float(len(inline_packages))
-    ) * 100.0
-
-    bonus_packages = syft_packages - inline_packages
-    missing_packages = inline_packages - syft_packages
-
-    inline_metadata_set = set()
-    for package in inline_packages:
-        metadata = inline_metadata[package.type][package]
-        inline_metadata_set.add((package, metadata))
-
-    syft_overlap_metadata_set = set()
-    for package in syft_packages:
-        metadata = syft_metadata[package.type][package]
-        # we only want to really count mismatched metadata for packages that are at least found by inline
-        if package in inline_metadata.get(package.type, []):
-            syft_overlap_metadata_set.add((package, metadata))
-
-    same_metadata = syft_overlap_metadata_set & inline_metadata_set
-    percent_overlap_metadata = (
-        float(len(same_metadata)) / float(len(inline_metadata_set))
-    ) * 100.0
-    missing_metadata = inline_metadata_set - same_metadata
-
-    if bonus_packages:
-        rows = []
-        print(colors.bold + "Syft found extra packages:", colors.reset)
-        for package in sorted(list(bonus_packages)):
-            rows.append([INDENT, repr(package)])
-        print_rows(rows)
-        print()
-
-    if missing_packages:
-        rows = []
-        print(colors.bold + "Syft missed packages:", colors.reset)
-        for package in sorted(list(missing_packages)):
-            rows.append([INDENT, repr(package)])
-        print_rows(rows)
-        print()
-
-    if missing_metadata:
-        rows = []
-        print(colors.bold + "Syft mismatched metadata:", colors.reset)
-        for inline_metadata_pair in sorted(list(missing_metadata)):
-            pkg, metadata = inline_metadata_pair
-            if pkg in syft_metadata[pkg.type]:
-                syft_metadata_item = syft_metadata[pkg.type][pkg]
-            else:
-                syft_metadata_item = "--- MISSING ---"
-            rows.append([INDENT, "for:", repr(pkg), ":", repr(syft_metadata_item), "!=", repr(metadata)])
-        print_rows(rows)
-        print()
-
-    print(colors.bold+"Summary:", colors.reset)
-    print("   Image: %s" % image)
-    print("   Inline Packages : %d" % len(inline_packages))
-    print("   Syft Packages   : %d" % len(syft_packages))
-    print("         (extra)   : %d" % len(bonus_packages))
-    print("       (missing)   : %d" % len(missing_packages))
-    print(
-        "   Baseline Packages Matched: %2.3f %% (%d/%d packages)"
-        % (percent_overlap_packages, len(same_packages), len(inline_packages))
-    )
-    print(
-        "   Baseline Metadata Matched: %2.3f %% (%d/%d metadata)"
-        % (percent_overlap_metadata, len(same_metadata), len(inline_metadata_set))
+    # analyze the raw data to generate all derivative data for the report and quality gate
+    analysis = utils.package.Analysis(
+        syft_data=syft.packages(), inline_data=inline.packages()
     )
 
-    overall_score = (percent_overlap_packages + percent_overlap_metadata) / 2.0
+    # show some useful report data for debugging / warm fuzzies
+    report(image, analysis)
 
-    print(colors.bold + "   Overall Score: %2.1f %%" % overall_score, colors.reset)
+    # enforce a quality gate based on the comparison of package values and metadata values
+    success = True
+    success &= enforce_quality_gate(
+        title="Package",
+        actual_value=analysis.percent_overlapping_packages,
+        lower_gate_value=PACKAGE_QUALITY_GATE[image] * 100,
+        upper_gate_value=PACKAGE_UPPER_THRESHOLD[image] * 100
+    )
+    success &= enforce_quality_gate(
+        title="Metadata",
+        actual_value=analysis.percent_overlapping_metadata,
+        lower_gate_value=METADATA_QUALITY_GATE[image] * 100,
+        upper_gate_value=METADATA_UPPER_THRESHOLD[image] * 100
+    )
 
-    upper_gate_value = IMAGE_UPPER_THRESHOLD[image] * 100
-    lower_gate_value = IMAGE_QUALITY_GATE[image] * 100
-    if overall_score < lower_gate_value:
-        print(colors.bold + "   Quality Gate: " + colors.fg.red + "FAILED (is not >= %d %%)\n" % lower_gate_value, colors.reset)
+    if not success:
         return 1
-    elif overall_score > upper_gate_value:
-        print(colors.bold + "   Quality Gate: " + colors.fg.orange + "FAILED (lower threshold is artificially low and should be updated)\n", colors.reset)
-        return 1
-    else:
-        print(colors.bold + "   Quality Gate: " + colors.fg.green + "pass (>= %d %%)\n" % lower_gate_value, colors.reset)
-
     return 0
 
-
-class colors:
-    reset='\033[0m'
-    bold='\033[01m'
-    disable='\033[02m'
-    underline='\033[04m'
-    reverse='\033[07m'
-    strikethrough='\033[09m'
-    invisible='\033[08m'
-    class fg:
-        black='\033[30m'
-        red='\033[31m'
-        green='\033[32m'
-        orange='\033[33m'
-        blue='\033[34m'
-        purple='\033[35m'
-        cyan='\033[36m'
-        lightgrey='\033[37m'
-        darkgrey='\033[90m'
-        lightred='\033[91m'
-        lightgreen='\033[92m'
-        yellow='\033[93m'
-        lightblue='\033[94m'
-        pink='\033[95m'
-        lightcyan='\033[96m'
-    class bg:
-        black='\033[40m'
-        red='\033[41m'
-        green='\033[42m'
-        orange='\033[43m'
-        blue='\033[44m'
-        purple='\033[45m'
-        cyan='\033[46m'
-        lightgrey='\033[47m'
-
-
 if __name__ == "__main__":
     if len(sys.argv) != 2:
         sys.exit("provide an image")

test/inline-compare/utils/format.py

@@ -0,0 +1,46 @@
+class Colors:
+    reset = "\033[0m"
+    bold = "\033[01m"
+    disable = "\033[02m"
+    underline = "\033[04m"
+    reverse = "\033[07m"
+    strikethrough = "\033[09m"
+    invisible = "\033[08m"
+
+    class FG:
+        black = "\033[30m"
+        red = "\033[31m"
+        green = "\033[32m"
+        orange = "\033[33m"
+        blue = "\033[34m"
+        purple = "\033[35m"
+        cyan = "\033[36m"
+        lightgrey = "\033[37m"
+        darkgrey = "\033[90m"
+        lightred = "\033[91m"
+        lightgreen = "\033[92m"
+        yellow = "\033[93m"
+        lightblue = "\033[94m"
+        pink = "\033[95m"
+        lightcyan = "\033[96m"
+
+    class BG:
+        black = "\033[40m"
+        red = "\033[41m"
+        green = "\033[42m"
+        orange = "\033[43m"
+        blue = "\033[44m"
+        purple = "\033[45m"
+        cyan = "\033[46m"
+        lightgrey = "\033[47m"
+
+
+def print_rows(rows):
+    if not rows:
+        return
+    widths = []
+    for col, _ in enumerate(rows[0]):
+        width = max(len(row[col]) for row in rows) + 2  # padding
+        widths.append(width)
+    for row in rows:
+        print("".join(word.ljust(widths[col_idx]) for col_idx, word in enumerate(row)))