refactor: move apk upstream logic to apk metadata (#1619)

* refactor: move apk upstream logic to apk metadata

Export the logic for parsing upstream APK package names
so it can be accessed from apk metadata objects directly.

This also tightens the upstream regex pattern as several
edge cases were being missed.

Signed-off-by: Weston Steimel <weston.steimel@anchore.com>

* fix: ensure correct handling for apk packages beginning with digits

Signed-off-by: Weston Steimel <weston.steimel@anchore.com>

* fix: upstream generation for ruby

Signed-off-by: Weston Steimel <weston.steimel@anchore.com>

---------

Signed-off-by: Weston Steimel <weston.steimel@anchore.com>
This commit is contained in:
Weston Steimel 2023-02-24 15:59:19 +00:00 committed by GitHub
parent 5e8aa4da5e
commit 0c5f03235e
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 226 additions and 31 deletions

View File

@ -4,18 +4,24 @@ import (
"encoding/json" "encoding/json"
"fmt" "fmt"
"reflect" "reflect"
"regexp"
"sort" "sort"
"strings" "strings"
"github.com/mitchellh/mapstructure" "github.com/mitchellh/mapstructure"
"github.com/scylladb/go-set/strset" "github.com/scylladb/go-set/strset"
"github.com/anchore/syft/internal"
"github.com/anchore/syft/syft/file" "github.com/anchore/syft/syft/file"
) )
const ApkDBGlob = "**/lib/apk/db/installed" const ApkDBGlob = "**/lib/apk/db/installed"
var _ FileOwner = (*ApkMetadata)(nil) var (
_ FileOwner = (*ApkMetadata)(nil)
prefixes = []string{"py-", "py2-", "py3-", "ruby-"}
upstreamPattern = regexp.MustCompile(`^(?P<upstream>[a-zA-Z][\w-]*?)\-?\d[\d\.]*$`)
)
// ApkMetadata represents all captured data for a Alpine DB package entry. // ApkMetadata represents all captured data for a Alpine DB package entry.
// See the following sources for more information: // See the following sources for more information:
@ -114,3 +120,24 @@ func (m ApkMetadata) OwnedFiles() (result []string) {
sort.Strings(result) sort.Strings(result)
return result return result
} }
func (m ApkMetadata) Upstream() string {
if m.OriginPackage != "" && m.OriginPackage != m.Package {
return m.OriginPackage
}
groups := internal.MatchNamedCaptureGroups(upstreamPattern, m.Package)
upstream, ok := groups["upstream"]
if !ok {
upstream = m.Package
}
for _, p := range prefixes {
if strings.HasPrefix(upstream, p) {
return strings.TrimPrefix(upstream, p)
}
}
return upstream
}

View File

@ -4,6 +4,7 @@ import (
"encoding/json" "encoding/json"
"testing" "testing"
"github.com/sergi/go-diff/diffmatchpatch"
"github.com/stretchr/testify/assert" "github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require" "github.com/stretchr/testify/require"
) )
@ -163,3 +164,169 @@ func TestSpaceDelimitedStringSlice_UnmarshalJSON(t *testing.T) {
}) })
} }
} }
func TestApkMetadata_Upstream(t *testing.T) {
tests := []struct {
name string
metadata ApkMetadata
expected string
}{
{
name: "gocase",
metadata: ApkMetadata{
Package: "p",
},
expected: "p",
},
{
name: "same package and origin",
metadata: ApkMetadata{
Package: "p",
OriginPackage: "p",
},
expected: "p",
},
{
name: "different package and origin",
metadata: ApkMetadata{
Package: "p",
OriginPackage: "origin",
},
expected: "origin",
},
{
name: "upstream python package information as qualifier",
metadata: ApkMetadata{
Package: "py3-potatoes",
OriginPackage: "py3-potatoes",
},
expected: "potatoes",
},
{
name: "python package with distinct origin package",
metadata: ApkMetadata{
Package: "py3-non-existant",
OriginPackage: "abcdefg",
},
expected: "abcdefg",
},
{
name: "upstream ruby package information as qualifier",
metadata: ApkMetadata{
Package: "ruby-something",
OriginPackage: "ruby-something",
},
expected: "something",
},
{
name: "python package with distinct origin package",
metadata: ApkMetadata{
Package: "ruby-something",
OriginPackage: "1234567",
},
expected: "1234567",
},
{
name: "postgesql-15 upstream postgresql",
metadata: ApkMetadata{
Package: "postgresql-15",
},
expected: "postgresql",
},
{
name: "postgesql15 upstream postgresql",
metadata: ApkMetadata{
Package: "postgresql15",
},
expected: "postgresql",
},
{
name: "go-1.19 upstream go",
metadata: ApkMetadata{
Package: "go-1.19",
},
expected: "go",
},
{
name: "go1.143 upstream go",
metadata: ApkMetadata{
Package: "go1.143",
},
expected: "go",
},
{
name: "abc-101.191.23456 upstream abc",
metadata: ApkMetadata{
Package: "abc-101.191.23456",
},
expected: "abc",
},
{
name: "abc101.191.23456 upstream abc",
metadata: ApkMetadata{
Package: "abc101.191.23456",
},
expected: "abc",
},
{
name: "abc101-12345-1045 upstream abc101-12345",
metadata: ApkMetadata{
Package: "abc101-12345-1045",
},
expected: "abc101-12345",
},
{
name: "abc101-a12345-1045 upstream abc101-a12345",
metadata: ApkMetadata{
Package: "abc101-a12345-1045",
},
expected: "abc101-a12345",
},
{
name: "package starting with single digit",
metadata: ApkMetadata{
Package: "3proxy",
},
expected: "3proxy",
},
{
name: "package starting with multiple digits",
metadata: ApkMetadata{
Package: "356proxy",
},
expected: "356proxy",
},
{
name: "package composed of only digits",
metadata: ApkMetadata{
Package: "123456",
},
expected: "123456",
},
{
name: "ruby-3.6 upstream ruby",
metadata: ApkMetadata{
Package: "ruby-3.6",
},
expected: "ruby",
},
{
name: "ruby3.6 upstream ruby",
metadata: ApkMetadata{
Package: "ruby3.6",
},
expected: "ruby",
},
}
for _, test := range tests {
t.Run(test.name, func(t *testing.T) {
actual := test.metadata.Upstream()
if actual != test.expected {
dmp := diffmatchpatch.New()
diffs := dmp.DiffMain(test.expected, actual, true)
t.Errorf("diff: %s", dmp.DiffPrettyText(diffs))
}
})
}
}

View File

@ -1,20 +1,14 @@
package apkdb package apkdb
import ( import (
"regexp"
"strings" "strings"
"github.com/anchore/packageurl-go" "github.com/anchore/packageurl-go"
"github.com/anchore/syft/internal"
"github.com/anchore/syft/syft/linux" "github.com/anchore/syft/syft/linux"
"github.com/anchore/syft/syft/pkg" "github.com/anchore/syft/syft/pkg"
"github.com/anchore/syft/syft/source" "github.com/anchore/syft/syft/source"
) )
var (
prefixes = []string{"py-", "py2-", "py3-", "ruby-"}
)
func newPackage(d pkg.ApkMetadata, release *linux.Release, locations ...source.Location) pkg.Package { func newPackage(d pkg.ApkMetadata, release *linux.Release, locations ...source.Location) pkg.Package {
p := pkg.Package{ p := pkg.Package{
Name: d.Package, Name: d.Package,
@ -32,28 +26,6 @@ func newPackage(d pkg.ApkMetadata, release *linux.Release, locations ...source.L
return p return p
} }
func generateUpstream(m pkg.ApkMetadata) string {
if m.OriginPackage != "" && m.OriginPackage != m.Package {
return m.OriginPackage
}
for _, p := range prefixes {
if strings.HasPrefix(m.Package, p) {
return strings.TrimPrefix(m.Package, p)
}
}
pattern := regexp.MustCompile(`(?P<upstream>\w+?)\-?\d[\d\.]*`)
groups := internal.MatchNamedCaptureGroups(pattern, m.Package)
upstream, ok := groups["upstream"]
if ok {
return upstream
}
return m.Package
}
// packageURL returns the PURL for the specific Alpine package (see https://github.com/package-url/purl-spec) // packageURL returns the PURL for the specific Alpine package (see https://github.com/package-url/purl-spec)
func packageURL(m pkg.ApkMetadata, distro *linux.Release) string { func packageURL(m pkg.ApkMetadata, distro *linux.Release) string {
if distro == nil || distro.ID != "alpine" { if distro == nil || distro.ID != "alpine" {
@ -65,8 +37,9 @@ func packageURL(m pkg.ApkMetadata, distro *linux.Release) string {
pkg.PURLQualifierArch: m.Architecture, pkg.PURLQualifierArch: m.Architecture,
} }
if m.OriginPackage != "" { upstream := m.Upstream()
qualifiers[pkg.PURLQualifierUpstream] = generateUpstream(m) if upstream != "" && upstream != m.Package {
qualifiers[pkg.PURLQualifierUpstream] = upstream
} }
return packageurl.NewPackageURL( return packageurl.NewPackageURL(

View File

@ -208,6 +208,34 @@ func Test_PackageURL(t *testing.T) {
}, },
expected: "pkg:apk/alpine/abc101.191.23456@101.191.23456?arch=a&upstream=abc&distro=alpine-3.4.6", expected: "pkg:apk/alpine/abc101.191.23456@101.191.23456?arch=a&upstream=abc&distro=alpine-3.4.6",
}, },
{
name: "abc101-12345-1045 upstream abc101-12345",
metadata: pkg.ApkMetadata{
Package: "abc101-12345-1045",
Version: "101.191.23456",
Architecture: "a",
OriginPackage: "abc101-12345-1045",
},
distro: linux.Release{
ID: "alpine",
VersionID: "3.4.6",
},
expected: "pkg:apk/alpine/abc101-12345-1045@101.191.23456?arch=a&upstream=abc101-12345&distro=alpine-3.4.6",
},
{
name: "abc101-a12345-1045 upstream abc101-a12345",
metadata: pkg.ApkMetadata{
Package: "abc101-a12345-1045",
Version: "101.191.23456",
Architecture: "a",
OriginPackage: "abc101-a12345-1045",
},
distro: linux.Release{
ID: "alpine",
VersionID: "3.4.6",
},
expected: "pkg:apk/alpine/abc101-a12345-1045@101.191.23456?arch=a&upstream=abc101-a12345&distro=alpine-3.4.6",
},
} }
for _, test := range tests { for _, test := range tests {