Fix parsing of apk databases with large entries (#1365)

Closes https://github.com/anchore/syft/issues/1354

syft/pkg/cataloger/apkdb/parse_apk_db.go

@@ -3,13 +3,10 @@ package apkdb
 import (
 	"bufio"
 	"fmt"
-	"io"
 	"path"
 	"strconv"
 	"strings"
 
-	"github.com/mitchellh/mapstructure"
-
 	"github.com/anchore/syft/internal/log"
 	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/file"
@@ -22,28 +19,73 @@ import (
 // integrity check
 var _ generic.Parser = parseApkDB
 
-// parseApkDb parses individual packages from a given Alpine DB file. For more information on specific fields
+// parseApkDB parses packages from a given APK installed DB file. For more
-// see https://wiki.alpinelinux.org/wiki/Apk_spec .
+// information on specific fields, see https://wiki.alpinelinux.org/wiki/Apk_spec.
+//
+//nolint:funlen
 func parseApkDB(_ source.FileResolver, env *generic.Environment, reader source.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
-	// larger capacity for the scanner.
-	const maxScannerCapacity = 1024 * 1024
-	// a new larger buffer for the scanner
-	bufScan := make([]byte, maxScannerCapacity)
-	pkgs := make([]pkg.Package, 0)
-
 	scanner := bufio.NewScanner(reader)
-	scanner.Buffer(bufScan, maxScannerCapacity)
+
-	onDoubleLF := func(data []byte, atEOF bool) (advance int, token []byte, err error) {
+	var apks []pkg.ApkMetadata
-		for i := 0; i < len(data); i++ {
+	var currentEntry pkg.ApkMetadata
-			if i > 0 && data[i-1] == '\n' && data[i] == '\n' {
+	entryParsingInProgress := false
-				return i + 1, data[:i-1], nil
+	fileParsingCtx := newApkFileParsingContext()
+
+	// creating a dedicated append-like function here instead of using `append(...)`
+	// below since there is nontrivial logic to be performed for each finalized apk
+	// entry.
+	appendApk := func(p pkg.ApkMetadata) {
+		if files := fileParsingCtx.files; len(files) >= 1 {
+			// attached accumulated files to current package
+			p.Files = files
+
+			// reset file parsing for next use
+			fileParsingCtx = newApkFileParsingContext()
+		}
+
+		nilFieldsToEmptySlice(&p)
+		apks = append(apks, p)
+	}
+
+	for scanner.Scan() {
+		line := scanner.Text()
+
+		if line == "" {
+			// i.e. apk entry separator
+
+			if entryParsingInProgress {
+				// current entry is complete
+				appendApk(currentEntry)
 			}
+
+			entryParsingInProgress = false
+
+			// zero-out currentEntry for use by any future entry
+			currentEntry = pkg.ApkMetadata{}
+
+			continue
 		}
-		if !atEOF {
+
-			return 0, nil, nil
+		field := parseApkField(line)
+		if field == nil {
+			log.Warnf("unable to parse field data from line %q", line)
+			continue
 		}
-		// deliver the last token (which could be an empty string)
+
-		return 0, data, bufio.ErrFinalToken
+		entryParsingInProgress = true
+
+		field.apply(&currentEntry, fileParsingCtx)
+	}
+
+	if entryParsingInProgress {
+		// There was no final empty line, so currentEntry hasn't been added to the
+		// collection yet; but we've now reached the end of scanning, so let's be sure to
+		// add currentEntry to the collection.
+		appendApk(currentEntry)
+	}
+
+	if err := scanner.Err(); err != nil {
+		return nil, nil, fmt.Errorf("failed to parse APK installed DB file: %w", err)
 	}
 
 	var r *linux.Release
@@ -51,128 +93,192 @@ func parseApkDB(_ source.FileResolver, env *generic.Environment, reader source.L
 		r = env.LinuxRelease
 	}
 
-	scanner.Split(onDoubleLF)
+	pkgs := make([]pkg.Package, 0, len(apks))
-	for scanner.Scan() {
+	for _, apk := range apks {
-		metadata, err := parseApkDBEntry(strings.NewReader(scanner.Text()))
+		pkgs = append(pkgs, newPackage(apk, r, reader.Location))
-		if err != nil {
-			return nil, nil, err
-		}
-		if metadata != nil {
-			pkgs = append(pkgs, newPackage(*metadata, r, reader.Location))
-		}
-	}
-
-	if err := scanner.Err(); err != nil {
-		return nil, nil, fmt.Errorf("failed to parse APK DB file: %w", err)
 	}
 
 	return pkgs, discoverPackageDependencies(pkgs), nil
 }
 
-// parseApkDBEntry reads and parses a single pkg.ApkMetadata element from the stream, returning nil if their are no more entries.
+func parseApkField(line string) *apkField {
-//
+	parts := strings.SplitN(line, ":", 2)
+	if len(parts) != 2 {
+		return nil
+	}
+
+	f := apkField{
+		name:  parts[0],
+		value: parts[1],
+	}
+
+	return &f
+}
+
+type apkField struct {
+	name  string
+	value string
+}
+
 //nolint:funlen
-func parseApkDBEntry(reader io.Reader) (*pkg.ApkMetadata, error) {
+func (f apkField) apply(p *pkg.ApkMetadata, ctx *apkFileParsingContext) {
-	var entry pkg.ApkMetadata
+	switch f.name {
-	pkgFields := make(map[string]interface{})
+	// APKINDEX field parsing
 
-	// We want sane defaults for collections, i.e. an empty array instead of null.
+	case "P":
-	pkgFields["D"] = []string{}
+		p.Package = f.value
-	pkgFields["p"] = []string{}
+	case "o":
-	files := make([]pkg.ApkFileRecord, 0)
+		p.OriginPackage = f.value
-
+	case "m":
-	var fileRecord *pkg.ApkFileRecord
+		p.Maintainer = f.value
-	lastFile := "/"
+	case "V":
-
+		p.Version = f.value
-	scanner := bufio.NewScanner(reader)
+	case "L":
-	for scanner.Scan() {
+		p.License = f.value
-		line := scanner.Text()
+	case "A":
-		fields := strings.SplitN(line, ":", 2)
+		p.Architecture = f.value
-		if len(fields) != 2 {
+	case "U":
-			continue
+		p.URL = f.value
+	case "T":
+		p.Description = f.value
+	case "S":
+		i, err := strconv.Atoi(f.value)
+		if err != nil {
+			log.Warnf("unable to parse value %q for field %q: %w", f.value, f.name, err)
+			return
 		}
 
-		key := fields[0]
+		p.Size = i
-		value := strings.TrimSpace(fields[1])
+	case "I":
-
+		i, err := strconv.Atoi(f.value)
-		switch key {
+		if err != nil {
-		case "D", "p":
+			log.Warnf("unable to parse value %q for field %q: %w", f.value, f.name, err)
-			entries := strings.Split(value, " ")
+			return
-			pkgFields[key] = entries
-		case "F":
-			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{
-				Path: path.Join(lastFile, value),
-			}
-			files = append(files, newFileRecord)
-			fileRecord = &files[len(files)-1]
-		case "a", "M":
-			ownershipFields := strings.Split(value, ":")
-			if len(ownershipFields) < 3 {
-				log.Warnf("unexpected APK ownership field: %q", value)
-				continue
-			}
-			if fileRecord == nil {
-				log.Warnf("ownership field with no parent record: %q", value)
-				continue
-			}
-			fileRecord.OwnerUID = ownershipFields[0]
-			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="
-		case "Z":
-			if fileRecord == nil {
-				log.Warnf("checksum field with no parent record: %q", value)
-				continue
-			}
-			fileRecord.Digest = processChecksum(value)
-		case "I", "S":
-			// coerce to integer
-			iVal, err := strconv.Atoi(value)
-			if err != nil {
-				return nil, fmt.Errorf("failed to parse APK int: '%+v'", value)
-			}
-			pkgFields[key] = iVal
-		default:
-			pkgFields[key] = value
 		}
+
+		p.InstalledSize = i
+	case "D":
+		deps := parseListValue(f.value)
+		p.Dependencies = deps
+	case "p":
+		provides := parseListValue(f.value)
+		p.Provides = provides
+	case "C":
+		p.Checksum = f.value
+	case "c":
+		p.GitCommit = f.value
+
+	// File/directory field parsing:
+
+	case "F":
+		directory := path.Join("/", f.value)
+
+		ctx.files = append(ctx.files, pkg.ApkFileRecord{Path: directory})
+		ctx.indexOfLatestDirectory = len(ctx.files) - 1
+	case "M":
+		i := ctx.indexOfLatestDirectory
+		latest := ctx.files[i]
+
+		var ok bool
+		latest.OwnerUID, latest.OwnerGID, latest.Permissions, ok = processFileInfo(f.value)
+		if !ok {
+			log.Warnf("unexpected value for APK ACL field %q: %q", f.name, f.value)
+			return
+		}
+
+		// save updated directory
+		ctx.files[i] = latest
+	case "R":
+		var regularFile string
+
+		dirIndex := ctx.indexOfLatestDirectory
+		if dirIndex < 0 {
+			regularFile = path.Join("/", f.value)
+		} else {
+			latestDirPath := ctx.files[dirIndex].Path
+			regularFile = path.Join(latestDirPath, f.value)
+		}
+
+		ctx.files = append(ctx.files, pkg.ApkFileRecord{Path: regularFile})
+		ctx.indexOfLatestRegularFile = len(ctx.files) - 1
+	case "a":
+		i := ctx.indexOfLatestRegularFile
+		latest := ctx.files[i]
+
+		var ok bool
+		latest.OwnerUID, latest.OwnerGID, latest.Permissions, ok = processFileInfo(f.value)
+		if !ok {
+			log.Warnf("unexpected value for APK ACL field %q: %q", f.name, f.value)
+			return
+		}
+
+		// save updated file
+		ctx.files[i] = latest
+	case "Z":
+		i := ctx.indexOfLatestRegularFile
+		latest := ctx.files[i]
+		latest.Digest = processChecksum(f.value)
+
+		// save updated file
+		ctx.files[i] = latest
+	}
+}
+
+func processFileInfo(v string) (uid, gid, perms string, ok bool) {
+	ok = false
+
+	fileInfo := strings.Split(v, ":")
+	if len(fileInfo) < 3 {
+		return
 	}
 
-	decoder, err := mapstructure.NewDecoder(&mapstructure.DecoderConfig{
+	uid = fileInfo[0]
-		// By default, mapstructure compares field names in a *case-insensitive* manner.
+	gid = fileInfo[1]
-		// That would be the wrong approach here, since these apk files use case
+	perms = fileInfo[2]
-		// *sensitive* field names (e.g. 'P' vs. 'p').
+
-		MatchName: func(mapKey, fieldName string) bool {
+	// note: there are more optional fields available that we are not capturing,
-			return mapKey == fieldName
+	// e.g.: "0:0:755:Q1JaDEHQHBbizhEzoWK1YxuraNU/4="
-		},
+
-		Result: &entry,
+	ok = true
-	})
+	return
-	if err != nil {
+}
-		return nil, err
+
+// apkFileParsingContext helps keep track of what file data has been captured so far for the APK currently being parsed.
+type apkFileParsingContext struct {
+	files                    []pkg.ApkFileRecord
+	indexOfLatestDirectory   int
+	indexOfLatestRegularFile int
+}
+
+func newApkFileParsingContext() *apkFileParsingContext {
+	return &apkFileParsingContext{
+		indexOfLatestDirectory:   -1, // no directories yet
+		indexOfLatestRegularFile: -1, // no regular files yet
+	}
+}
+
+// parseListValue parses a space-separated list from an apk entry field value.
+func parseListValue(value string) []string {
+	items := strings.Split(value, " ")
+	if len(items) >= 1 {
+		return items
 	}
 
-	if err := decoder.Decode(pkgFields); err != nil {
+	return nil
-		return nil, fmt.Errorf("unable to parse APK metadata: %w", err)
+}
-	}
+
-	if entry.Package == "" {
+func nilFieldsToEmptySlice(p *pkg.ApkMetadata) {
-		return nil, nil
+	if p.Dependencies == nil {
+		p.Dependencies = []string{}
 	}
 
-	entry.Files = files
+	if p.Provides == nil {
+		p.Provides = []string{}
+	}
 
-	return &entry, nil
+	if p.Files == nil {
+		p.Files = []pkg.ApkFileRecord{}
+	}
 }
 
 func processChecksum(value string) *file.Digest {

syft/pkg/cataloger/apkdb/parse_apk_db_test.go

@@ -1,8 +1,8 @@
 package apkdb
 
 import (
-	"bufio"
 	"os"
+	"path/filepath"
 	"testing"
 
 	"github.com/google/go-cmp/cmp"
@@ -60,16 +60,15 @@ func TestExtraFileAttributes(t *testing.T) {
 
 	for _, test := range tests {
 		t.Run(test.name, func(t *testing.T) {
-			f, err := os.Open("test-fixtures/extra-file-attributes")
+			fixturePath := "test-fixtures/extra-file-attributes"
-			require.NoError(t, err)
+			lrc := newLocationReadCloser(t, fixturePath)
-			t.Cleanup(func() { require.NoError(t, f.Close()) })
 
-			reader := bufio.NewReader(f)
+			pkgs, _, err := parseApkDB(nil, new(generic.Environment), lrc)
+			assert.NoError(t, err)
+			require.Len(t, pkgs, 1)
+			metadata := pkgs[0].Metadata.(pkg.ApkMetadata)
 
-			entry, err := parseApkDBEntry(reader)
+			if diff := cmp.Diff(test.expected.Files, metadata.Files); diff != "" {
-			require.NoError(t, err)
-
-			if diff := cmp.Diff(entry.Files, test.expected.Files); diff != "" {
 				t.Errorf("Files mismatch (-want +got):\n%s", diff)
 			}
 		})
@@ -615,17 +614,14 @@ func TestSinglePackageDetails(t *testing.T) {
 
 	for _, test := range tests {
 		t.Run(test.fixture, func(t *testing.T) {
-			f, err := os.Open(test.fixture)
+			lrc := newLocationReadCloser(t, test.fixture)
+
+			pkgs, _, err := parseApkDB(nil, new(generic.Environment), lrc)
 			require.NoError(t, err)
-			t.Cleanup(func() { require.NoError(t, f.Close()) })
+			require.Len(t, pkgs, 1)
+			metadata := pkgs[0].Metadata.(pkg.ApkMetadata)
 
-			reader := bufio.NewReader(f)
+			if diff := cmp.Diff(test.expected, metadata); diff != "" {
-
-			entry, err := parseApkDBEntry(reader)
-			require.NoError(t, err)
-			require.NotNil(t, entry)
-
-			if diff := cmp.Diff(*entry, test.expected); diff != "" {
 				t.Errorf("Entry mismatch (-want +got):\n%s", diff)
 			}
 		})
@@ -765,7 +761,6 @@ func TestMultiplePackages(t *testing.T) {
 	}}
 
 	pkgtest.TestFileParserWithEnv(t, fixture, parseApkDB, &env, expectedPkgs, expectedRelationships)
-
 }
 
 func Test_processChecksum(t *testing.T) {
@@ -791,15 +786,15 @@ func Test_processChecksum(t *testing.T) {
 			},
 		},
 	}
-	for _, tt := range tests {
+
-		t.Run(tt.name, func(t *testing.T) {
+	for _, test := range tests {
-			assert.Equal(t, &tt.want, processChecksum(tt.value))
+		t.Run(test.name, func(t *testing.T) {
+			assert.Equal(t, &test.want, processChecksum(test.value))
 		})
 	}
 }
 
 func Test_discoverPackageDependencies(t *testing.T) {
-
 	tests := []struct {
 		name  string
 		genFn func() ([]pkg.Package, []artifact.Relationship)
@@ -934,9 +929,10 @@ func Test_discoverPackageDependencies(t *testing.T) {
 			},
 		},
 	}
-	for _, tt := range tests {
+
-		t.Run(tt.name, func(t *testing.T) {
+	for _, test := range tests {
-			pkgs, wantRelationships := tt.genFn()
+		t.Run(test.name, func(t *testing.T) {
+			pkgs, wantRelationships := test.genFn()
 			gotRelationships := discoverPackageDependencies(pkgs)
 			d := cmp.Diff(wantRelationships, gotRelationships, cmpopts.IgnoreUnexported(pkg.Package{}, source.LocationSet{}))
 			if d != "" {
@@ -1007,3 +1003,89 @@ func TestPackageDbDependenciesByParse(t *testing.T) {
 		})
 	}
 }
+
+func Test_parseApkDB_expectedPkgNames(t *testing.T) {
+	tests := []struct {
+		fixture      string
+		wantPkgNames []string
+		wantErr      assert.ErrorAssertionFunc
+	}{
+		{
+			fixture: "very-large-entries",
+			wantPkgNames: []string{
+				"ca-certificates-bundle",
+				"glibc-locale-posix",
+				"wolfi-baselayout",
+				"glibc",
+				"libcrypto3",
+				"libssl3",
+				"zlib",
+				"apk-tools",
+				"ncurses-terminfo-base",
+				"ncurses",
+				"bash",
+				"libcap",
+				"bubblewrap",
+				"busybox",
+				"libbrotlicommon1",
+				"libbrotlidec1",
+				"libnghttp2-14",
+				"libcurl4",
+				"curl",
+				"expat",
+				"libpcre2-8-0",
+				"git",
+				"binutils",
+				"libstdc++-dev",
+				"libgcc",
+				"libstdc++",
+				"gmp",
+				"isl",
+				"mpfr",
+				"mpc",
+				"gcc",
+				"linux-headers",
+				"glibc-dev",
+				"make",
+				"pkgconf",
+				"build-base",
+				"go",
+				"tree",
+				"sdk",
+			},
+			wantErr: assert.NoError,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.fixture, func(t *testing.T) {
+			fixturePath := filepath.Join("test-fixtures", test.fixture)
+			lrc := newLocationReadCloser(t, fixturePath)
+
+			pkgs, _, err := parseApkDB(nil, new(generic.Environment), lrc)
+			test.wantErr(t, err)
+
+			names := toPackageNames(pkgs)
+			if diff := cmp.Diff(test.wantPkgNames, names); diff != "" {
+				t.Errorf("Packages mismatch (-want +got):\n%s", diff)
+			}
+		})
+	}
+}
+
+func toPackageNames(pkgs []pkg.Package) []string {
+	names := make([]string, 0, len(pkgs))
+	for _, p := range pkgs {
+		names = append(names, p.Name)
+	}
+
+	return names
+}
+
+func newLocationReadCloser(t *testing.T, path string) source.LocationReadCloser {
+	f, err := os.Open(path)
+	require.NoError(t, err)
+	t.Cleanup(func() { f.Close() })
+
+	return source.NewLocationReadCloser(source.NewLocation(path), f)
+}