Add support for scanning IPK packages using the deb-archive-cataloger (#5111)

* Add support for scanning IPK packages using the deb-archive-cataloger

Signed-off-by: Silas Parker <silas@srp.me.uk>

* fix(debian): share License field parsing between opkg status and archive control files

- the control-file `License:` fallback now goes through the same helper as the opkg status DB path, so a package reports the same licenses whether found installed or as an archive
- bitbake `&` / `|` are rewritten to SPDX `AND` / `OR` and the whole value is validated as one expression instead of split by hand, which kept grouping like `(A & B) | C` intact
- non-SPDX values fall back to a deduped list of names with operators and parens dropped (previously the status DB path emitted `&`, `(BSD`, etc as licenses)
- fields over 4KB are ignored, bounding work per archive and keeping deep paren nesting away from the recursive expression parser

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* fix(debian): split Yocto recipe filenames in the Source field

Yocto writes the bitbake recipe filename (`Source: zlib_1.3.2.bb`) where dpkg expects a source package name. This is now split into `Source: zlib` / `SourceVersion: 1.3.2` for both ipk archives and opkg status DB entries, instead of surfacing the `.bb` filename as the upstream package.

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* fix(debian): do not label ipk archives as Debian packages

- `.ipk` archives no longer get a `pkg:deb` PURL from the assumed debian distro; opkg-based distros (Yocto, OpenWrt) have no purl type, so these now match packages found in an opkg status DB, which already get no PURL
- docs now cover `.ipk` and note that only ar-framed ipks are supported (OpenWrt's older gzipped-tar framing is rejected as an invalid archive)

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* test(debian): assemble the ipk fixture at build time instead of committing it

The fixture Dockerfile now builds a Yocto-shaped `.ipk` from a checked-in `control/` directory (copied from the original poky-tiny build) using `tar`, `zstd` and binutils `ar`, the same way bitbake's `package_ipk` packs one. This drops the prebuilt binary and the 20-30 minute Yocto `Dockerfile.build` while keeping the same archive layout and control metadata.

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

---------

Signed-off-by: Silas Parker <silas@srp.me.uk>
Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>
Co-authored-by: Alex Goodman <wagoodman@users.noreply.github.com>

syft/pkg/cataloger/debian/capabilities.yaml

@@ -75,6 +75,7 @@ catalogers:
           method: glob # AUTO-GENERATED
           criteria: # AUTO-GENERATED
             - '**/*.deb'
+            - '**/*.ipk'
         metadata_types: # AUTO-GENERATED
           - pkg.DpkgArchiveEntry
         package_types: # AUTO-GENERATED

syft/pkg/cataloger/debian/cataloger.go

@@ -18,8 +18,8 @@ func NewDBCataloger() pkg.Cataloger {
 		WithProcessors(dependency.Processor(dbEntryDependencySpecifier))
 }
 
-// NewArchiveCataloger returns a new Debian package cataloger object capable of parsing .deb archive files
+// NewArchiveCataloger returns a new Debian package cataloger object capable of parsing .deb and .ipk (opkg) archive files
 func NewArchiveCataloger() pkg.Cataloger {
 	return generic.NewCataloger("deb-archive-cataloger").
-		WithParserByGlobs(parseDebArchive, "**/*.deb")
+		WithParserByGlobs(parseDebArchive, "**/*.deb", "**/*.ipk")
 }

syft/pkg/cataloger/debian/cataloger_test.go

@@ -276,6 +276,36 @@ func TestDpkgArchiveCataloger(t *testing.T) {
 					},
 				},
 			},
+		}, {
+			name: "image-single-ipk",
+			expected: []pkg.Package{
+				{
+					Name:    "libz1",
+					Version: "1.3.2-r0",
+					FoundBy: "deb-archive-cataloger",
+					Locations: file.NewLocationSet(
+						file.NewLocation("/libz1_1.3.2-r0_x86-64-v3.ipk"),
+					),
+					Type: pkg.DebPkg,
+					Metadata: pkg.DpkgArchiveEntry{
+						Package:       "libz1",
+						Source:        "zlib",
+						SourceVersion: "1.3.2",
+						Version:       "1.3.2-r0",
+						Architecture:  "x86-64-v3",
+						Maintainer:    "Poky Maintainers <poky@lists.yoctoproject.org>",
+						InstalledSize: 0,
+						Description: `Zlib Compression Library
+ Zlib is a general-purpose, patent-free, lossless data compression library
+ which is used by many different programs.`,
+						Provides: []string{"zlib"},
+						Depends:  []string{"musl (>= 1.2.6+git0+9fa28ece75)"},
+					},
+					Licenses: pkg.NewLicenseSet(
+						pkg.NewLicenseFromLocationsWithContext(ctx, "Zlib"),
+					),
+				},
+			},
 		},
 	}
 

syft/pkg/cataloger/debian/package.go

@@ -9,6 +9,8 @@ import (
 	"sort"
 	"strings"
 
+	"github.com/scylladb/go-set/strset"
+
 	"github.com/anchore/packageurl-go"
 	"github.com/anchore/syft/internal"
 	"github.com/anchore/syft/internal/log"
@@ -61,32 +63,65 @@ func newDpkgPackage(ctx context.Context, d dpkgExtractedMetadata, dbLocation fil
 	return p
 }
 
+// a declared License field holds a single expression, a few hundred bytes at most in practice. Anything
+// much larger only feeds the expression parser (which recurses on parentheses) attacker-sized input.
+const maxDeclaredLicenseLen = 4096
+
+var (
+	// opkg/ipkg (Yocto via bitbake) declare licenses with "&" for AND and "|" for OR
+	bitbakeLicenseOperators = strings.NewReplacer("&", " AND ", "|", " OR ")
+	tightenParens           = strings.NewReplacer("( ", "(", " )", ")")
+	licenseListSeparators   = strings.NewReplacer("(", " ", ")", " ")
+)
+
 // extractDeclaredLicenses converts a License field from the status DB into a license set. Returns nil
 // for empty input so standard dpkg entries (which never declare License inline) incur no allocation.
-// Mirrors the alpine cataloger's approach: keep the value whole if it parses as a valid SPDX expression,
-// otherwise split on whitespace to handle space-separated lists.
 func extractDeclaredLicenses(ctx context.Context, raw string, dbLocation file.Location) []pkg.License {
+	values := declaredLicenseValues(raw)
+	if len(values) == 0 {
+		return nil
+	}
+	return pkg.NewLicensesFromLocationWithContext(ctx, dbLocation, values...)
+}
+
+// declaredLicenseValues converts an inline License field (opkg status DB or an archive control file) into
+// license values. Bitbake operators are rewritten to SPDX ones and the value is kept whole if it then parses
+// as a valid SPDX expression (mirroring the alpine cataloger), otherwise it is treated as a list of license
+// names with operators and grouping dropped.
+func declaredLicenseValues(raw string) []string {
+	raw = strings.TrimSpace(raw)
 	if raw == "" {
 		return nil
 	}
-	licenseStrings := []string{raw}
+	if len(raw) > maxDeclaredLicenseLen {
-	if _, err := license.ParseExpression(raw); err != nil {
+		log.WithFields("length", len(raw)).Debug("ignoring oversized declared license field")
-		licenseStrings = strings.Fields(raw)
+		return nil
 	}
-	return pkg.NewLicensesFromLocationWithContext(ctx, dbLocation, licenseStrings...)
+
+	expr := tightenParens.Replace(strings.Join(strings.Fields(bitbakeLicenseOperators.Replace(raw)), " "))
+	if _, err := license.ParseExpression(expr); err == nil {
+		return []string{expr}
+	}
+
+	var values []string
+	seen := strset.New()
+	for _, field := range strings.Fields(licenseListSeparators.Replace(expr)) {
+		if field == "AND" || field == "OR" || seen.Has(field) {
+			continue
+		}
+		seen.Add(field)
+		values = append(values, field)
+	}
+	return values
 }
 
 func newDebArchivePackage(ctx context.Context, location file.Location, metadata pkg.DpkgArchiveEntry, licenseStrings []string) pkg.Package {
 	p := pkg.Package{
-		Name:     metadata.Package,
+		Name:      metadata.Package,
-		Version:  metadata.Version,
+		Version:   metadata.Version,
-		Licenses: pkg.NewLicenseSet(pkg.NewLicensesFromValuesWithContext(ctx, licenseStrings...)...),
+		Licenses:  pkg.NewLicenseSet(pkg.NewLicensesFromValuesWithContext(ctx, licenseStrings...)...),
-		Type:     pkg.DebPkg,
+		Type:      pkg.DebPkg,
-		PURL: packageURL(
+		PURL:      packageURL(pkg.DpkgDBEntry(metadata), archiveDistro(location)),
-			pkg.DpkgDBEntry(metadata),
-			// we don't know the distro information, but since this is a deb file then we can reasonably assume it is a debian-based distro
-			&linux.Release{IDLike: []string{"debian"}},
-		),
 		Metadata:  metadata,
 		Locations: file.NewLocationSet(location.WithAnnotation(pkg.EvidenceAnnotationKey, pkg.PrimaryEvidenceAnnotation)),
 	}
@@ -95,6 +130,17 @@ func newDebArchivePackage(ctx context.Context, location file.Location, metadata
 	return p
 }
 
+// archiveDistro guesses the distro an archive was built for. We don't know the distro information, but a .deb
+// can reasonably be assumed to target a debian-based distro. An .ipk targets opkg-based distros (Yocto,
+// OpenWrt) which have no purl type, so it gets no distro and therefore no PURL, the same as packages found in
+// an opkg status DB.
+func archiveDistro(location file.Location) *linux.Release {
+	if strings.EqualFold(path.Ext(location.Path()), ".ipk") {
+		return nil
+	}
+	return &linux.Release{IDLike: []string{"debian"}}
+}
+
 // PackageURL returns the PURL for the specific Debian package (see https://github.com/package-url/purl-spec)
 func packageURL(m pkg.DpkgDBEntry, distro *linux.Release) string {
 	if distro == nil {

syft/pkg/cataloger/debian/package_test.go

@@ -3,6 +3,7 @@ package debian
 import (
 	"context"
 	"sort"
+	"strings"
 	"testing"
 
 	"github.com/google/go-cmp/cmp"
@@ -145,6 +146,36 @@ func Test_extractDeclaredLicenses(t *testing.T) {
 			raw:      "GPL-2.0 BSD-3-Clause",
 			expected: []string{"GPL-2.0", "BSD-3-Clause"},
 		},
+		{
+			name:     "bitbake operators become an SPDX expression",
+			raw:      "MIT & (BSD-3-Clause | Apache-2.0) & ( GPL-2.0-or-later )",
+			expected: []string{"MIT AND (BSD-3-Clause OR Apache-2.0) AND (GPL-2.0-or-later)"},
+		},
+		{
+			name:     "bitbake grouping is preserved rather than split",
+			raw:      "(MIT & BSD-3-Clause) | Apache-2.0",
+			expected: []string{"(MIT AND BSD-3-Clause) OR Apache-2.0"},
+		},
+		{
+			name:     "parenthesized alternatives are preserved",
+			raw:      "(MIT) | (Apache-2.0)",
+			expected: []string{"(MIT) OR (Apache-2.0)"},
+		},
+		{
+			name:     "non-SPDX names drop operators and grouping",
+			raw:      "GPLv2+ & (LGPLv2.1 | BSD) & GPLv2+",
+			expected: []string{"GPLv2+", "LGPLv2.1", "BSD"},
+		},
+		{
+			name:     "oversized field is ignored",
+			raw:      strings.Repeat("MIT & ", maxDeclaredLicenseLen),
+			expected: nil,
+		},
+		{
+			name:     "deeply nested parens are ignored",
+			raw:      strings.Repeat("(", maxDeclaredLicenseLen) + "MIT" + strings.Repeat(")", maxDeclaredLicenseLen),
+			expected: nil,
+		},
 	}
 
 	for _, test := range tests {

syft/pkg/cataloger/debian/parse_deb_archive.go

@@ -2,6 +2,7 @@ package debian
 
 import (
 	"archive/tar"
+	"bytes"
 	"context"
 	"errors"
 	"fmt"
@@ -23,18 +24,27 @@ import (
 	"github.com/anchore/syft/syft/pkg/cataloger/generic"
 )
 
-// parseDebArchive parses a Debian package archive (.deb) file and returns the packages it contains.
+// parseDebArchive parses a Debian package archive (.deb) or opkg package (.ipk) file and returns the packages it
-// A .deb file is an ar archive containing three main files:
+// contains. Both are ar archives containing three main files:
-// - debian-binary: Version of the .deb format (usually "2.0")
+// - debian-binary: Version of the archive format (usually "2.0")
 // - control.tar.gz/xz/zst: Contains package metadata (control file, md5sums, conffiles)
 // - data.tar.gz/xz/zst: Contains the actual files to be installed (not processed by this cataloger)
 //
-// This function extracts and processes the control information to create package metadata.
+// This function extracts and processes the control information to create package metadata. Licenses come
+// from copyright files in the data tar, falling back to the control file's License field (common for .ipk).
+// Only ar-framed .ipk files are supported; the older gzipped-tar framing (e.g. OpenWrt's ipkg-build) is
+// rejected as an invalid archive.
 func parseDebArchive(ctx context.Context, _ file.Resolver, _ *generic.Environment, reader file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
-	arReader := ar.NewReader(reader)
+	validatedReader, err := newValidatedArReader(reader)
+	if err != nil {
+		return nil, nil, fmt.Errorf("invalid ar archive: %w", err)
+	}
+
+	arReader := ar.NewReader(validatedReader)
 
 	var metadata *pkg.DpkgArchiveEntry
 	var licenses []string
+	var ctrlLicenses []string
 	var unknownErr error
 	var sawControl, sawData bool
 	for {
@@ -56,7 +66,7 @@ func parseDebArchive(ctx context.Context, _ file.Resolver, _ *generic.Environmen
 			if err != nil {
 				return nil, nil, unknown.New(reader.Location, fmt.Errorf("failed to decompress control.tar.* file: %w", err))
 			}
-			metadata, err = processControlTar(dcReader)
+			metadata, ctrlLicenses, err = processControlTar(dcReader)
 			switch {
 			case err != nil && metadata == nil:
 				return nil, nil, unknown.New(reader.Location, fmt.Errorf("failed to process control.tar.* file: %w", err))
@@ -87,6 +97,10 @@ func parseDebArchive(ctx context.Context, _ file.Resolver, _ *generic.Environmen
 		return nil, nil, unknown.New(reader.Location, fmt.Errorf("no application found described in .dpkg archive"))
 	}
 
+	if len(licenses) == 0 && len(ctrlLicenses) > 0 {
+		licenses = ctrlLicenses
+	}
+
 	// a partial parse still yields a usable package, so report what went wrong alongside it rather than
 	// dropping either one
 	return []pkg.Package{
@@ -128,10 +142,10 @@ func processDataTar(dcReader io.ReadCloser) ([]string, error) {
 }
 
 // processControlTar always returns whatever metadata it managed to parse, even alongside a non-nil
-// error. A non-nil error with non-nil metadata means the package is usable but incomplete (e.g. a
+// error, along with any licenses declared in the control file's License field. A non-nil error with
-// clipped file listing); nil metadata means nothing usable was found and the caller should treat it
+// non-nil metadata means the package is usable but incomplete (e.g. a clipped file listing); nil
-// as fatal.
+// metadata means nothing usable was found and the caller should treat it as fatal.
-func processControlTar(dcReader io.ReadCloser) (*pkg.DpkgArchiveEntry, error) {
+func processControlTar(dcReader io.ReadCloser) (*pkg.DpkgArchiveEntry, []string, error) {
 	defer internal.CloseAndLogError(dcReader, "")
 
 	tarReader := tar.NewReader(dcReader)
@@ -140,6 +154,7 @@ func processControlTar(dcReader io.ReadCloser) (*pkg.DpkgArchiveEntry, error) {
 	var files []pkg.DpkgFileRecord
 	var confFileRecords []pkg.DpkgFileRecord
 	var listingErr error
+	var licenses []string
 
 	for {
 		header, err := tarReader.Next()
@@ -147,7 +162,7 @@ func processControlTar(dcReader io.ReadCloser) (*pkg.DpkgArchiveEntry, error) {
 			break
 		}
 		if err != nil {
-			return metadata, fmt.Errorf("failed to read control tar: %w", err)
+			return metadata, nil, fmt.Errorf("failed to read control tar: %w", err)
 		}
 
 		switch filepath.Base(header.Name) {
@@ -155,12 +170,13 @@ func processControlTar(dcReader io.ReadCloser) (*pkg.DpkgArchiveEntry, error) {
 			// parseDpkgStatus already streams via bufio.Reader
 			entries, err := parseDpkgStatus(tarReader)
 			if err != nil {
-				return nil, fmt.Errorf("failed to parse control file: %w", err)
+				return nil, nil, fmt.Errorf("failed to parse control file: %w", err)
 			}
 			if len(entries) == 0 {
-				return nil, fmt.Errorf("no package entries found in control file")
+				return nil, nil, fmt.Errorf("no package entries found in control file")
 			}
 			entry := pkg.DpkgArchiveEntry(entries[0].toDpkgEntry())
+			licenses = declaredLicenseValues(entries[0].License)
 			metadata = &entry
 		case "md5sums":
 			// parseDpkgMD5Info streams via bufio.Scanner and reports its own clipping/scan errors
@@ -176,7 +192,7 @@ func processControlTar(dcReader io.ReadCloser) (*pkg.DpkgArchiveEntry, error) {
 	}
 
 	if metadata == nil {
-		return nil, fmt.Errorf("control file not found in archive")
+		return nil, nil, fmt.Errorf("control file not found in archive")
 	}
 
 	if len(confFileRecords) > 0 && len(files) > 0 {
@@ -195,7 +211,22 @@ func processControlTar(dcReader io.ReadCloser) (*pkg.DpkgArchiveEntry, error) {
 
 	// a clipped or unreadable file listing leaves the package usable, so hand both back and let the
 	// caller decide
-	return metadata, listingErr
+	return metadata, licenses, listingErr
+}
+
+// newValidatedArReader checks for the ar global header before handing the stream to the ar reader, which
+// otherwise skips those bytes unchecked and reads garbage headers from non-ar input.
+func newValidatedArReader(reader io.ReadCloser) (io.ReadCloser, error) {
+	prefix := make([]byte, len(ar.GLOBAL_HEADER))
+	if _, err := io.ReadFull(reader, prefix); err != nil {
+		return nil, fmt.Errorf("failed to read ar header: %w", err)
+	}
+
+	if !bytes.Equal(prefix, []byte(ar.GLOBAL_HEADER)) {
+		return nil, fmt.Errorf("expected ar header %q, got %q", string(ar.GLOBAL_HEADER), string(prefix))
+	}
+
+	return io.NopCloser(io.MultiReader(bytes.NewReader(prefix), reader)), nil
 }
 
 // a .deb's member sizes are bounded by the archive, but what they decompress to is not: xz and zstd

syft/pkg/cataloger/debian/parse_deb_archive_test.go

@@ -26,12 +26,13 @@ import (
 func TestProcessControlTar(t *testing.T) {
 	tarBytes := createTestTarWithControlFiles(t)
 
-	metadata, err := processControlTar(io.NopCloser(bytes.NewReader(tarBytes)))
+	metadata, licenses, err := processControlTar(io.NopCloser(bytes.NewReader(tarBytes)))
 
 	require.NoError(t, err)
 	require.NotNil(t, metadata)
 
 	assert.Equal(t, "test-package", metadata.Package)
+	assert.Equal(t, []string{"MIT"}, licenses)
 	assert.Equal(t, "1.0.0", metadata.Version)
 
 	// md5sums should have been parsed into file records
@@ -43,6 +44,14 @@ func TestProcessControlTar(t *testing.T) {
 	assert.True(t, metadata.Files[0].IsConfigFile, "file listed in conffiles should be marked as config")
 }
 
+func TestParseDebArchive_RejectsInvalidArchiveHeader(t *testing.T) {
+	reader := file.NewLocationReadCloser(file.NewLocation("test.deb"), io.NopCloser(bytes.NewReader([]byte("not-an-ar-archive"))))
+
+	_, _, err := parseDebArchive(context.Background(), nil, nil, reader)
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "ar archive")
+}
+
 func TestProcessControlTar_ConfigFileMarking(t *testing.T) {
 	// Create a tar where conffiles lists paths that overlap with md5sums entries
 	var buf bytes.Buffer
@@ -61,9 +70,10 @@ func TestProcessControlTar_ConfigFileMarking(t *testing.T) {
 
 	require.NoError(t, tw.Close())
 
-	metadata, err := processControlTar(io.NopCloser(bytes.NewReader(buf.Bytes())))
+	metadata, licenses, err := processControlTar(io.NopCloser(bytes.NewReader(buf.Bytes())))
 	require.NoError(t, err)
 	require.Len(t, metadata.Files, 3)
+	assert.Empty(t, licenses)
 
 	assert.True(t, metadata.Files[0].IsConfigFile, "first file should be marked as config file")
 	assert.True(t, metadata.Files[1].IsConfigFile, "second file should be marked as config file")
@@ -75,7 +85,7 @@ func createTestTarWithControlFiles(t *testing.T) []byte {
 	var buf bytes.Buffer
 	tw := tar.NewWriter(&buf)
 
-	controlContent := "Package: test-package\nVersion: 1.0.0\nArchitecture: all\nMaintainer: Test <test@example.com>\nDescription: Test package\n"
+	controlContent := "Package: test-package\nVersion: 1.0.0\nArchitecture: all\nMaintainer: Test <test@example.com>\nDescription: Test package\nLicense: MIT\n"
 	writeTarEntry(t, tw, "control", controlContent)
 
 	md5Content := "d41d8cd98f00b204e9800998ecf8427e  usr/bin/test-command\n"
@@ -165,7 +175,7 @@ func Test_processControlTar_rejectsABombedControlMember(t *testing.T) {
 	writeTarEntry(t, tw, "md5sums", strings.Repeat("d41d8cd98f00b204e9800998ecf8427e  usr/bin/x\n", 20000))
 	require.NoError(t, tw.Close())
 
-	metadata, err := processControlTar(newBoundedReadCloser(io.NopCloser(bytes.NewReader(tarBuf.Bytes())), 4096))
+	metadata, _, err := processControlTar(newBoundedReadCloser(io.NopCloser(bytes.NewReader(tarBuf.Bytes())), 4096))
 
 	require.ErrorIs(t, err, errDecompressedTooLarge)
 	require.NotNil(t, metadata, "a control tar read failure after \"control\" was parsed must not drop the package")
@@ -181,7 +191,7 @@ func Test_processControlTar_clippedListingIsUsableButReported(t *testing.T) {
 	writeTarEntry(t, tw, "md5sums", strings.Repeat("d41d8cd98f00b204e9800998ecf8427e  usr/bin/x\n", maxDpkgFileRecords+10))
 	require.NoError(t, tw.Close())
 
-	metadata, err := processControlTar(io.NopCloser(bytes.NewReader(tarBuf.Bytes())))
+	metadata, _, err := processControlTar(io.NopCloser(bytes.NewReader(tarBuf.Bytes())))
 
 	require.Error(t, err, "clipping must be reported")
 	require.ErrorIs(t, err, errClippedFileListing)
@@ -418,3 +428,18 @@ func Test_decompressionStream_failsEarlyInsteadOfBuffering(t *testing.T) {
 	assert.Less(t, bounded, uint64(8*intFile.MB),
 		"a bounded stream must stop at the cap, not buffer the whole member and check afterwards")
 }
+
+func TestProcessControlTar_DeclaredLicense(t *testing.T) {
+	var buf bytes.Buffer
+	tw := tar.NewWriter(&buf)
+
+	controlContent := "Package: test-package\nVersion: 1.0.0\nArchitecture: all\nLicense: MIT & (BSD-3-Clause | Apache-2.0) & ( GPL-2.0-or-later )\n"
+	writeTarEntry(t, tw, "control", controlContent)
+
+	require.NoError(t, tw.Close())
+
+	metadata, licenses, err := processControlTar(io.NopCloser(bytes.NewReader(buf.Bytes())))
+	require.NoError(t, err)
+	require.NotNil(t, metadata)
+	assert.Equal(t, []string{"MIT AND (BSD-3-Clause OR Apache-2.0) AND (GPL-2.0-or-later)"}, licenses)
+}

syft/pkg/cataloger/debian/parse_dpkg_db.go

@@ -183,9 +183,11 @@ func parseDpkgStatusEntry(reader *bufio.Reader) (*dpkgExtractedMetadata, error)
 	}
 
 	sourceName, sourceVersion := extractSourceVersion(raw.Source)
+	if sourceName != "" {
+		raw.Source = sourceName
+	}
 	if sourceVersion != "" {
 		raw.SourceVersion = sourceVersion
-		raw.Source = sourceName
 	}
 
 	if raw.Package == "" {
@@ -263,8 +265,14 @@ func extractAllFields(reader *bufio.Reader) (map[string]any, error) {
 }
 
 // If the source entry string is of the form "<name> (<version>)" then parse and return the components, if
-// of the "<name>" form, then return name and nil
+// of the "<name>" form, then return name and nil. Yocto (opkg/ipk) instead writes the bitbake recipe filename
+// ("<name>_<version>.bb"), which is split the same way.
 func extractSourceVersion(source string) (string, string) {
+	if recipe, ok := strings.CutSuffix(source, ".bb"); ok && !strings.ContainsAny(recipe, " (") {
+		// bitbake reserves "_" as the recipe name / version separator
+		name, version, _ := strings.Cut(recipe, "_")
+		return name, version
+	}
 	// special handling for the Source field since it has formatted data
 	match := internal.MatchNamedCaptureGroups(sourceRegexp, source)
 	return match["name"], match["version"]

syft/pkg/cataloger/debian/parse_dpkg_db_test.go

@@ -366,6 +366,16 @@ func TestSourceVersionExtract(t *testing.T) {
 			input:    "",
 			expected: []string{"", ""},
 		},
+		{
+			name:     "bitbake recipe with version",
+			input:    "zlib_1.3.2.bb",
+			expected: []string{"zlib", "1.3.2"},
+		},
+		{
+			name:     "bitbake recipe without version",
+			input:    "packagegroup-core-boot.bb",
+			expected: []string{"packagegroup-core-boot", ""},
+		},
 	}
 
 	for _, test := range tests {

syft/pkg/cataloger/debian/testdata/image-single-ipk/Dockerfile

@@ -0,0 +1,21 @@
+# assembles a Yocto-style .ipk the same way bitbake's package_ipk does (opkg-build: a GNU ar archive of
+# debian-binary, control.tar.gz and data.tar.zst) without paying for a full Yocto build. control/ is copied
+# verbatim from libz1_1.3.2-r0_x86-64-v3.ipk built by poky-tiny (wrynose); the library itself is a stub
+# since only the archive shape and control metadata matter here.
+FROM alpine:3.18.4@sha256:48d9183eb12a05c99bcc0bf44a003607b8e941e1d4f41f9ad12bdcc4b5672f86 AS build
+
+RUN apk add --no-cache binutils zstd
+
+COPY control /pkg/control
+
+RUN mkdir -p /pkg/data/usr/lib \
+    && echo stub > /pkg/data/usr/lib/libz.so.1.3.2 \
+    && ln -s libz.so.1.3.2 /pkg/data/usr/lib/libz.so.1 \
+    && cd /pkg \
+    && echo 2.0 > debian-binary \
+    && tar -C control -czf control.tar.gz . \
+    && tar -C data -cf - . | zstd -q -o data.tar.zst \
+    && ar rcD /libz1_1.3.2-r0_x86-64-v3.ipk debian-binary control.tar.gz data.tar.zst
+
+FROM scratch
+COPY --from=build /libz1_1.3.2-r0_x86-64-v3.ipk /

syft/pkg/cataloger/debian/testdata/image-single-ipk/control/control

@@ -0,0 +1,15 @@
+Package: libz1
+Version: 1.3.2-r0
+Description: Zlib Compression Library
+ Zlib is a general-purpose, patent-free, lossless data compression library
+ which is used by many different programs.
+Section: libs
+Priority: optional
+Maintainer: Poky Maintainers <poky@lists.yoctoproject.org>
+License: Zlib
+Architecture: x86-64-v3
+OE: zlib
+Homepage: http://zlib.net/
+Depends: musl (>= 1.2.6+git0+9fa28ece75)
+Provides: zlib
+Source: zlib_1.3.2.bb

syft/pkg/cataloger/debian/testdata/image-single-ipk/control/postinst

@@ -0,0 +1,5 @@
+#!/bin/sh
+set -e
+if [ x"$D" = "x" ]; then
+	if [ -x /sbin/ldconfig ]; then /sbin/ldconfig ; fi
+fi