feat: Add support for scanning GGUF models from OCI registries (#4335)

---------
Signed-off-by: Christopher Phillips <32073428+spiffcs@users.noreply.github.com>
Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>
Co-authored-by: Alex Goodman <wagoodman@users.noreply.github.com>

syft/pkg/cataloger/generic/cataloger.go

@@ -114,6 +114,26 @@ func (c *Cataloger) WithParserByPath(parser Parser, paths ...string) *Cataloger
 	return c
 }
 
+func (c *Cataloger) WithParserByMediaType(parser Parser, types ...string) *Cataloger {
+	c.requesters = append(c.requesters,
+		func(resolver file.Resolver, _ Environment) []request {
+			var requests []request
+			log.WithFields("mediatypes", types).Trace("searching content matching mediatypes")
+			ociResolver, ok := resolver.(file.OCIMediaTypeResolver)
+			if !ok {
+				return nil
+			}
+			matches, err := ociResolver.FilesByMediaType(types...)
+			if err != nil {
+				return nil
+			}
+			requests = append(requests, makeRequests(parser, matches)...)
+			return requests
+		},
+	)
+	return c
+}
+
 func (c *Cataloger) WithProcessors(processors ...Processor) *Cataloger {
 	for _, p := range processors {
 		c.processors = append(c.processors, processorWrapper{Processor: p})

syft/pkg/cataloger/generic/cataloger_test.go

@@ -138,6 +138,10 @@ func (m spyReturningFileResolver) FilesByMIMEType(types ...string) ([]file.Locat
 	return m.m.FilesByMIMEType(types...)
 }
 
+func (m spyReturningFileResolver) FilesByMediaType(types ...string) ([]file.Location, error) {
+	return m.m.FilesByMediaType(types...)
+}
+
 func (m spyReturningFileResolver) RelativeFileByPath(f file.Location, path string) *file.Location {
 	return m.m.RelativeFileByPath(f, path)
 }
@@ -189,6 +193,55 @@ func TestClosesFileOnParserPanic(t *testing.T) {
 	require.True(t, spy.closed)
 }
 
+func Test_CatalogerWithParserByMediaType(t *testing.T) {
+	allParsedPaths := make(map[string]bool)
+	parser := func(_ context.Context, resolver file.Resolver, env *Environment, reader file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+		allParsedPaths[reader.Path()] = true
+		contents, err := io.ReadAll(reader)
+		require.NoError(t, err)
+
+		if len(contents) == 0 {
+			return nil, nil, nil
+		}
+
+		p := pkg.Package{
+			Name:      string(contents),
+			Locations: file.NewLocationSet(reader.Location),
+		}
+
+		return []pkg.Package{p}, nil, nil
+	}
+
+	upstream := "media-type-cataloger"
+
+	// Create locations with test fixtures that exist on disk
+	loc1 := file.NewLocation("test-fixtures/a-path.txt")
+	loc2 := file.NewLocation("test-fixtures/another-path.txt")
+
+	// Create a mock resolver that maps media types to locations
+	resolver := file.NewMockResolverForMediaTypes(map[string][]file.Location{
+		"application/vnd.test.model": {loc1, loc2},
+	})
+
+	cataloger := NewCataloger(upstream).
+		WithParserByMediaType(parser, "application/vnd.test.model")
+
+	actualPkgs, _, err := cataloger.Catalog(context.Background(), resolver)
+	assert.NoError(t, err)
+
+	// Verify both files were parsed
+	assert.True(t, allParsedPaths["test-fixtures/a-path.txt"], "expected a-path.txt to be parsed")
+	assert.True(t, allParsedPaths["test-fixtures/another-path.txt"], "expected another-path.txt to be parsed")
+
+	// Verify packages were created
+	assert.Len(t, actualPkgs, 2)
+
+	// Verify FoundBy is set correctly
+	for _, p := range actualPkgs {
+		assert.Equal(t, upstream, p.FoundBy)
+	}
+}
+
 func Test_genericCatalogerReturnsErrors(t *testing.T) {
 	genericErrorReturning := NewCataloger("error returning").WithParserByGlobs(func(ctx context.Context, resolver file.Resolver, environment *Environment, locationReader file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
 		return []pkg.Package{