Skip to content

Commit

Permalink
fix: apk product/vendor generation for old metadata (#1635)
Browse files Browse the repository at this point in the history 
This fixes some instances where the improved APK CPE generation
logic caused regressions for older alpine package APK metadata.
It now generates multiple "upstream" candidates with both name
and package type which reduces the amount of duplicated code in
the apk cpe gen logic.  This also improves the handling of stream
version packages, so now we can correctly identify packages such
as ruby3.2-rexml as the rexml ruby gem.

Signed-off-by: Weston Steimel <weston.steimel@anchore.com>
  • Loading branch information
@westonsteimel
westonsteimel authored Mar 1, 2023
1 parent e92b0fa commit 8e1205f
Show file tree
Hide file tree
Showing 7 changed files with 287 additions and 206 deletions.
50 changes: 37 additions & 13 deletions syft/pkg/apk_metadata.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -18,9 +18,12 @@ import (
const ApkDBGlob = "**/lib/apk/db/installed"

var (
_ FileOwner = (*ApkMetadata)(nil)
prefixes = []string{"py-", "py2-", "py3-", "ruby-"}
upstreamPattern = regexp.MustCompile(`^(?P<upstream>[a-zA-Z][\w-]*?)\-?\d[\d\.]*$`)
_ FileOwner = (*ApkMetadata)(nil)
prefixesToPackageType = map[string]Type{
"py-": PythonPkg,
"ruby-": GemPkg,
}
streamVersionPkgNamePattern = regexp.MustCompile(`^(?P<stream>[a-zA-Z][\w-]*?)(?P<streamVersion>\-?\d[\d\.]*?)($|-(?P<subPackage>[a-zA-Z][\w-]*?)?)$`)
)

// ApkMetadata represents all captured data for a Alpine DB package entry.
Expand Down Expand Up @@ -121,23 +124,44 @@ func (m ApkMetadata) OwnedFiles() (result []string) {
return result
}

func (m ApkMetadata) Upstream() string {
type UpstreamCandidate struct {
Name string
Type Type
}

func (m ApkMetadata) UpstreamCandidates() (candidates []UpstreamCandidate) {
name := m.Package
if m.OriginPackage != "" && m.OriginPackage != m.Package {
return m.OriginPackage
candidates = append(candidates, UpstreamCandidate{Name: m.OriginPackage, Type: ApkPkg})
}

groups := internal.MatchNamedCaptureGroups(upstreamPattern, m.Package)
groups := internal.MatchNamedCaptureGroups(streamVersionPkgNamePattern, m.Package)
stream, ok := groups["stream"]

upstream, ok := groups["upstream"]
if !ok {
upstream = m.Package
if ok && stream != "" {
sub, ok := groups["subPackage"]

if ok && sub != "" {
name = fmt.Sprintf("%s-%s", stream, sub)
} else {
name = stream
}
}

for _, p := range prefixes {
if strings.HasPrefix(upstream, p) {
return strings.TrimPrefix(upstream, p)
for prefix, typ := range prefixesToPackageType {
if strings.HasPrefix(name, prefix) {
t := strings.TrimPrefix(name, prefix)
if t != "" {
candidates = append(candidates, UpstreamCandidate{Name: t, Type: typ})
return candidates
}
}
}

return upstream
if name != "" {
candidates = append(candidates, UpstreamCandidate{Name: name, Type: UnknownPkg})
return candidates
}

return candidates
}
139 changes: 107 additions & 32 deletions syft/pkg/apk_metadata_test.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,6 @@ import (
"encoding/json"
"testing"

"github.com/sergi/go-diff/diffmatchpatch"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
Expand Down Expand Up @@ -165,168 +164,244 @@ func TestSpaceDelimitedStringSlice_UnmarshalJSON(t *testing.T) {
}
}

func TestApkMetadata_Upstream(t *testing.T) {
func TestApkMetadata_UpstreamCandidates(t *testing.T) {
tests := []struct {
name string
metadata ApkMetadata
expected string
expected []UpstreamCandidate
}{
{
name: "gocase",
metadata: ApkMetadata{
Package: "p",
},
expected: "p",
expected: []UpstreamCandidate{
{Name: "p", Type: UnknownPkg},
},
},
{
name: "same package and origin",
name: "same package and origin simple case",
metadata: ApkMetadata{
Package: "p",
OriginPackage: "p",
},
expected: "p",
expected: []UpstreamCandidate{
{Name: "p", Type: UnknownPkg},
},
},
{
name: "different package and origin",
metadata: ApkMetadata{
Package: "p",
OriginPackage: "origin",
},
expected: "origin",
expected: []UpstreamCandidate{
{Name: "origin", Type: ApkPkg},
{Name: "p", Type: UnknownPkg},
},
},
{
name: "upstream python package information as qualifier",
name: "upstream python package information as qualifier py- prefix",
metadata: ApkMetadata{
Package: "py-potatoes",
OriginPackage: "py-potatoes",
},
expected: []UpstreamCandidate{
{Name: "potatoes", Type: PythonPkg},
},
},
{
name: "upstream python package information as qualifier py3- prefix",
metadata: ApkMetadata{
Package: "py3-potatoes",
OriginPackage: "py3-potatoes",
},
expected: "potatoes",
expected: []UpstreamCandidate{
{Name: "potatoes", Type: PythonPkg},
},
},
{
name: "python package with distinct origin package",
metadata: ApkMetadata{
Package: "py3-non-existant",
OriginPackage: "abcdefg",
},
expected: "abcdefg",
expected: []UpstreamCandidate{
{Name: "abcdefg", Type: ApkPkg},
{Name: "non-existant", Type: PythonPkg},
},
},
{
name: "upstream ruby package information as qualifier",
metadata: ApkMetadata{
Package: "ruby-something",
OriginPackage: "ruby-something",
},
expected: "something",
expected: []UpstreamCandidate{
{Name: "something", Type: GemPkg},
},
},
{
name: "python package with distinct origin package",
name: "ruby package with distinct origin package",
metadata: ApkMetadata{
Package: "ruby-something",
OriginPackage: "1234567",
},
expected: "1234567",
expected: []UpstreamCandidate{
{Name: "1234567", Type: ApkPkg},
{Name: "something", Type: GemPkg},
},
},
{
name: "postgesql-15 upstream postgresql",
metadata: ApkMetadata{
Package: "postgresql-15",
},
expected: "postgresql",
expected: []UpstreamCandidate{
{Name: "postgresql", Type: UnknownPkg},
},
},
{
name: "postgesql15 upstream postgresql",
metadata: ApkMetadata{
Package: "postgresql15",
},
expected: "postgresql",
expected: []UpstreamCandidate{
{Name: "postgresql", Type: UnknownPkg},
},
},
{
name: "go-1.19 upstream go",
metadata: ApkMetadata{
Package: "go-1.19",
},
expected: "go",
expected: []UpstreamCandidate{
{Name: "go", Type: UnknownPkg},
},
},
{
name: "go1.143 upstream go",
metadata: ApkMetadata{
Package: "go1.143",
},
expected: "go",
expected: []UpstreamCandidate{
{Name: "go", Type: UnknownPkg},
},
},
{
name: "abc-101.191.23456 upstream abc",
metadata: ApkMetadata{
Package: "abc-101.191.23456",
},
expected: "abc",
expected: []UpstreamCandidate{
{Name: "abc", Type: UnknownPkg},
},
},
{
name: "abc101.191.23456 upstream abc",
metadata: ApkMetadata{
Package: "abc101.191.23456",
},
expected: "abc",
expected: []UpstreamCandidate{
{Name: "abc", Type: UnknownPkg},
},
},
{
name: "abc101-12345-1045 upstream abc101-12345",
metadata: ApkMetadata{
Package: "abc101-12345-1045",
},
expected: "abc101-12345",
expected: []UpstreamCandidate{
{Name: "abc101-12345", Type: UnknownPkg},
},
},
{
name: "abc101-a12345-1045 upstream abc101-a12345",
metadata: ApkMetadata{
Package: "abc101-a12345-1045",
},
expected: "abc101-a12345",
expected: []UpstreamCandidate{
{Name: "abc-a12345-1045", Type: UnknownPkg},
},
},
{
name: "package starting with single digit",
metadata: ApkMetadata{
Package: "3proxy",
},
expected: "3proxy",
expected: []UpstreamCandidate{
{Name: "3proxy", Type: UnknownPkg},
},
},
{
name: "package starting with multiple digits",
metadata: ApkMetadata{
Package: "356proxy",
},
expected: "356proxy",
expected: []UpstreamCandidate{
{Name: "356proxy", Type: UnknownPkg},
},
},
{
name: "package composed of only digits",
metadata: ApkMetadata{
Package: "123456",
},
expected: "123456",
expected: []UpstreamCandidate{
{Name: "123456", Type: UnknownPkg},
},
},
{
name: "ruby-3.6 upstream ruby",
metadata: ApkMetadata{
Package: "ruby-3.6",
},
expected: "ruby",
expected: []UpstreamCandidate{
{Name: "ruby", Type: UnknownPkg},
},
},
{
name: "ruby3.6 upstream ruby",
metadata: ApkMetadata{
Package: "ruby3.6",
},
expected: "ruby",
expected: []UpstreamCandidate{
{Name: "ruby", Type: UnknownPkg},
},
},
{
name: "ruby3.6-tacos upstream tacos",
metadata: ApkMetadata{
Package: "ruby3.6-tacos",
},
expected: []UpstreamCandidate{
{Name: "tacos", Type: GemPkg},
},
},
{
name: "ruby-3.6-tacos upstream tacos",
metadata: ApkMetadata{
Package: "ruby-3.6-tacos",
},
expected: []UpstreamCandidate{
{Name: "tacos", Type: GemPkg},
},
},
{
name: "abc1234jksajflksa",
metadata: ApkMetadata{
Package: "abc1234jksajflksa",
},
expected: []UpstreamCandidate{
{Name: "abc1234jksajflksa", Type: UnknownPkg},
},
},
}

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))
}
actual := test.metadata.UpstreamCandidates()
assert.Equal(t, test.expected, actual)
})
}
}
11 changes: 8 additions & 3 deletions syft/pkg/cataloger/apkdb/package.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -36,9 +36,14 @@ func packageURL(m pkg.ApkMetadata, distro *linux.Release) string {
pkg.PURLQualifierArch: m.Architecture,
}

upstream := m.Upstream()
if upstream != "" && upstream != m.Package {
qualifiers[pkg.PURLQualifierUpstream] = upstream
upstreams := m.UpstreamCandidates()
if len(upstreams) > 0 {
// only room for one value so for now just take the first one
upstream := upstreams[0]

if upstream.Name != "" && upstream.Name != m.Package {
qualifiers[pkg.PURLQualifierUpstream] = upstream.Name
}
}

return packageurl.NewPackageURL(
Expand Down
2 changes: 1 addition & 1 deletion syft/pkg/cataloger/apkdb/package_test.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -234,7 +234,7 @@ func Test_PackageURL(t *testing.T) {
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",
expected: "pkg:apk/alpine/abc101-a12345-1045@101.191.23456?arch=a&upstream=abc-a12345-1045&distro=alpine-3.4.6",
},
{
name: "wolfi distro",
Expand Down
Loading

0 comments on commit 8e1205f

Please sign in to comment.