feat: Add license enrichment from pypi to python packages (#4295)

* feat: Add license enrichment from pypi to python packages
* Implement license caching and improve test coverage
---------
Signed-off-by: Tim Olshansky <456103+timols@users.noreply.github.com>
This commit is contained in:
Tim Olshansky
2025-11-06 16:05:08 -05:00
committed by GitHub
parent 4e06a7ab32
commit bbef262b8f
30 changed files with 1038 additions and 67 deletions
+5 -3
View File
@@ -198,9 +198,10 @@ func (cfg Catalog) ToPackagesConfig() pkgcataloging.Config {
},
Nix: nix.DefaultConfig().
WithCaptureOwnedFiles(cfg.Nix.CaptureOwnedFiles),
Python: python.CatalogerConfig{
GuessUnpinnedRequirements: cfg.Python.GuessUnpinnedRequirements,
},
Python: python.DefaultCatalogerConfig().
WithSearchRemoteLicenses(*multiLevelOption(false, enrichmentEnabled(cfg.Enrich, task.Python), cfg.Python.SearchRemoteLicenses)).
WithPypiBaseURL(cfg.Python.PypiBaseURL).
WithGuessUnpinnedRequirements(*multiLevelOption(false, enrichmentEnabled(cfg.Enrich, task.Python), cfg.Python.GuessUnpinnedRequirements)),
JavaArchive: java.DefaultArchiveCatalogerConfig().
WithUseMavenLocalRepository(*multiLevelOption(false, enrichmentEnabled(cfg.Enrich, task.Java, task.Maven), cfg.Java.UseMavenLocalRepository)).
WithMavenLocalRepositoryDir(cfg.Java.MavenLocalRepositoryDir).
@@ -320,6 +321,7 @@ var publicisedEnrichmentOptions = []string{
task.Golang,
task.Java,
task.JavaScript,
task.Python,
}
func enrichmentEnabled(enrichDirectives []string, features ...string) *bool {
+5 -1
View File
@@ -3,7 +3,9 @@ package options
import "github.com/anchore/clio"
type pythonConfig struct {
GuessUnpinnedRequirements bool `json:"guess-unpinned-requirements" yaml:"guess-unpinned-requirements" mapstructure:"guess-unpinned-requirements"`
SearchRemoteLicenses *bool `json:"search-remote-licenses" yaml:"search-remote-licenses" mapstructure:"search-remote-licenses"`
PypiBaseURL string `json:"pypi-base-url" yaml:"pypi-base-url" mapstructure:"pypi-base-url"`
GuessUnpinnedRequirements *bool `json:"guess-unpinned-requirements" yaml:"guess-unpinned-requirements" mapstructure:"guess-unpinned-requirements"`
}
var _ interface {
@@ -11,6 +13,8 @@ var _ interface {
} = (*pythonConfig)(nil)
func (o *pythonConfig) DescribeFields(descriptions clio.FieldDescriptionSet) {
descriptions.Add(&o.SearchRemoteLicenses, `enables Syft to use the network to fill in more detailed license information`)
descriptions.Add(&o.PypiBaseURL, `base Pypi url to use`)
descriptions.Add(&o.GuessUnpinnedRequirements, `when running across entries in requirements.txt that do not specify a specific version
(e.g. "sqlalchemy >= 1.0.0, <= 2.0.0, != 3.0.0, <= 3.0.0"), attempt to guess what the version could
be based on the version requirements specified (e.g. "1.0.0"). When enabled the lowest expressible version