add single license scanner instance

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

internal/licenses/context.go

@@ -0,0 +1,18 @@
+package licenses
+
+import (
+	"context"
+)
+
+type licenseScannerKey struct{}
+
+func SetContextLicenseScanner(ctx context.Context, s Scanner) context.Context {
+	return context.WithValue(ctx, licenseScannerKey{}, s)
+}
+
+func ContextLicenseScanner(ctx context.Context) Scanner {
+	if s, ok := ctx.Value(licenseScannerKey{}).(Scanner); ok {
+		return s
+	}
+	return NewDefaultScanner()
+}

internal/licenses/parser.go

@@ -1,39 +0,0 @@
-package licenses
-
-import (
-	"io"
-
-	"github.com/google/licensecheck"
-
-	"github.com/anchore/syft/syft/file"
-	"github.com/anchore/syft/syft/license"
-	"github.com/anchore/syft/syft/pkg"
-)
-
-const (
-	coverageThreshold  = 75
-	unknownLicenseType = "UNKNOWN"
-)
-
-// Parse scans the contents of a license file to attempt to determine the type of license it is
-func Parse(reader io.Reader, l file.Location) (licenses []pkg.License, err error) {
-	licenses = make([]pkg.License, 0)
-	contents, err := io.ReadAll(reader)
-	if err != nil {
-		return nil, err
-	}
-	cov := licensecheck.Scan(contents)
-	if cov.Percent < coverageThreshold {
-		// unknown or no licenses here?
-		return licenses, nil
-	}
-
-	for _, m := range cov.Match {
-		lic := pkg.NewLicenseFromLocations(m.ID, l)
-		lic.Type = license.Concluded
-
-		licenses = append(licenses, lic)
-	}
-
-	return licenses, nil
-}

internal/licenses/scanner.go

@@ -0,0 +1,68 @@
+package licenses
+
+import (
+	"context"
+	"io"
+
+	"github.com/google/licensecheck"
+
+	"github.com/anchore/syft/internal/log"
+)
+
+const coverageThreshold = 75 // determined by experimentation
+
+type Scanner interface {
+	IdentifyLicenseIDs(context.Context, io.Reader) ([]string, error)
+}
+
+var _ Scanner = (*scanner)(nil)
+
+type scanner struct {
+	coverageThreshold float64 // between 0 and 100
+	scanner           func([]byte) licensecheck.Coverage
+}
+
+// NewDefaultScanner returns a scanner that uses a new instance of the default licensecheck package scanner.
+func NewDefaultScanner() Scanner {
+	s, err := licensecheck.NewScanner(licensecheck.BuiltinLicenses())
+	if err != nil {
+		log.WithFields("error", err).Trace("unable to create default license scanner")
+		s = nil
+	}
+	return &scanner{
+		coverageThreshold: coverageThreshold,
+		scanner:           s.Scan,
+	}
+}
+
+// StaticScanner returns a scanner that uses the built-in license scanner from the licensecheck package.
+// THIS IS ONLY MEANT FOR TEST CODE, NOT PRODUCTION CODE.
+func StaticScanner() Scanner {
+	return &scanner{
+		coverageThreshold: coverageThreshold,
+		scanner:           licensecheck.Scan,
+	}
+}
+
+func (s scanner) IdentifyLicenseIDs(_ context.Context, reader io.Reader) ([]string, error) {
+	if s.scanner == nil {
+		return nil, nil
+	}
+
+	content, err := io.ReadAll(reader)
+	if err != nil {
+		return nil, err
+	}
+
+	cov := s.scanner(content)
+	if cov.Percent < s.coverageThreshold {
+		// unknown or no licenses here?
+		return nil, nil
+	}
+
+	var ids []string
+	for _, m := range cov.Match {
+		ids = append(ids, m.ID)
+	}
+	return ids, nil
+}

internal/licenses/search.go

@@ -0,0 +1,28 @@
+package licenses
+
+import (
+	"context"
+
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/license"
+	"github.com/anchore/syft/syft/pkg"
+)
+
+// Search scans the contents of a license file to attempt to determine the type of license it is
+func Search(ctx context.Context, scanner Scanner, reader file.LocationReadCloser) (licenses []pkg.License, err error) {
+	licenses = make([]pkg.License, 0)
+
+	ids, err := scanner.IdentifyLicenseIDs(ctx, reader)
+	if err != nil {
+		return nil, err
+	}
+
+	for _, id := range ids {
+		lic := pkg.NewLicenseFromLocations(id, reader.Location)
+		lic.Type = license.Concluded
+
+		licenses = append(licenses, lic)
+	}
+
+	return licenses, nil
+}