feat(v2/csv): use module dir as find license boundary

v2/csv.go

@@ -18,7 +18,6 @@ import (
 	"context"
 	"encoding/csv"
 	"os"
-	"strings"
 
 	"github.com/golang/glog"
 	"github.com/google/go-licenses/v2/licenses"
@@ -56,41 +55,12 @@ func csvMain(_ *cobra.Command, args []string) error {
 	}
 	for _, lib := range libs {
 		licenseURL := "Unknown"
-		licenseName := "Unknown"
-		if lib.LicensePath != "" {
-			// Find a URL for the license file, based on the URL of a remote for the Git repository.
-			var errs []string
-			repo, err := licenses.FindGitRepo(lib.LicensePath)
-			if err != nil {
-				// Can't find Git repo (possibly a Go Module?) - derive URL from lib name instead.
-				lURL, err := lib.FileURL(lib.LicensePath)
-				if err != nil {
-					errs = append(errs, err.Error())
-				} else {
-					licenseURL = lURL.String()
-				}
-			} else {
-				for _, remote := range gitRemotes {
-					url, err := repo.FileURL(lib.LicensePath, remote)
-					if err != nil {
-						errs = append(errs, err.Error())
-						continue
-					}
-					licenseURL = url.String()
-					break
-				}
-			}
-			if licenseURL == "Unknown" {
-				glog.Errorf("Error discovering URL for %q:\n- %s", lib.LicensePath, strings.Join(errs, "\n- "))
-			}
-			licenseName, _, err = classifier.Identify(lib.LicensePath)
-			if err != nil {
-				glog.Errorf("Error identifying license in %q: %v", lib.LicensePath, err)
-				licenseName = "Unknown"
-			}
+		licenseName, _, err := classifier.Identify(lib.LicensePath)
+		if err != nil {
+			glog.Errorf("Error identifying license in %q: %v", lib.LicensePath, err)
+			licenseName = "Unknown"
 		}
-		// Remove the "*/vendor/" prefix from the library name for conciseness.
-		if err := writer.Write([]string{unvendor(lib.Name()), licenseURL, licenseName}); err != nil {
+		if err := writer.Write([]string{lib.Name(), licenseURL, licenseName}); err != nil {
 			return err
 		}
 	}

v2/licenses/find.go

@@ -16,30 +16,34 @@ package licenses
 
 import (
 	"fmt"
-	"go/build"
 	"io/ioutil"
 	"path/filepath"
 	"regexp"
+	"strings"
 )
 
 var (
 	licenseRegexp = regexp.MustCompile(`^(?i)(LICEN(S|C)E|COPYING|README|NOTICE).*$`)
-	srcDirRegexps = func() []*regexp.Regexp {
-		var rs []*regexp.Regexp
-		for _, s := range build.Default.SrcDirs() {
-			rs = append(rs, regexp.MustCompile("^"+regexp.QuoteMeta(s)+"$"))
-		}
-		return rs
-	}()
-	vendorRegexp = regexp.MustCompile(`.+/vendor(/)?$`)
 )
 
 // Find returns the file path of the license for this package.
-func Find(dir string, classifier Classifier) (string, error) {
-	var stopAt []*regexp.Regexp
-	stopAt = append(stopAt, srcDirRegexps...)
-	stopAt = append(stopAt, vendorRegexp)
-	return findUpwards(dir, licenseRegexp, stopAt, func(path string) bool {
+//
+// dir is path of the directory where we want to find a license.
+// rootDir is path of the module containing this package. Find will not search out of the
+// rootDir.
+func Find(dir string, rootDir string, classifier Classifier) (string, error) {
+	dir, err := filepath.Abs(dir)
+	if err != nil {
+		return "", err
+	}
+	rootDir, err = filepath.Abs(rootDir)
+	if err != nil {
+		return "", err
+	}
+	if !strings.HasPrefix(dir, rootDir) {
+		return "", fmt.Errorf("licenses.Find: rootDir %s should contain dir %s", rootDir, dir)
+	}
+	return findUpwards(dir, licenseRegexp, rootDir, func(path string) bool {
 		// TODO(RJPercival): Return license details
 		if _, _, err := classifier.Identify(path); err != nil {
 			return false
@@ -48,15 +52,15 @@ func Find(dir string, classifier Classifier) (string, error) {
 	})
 }
 
-func findUpwards(dir string, r *regexp.Regexp, stopAt []*regexp.Regexp, predicate func(path string) bool) (string, error) {
+func findUpwards(dir string, r *regexp.Regexp, stopAt string, predicate func(path string) bool) (string, error) {
 	// Dir must be made absolute for reliable matching with stopAt regexps
 	dir, err := filepath.Abs(dir)
 	if err != nil {
 		return "", err
 	}
 	start := dir
-	// Stop once dir matches a stopAt regexp or dir is the filesystem root
-	for !matchAny(stopAt, dir) {
+	// Stop once we go out of the stopAt dir.
+	for strings.HasPrefix(dir, stopAt) {
 		dirContents, err := ioutil.ReadDir(dir)
 		if err != nil {
 			return "", err
@@ -79,12 +83,3 @@ func findUpwards(dir string, r *regexp.Regexp, stopAt []*regexp.Regexp, predicat
 	}
 	return "", fmt.Errorf("no file/directory matching regexp %q found for %s", r, start)
 }
-
-func matchAny(patterns []*regexp.Regexp, s string) bool {
-	for _, p := range patterns {
-		if p.MatchString(s) {
-			return true
-		}
-	}
-	return false
-}

v2/licenses/git.go

@@ -46,11 +46,12 @@ type GitRepo struct {
 // FindGitRepo finds the Git repository that contains the specified filePath
 // by searching upwards through the directory tree for a ".git" directory.
 func FindGitRepo(filePath string) (*GitRepo, error) {
-	path, err := findUpwards(filepath.Dir(filePath), gitRegexp, srcDirRegexps, nil)
-	if err != nil {
-		return nil, err
-	}
-	return &GitRepo{dotGitPath: path}, nil
+	return nil, nil
+	// path, err := findUpwards(filepath.Dir(filePath), gitRegexp, srcDirRegexps, nil)
+	// if err != nil {
+	// 	return nil, err
+	// }
+	// return &GitRepo{dotGitPath: path}, nil
 }
 
 // FileURL returns the URL of a file stored in a Git repository.

v2/licenses/library.go

@@ -68,7 +68,7 @@ func (e PackagesError) Error() string {
 func Libraries(ctx context.Context, classifier Classifier, importPaths ...string) ([]*Library, error) {
 	cfg := &packages.Config{
 		Context: ctx,
-		Mode:    packages.NeedImports | packages.NeedDeps | packages.NeedFiles | packages.NeedName,
+		Mode:    packages.NeedImports | packages.NeedDeps | packages.NeedFiles | packages.NeedName | packages.NeedModule,
 	}
 
 	rootPkgs, err := packages.Load(cfg, importPaths...)
@@ -103,7 +103,7 @@ func Libraries(ctx context.Context, classifier Classifier, importPaths ...string
 			// This package is empty - nothing to do.
 			return true
 		}
-		licensePath, err := Find(pkgDir, classifier)
+		licensePath, err := Find(pkgDir, p.Module.Dir, classifier)
 		if err != nil {
 			glog.Errorf("Failed to find license for %s: %v", p.PkgPath, err)
 		}
@@ -202,6 +202,10 @@ func (l *Library) FileURL(filePath string) (*url.URL, error) {
 
 // isStdLib returns true if this package is part of the Go standard library.
 func isStdLib(pkg *packages.Package) bool {
+	if pkg.Name == "unsafe" {
+		// Special case unsafe stdlib, because it does not contain go files.
+		return true
+	}
 	if len(pkg.GoFiles) == 0 {
 		return false
 	}

v2/third_party/licenses.csv

@@ -1,17 +1,18 @@
-github.com/emirpasic/gods,https://github.com/emirpasic/gods/blob/master/LICENSE,BSD-2-Clause
-github.com/golang/glog,https://github.com/golang/glog/blob/master/LICENSE,Apache-2.0
-github.com/google/go-licenses/v2,Unknown,Apache-2.0
-github.com/google/licenseclassifier,https://github.com/google/licenseclassifier/blob/master/LICENSE,Apache-2.0
-github.com/google/licenseclassifier/stringclassifier,https://github.com/google/licenseclassifier/blob/master/stringclassifier/LICENSE,Apache-2.0
-github.com/jbenet/go-context/io,https://github.com/jbenet/go-context/blob/master/io/LICENSE,MIT
-github.com/kevinburke/ssh_config,https://github.com/kevinburke/ssh_config/blob/master/LICENSE,MIT
-github.com/mitchellh/go-homedir,https://github.com/mitchellh/go-homedir/blob/master/LICENSE,MIT
-github.com/otiai10/copy,https://github.com/otiai10/copy/blob/master/LICENSE,MIT
-github.com/sergi/go-diff/diffmatchpatch,https://github.com/sergi/go-diff/blob/master/diffmatchpatch/LICENSE,MIT
-github.com/spf13/cobra,https://github.com/spf13/cobra/blob/master/LICENSE.txt,Apache-2.0
-github.com/spf13/pflag,https://github.com/spf13/pflag/blob/master/LICENSE,BSD-3-Clause
-github.com/src-d/gcfg,https://github.com/src-d/gcfg/blob/master/LICENSE,BSD-3-Clause
-github.com/xanzy/ssh-agent,https://github.com/xanzy/ssh-agent/blob/master/LICENSE,Apache-2.0
+github.com/emirpasic/gods,Unknown,BSD-2-Clause
+github.com/golang/glog,Unknown,Apache-2.0
+github.com/google/go-licenses/v2,Unknown,
+github.com/google/go-licenses/v2/licenses,Unknown,
+github.com/google/licenseclassifier,Unknown,Apache-2.0
+github.com/google/licenseclassifier/stringclassifier,Unknown,Apache-2.0
+github.com/jbenet/go-context/io,Unknown,MIT
+github.com/kevinburke/ssh_config,Unknown,MIT
+github.com/mitchellh/go-homedir,Unknown,MIT
+github.com/otiai10/copy,Unknown,MIT
+github.com/sergi/go-diff/diffmatchpatch,Unknown,MIT
+github.com/spf13/cobra,Unknown,Apache-2.0
+github.com/spf13/pflag,Unknown,BSD-3-Clause
+github.com/src-d/gcfg,Unknown,BSD-3-Clause
+github.com/xanzy/ssh-agent,Unknown,Apache-2.0
 golang.org/x/crypto,Unknown,BSD-3-Clause
 golang.org/x/mod/semver,Unknown,BSD-3-Clause
 golang.org/x/net,Unknown,BSD-3-Clause