fix: java archives excluded due to incorrect license glob results (#4449)

Signed-off-by: Keith Zantow <kzantow@gmail.com>

internal/file/glob_match.go

@@ -1,46 +0,0 @@
-package file
-
-// GlobMatch evaluates the given glob pattern against the given "name" string, indicating if there is a match or not.
-// Source: https://research.swtch.com/glob.go
-func GlobMatch(pattern, name string) bool {
-	px := 0
-	nx := 0
-	nextPx := 0
-	nextNx := 0
-	for px < len(pattern) || nx < len(name) {
-		if px < len(pattern) {
-			c := pattern[px]
-			switch c {
-			default: // ordinary character
-				if nx < len(name) && name[nx] == c {
-					px++
-					nx++
-					continue
-				}
-			case '?': // single-character wildcard
-				if nx < len(name) {
-					px++
-					nx++
-					continue
-				}
-			case '*': // zero-or-more-character wildcard
-				// Try to match at nx.
-				// If that doesn't work out,
-				// restart at nx+1 next.
-				nextPx = px
-				nextNx = nx + 1
-				px++
-				continue
-			}
-		}
-		// Mismatch. Maybe restart.
-		if 0 < nextNx && nextNx <= len(name) {
-			px = nextPx
-			nx = nextNx
-			continue
-		}
-		return false
-	}
-	// Matched all of pattern to all of name. Success.
-	return true
-}

internal/file/glob_match_test.go

@@ -1,39 +0,0 @@
-package file
-
-import (
-	"strings"
-	"testing"
-)
-
-func TestGlobMatch(t *testing.T) {
-	var tests = []struct {
-		pattern string
-		data    string
-		ok      bool
-	}{
-		{"", "", true},
-		{"x", "", false},
-		{"", "x", false},
-		{"abc", "abc", true},
-		{"*", "abc", true},
-		{"*c", "abc", true},
-		{"*b", "abc", false},
-		{"a*", "abc", true},
-		{"b*", "abc", false},
-		{"a*", "a", true},
-		{"*a", "a", true},
-		{"a*b*c*d*e*", "axbxcxdxe", true},
-		{"a*b*c*d*e*", "axbxcxdxexxx", true},
-		{"a*b?c*x", "abxbbxdbxebxczzx", true},
-		{"a*b?c*x", "abxbbxdbxebxczzy", false},
-		{"a*a*a*a*b", strings.Repeat("a", 100), false},
-		{"*x", "xxx", true},
-		{"/home/place/**", "/home/place/a/thing", true},
-	}
-
-	for _, test := range tests {
-		if GlobMatch(test.pattern, test.data) != test.ok {
-			t.Errorf("failed glob='%s' data='%s'", test.pattern, test.data)
-		}
-	}
-}

internal/file/test-fixtures/zip-source/b-file/in-subdir.txt

@@ -0,0 +1 @@
+this file is in a subdirectory

internal/file/zip_file_helpers_test.go

@@ -7,14 +7,14 @@ import (
 	"path/filepath"
 	"syscall"
 	"testing"
-
-	"github.com/stretchr/testify/assert"
 )
 
 var expectedZipArchiveEntries = []string{
-	"some-dir" + string(os.PathSeparator),
-	filepath.Join("some-dir", "a-file.txt"),
+	"some-dir/",
+	"some-dir/a-file.txt",
 	"b-file.txt",
+	"b-file/",
+	"b-file/in-subdir.txt",
 	"nested.zip",
 }
 
@@ -59,12 +59,6 @@ func createZipArchive(t testing.TB, sourceDirPath, destinationArchivePath string
 
 }
 
-func assertNoError(t testing.TB, fn func() error) func() {
-	return func() {
-		assert.NoError(t, fn())
-	}
-}
-
 // setupZipFileTest encapsulates common test setup work for zip file tests. It returns a cleanup function,
 // which should be called (typically deferred) by the caller, the path of the created zip archive, and an error,
 // which should trigger a fatal test failure in the consuming test. The returned cleanup function will never be nil

internal/file/zip_file_manifest.go

@@ -6,6 +6,7 @@ import (
 	"sort"
 	"strings"
 
+	"github.com/bmatcuk/doublestar/v4"
 	"github.com/mholt/archives"
 	"github.com/scylladb/go-set/strset"
 
@@ -44,12 +45,17 @@ func (z ZipFileManifest) Add(entry string, info os.FileInfo) {
 	z[entry] = info
 }
 
-// GlobMatch returns the path keys that match the given value(s).
+// GlobMatch returns the path keys to files (not directories) that match the given value(s).
 func (z ZipFileManifest) GlobMatch(caseInsensitive bool, patterns ...string) []string {
 	uniqueMatches := strset.New()
 
 	for _, pattern := range patterns {
 		for entry := range z {
+			fileInfo := z[entry]
+			if fileInfo != nil && fileInfo.IsDir() {
+				continue
+			}
+
 			// We want to match globs as if entries begin with a leading slash (akin to an absolute path)
 			// so that glob logic is consistent inside and outside of ZIP archives
 			normalizedEntry := normalizeZipEntryName(caseInsensitive, entry)
@@ -57,7 +63,13 @@ func (z ZipFileManifest) GlobMatch(caseInsensitive bool, patterns ...string) []s
 			if caseInsensitive {
 				pattern = strings.ToLower(pattern)
 			}
-			if GlobMatch(pattern, normalizedEntry) {
+
+			matches, err := doublestar.Match(pattern, normalizedEntry)
+			if err != nil {
+				log.Debugf("error with match pattern '%s', including by default: %v", pattern, err)
+				matches = true
+			}
+			if matches {
 				uniqueMatches.Add(entry)
 			}
 		}

internal/file/zip_file_manifest_test.go

@@ -9,6 +9,8 @@ import (
 	"os"
 	"path"
 	"testing"
+
+	"github.com/stretchr/testify/require"
 )
 
 func TestNewZipFileManifest(t *testing.T) {
@@ -107,23 +109,27 @@ func TestZipFileManifest_GlobMatch(t *testing.T) {
 
 	cases := []struct {
 		glob     string
-		expected string
+		expected []string
 	}{
 		{
 			"/b*",
-			"b-file.txt",
+			[]string{"b-file.txt"},
 		},
 		{
-			"*/a-file.txt",
-			"some-dir/a-file.txt",
+			"/b*/**",
+			[]string{"b-file.txt", "b-file/in-subdir.txt"},
 		},
 		{
-			"*/A-file.txt",
-			"some-dir/a-file.txt",
+			"**/a-file.txt",
+			[]string{"some-dir/a-file.txt"},
+		},
+		{
+			"**/A-file.txt",
+			[]string{"some-dir/a-file.txt"},
 		},
 		{
 			"**/*.zip",
-			"nested.zip",
+			[]string{"nested.zip"},
 		},
 	}
 
@@ -133,11 +139,7 @@ func TestZipFileManifest_GlobMatch(t *testing.T) {
 
 			results := z.GlobMatch(true, glob)
 
-			if len(results) == 1 && results[0] == tc.expected {
-				return
-			}
-
-			t.Errorf("unexpected results for glob '%s': %+v", glob, results)
+			require.ElementsMatch(t, tc.expected, results)
 		})
 	}
 }