fix: do better parsing of file modes (#9161)

- No longer hardcode the file modes we expect, parse them as numbers and do bitmask tricks that Git does so we allow a more variety of _weird_ file modes that can happen in the wild.
- Ref: https://codeberg.org/forgejo/forgejo/pulls/8900#issuecomment-6429175
- Resolves Codeberg/Community#2111

Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/9161
Reviewed-by: Earl Warren <earl-warren@noreply.codeberg.org>
Co-authored-by: Gusted <postmaster@gusted.xyz>
Co-committed-by: Gusted <postmaster@gusted.xyz>

modules/git/parse.go

@@ -10,8 +10,6 @@ import (
 	"io"
 	"strconv"
 	"strings"
-
-	"forgejo.org/modules/log"
 )
 
 // ParseTreeEntries parses the output of a `git ls-tree -l` command.
@@ -55,19 +53,9 @@ func parseTreeEntries(data []byte, ptree *Tree) ([]*TreeEntry, error) {
 			entry.sized = true
 		}
 
-		switch string(entryMode) {
-		case "100644":
-			entry.entryMode = EntryModeBlob
-		case "100755":
-			entry.entryMode = EntryModeExec
-		case "120000":
-			entry.entryMode = EntryModeSymlink
-		case "160000":
-			entry.entryMode = EntryModeCommit
-		case "040000", "040755", "040775": // git uses 040000 for tree object, but some users may get 040755 or 040775 for unknown reasons
-			entry.entryMode = EntryModeTree
-		default:
-			return nil, fmt.Errorf("unknown type: %v", string(entryMode))
+		entry.entryMode, err = parseMode(string(entryMode))
+		if err != nil {
+			return nil, err
 		}
 
 		entry.ID, err = NewIDFromString(string(entryObjectID))
@@ -108,23 +96,10 @@ loop:
 		sz -= int64(count)
 		entry := new(TreeEntry)
 		entry.ptree = ptree
-
-		switch string(mode) {
-		case "100644":
-			entry.entryMode = EntryModeBlob
-		case "100755":
-			entry.entryMode = EntryModeExec
-		case "120000":
-			entry.entryMode = EntryModeSymlink
-		case "160000":
-			entry.entryMode = EntryModeCommit
-		case "40000", "40755", "40775": // git uses 40000 for tree object, but some users may get 40755 or 40775 for unknown reasons
-			entry.entryMode = EntryModeTree
-		default:
-			log.Debug("Unknown mode: %v", string(mode))
-			return nil, fmt.Errorf("unknown mode: %v", string(mode))
+		entry.entryMode, err = parseMode(string(mode))
+		if err != nil {
+			return nil, err
 		}
-
 		entry.ID = objectFormat.MustID(sha)
 		entry.name = string(fname)
 		entries = append(entries, entry)
@@ -135,3 +110,31 @@ loop:
 
 	return entries, nil
 }
+
+// Parse the file mode, we cannot hardcode the modes that we expect for
+// a variety of reasons (that is not known to us) the permissions bits
+// of files can vary, usually the result because of tooling that uses Git in
+// a funny way. So we have to parse the mode as a integer and do bit tricks.
+func parseMode(modeStr string) (EntryMode, error) {
+	mode, err := strconv.ParseUint(modeStr, 8, 64)
+	if err != nil {
+		return 0, fmt.Errorf("cannot parse mode: %v", err)
+	}
+
+	switch mode & 0o170000 {
+	case 0o040000:
+		return EntryModeTree, nil
+	case 0o120000:
+		return EntryModeSymlink, nil
+	case 0o160000:
+		return EntryModeCommit, nil
+	case 0o100000:
+		// Check for the permission bit on the owner.
+		if mode&0o100 == 0o100 {
+			return EntryModeExec, nil
+		}
+		return EntryModeBlob, nil
+	}
+
+	return 0, fmt.Errorf("unknown mode: %o", mode)
+}

modules/git/parse_test.go

@@ -101,3 +101,38 @@ func TestParseTreeEntriesInvalid(t *testing.T) {
 	require.Error(t, err)
 	assert.Empty(t, entries)
 }
+
+func TestParseMode(t *testing.T) {
+	ok := func(t *testing.T, mode string, entry EntryMode) {
+		t.Helper()
+		actualEntry, err := parseMode(mode)
+		require.NoError(t, err)
+		assert.Equal(t, entry, actualEntry)
+	}
+
+	fail := func(t *testing.T, mode string) {
+		t.Helper()
+		entry, err := parseMode(mode)
+		require.Error(t, err)
+		assert.Zero(t, entry)
+	}
+
+	ok(t, "100644", EntryModeBlob)
+	ok(t, "100755", EntryModeExec)
+	ok(t, "100754", EntryModeExec)
+	ok(t, "100700", EntryModeExec)
+	ok(t, "100744", EntryModeExec)
+	ok(t, "120000", EntryModeSymlink)
+	ok(t, "120644", EntryModeSymlink)
+	ok(t, "160000", EntryModeCommit)
+	ok(t, "160644", EntryModeCommit)
+	ok(t, "040000", EntryModeTree)
+	ok(t, "040755", EntryModeTree)
+	ok(t, "040775", EntryModeTree)
+	ok(t, "040754", EntryModeTree)
+
+	fail(t, "not-a-number")
+	fail(t, "000000")
+	fail(t, "400000")
+	fail(t, "111111")
+}