Backport: skip non-regular files (e.g. submodules) on repo indexing (#7717)

* Backport: skip non-regular files (e.g. submodules) on repo indexing

* Include "executable" files in the index, as they are not necessarily binary

Author: guillep2k

models/repo_indexer.go

@@ -231,20 +231,28 @@ func addDelete(filename string, repo *Repository, batch rupture.FlushingBatch) e
 	return indexerUpdate.AddToFlushingBatch(batch)
 }
 
+func isIndexable(entry *git.TreeEntry) bool {
+	return entry.IsRegular() || entry.IsExecutable()
+}
+
 // parseGitLsTreeOutput parses the output of a `git ls-tree -r --full-name` command
 func parseGitLsTreeOutput(stdout []byte) ([]fileUpdate, error) {
 	entries, err := git.ParseTreeEntries(stdout)
 	if err != nil {
 		return nil, err
 	}
+	var idxCount = 0
 	updates := make([]fileUpdate, len(entries))
-	for i, entry := range entries {
-		updates[i] = fileUpdate{
-			Filename: entry.Name(),
-			BlobSha:  entry.ID.String(),
+	for _, entry := range entries {
+		if isIndexable(entry) {
+			updates[idxCount] = fileUpdate{
+				Filename: entry.Name(),
+				BlobSha:  entry.ID.String(),
+			}
+			idxCount++
 		}
 	}
-	return updates, nil
+	return updates[:idxCount], nil
 }
 
 // genesisChanges get changes to add repo to the indexer for the first time

modules/git/tree_entry.go

@@ -108,6 +108,11 @@ func (te *TreeEntry) IsRegular() bool {
 	return te.gogitTreeEntry.Mode == filemode.Regular
 }
 
+// IsExecutable if the entry is an executable file (not necessarily binary)
+func (te *TreeEntry) IsExecutable() bool {
+	return te.gogitTreeEntry.Mode == filemode.Executable
+}
+
 // Blob returns the blob object the entry
 func (te *TreeEntry) Blob() *Blob {
 	encodedObj, err := te.ptree.repo.gogitRepo.Storer.EncodedObject(plumbing.AnyObject, te.gogitTreeEntry.Hash)