feat: vendor gitea 1.16.2

This commit is contained in:
2026-06-02 08:36:01 +00:00
parent 247cd60f69
commit 54798b4615
2145 changed files with 162965 additions and 126447 deletions
@@ -4,7 +4,6 @@
package bleve
import (
"bufio"
"context"
"fmt"
"io"
@@ -17,6 +16,7 @@ import (
"code.gitea.io/gitea/modules/charset"
"code.gitea.io/gitea/modules/git"
"code.gitea.io/gitea/modules/git/gitcmd"
"code.gitea.io/gitea/modules/gitrepo"
"code.gitea.io/gitea/modules/indexer"
path_filter "code.gitea.io/gitea/modules/indexer/code/bleve/token/path"
"code.gitea.io/gitea/modules/indexer/code/internal"
@@ -150,7 +150,7 @@ func NewIndexer(indexDir string) *Indexer {
}
}
func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserError, batchReader *bufio.Reader, commitSha string,
func (b *Indexer) addUpdate(ctx context.Context, catFileBatch git.CatFileBatch, commitSha string,
update internal.FileUpdate, repo *repo_model.Repository, batch *inner_bleve.FlushingBatch,
) error {
// Ignore vendored files in code search
@@ -163,7 +163,7 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
var err error
if !update.Sized {
var stdout string
stdout, _, err = gitcmd.NewCommand("cat-file", "-s").AddDynamicArguments(update.BlobSha).RunStdString(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
stdout, _, err = gitrepo.RunCmdString(ctx, repo, gitcmd.NewCommand("cat-file", "-s").AddDynamicArguments(update.BlobSha))
if err != nil {
return err
}
@@ -176,16 +176,11 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
return b.addDelete(update.Filename, repo, batch)
}
if _, err := batchWriter.Write([]byte(update.BlobSha + "\n")); err != nil {
return err
}
_, _, size, err = git.ReadBatchLine(batchReader)
info, batchReader, err := catFileBatch.QueryContent(update.BlobSha)
if err != nil {
return err
}
fileContents, err := io.ReadAll(io.LimitReader(batchReader, size))
fileContents, err := io.ReadAll(io.LimitReader(batchReader, info.Size))
if err != nil {
return err
} else if !typesniffer.DetectContentType(fileContents).IsText() {
@@ -202,7 +197,7 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
RepoID: repo.ID,
CommitID: commitSha,
Filename: update.Filename,
Content: string(charset.ToUTF8DropErrors(fileContents, charset.ConvertOpts{})),
Content: string(charset.ToUTF8DropErrors(fileContents)),
Language: analyze.GetCodeLanguage(update.Filename, fileContents),
UpdatedAt: time.Now().UTC(),
})
@@ -217,18 +212,17 @@ func (b *Indexer) addDelete(filename string, repo *repo_model.Repository, batch
func (b *Indexer) Index(ctx context.Context, repo *repo_model.Repository, sha string, changes *internal.RepoChanges) error {
batch := inner_bleve.NewFlushingBatch(b.inner.Indexer, maxBatchSize)
if len(changes.Updates) > 0 {
gitBatch, err := git.NewBatch(ctx, repo.RepoPath())
catfileBatch, err := gitrepo.NewBatch(ctx, repo)
if err != nil {
return err
}
defer gitBatch.Close()
defer catfileBatch.Close()
for _, update := range changes.Updates {
if err := b.addUpdate(ctx, gitBatch.Writer, gitBatch.Reader, sha, update, repo, batch); err != nil {
if err := b.addUpdate(ctx, catfileBatch, sha, update, repo, batch); err != nil {
return err
}
}
gitBatch.Close()
}
for _, filename := range changes.RemovedFilenames {
if err := b.addDelete(filename, repo, batch); err != nil {
@@ -27,7 +27,7 @@ func TokenFilterConstructor(config map[string]any, cache *registry.Cache) (analy
func (s *TokenFilter) Filter(input analysis.TokenStream) analysis.TokenStream {
if len(input) == 1 {
// if there is only one token, we dont need to generate the reversed chain
// if there is only one token, we don't need to generate the reversed chain
return generatePathTokens(input, false)
}
@@ -4,7 +4,6 @@
package elasticsearch
import (
"bufio"
"context"
"fmt"
"io"
@@ -16,6 +15,7 @@ import (
"code.gitea.io/gitea/modules/charset"
"code.gitea.io/gitea/modules/git"
"code.gitea.io/gitea/modules/git/gitcmd"
"code.gitea.io/gitea/modules/gitrepo"
"code.gitea.io/gitea/modules/indexer"
"code.gitea.io/gitea/modules/indexer/code/internal"
indexer_internal "code.gitea.io/gitea/modules/indexer/internal"
@@ -138,7 +138,7 @@ const (
}`
)
func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserError, batchReader *bufio.Reader, sha string, update internal.FileUpdate, repo *repo_model.Repository) ([]elastic.BulkableRequest, error) {
func (b *Indexer) addUpdate(ctx context.Context, catFileBatch git.CatFileBatch, sha string, update internal.FileUpdate, repo *repo_model.Repository) ([]elastic.BulkableRequest, error) {
// Ignore vendored files in code search
if setting.Indexer.ExcludeVendored && analyze.IsVendor(update.Filename) {
return nil, nil
@@ -148,7 +148,7 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
var err error
if !update.Sized {
var stdout string
stdout, _, err = gitcmd.NewCommand("cat-file", "-s").AddDynamicArguments(update.BlobSha).RunStdString(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
stdout, _, err = gitrepo.RunCmdString(ctx, repo, gitcmd.NewCommand("cat-file", "-s").AddDynamicArguments(update.BlobSha))
if err != nil {
return nil, err
}
@@ -161,16 +161,12 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
return []elastic.BulkableRequest{b.addDelete(update.Filename, repo)}, nil
}
if _, err := batchWriter.Write([]byte(update.BlobSha + "\n")); err != nil {
return nil, err
}
_, _, size, err = git.ReadBatchLine(batchReader)
info, batchReader, err := catFileBatch.QueryContent(update.BlobSha)
if err != nil {
return nil, err
}
fileContents, err := io.ReadAll(io.LimitReader(batchReader, size))
fileContents, err := io.ReadAll(io.LimitReader(batchReader, info.Size))
if err != nil {
return nil, err
} else if !typesniffer.DetectContentType(fileContents).IsText() {
@@ -190,7 +186,7 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
Doc(map[string]any{
"repo_id": repo.ID,
"filename": update.Filename,
"content": string(charset.ToUTF8DropErrors(fileContents, charset.ConvertOpts{})),
"content": string(charset.ToUTF8DropErrors(fileContents)),
"commit_id": sha,
"language": analyze.GetCodeLanguage(update.Filename, fileContents),
"updated_at": timeutil.TimeStampNow(),
@@ -209,14 +205,14 @@ func (b *Indexer) addDelete(filename string, repo *repo_model.Repository) elasti
func (b *Indexer) Index(ctx context.Context, repo *repo_model.Repository, sha string, changes *internal.RepoChanges) error {
reqs := make([]elastic.BulkableRequest, 0)
if len(changes.Updates) > 0 {
batch, err := git.NewBatch(ctx, repo.RepoPath())
batch, err := gitrepo.NewBatch(ctx, repo)
if err != nil {
return err
}
defer batch.Close()
for _, update := range changes.Updates {
updateReqs, err := b.addUpdate(ctx, batch.Writer, batch.Reader, sha, update, repo)
updateReqs, err := b.addUpdate(ctx, batch, sha, update, repo)
if err != nil {
return err
}
@@ -224,7 +220,6 @@ func (b *Indexer) Index(ctx context.Context, repo *repo_model.Repository, sha st
reqs = append(reqs, updateReqs...)
}
}
batch.Close()
}
for _, filename := range changes.RemovedFilenames {
@@ -398,7 +393,7 @@ func (b *Indexer) Search(ctx context.Context, opts *internal.SearchOptions) (int
elastic.NewHighlight().
Field("content").
Field("filename").
NumOfFragments(0). // return all highting content on fragments
NumOfFragments(0). // return all highlighting content on fragments
HighlighterType("fvh"),
).
Sort("_score", false).
@@ -431,7 +426,7 @@ func (b *Indexer) Search(ctx context.Context, opts *internal.SearchOptions) (int
elastic.NewHighlight().
Field("content").
Field("filename").
NumOfFragments(0). // return all highting content on fragments
NumOfFragments(0). // return all highlighting content on fragments
HighlighterType("fvh"),
).
Sort("_score", false).
@@ -11,13 +11,14 @@ import (
repo_model "code.gitea.io/gitea/models/repo"
"code.gitea.io/gitea/modules/git"
"code.gitea.io/gitea/modules/git/gitcmd"
"code.gitea.io/gitea/modules/gitrepo"
"code.gitea.io/gitea/modules/indexer/code/internal"
"code.gitea.io/gitea/modules/log"
"code.gitea.io/gitea/modules/setting"
)
func getDefaultBranchSha(ctx context.Context, repo *repo_model.Repository) (string, error) {
stdout, _, err := gitcmd.NewCommand("show-ref", "-s").AddDynamicArguments(git.BranchPrefix+repo.DefaultBranch).RunStdString(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
stdout, _, err := gitrepo.RunCmdString(ctx, repo, gitcmd.NewCommand("show-ref", "-s").AddDynamicArguments(git.BranchPrefix+repo.DefaultBranch))
if err != nil {
return "", err
}
@@ -34,7 +35,7 @@ func getRepoChanges(ctx context.Context, repo *repo_model.Repository, revision s
needGenesis := len(status.CommitSha) == 0
if !needGenesis {
hasAncestorCmd := gitcmd.NewCommand("merge-base").AddDynamicArguments(status.CommitSha, revision)
stdout, _, _ := hasAncestorCmd.RunStdString(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
stdout, _, _ := gitrepo.RunCmdString(ctx, repo, hasAncestorCmd) // FIXME: error is not handled
needGenesis = len(stdout) == 0
}
@@ -87,7 +88,7 @@ func parseGitLsTreeOutput(stdout []byte) ([]internal.FileUpdate, error) {
// genesisChanges get changes to add repo to the indexer for the first time
func genesisChanges(ctx context.Context, repo *repo_model.Repository, revision string) (*internal.RepoChanges, error) {
var changes internal.RepoChanges
stdout, _, runErr := gitcmd.NewCommand("ls-tree", "--full-tree", "-l", "-r").AddDynamicArguments(revision).RunStdBytes(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
stdout, _, runErr := gitrepo.RunCmdBytes(ctx, repo, gitcmd.NewCommand("ls-tree", "--full-tree", "-l", "-r").AddDynamicArguments(revision))
if runErr != nil {
return nil, runErr
}
@@ -100,7 +101,7 @@ func genesisChanges(ctx context.Context, repo *repo_model.Repository, revision s
// nonGenesisChanges get changes since the previous indexer update
func nonGenesisChanges(ctx context.Context, repo *repo_model.Repository, revision string) (*internal.RepoChanges, error) {
diffCmd := gitcmd.NewCommand("diff", "--name-status").AddDynamicArguments(repo.CodeIndexerStatus.CommitSha, revision)
stdout, _, runErr := diffCmd.RunStdString(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
stdout, _, runErr := gitrepo.RunCmdString(ctx, repo, diffCmd)
if runErr != nil {
// previous commit sha may have been removed by a force push, so
// try rebuilding from scratch
@@ -118,7 +119,7 @@ func nonGenesisChanges(ctx context.Context, repo *repo_model.Repository, revisio
updateChanges := func() error {
cmd := gitcmd.NewCommand("ls-tree", "--full-tree", "-l").AddDynamicArguments(revision).
AddDashesAndList(updatedFilenames...)
lsTreeStdout, _, err := cmd.RunStdBytes(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
lsTreeStdout, _, err := gitrepo.RunCmdBytes(ctx, repo, cmd)
if err != nil {
return err
}
@@ -24,7 +24,7 @@ func indexSettingToGitGrepPathspecList() (list []string) {
return list
}
func PerformSearch(ctx context.Context, page int, repoID int64, gitRepo *git.Repository, ref git.RefName, keyword string, searchMode indexer.SearchModeType) (searchResults []*code_indexer.Result, total int, err error) {
func PerformSearch(ctx context.Context, page int, repoID int64, gitRepo *git.Repository, ref git.RefName, keyword string, searchMode indexer.SearchModeType) (searchResults []*code_indexer.Result, total int64, err error) {
grepMode := git.GrepModeWords
switch searchMode {
case indexer.SearchModeExact:
@@ -47,7 +47,7 @@ func PerformSearch(ctx context.Context, page int, repoID int64, gitRepo *git.Rep
return nil, 0, fmt.Errorf("gitRepo.GetRefCommitID: %w", err)
}
total = len(res)
total = int64(len(res))
pageStart := min((page-1)*setting.UI.RepoSearchPagingNum, len(res))
pageEnd := min(page*setting.UI.RepoSearchPagingNum, len(res))
res = res[pageStart:pageEnd]
@@ -17,20 +17,20 @@ func FilenameIndexerID(repoID int64, filename string) string {
}
func ParseIndexerID(indexerID string) (int64, string) {
index := strings.IndexByte(indexerID, '_')
if index == -1 {
before, after, ok := strings.Cut(indexerID, "_")
if !ok {
log.Error("Unexpected ID in repo indexer: %s", indexerID)
}
repoID, _ := internal.ParseBase36(indexerID[:index])
return repoID, indexerID[index+1:]
repoID, _ := internal.ParseBase36(before)
return repoID, after
}
func FilenameOfIndexerID(indexerID string) string {
index := strings.IndexByte(indexerID, '_')
if index == -1 {
_, after, ok := strings.Cut(indexerID, "_")
if !ok {
log.Error("Unexpected ID in repo indexer: %s", indexerID)
}
return indexerID[index+1:]
return after
}
// FilenameMatchIndexPos returns the boundaries of its first seven lines.
@@ -72,10 +72,11 @@ func writeStrings(buf *bytes.Buffer, strs ...string) error {
func HighlightSearchResultCode(filename, language string, lineNums []int, code string) []*ResultLine {
// we should highlight the whole code block first, otherwise it doesn't work well with multiple line highlighting
hl, _ := highlight.Code(filename, language, code)
highlightedLines := strings.Split(string(hl), "\n")
lexer := highlight.DetectChromaLexerByFileName(filename, language)
hl := highlight.RenderCodeByLexer(lexer, code)
highlightedLines := highlight.UnsafeSplitHighlightedLines(hl)
// The lineNums outputted by highlight.Code might not match the original lineNums, because "highlight" removes the last `\n`
// The lineNums outputted by render might not match the original lineNums, because "highlight" removes the last `\n`
lines := make([]*ResultLine, min(len(highlightedLines), len(lineNums)))
for i := range lines {
lines[i] = &ResultLine{
@@ -129,7 +130,7 @@ func searchResult(result *internal.SearchResult, startIndex, endIndex int) (*Res
}
// PerformSearch perform a search on a repository
func PerformSearch(ctx context.Context, opts *SearchOptions) (int, []*Result, []*SearchResultLanguages, error) {
func PerformSearch(ctx context.Context, opts *SearchOptions) (int64, []*Result, []*SearchResultLanguages, error) {
if opts == nil || len(opts.Keyword) == 0 {
return 0, nil, nil, nil
}
@@ -148,5 +149,5 @@ func PerformSearch(ctx context.Context, opts *SearchOptions) (int, []*Result, []
return 0, nil, nil, err
}
}
return int(total), displayResults, resultLanguages, nil
return total, displayResults, resultLanguages, nil
}