fork of https://github.com/sourcegraph/zoekt
1// Copyright 2016 Google Inc. All rights reserved.
2//
3// Licensed under the Apache License, Version 2.0 (the "License");
4// you may not use this file except in compliance with the License.
5// You may obtain a copy of the License at
6//
7// http://www.apache.org/licenses/LICENSE-2.0
8//
9// Unless required by applicable law or agreed to in writing, software
10// distributed under the License is distributed on an "AS IS" BASIS,
11// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12// See the License for the specific language governing permissions and
13// limitations under the License.
14
15// Package gitindex provides functions for indexing Git repositories.
16package gitindex
17
18import (
19 "bytes"
20 "context"
21 "errors"
22 "fmt"
23 "io"
24 "log"
25 "math"
26 "net/url"
27 "os"
28 "path/filepath"
29 "regexp"
30 "sort"
31 "strconv"
32 "strings"
33
34 "github.com/go-git/go-billy/v5/osfs"
35 "github.com/go-git/go-git/v5/plumbing/cache"
36 "github.com/go-git/go-git/v5/storage/filesystem"
37 "github.com/sourcegraph/zoekt"
38 "github.com/sourcegraph/zoekt/build"
39 "github.com/sourcegraph/zoekt/ignore"
40
41 "github.com/go-git/go-git/v5/config"
42 "github.com/go-git/go-git/v5/plumbing"
43 "github.com/go-git/go-git/v5/plumbing/object"
44
45 git "github.com/go-git/go-git/v5"
46)
47
48// FindGitRepos finds directories holding git repositories below the
49// given directory. It will find both bare and the ".git" dirs in
50// non-bare repositories. It returns the full path including the dir
51// passed in.
52func FindGitRepos(dir string) ([]string, error) {
53 arg, err := filepath.Abs(dir)
54 if err != nil {
55 return nil, err
56 }
57 var dirs []string
58 if err := filepath.Walk(arg, func(name string, fi os.FileInfo, err error) error {
59 // Best-effort, ignore filepath.Walk failing
60 if err != nil {
61 return nil
62 }
63
64 if fi, err := os.Lstat(filepath.Join(name, ".git")); err == nil && fi.IsDir() {
65 dirs = append(dirs, filepath.Join(name, ".git"))
66 return filepath.SkipDir
67 }
68
69 if !strings.HasSuffix(name, ".git") || !fi.IsDir() {
70 return nil
71 }
72
73 fi, err = os.Lstat(filepath.Join(name, "objects"))
74 if err != nil || !fi.IsDir() {
75 return nil
76 }
77
78 dirs = append(dirs, name)
79 return filepath.SkipDir
80 }); err != nil {
81 return nil, err
82 }
83
84 return dirs, nil
85}
86
87// setTemplates fills in URL templates for known git hosting
88// sites.
89func setTemplates(repo *zoekt.Repository, u *url.URL, typ string) error {
90 if u.Scheme == "ssh+git" {
91 u.Scheme = "https"
92 u.User = nil
93 }
94
95 // helper to generate u.JoinPath as a template
96 varVersion := ".Version"
97 varPath := ".Path"
98 urlJoinPath := func(elem ...string) string {
99 elem = append([]string{u.String()}, elem...)
100 var parts []string
101 for _, e := range elem {
102 if e == varVersion || e == varPath {
103 parts = append(parts, e)
104 } else {
105 parts = append(parts, strconv.Quote(e))
106 }
107 }
108 return fmt.Sprintf("{{URLJoinPath %s}}", strings.Join(parts, " "))
109 }
110
111 repo.URL = u.String()
112 switch typ {
113 case "gitiles":
114 // eg. https://gerrit.googlesource.com/gitiles/+/master/tools/run_dev.sh#20
115 repo.CommitURLTemplate = urlJoinPath("+", varVersion)
116 repo.FileURLTemplate = urlJoinPath("+", varVersion, varPath)
117 repo.LineFragmentTemplate = "#{{.LineNumber}}"
118 case "github":
119 // eg. https://github.com/hanwen/go-fuse/blob/notify/genversion.sh#L10
120 repo.CommitURLTemplate = urlJoinPath("commit", varVersion)
121 repo.FileURLTemplate = urlJoinPath("blob", varVersion, varPath)
122 repo.LineFragmentTemplate = "#L{{.LineNumber}}"
123 case "cgit":
124 // http://git.savannah.gnu.org/cgit/lilypond.git/tree/elisp/lilypond-mode.el?h=dev/philh&id=b2ca0fefe3018477aaca23b6f672c7199ba5238e#n100
125 repo.CommitURLTemplate = urlJoinPath("commit") + "/?id={{.Version}}"
126 repo.FileURLTemplate = urlJoinPath("tree", varPath) + "/?id={{.Version}}"
127 repo.LineFragmentTemplate = "#n{{.LineNumber}}"
128 case "gitweb":
129 // https://gerrit.libreoffice.org/gitweb?p=online.git;a=blob;f=Makefile.am;h=cfcfd7c36fbae10e269653dc57a9b68c92d4c10b;hb=848145503bf7b98ce4a4aa0a858a0d71dd0dbb26#l10
130 repo.FileURLTemplate = u.String() + ";a=blob;f={{.Path}};hb={{.Version}}"
131 repo.CommitURLTemplate = u.String() + ";a=commit;h={{.Version}}"
132 repo.LineFragmentTemplate = "#l{{.LineNumber}}"
133 case "source.bazel.build":
134 // https://source.bazel.build/bazel/+/57bc201346e61c62a921c1cbf32ad24f185c10c9
135 // https://source.bazel.build/bazel/+/57bc201346e61c62a921c1cbf32ad24f185c10c9:tools/cpp/BUILD.empty;l=10
136 repo.CommitURLTemplate = u.String() + "/%2B/{{.Version}}"
137 repo.FileURLTemplate = u.String() + "/%2B/{{.Version}}:{{.Path}}"
138 repo.LineFragmentTemplate = ";l={{.LineNumber}}"
139 case "bitbucket-server":
140 // https://<bitbucketserver-host>/projects/<project>/repos/<repo>/commits/5be7ca73b898bf17a08e607918accfdeafe1e0bc
141 // https://<bitbucketserver-host>/projects/<project>/repos/<repo>/browse/<file>?at=5be7ca73b898bf17a08e607918accfdeafe1e0bc
142 repo.CommitURLTemplate = urlJoinPath("commits", varVersion)
143 repo.FileURLTemplate = urlJoinPath(varPath) + "?at={{.Version}}"
144 repo.LineFragmentTemplate = "#{{.LineNumber}}"
145 case "gitlab":
146 // https://gitlab.com/gitlab-org/omnibus-gitlab/-/commit/b152c864303dae0e55377a1e2c53c9592380ffed
147 // https://gitlab.com/gitlab-org/omnibus-gitlab/-/blob/aad04155b3f6fc50ede88aedaee7fc624d481149/files/gitlab-config-template/gitlab.rb.template
148 repo.CommitURLTemplate = urlJoinPath("-/commit", varVersion)
149 repo.FileURLTemplate = urlJoinPath("-/blob", varVersion, varPath)
150 repo.LineFragmentTemplate = "#L{{.LineNumber}}"
151 case "gitea":
152 repo.CommitURLTemplate = urlJoinPath("commit", varVersion)
153 // NOTE The `display=source` query parameter is required to disable file rendering.
154 // Since line numbers are disabled in rendered files, you wouldn't be able to jump to
155 // a line without `display=source`. This is supported since gitea 1.17.0.
156 // When /src/{{.Version}} is used it will redirect to /src/commit/{{.Version}},
157 // but the query parameters are obmitted.
158 repo.FileURLTemplate = urlJoinPath("src/commit", varVersion, varPath) + "?display=source"
159 repo.LineFragmentTemplate = "#L{{.LineNumber}}"
160 default:
161 return fmt.Errorf("URL scheme type %q unknown", typ)
162 }
163 return nil
164}
165
166// getCommit returns a tree object for the given reference.
167func getCommit(repo *git.Repository, prefix, ref string) (*object.Commit, error) {
168 sha1, err := repo.ResolveRevision(plumbing.Revision(ref))
169 // ref might be a branch name (e.g. "master") add branch prefix and try again.
170 if err != nil {
171 sha1, err = repo.ResolveRevision(plumbing.Revision(filepath.Join(prefix, ref)))
172 }
173 if err != nil {
174 return nil, err
175 }
176
177 commitObj, err := repo.CommitObject(*sha1)
178 if err != nil {
179 return nil, err
180 }
181 return commitObj, nil
182}
183
184func configLookupRemoteURL(cfg *config.Config, key string) string {
185 rc := cfg.Remotes[key]
186 if rc == nil || len(rc.URLs) == 0 {
187 return ""
188 }
189 return rc.URLs[0]
190}
191
192var sshRelativeURLRegexp = regexp.MustCompile(`^([^@]+)@([^:]+):(.*)$`)
193
194func setTemplatesFromConfig(desc *zoekt.Repository, repoDir string) error {
195 repo, err := git.PlainOpen(repoDir)
196 if err != nil {
197 return err
198 }
199
200 cfg, err := repo.Config()
201 if err != nil {
202 return err
203 }
204
205 sec := cfg.Raw.Section("zoekt")
206
207 webURLStr := sec.Options.Get("web-url")
208 webURLType := sec.Options.Get("web-url-type")
209
210 if webURLType != "" && webURLStr != "" {
211 webURL, err := url.Parse(webURLStr)
212 if err != nil {
213 return err
214 }
215 if err := setTemplates(desc, webURL, webURLType); err != nil {
216 return err
217 }
218 } else if webURLStr != "" {
219 desc.URL = webURLStr
220 }
221
222 name := sec.Options.Get("name")
223 if name != "" {
224 desc.Name = name
225 } else {
226 remoteURL := configLookupRemoteURL(cfg, "origin")
227 if remoteURL == "" {
228 return nil
229 }
230 if sm := sshRelativeURLRegexp.FindStringSubmatch(remoteURL); sm != nil {
231 user := sm[1]
232 host := sm[2]
233 path := sm[3]
234
235 remoteURL = fmt.Sprintf("ssh+git://%s@%s/%s", user, host, path)
236 }
237
238 u, err := url.Parse(remoteURL)
239 if err != nil {
240 return err
241 }
242 if err := SetTemplatesFromOrigin(desc, u); err != nil {
243 return err
244 }
245 }
246
247 id, _ := strconv.ParseUint(sec.Options.Get("repoid"), 10, 32)
248 desc.ID = uint32(id)
249
250 if desc.RawConfig == nil {
251 desc.RawConfig = map[string]string{}
252 }
253 for _, o := range sec.Options {
254 desc.RawConfig[o.Key] = o.Value
255 }
256
257 // Ranking info.
258
259 // Github:
260 traction := 0
261 for _, s := range []string{"github-stars", "github-forks", "github-watchers", "github-subscribers"} {
262 f, err := strconv.Atoi(sec.Options.Get(s))
263 if err == nil {
264 traction += f
265 }
266 }
267
268 if strings.Contains(desc.Name, "googlesource.com/") && traction == 0 {
269 // Pretend everything on googlesource.com has 1000
270 // github stars.
271 traction = 1000
272 }
273
274 if traction > 0 {
275 l := math.Log(float64(traction))
276 desc.Rank = uint16((1.0 - 1.0/math.Pow(1+l, 0.6)) * 10000)
277 }
278
279 return nil
280}
281
282// SetTemplatesFromOrigin fills in templates based on the origin URL.
283func SetTemplatesFromOrigin(desc *zoekt.Repository, u *url.URL) error {
284 desc.Name = filepath.Join(u.Host, strings.TrimSuffix(u.Path, ".git"))
285
286 if strings.HasSuffix(u.Host, ".googlesource.com") {
287 return setTemplates(desc, u, "gitiles")
288 } else if u.Host == "github.com" {
289 u.Path = strings.TrimSuffix(u.Path, ".git")
290 return setTemplates(desc, u, "github")
291 } else {
292 return fmt.Errorf("unknown git hosting site %q", u)
293 }
294}
295
296// The Options structs controls details of the indexing process.
297type Options struct {
298 // The repository to be indexed.
299 RepoDir string
300
301 // If set, follow submodule links. This requires RepoCacheDir to be set.
302 Submodules bool
303
304 // If set, skip indexing if the existing index shard is newer
305 // than the refs in the repository.
306 Incremental bool
307
308 // Don't error out if some branch is missing
309 AllowMissingBranch bool
310
311 // Specifies the root of a Repository cache. Needed for submodule indexing.
312 RepoCacheDir string
313
314 // Indexing options.
315 BuildOptions build.Options
316
317 // Prefix of the branch to index, e.g. `remotes/origin`.
318 BranchPrefix string
319
320 // List of branch names to index, e.g. []string{"HEAD", "stable"}
321 Branches []string
322
323 // DeltaShardNumberFallbackThreshold defines an upper limit (inclusive) on the number of preexisting shards
324 // that can exist before attempting another delta build. If the number of preexisting shards exceeds this threshold,
325 // then a normal build will be performed instead.
326 //
327 // If DeltaShardNumberFallbackThreshold is 0, then this fallback behavior is disabled:
328 // a delta build will always be performed regardless of the number of preexisting shards.
329 DeltaShardNumberFallbackThreshold uint64
330}
331
332func expandBranches(repo *git.Repository, bs []string, prefix string) ([]string, error) {
333 var result []string
334 for _, b := range bs {
335 // Sourcegraph: We disable resolving refs. We want to return the exact ref
336 // requested so we can match it up.
337 if b == "HEAD" && false {
338 ref, err := repo.Head()
339 if err != nil {
340 return nil, err
341 }
342
343 result = append(result, strings.TrimPrefix(ref.Name().String(), prefix))
344 continue
345 }
346
347 if strings.Contains(b, "*") {
348 iter, err := repo.Branches()
349 if err != nil {
350 return nil, err
351 }
352
353 defer iter.Close()
354 for {
355 ref, err := iter.Next()
356 if err == io.EOF {
357 break
358 }
359 if err != nil {
360 return nil, err
361 }
362
363 name := ref.Name().Short()
364 if matched, err := filepath.Match(b, name); err != nil {
365 return nil, err
366 } else if !matched {
367 continue
368 }
369
370 result = append(result, strings.TrimPrefix(name, prefix))
371 }
372 continue
373 }
374
375 result = append(result, b)
376 }
377
378 return result, nil
379}
380
381// IndexGitRepo indexes the git repository as specified by the options.
382// The returned bool indicates whether the index was updated as a result. This
383// can be informative if doing incremental indexing.
384func IndexGitRepo(opts Options) (bool, error) {
385 return indexGitRepo(opts, gitIndexConfig{})
386}
387
388// indexGitRepo indexes the git repository as specified by the options and the provided gitIndexConfig.
389// The returned bool indicates whether the index was updated as a result. This
390// can be informative if doing incremental indexing.
391func indexGitRepo(opts Options, config gitIndexConfig) (bool, error) {
392 prepareDeltaBuild := prepareDeltaBuild
393 if config.prepareDeltaBuild != nil {
394 prepareDeltaBuild = config.prepareDeltaBuild
395 }
396
397 prepareNormalBuild := prepareNormalBuild
398 if config.prepareNormalBuild != nil {
399 prepareNormalBuild = config.prepareNormalBuild
400 }
401
402 // Set max thresholds, since we use them in this function.
403 opts.BuildOptions.SetDefaults()
404 if opts.RepoDir == "" {
405 return false, fmt.Errorf("gitindex: must set RepoDir")
406 }
407
408 opts.BuildOptions.RepositoryDescription.Source = opts.RepoDir
409
410 var repo *git.Repository
411 // TODO: remove this feature flag once we test this on a large-scale instance.
412 legacyRepoOpen := os.Getenv("ZOEKT_DISABLE_GOGIT_OPTIMIZATION")
413 if b, err := strconv.ParseBool(legacyRepoOpen); b && err == nil {
414 repo, err = git.PlainOpen(opts.RepoDir)
415 if err != nil {
416 return false, fmt.Errorf("git.PlainOpen: %w", err)
417 }
418 } else {
419 var repoCloser io.Closer
420 repo, repoCloser, err = openRepo(opts.RepoDir)
421 if err != nil {
422 return false, fmt.Errorf("openRepo: %w", err)
423 }
424 defer repoCloser.Close()
425 }
426
427 if err := setTemplatesFromConfig(&opts.BuildOptions.RepositoryDescription, opts.RepoDir); err != nil {
428 log.Printf("setTemplatesFromConfig(%s): %s", opts.RepoDir, err)
429 }
430
431 branches, err := expandBranches(repo, opts.Branches, opts.BranchPrefix)
432 if err != nil {
433 return false, fmt.Errorf("expandBranches: %w", err)
434 }
435 for _, b := range branches {
436 commit, err := getCommit(repo, opts.BranchPrefix, b)
437 if err != nil {
438 if opts.AllowMissingBranch && err.Error() == "reference not found" {
439 continue
440 }
441
442 return false, fmt.Errorf("getCommit(%q, %q): %w", opts.BranchPrefix, b, err)
443 }
444
445 opts.BuildOptions.RepositoryDescription.Branches = append(opts.BuildOptions.RepositoryDescription.Branches, zoekt.RepositoryBranch{
446 Name: b,
447 Version: commit.Hash.String(),
448 })
449
450 if when := commit.Committer.When; when.After(opts.BuildOptions.RepositoryDescription.LatestCommitDate) {
451 opts.BuildOptions.RepositoryDescription.LatestCommitDate = when
452 }
453 }
454
455 if opts.Incremental && opts.BuildOptions.IncrementalSkipIndexing() {
456 return false, nil
457 }
458
459 // branch => (path, sha1) => repo.
460 var repos map[fileKey]BlobLocation
461
462 // Branch => Repo => SHA1
463 var branchVersions map[string]map[string]plumbing.Hash
464
465 // set of file paths that have been changed or deleted since
466 // the last indexed commit
467 //
468 // These only have an effect on delta builds
469 var changedOrRemovedFiles []string
470
471 if opts.BuildOptions.IsDelta {
472 repos, branchVersions, changedOrRemovedFiles, err = prepareDeltaBuild(opts, repo)
473 if err != nil {
474 log.Printf("delta build: falling back to normal build since delta build failed, repository=%q, err=%s", opts.BuildOptions.RepositoryDescription.Name, err)
475 opts.BuildOptions.IsDelta = false
476 }
477 }
478
479 if !opts.BuildOptions.IsDelta {
480 repos, branchVersions, err = prepareNormalBuild(opts, repo)
481 if err != nil {
482 return false, fmt.Errorf("preparing normal build: %w", err)
483 }
484 }
485
486 reposByPath := map[string]BlobLocation{}
487 for key, info := range repos {
488 reposByPath[key.SubRepoPath] = info
489 }
490
491 opts.BuildOptions.SubRepositories = map[string]*zoekt.Repository{}
492 for path, info := range reposByPath {
493 tpl := opts.BuildOptions.RepositoryDescription
494 if path != "" {
495 tpl = zoekt.Repository{URL: info.URL.String()}
496 if err := SetTemplatesFromOrigin(&tpl, info.URL); err != nil {
497 log.Printf("setTemplatesFromOrigin(%s, %s): %s", path, info.URL, err)
498 }
499 }
500 opts.BuildOptions.SubRepositories[path] = &tpl
501 }
502
503 for _, br := range opts.BuildOptions.RepositoryDescription.Branches {
504 for path, repo := range opts.BuildOptions.SubRepositories {
505 id := branchVersions[br.Name][path]
506 repo.Branches = append(repo.Branches, zoekt.RepositoryBranch{
507 Name: br.Name,
508 Version: id.String(),
509 })
510 }
511 }
512
513 builder, err := build.NewBuilder(opts.BuildOptions)
514 if err != nil {
515 return false, fmt.Errorf("build.NewBuilder: %w", err)
516 }
517
518 // Preparing the build can consume substantial memory, so check usage before starting to index.
519 builder.CheckMemoryUsage()
520
521 // we don't need to check error, since we either already have an error, or
522 // we returning the first call to builder.Finish.
523 defer builder.Finish() // nolint:errcheck
524
525 for _, f := range changedOrRemovedFiles {
526 builder.MarkFileAsChangedOrRemoved(f)
527 }
528
529 var names []string
530 fileKeys := map[string][]fileKey{}
531 totalFiles := 0
532
533 for key := range repos {
534 n := key.FullPath()
535 fileKeys[n] = append(fileKeys[n], key)
536 names = append(names, n)
537 totalFiles++
538 }
539
540 sort.Strings(names)
541 names = uniq(names)
542
543 log.Printf("attempting to index %d total files", totalFiles)
544 for idx, name := range names {
545 keys := fileKeys[name]
546
547 for _, key := range keys {
548 doc, err := createDocument(key, repos, opts.BuildOptions)
549 if err != nil {
550 return false, err
551 }
552
553 if err := builder.Add(doc); err != nil {
554 return false, fmt.Errorf("error adding document with name %s: %w", key.FullPath(), err)
555 }
556
557 if idx%10_000 == 0 {
558 builder.CheckMemoryUsage()
559 }
560 }
561 }
562 return true, builder.Finish()
563}
564
565// openRepo opens a git repository in a way that's optimized for indexing.
566//
567// It copies the relevant logic from git.PlainOpen, and tweaks certain filesystem options.
568func openRepo(repoDir string) (*git.Repository, io.Closer, error) {
569 fs := osfs.New(repoDir)
570
571 // Check if the root directory exists.
572 if _, err := fs.Stat(""); err != nil {
573 if os.IsNotExist(err) {
574 return nil, nil, git.ErrRepositoryNotExists
575 }
576 return nil, nil, err
577 }
578
579 // If there's a .git directory, use that as the new root.
580 if fi, err := fs.Stat(git.GitDirName); err == nil && fi.IsDir() {
581 if fs, err = fs.Chroot(git.GitDirName); err != nil {
582 return nil, nil, fmt.Errorf("fs.Chroot: %w", err)
583 }
584 }
585
586 s := filesystem.NewStorageWithOptions(fs, cache.NewObjectLRUDefault(), filesystem.Options{
587 // Cache the packfile handles, preventing the packfile from being opened then closed on every object access
588 KeepDescriptors: true,
589 // Disable caching for most objects, by setting the threshold to 1 byte. This avoids allocating a bunch of
590 // in-memory objects that are unlikely to be reused, since we only read each file once. Note: go-git still
591 // proactively caches objects under 16KB (see smallObjectThreshold in packfile logic).
592 LargeObjectThreshold: 1,
593 })
594
595 // Because we're keeping descriptors open, we need to close the storage object when we're done.
596 repo, err := git.Open(s, fs)
597 return repo, s, err
598}
599
600type repoPathRanks struct {
601 MeanRank float64 `json:"mean_reference_count"`
602 Paths map[string]float64 `json:"paths"`
603}
604
605// rank returns the rank for a given path. It uses these rules:
606// - If we have a concrete rank for this file, always use it
607// - If there's no rank, and it's a low priority file like a test, then use rank 0
608// - Otherwise use the mean rank of this repository, to avoid giving it a big disadvantage
609func (r repoPathRanks) rank(path string, content []byte) float64 {
610 if rank, ok := r.Paths[path]; ok {
611 return rank
612 } else if build.IsLowPriority(path, content) {
613 return 0.0
614 } else {
615 return r.MeanRank
616 }
617}
618
619func newIgnoreMatcher(tree *object.Tree) (*ignore.Matcher, error) {
620 ignoreFile, err := tree.File(ignore.IgnoreFile)
621 if err == object.ErrFileNotFound {
622 return &ignore.Matcher{}, nil
623 }
624 if err != nil {
625 return nil, err
626 }
627 content, err := ignoreFile.Contents()
628 if err != nil {
629 return nil, err
630 }
631 return ignore.ParseIgnoreFile(strings.NewReader(content))
632}
633
634// prepareDeltaBuildFunc is a function that calculates the necessary metadata for preparing
635// a build.Builder instance for generating a delta build.
636type prepareDeltaBuildFunc func(options Options, repository *git.Repository) (repos map[fileKey]BlobLocation, branchVersions map[string]map[string]plumbing.Hash, changedOrDeletedPaths []string, err error)
637
638// prepareNormalBuildFunc is a function that calculates the necessary metadata for preparing
639// a build.Builder instance for generating a normal build.
640type prepareNormalBuildFunc func(options Options, repository *git.Repository) (repos map[fileKey]BlobLocation, branchVersions map[string]map[string]plumbing.Hash, err error)
641
642type gitIndexConfig struct {
643 // prepareDeltaBuild, if not nil, is the function that is used to calculate the metadata that will be used to
644 // prepare the build.Builder instance for generating a delta build.
645 //
646 // If prepareDeltaBuild is nil, gitindex.prepareDeltaBuild will be used instead.
647 prepareDeltaBuild prepareDeltaBuildFunc
648
649 // prepareNormalBuild, if not nil, is the function that is used to calculate the metadata that will be used to
650 // prepare the build.Builder instance for generating a normal build.
651 //
652 // If prepareNormalBuild is nil, gitindex.prepareNormalBuild will be used instead.
653 prepareNormalBuild prepareNormalBuildFunc
654}
655
656func prepareDeltaBuild(options Options, repository *git.Repository) (repos map[fileKey]BlobLocation, branchVersions map[string]map[string]plumbing.Hash, changedOrDeletedPaths []string, err error) {
657 if options.Submodules {
658 return nil, nil, nil, fmt.Errorf("delta builds currently don't support submodule indexing")
659 }
660
661 // discover what commits we indexed during our last build
662 existingRepository, _, ok, err := options.BuildOptions.FindRepositoryMetadata()
663 if err != nil {
664 return nil, nil, nil, fmt.Errorf("failed to get repository metadata: %w", err)
665 }
666
667 if !ok {
668 return nil, nil, nil, fmt.Errorf("no existing shards found for repository")
669 }
670
671 if options.DeltaShardNumberFallbackThreshold > 0 {
672 // HACK: For our interim compaction strategy, we force a full normal index once
673 // the number of shards on disk for this repository exceeds the provided threshold.
674 //
675 // This strategy obviously isn't optimal (as an example: we currently can't differentiate
676 // between "normal" and "delta" shards, so repositories like the gigarepo that generate a large number of shards per
677 // build would be disproportionately affected by this), but it'll allow us to continue experimenting on real workloads
678 // while we create a better compaction strategy).
679
680 oldShards := options.BuildOptions.FindAllShards()
681 if uint64(len(oldShards)) > options.DeltaShardNumberFallbackThreshold {
682 return nil, nil, nil, fmt.Errorf("number of existing shards (%d) > requested shard threshold (%d)", len(oldShards), options.DeltaShardNumberFallbackThreshold)
683 }
684 }
685
686 // Check to see if the set of branch names is consistent with what we last indexed.
687 // If it isn't consistent, that we can't proceed with a delta build (and the caller should fall back to a
688 // normal one).
689
690 if !build.BranchNamesEqual(existingRepository.Branches, options.BuildOptions.RepositoryDescription.Branches) {
691 var existingBranchNames []string
692 for _, b := range existingRepository.Branches {
693 existingBranchNames = append(existingBranchNames, b.Name)
694 }
695
696 var optionsBranchNames []string
697 for _, b := range options.BuildOptions.RepositoryDescription.Branches {
698 optionsBranchNames = append(optionsBranchNames, b.Name)
699 }
700
701 existingBranchList := strings.Join(existingBranchNames, ", ")
702 optionsBranchList := strings.Join(optionsBranchNames, ", ")
703
704 return nil, nil, nil, fmt.Errorf("requested branch set in build options (%q) != branch set found on disk (%q) - branch set must be the same for delta shards", optionsBranchList, existingBranchList)
705 }
706
707 // Check if the build options hash does not match the repository metadata's hash
708 // If it does not match then one or more index options has changed and will require a normal build instead of a delta build
709 if options.BuildOptions.GetHash() != existingRepository.IndexOptions {
710 return nil, nil, nil, fmt.Errorf("one or more index options previously stored for repository %s (ID: %d) does not match the index options for this requested build; These index option updates are incompatible with delta build. new index options: %+v", existingRepository.Name, existingRepository.ID, options.BuildOptions.HashOptions())
711 }
712
713 // branch => (path, sha1) => repo.
714 repos = map[fileKey]BlobLocation{}
715
716 // branch name -> git worktree at most current commit
717 branchToCurrentTree := make(map[string]*object.Tree, len(options.Branches))
718
719 for _, b := range options.Branches {
720 commit, err := getCommit(repository, options.BranchPrefix, b)
721 if err != nil {
722 return nil, nil, nil, fmt.Errorf("getting last current commit for branch %q: %w", b, err)
723 }
724
725 tree, err := commit.Tree()
726 if err != nil {
727 return nil, nil, nil, fmt.Errorf("getting current git tree for branch %q: %w", b, err)
728 }
729
730 branchToCurrentTree[b] = tree
731 }
732
733 rawURL := options.BuildOptions.RepositoryDescription.URL
734 u, err := url.Parse(rawURL)
735 if err != nil {
736 return nil, nil, nil, fmt.Errorf("parsing repository URL %q: %w", rawURL, err)
737 }
738
739 // TODO: Support repository submodules for delta builds
740
741 // loop over all branches, calculate the diff between our
742 // last indexed commit and the current commit, and add files mentioned in the diff
743 for _, branch := range existingRepository.Branches {
744 lastIndexedCommit, err := getCommit(repository, "", branch.Version)
745 if err != nil {
746 return nil, nil, nil, fmt.Errorf("getting last indexed commit for branch %q: %w", branch.Name, err)
747 }
748
749 lastIndexedTree, err := lastIndexedCommit.Tree()
750 if err != nil {
751 return nil, nil, nil, fmt.Errorf("getting lasted indexed git tree for branch %q: %w", branch.Name, err)
752 }
753
754 changes, err := object.DiffTreeWithOptions(context.Background(), lastIndexedTree, branchToCurrentTree[branch.Name], &object.DiffTreeOptions{DetectRenames: false})
755 if err != nil {
756 return nil, nil, nil, fmt.Errorf("generating changeset for branch %q: %w", branch.Name, err)
757 }
758
759 for i, c := range changes {
760 oldFile, newFile, err := c.Files()
761 if err != nil {
762 return nil, nil, nil, fmt.Errorf("change #%d: getting files before and after change: %w", i, err)
763 }
764
765 if newFile != nil {
766 // note: newFile.Name could be a path that isn't relative to the repository root - using the
767 // change's Name field is the only way that @ggilmore saw to get the full path relative to the root
768 newFileRelativeRootPath := c.To.Name
769
770 // TODO@ggilmore: HACK - remove once ignore files are supported in delta builds
771 if newFileRelativeRootPath == ignore.IgnoreFile {
772 return nil, nil, nil, fmt.Errorf("%q file is not yet supported in delta builds", ignore.IgnoreFile)
773 }
774
775 // either file is added or renamed, so we need to add the new version to the build
776 file := fileKey{Path: newFileRelativeRootPath, ID: newFile.Hash}
777 if existing, ok := repos[file]; ok {
778 existing.Branches = append(existing.Branches, branch.Name)
779 repos[file] = existing
780 } else {
781 repos[file] = BlobLocation{
782 GitRepo: repository,
783 URL: u,
784 Branches: []string{branch.Name},
785 }
786 }
787 }
788
789 if oldFile == nil {
790 // file added - nothing more to do
791 continue
792 }
793
794 // Note: oldFile.Name could be a path that isn't relative to the repository root - using the
795 // change's "Name" field is the only way that ggilmore saw to get the full path relative to the root
796 oldFileRelativeRootPath := c.From.Name
797
798 if oldFileRelativeRootPath == ignore.IgnoreFile {
799 return nil, nil, nil, fmt.Errorf("%q file is not yet supported in delta builds", ignore.IgnoreFile)
800 }
801
802 // The file is either modified or deleted. So, we need to add ALL versions
803 // of the old file (across all branches) to the build.
804 for b, currentTree := range branchToCurrentTree {
805 f, err := currentTree.File(oldFileRelativeRootPath)
806 if err != nil {
807 // the file doesn't exist in this branch
808 if errors.Is(err, object.ErrFileNotFound) {
809 continue
810 }
811
812 return nil, nil, nil, fmt.Errorf("getting hash for file %q in branch %q: %w", oldFile.Name, b, err)
813 }
814
815 file := fileKey{Path: oldFileRelativeRootPath, ID: f.ID()}
816 if existing, ok := repos[file]; ok {
817 existing.Branches = append(existing.Branches, b)
818 repos[file] = existing
819 } else {
820 repos[file] = BlobLocation{
821 GitRepo: repository,
822 URL: u,
823 Branches: []string{b},
824 }
825 }
826 }
827
828 changedOrDeletedPaths = append(changedOrDeletedPaths, oldFileRelativeRootPath)
829 }
830 }
831
832 // we need to de-duplicate the branch map before returning it - it's possible for the same
833 // branch to have been added multiple times if a file has been modified across multiple commits
834 for _, info := range repos {
835 sort.Strings(info.Branches)
836 info.Branches = uniq(info.Branches)
837 }
838
839 // we also need to de-duplicate the list of changed or deleted file paths, it's also possible to have duplicates
840 // for the same reasoning as above
841 sort.Strings(changedOrDeletedPaths)
842 changedOrDeletedPaths = uniq(changedOrDeletedPaths)
843
844 return repos, nil, changedOrDeletedPaths, nil
845}
846
847func prepareNormalBuild(options Options, repository *git.Repository) (repos map[fileKey]BlobLocation, branchVersions map[string]map[string]plumbing.Hash, err error) {
848 var repoCache *RepoCache
849 if options.Submodules {
850 repoCache = NewRepoCache(options.RepoCacheDir)
851 }
852
853 // Branch => Repo => SHA1
854 branchVersions = map[string]map[string]plumbing.Hash{}
855
856 branches, err := expandBranches(repository, options.Branches, options.BranchPrefix)
857 if err != nil {
858 return nil, nil, fmt.Errorf("expandBranches: %w", err)
859 }
860
861 rw := NewRepoWalker(repository, options.BuildOptions.RepositoryDescription.URL, repoCache)
862 for _, b := range branches {
863 commit, err := getCommit(repository, options.BranchPrefix, b)
864 if err != nil {
865 if options.AllowMissingBranch && err.Error() == "reference not found" {
866 continue
867 }
868
869 return nil, nil, fmt.Errorf("getCommit: %w", err)
870 }
871
872 tree, err := commit.Tree()
873 if err != nil {
874 return nil, nil, fmt.Errorf("commit.Tree: %w", err)
875 }
876
877 ig, err := newIgnoreMatcher(tree)
878 if err != nil {
879 return nil, nil, fmt.Errorf("newIgnoreMatcher: %w", err)
880 }
881
882 subVersions, err := rw.CollectFiles(tree, b, ig)
883 if err != nil {
884 return nil, nil, fmt.Errorf("CollectFiles: %w", err)
885 }
886
887 branchVersions[b] = subVersions
888 }
889
890 return rw.Files, branchVersions, nil
891}
892
893func createDocument(key fileKey,
894 repos map[fileKey]BlobLocation,
895 opts build.Options,
896) (zoekt.Document, error) {
897 repo := repos[key]
898 blob, err := repo.GitRepo.BlobObject(key.ID)
899 branches := repos[key].Branches
900
901 // We filter out large documents when fetching the repo. So if an object is too large, it will not be found.
902 if errors.Is(err, plumbing.ErrObjectNotFound) {
903 return skippedLargeDoc(key, branches, opts), nil
904 }
905
906 if err != nil {
907 return zoekt.Document{}, err
908 }
909
910 keyFullPath := key.FullPath()
911 if blob.Size > int64(opts.SizeMax) && !opts.IgnoreSizeMax(keyFullPath) {
912 return skippedLargeDoc(key, branches, opts), nil
913 }
914
915 contents, err := blobContents(blob)
916 if err != nil {
917 return zoekt.Document{}, err
918 }
919
920 return zoekt.Document{
921 SubRepositoryPath: key.SubRepoPath,
922 Name: keyFullPath,
923 Content: contents,
924 Branches: branches,
925 }, nil
926}
927
928func skippedLargeDoc(key fileKey, branches []string, opts build.Options) zoekt.Document {
929 return zoekt.Document{
930 SkipReason: fmt.Sprintf("file size exceeds maximum size %d", opts.SizeMax),
931 Name: key.FullPath(),
932 Branches: branches,
933 SubRepositoryPath: key.SubRepoPath,
934 }
935}
936
937func blobContents(blob *object.Blob) ([]byte, error) {
938 r, err := blob.Reader()
939 if err != nil {
940 return nil, err
941 }
942 defer r.Close()
943
944 var buf bytes.Buffer
945 buf.Grow(int(blob.Size))
946 _, err = buf.ReadFrom(r)
947 if err != nil {
948 return nil, err
949 }
950 return buf.Bytes(), nil
951}
952
953func uniq(ss []string) []string {
954 result := ss[:0]
955 var last string
956 for i, s := range ss {
957 if i == 0 || s != last {
958 result = append(result, s)
959 }
960 last = s
961 }
962 return result
963}