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 "cmp"
21 "context"
22 "errors"
23 "fmt"
24 "io"
25 "log"
26 "math"
27 "net/url"
28 "os"
29 "path/filepath"
30 "regexp"
31 "sort"
32 "strconv"
33 "strings"
34
35 "github.com/go-git/go-billy/v5/osfs"
36 "github.com/go-git/go-git/v5/config"
37 "github.com/go-git/go-git/v5/plumbing"
38 "github.com/go-git/go-git/v5/plumbing/cache"
39 "github.com/go-git/go-git/v5/plumbing/object"
40 "github.com/go-git/go-git/v5/storage/filesystem"
41
42 "github.com/sourcegraph/zoekt"
43 "github.com/sourcegraph/zoekt/ignore"
44 "github.com/sourcegraph/zoekt/index"
45
46 git "github.com/go-git/go-git/v5"
47)
48
49// FindGitRepos finds directories holding git repositories below the
50// given directory. It will find both bare and the ".git" dirs in
51// non-bare repositories. It returns the full path including the dir
52// passed in.
53func FindGitRepos(dir string) ([]string, error) {
54 arg, err := filepath.Abs(dir)
55 if err != nil {
56 return nil, err
57 }
58 var dirs []string
59 if err := filepath.Walk(arg, func(name string, fi os.FileInfo, err error) error {
60 // Best-effort, ignore filepath.Walk failing
61 if err != nil {
62 return nil
63 }
64
65 if fi, err := os.Lstat(filepath.Join(name, ".git")); err == nil && fi.IsDir() {
66 dirs = append(dirs, filepath.Join(name, ".git"))
67 return filepath.SkipDir
68 }
69
70 if !strings.HasSuffix(name, ".git") || !fi.IsDir() {
71 return nil
72 }
73
74 fi, err = os.Lstat(filepath.Join(name, "objects"))
75 if err != nil || !fi.IsDir() {
76 return nil
77 }
78
79 dirs = append(dirs, name)
80 return filepath.SkipDir
81 }); err != nil {
82 return nil, err
83 }
84
85 return dirs, nil
86}
87
88// setTemplates fills in URL templates for known git hosting
89// sites.
90func setTemplates(repo *zoekt.Repository, u *url.URL, typ string) error {
91 if u.Scheme == "ssh+git" {
92 u.Scheme = "https"
93 u.User = nil
94 }
95
96 // helper to generate u.JoinPath as a template
97 varVersion := ".Version"
98 varPath := ".Path"
99 urlJoinPath := func(elem ...string) string {
100 elem = append([]string{u.String()}, elem...)
101 var parts []string
102 for _, e := range elem {
103 if e == varVersion || e == varPath {
104 parts = append(parts, e)
105 } else {
106 parts = append(parts, strconv.Quote(e))
107 }
108 }
109 return fmt.Sprintf("{{URLJoinPath %s}}", strings.Join(parts, " "))
110 }
111
112 repo.URL = u.String()
113 switch typ {
114 case "gitiles":
115 // eg. https://gerrit.googlesource.com/gitiles/+/master/tools/run_dev.sh#20
116 repo.CommitURLTemplate = urlJoinPath("+", varVersion)
117 repo.FileURLTemplate = urlJoinPath("+", varVersion, varPath)
118 repo.LineFragmentTemplate = "#{{.LineNumber}}"
119 case "github":
120 // eg. https://github.com/hanwen/go-fuse/blob/notify/genversion.sh#L10
121 repo.CommitURLTemplate = urlJoinPath("commit", varVersion)
122 repo.FileURLTemplate = urlJoinPath("blob", varVersion, varPath)
123 repo.LineFragmentTemplate = "#L{{.LineNumber}}"
124 case "cgit":
125 // http://git.savannah.gnu.org/cgit/lilypond.git/tree/elisp/lilypond-mode.el?h=dev/philh&id=b2ca0fefe3018477aaca23b6f672c7199ba5238e#n100
126 repo.CommitURLTemplate = urlJoinPath("commit") + "/?id={{.Version}}"
127 repo.FileURLTemplate = urlJoinPath("tree", varPath) + "/?id={{.Version}}"
128 repo.LineFragmentTemplate = "#n{{.LineNumber}}"
129 case "gitweb":
130 // https://gerrit.libreoffice.org/gitweb?p=online.git;a=blob;f=Makefile.am;h=cfcfd7c36fbae10e269653dc57a9b68c92d4c10b;hb=848145503bf7b98ce4a4aa0a858a0d71dd0dbb26#l10
131 repo.FileURLTemplate = u.String() + ";a=blob;f={{.Path}};hb={{.Version}}"
132 repo.CommitURLTemplate = u.String() + ";a=commit;h={{.Version}}"
133 repo.LineFragmentTemplate = "#l{{.LineNumber}}"
134 case "source.bazel.build":
135 // https://source.bazel.build/bazel/+/57bc201346e61c62a921c1cbf32ad24f185c10c9
136 // https://source.bazel.build/bazel/+/57bc201346e61c62a921c1cbf32ad24f185c10c9:tools/cpp/BUILD.empty;l=10
137 repo.CommitURLTemplate = u.String() + "/%2B/{{.Version}}"
138 repo.FileURLTemplate = u.String() + "/%2B/{{.Version}}:{{.Path}}"
139 repo.LineFragmentTemplate = ";l={{.LineNumber}}"
140 case "bitbucket-server":
141 // https://<bitbucketserver-host>/projects/<project>/repos/<repo>/commits/5be7ca73b898bf17a08e607918accfdeafe1e0bc
142 // https://<bitbucketserver-host>/projects/<project>/repos/<repo>/browse/<file>?at=5be7ca73b898bf17a08e607918accfdeafe1e0bc
143 repo.CommitURLTemplate = urlJoinPath("commits", varVersion)
144 repo.FileURLTemplate = urlJoinPath(varPath) + "?at={{.Version}}"
145 repo.LineFragmentTemplate = "#{{.LineNumber}}"
146 case "gitlab":
147 // https://gitlab.com/gitlab-org/omnibus-gitlab/-/commit/b152c864303dae0e55377a1e2c53c9592380ffed
148 // https://gitlab.com/gitlab-org/omnibus-gitlab/-/blob/aad04155b3f6fc50ede88aedaee7fc624d481149/files/gitlab-config-template/gitlab.rb.template
149 repo.CommitURLTemplate = urlJoinPath("-/commit", varVersion)
150 repo.FileURLTemplate = urlJoinPath("-/blob", varVersion, varPath)
151 repo.LineFragmentTemplate = "#L{{.LineNumber}}"
152 case "gitea":
153 repo.CommitURLTemplate = urlJoinPath("commit", varVersion)
154 // NOTE The `display=source` query parameter is required to disable file rendering.
155 // Since line numbers are disabled in rendered files, you wouldn't be able to jump to
156 // a line without `display=source`. This is supported since gitea 1.17.0.
157 // When /src/{{.Version}} is used it will redirect to /src/commit/{{.Version}},
158 // but the query parameters are obmitted.
159 repo.FileURLTemplate = urlJoinPath("src/commit", varVersion, varPath) + "?display=source"
160 repo.LineFragmentTemplate = "#L{{.LineNumber}}"
161 default:
162 return fmt.Errorf("URL scheme type %q unknown", typ)
163 }
164 return nil
165}
166
167// getCommit returns a tree object for the given reference.
168func getCommit(repo *git.Repository, prefix, ref string) (*object.Commit, error) {
169 sha1, err := repo.ResolveRevision(plumbing.Revision(ref))
170 // ref might be a branch name (e.g. "master") add branch prefix and try again.
171 if err != nil {
172 sha1, err = repo.ResolveRevision(plumbing.Revision(filepath.Join(prefix, ref)))
173 }
174 if err != nil {
175 return nil, err
176 }
177
178 commitObj, err := repo.CommitObject(*sha1)
179 if err != nil {
180 return nil, err
181 }
182 return commitObj, nil
183}
184
185func configLookupRemoteURL(cfg *config.Config, key string) string {
186 rc := cfg.Remotes[key]
187 if rc == nil || len(rc.URLs) == 0 {
188 return ""
189 }
190 return rc.URLs[0]
191}
192
193var sshRelativeURLRegexp = regexp.MustCompile(`^([^@]+)@([^:]+):(.*)$`)
194
195func setTemplatesFromConfig(desc *zoekt.Repository, repoDir string) error {
196 repo, err := git.PlainOpen(repoDir)
197 if err != nil {
198 return err
199 }
200
201 cfg, err := repo.Config()
202 if err != nil {
203 return err
204 }
205
206 sec := cfg.Raw.Section("zoekt")
207
208 webURLStr := sec.Options.Get("web-url")
209 webURLType := sec.Options.Get("web-url-type")
210
211 if webURLType != "" && webURLStr != "" {
212 webURL, err := url.Parse(webURLStr)
213 if err != nil {
214 return err
215 }
216 if err := setTemplates(desc, webURL, webURLType); err != nil {
217 return err
218 }
219 } else if webURLStr != "" {
220 desc.URL = webURLStr
221 }
222
223 name := sec.Options.Get("name")
224 if name != "" {
225 desc.Name = name
226 } else {
227 remoteURL := configLookupRemoteURL(cfg, "origin")
228 if remoteURL == "" {
229 return nil
230 }
231 if sm := sshRelativeURLRegexp.FindStringSubmatch(remoteURL); sm != nil {
232 user := sm[1]
233 host := sm[2]
234 path := sm[3]
235
236 remoteURL = fmt.Sprintf("ssh+git://%s@%s/%s", user, host, path)
237 }
238
239 u, err := url.Parse(remoteURL)
240 if err != nil {
241 return err
242 }
243 if err := SetTemplatesFromOrigin(desc, u); err != nil {
244 return err
245 }
246 }
247
248 id, _ := strconv.ParseUint(sec.Options.Get("repoid"), 10, 32)
249 desc.ID = uint32(id)
250
251 if desc.RawConfig == nil {
252 desc.RawConfig = map[string]string{}
253 }
254 for _, o := range sec.Options {
255 desc.RawConfig[o.Key] = o.Value
256 }
257
258 // Ranking info.
259
260 // Github:
261 traction := 0
262 for _, s := range []string{"github-stars", "github-forks", "github-watchers", "github-subscribers"} {
263 f, err := strconv.Atoi(sec.Options.Get(s))
264 if err == nil {
265 traction += f
266 }
267 }
268
269 if strings.Contains(desc.Name, "googlesource.com/") && traction == 0 {
270 // Pretend everything on googlesource.com has 1000
271 // github stars.
272 traction = 1000
273 }
274
275 if traction > 0 {
276 l := math.Log(float64(traction))
277 desc.Rank = uint16((1.0 - 1.0/math.Pow(1+l, 0.6)) * 10000)
278 }
279
280 return nil
281}
282
283// SetTemplatesFromOrigin fills in templates based on the origin URL.
284func SetTemplatesFromOrigin(desc *zoekt.Repository, u *url.URL) error {
285 desc.Name = filepath.Join(u.Host, strings.TrimSuffix(u.Path, ".git"))
286
287 if strings.HasSuffix(u.Host, ".googlesource.com") {
288 return setTemplates(desc, u, "gitiles")
289 } else if u.Host == "github.com" {
290 u.Path = strings.TrimSuffix(u.Path, ".git")
291 return setTemplates(desc, u, "github")
292 } else {
293 return fmt.Errorf("unknown git hosting site %q", u)
294 }
295}
296
297// The Options structs controls details of the indexing process.
298type Options struct {
299 // The repository to be indexed.
300 RepoDir string
301
302 // If set, follow submodule links. This requires RepoCacheDir to be set.
303 Submodules bool
304
305 // If set, skip indexing if the existing index shard is newer
306 // than the refs in the repository.
307 Incremental bool
308
309 // Don't error out if some branch is missing
310 AllowMissingBranch bool
311
312 // Specifies the root of a Repository cache. Needed for submodule indexing.
313 RepoCacheDir string
314
315 // Indexing options.
316 BuildOptions index.Options
317
318 // Prefix of the branch to index, e.g. `remotes/origin`.
319 BranchPrefix string
320
321 // List of branch names to index, e.g. []string{"HEAD", "stable"}
322 Branches []string
323
324 // DeltaShardNumberFallbackThreshold defines an upper limit (inclusive) on the number of preexisting shards
325 // that can exist before attempting another delta build. If the number of preexisting shards exceeds this threshold,
326 // then a normal build will be performed instead.
327 //
328 // If DeltaShardNumberFallbackThreshold is 0, then this fallback behavior is disabled:
329 // a delta build will always be performed regardless of the number of preexisting shards.
330 DeltaShardNumberFallbackThreshold uint64
331}
332
333func expandBranches(repo *git.Repository, bs []string, prefix string) ([]string, error) {
334 var result []string
335 for _, b := range bs {
336 // Sourcegraph: We disable resolving refs. We want to return the exact ref
337 // requested so we can match it up.
338 if b == "HEAD" && false {
339 ref, err := repo.Head()
340 if err != nil {
341 return nil, err
342 }
343
344 result = append(result, strings.TrimPrefix(ref.Name().String(), prefix))
345 continue
346 }
347
348 if strings.Contains(b, "*") {
349 iter, err := repo.Branches()
350 if err != nil {
351 return nil, err
352 }
353
354 defer iter.Close()
355 for {
356 ref, err := iter.Next()
357 if err == io.EOF {
358 break
359 }
360 if err != nil {
361 return nil, err
362 }
363
364 name := ref.Name().Short()
365 if matched, err := filepath.Match(b, name); err != nil {
366 return nil, err
367 } else if !matched {
368 continue
369 }
370
371 result = append(result, strings.TrimPrefix(name, prefix))
372 }
373 continue
374 }
375
376 result = append(result, b)
377 }
378
379 return result, nil
380}
381
382// IndexGitRepo indexes the git repository as specified by the options.
383// The returned bool indicates whether the index was updated as a result. This
384// can be informative if doing incremental indexing.
385func IndexGitRepo(opts Options) (bool, error) {
386 return indexGitRepo(opts, gitIndexConfig{})
387}
388
389// indexGitRepo indexes the git repository as specified by the options and the provided gitIndexConfig.
390// The returned bool indicates whether the index was updated as a result. This
391// can be informative if doing incremental indexing.
392func indexGitRepo(opts Options, config gitIndexConfig) (bool, error) {
393 prepareDeltaBuild := prepareDeltaBuild
394 if config.prepareDeltaBuild != nil {
395 prepareDeltaBuild = config.prepareDeltaBuild
396 }
397
398 prepareNormalBuild := prepareNormalBuild
399 if config.prepareNormalBuild != nil {
400 prepareNormalBuild = config.prepareNormalBuild
401 }
402
403 // Set max thresholds, since we use them in this function.
404 opts.BuildOptions.SetDefaults()
405 if opts.RepoDir == "" {
406 return false, fmt.Errorf("gitindex: must set RepoDir")
407 }
408
409 opts.BuildOptions.RepositoryDescription.Source = opts.RepoDir
410
411 var repo *git.Repository
412 legacyRepoOpen := cmp.Or(os.Getenv("ZOEKT_DISABLE_GOGIT_OPTIMIZATION"), "false")
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 := index.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 })
590
591 // Because we're keeping descriptors open, we need to close the storage object when we're done.
592 repo, err := git.Open(s, fs)
593 return repo, s, err
594}
595
596func newIgnoreMatcher(tree *object.Tree) (*ignore.Matcher, error) {
597 ignoreFile, err := tree.File(ignore.IgnoreFile)
598 if err == object.ErrFileNotFound {
599 return &ignore.Matcher{}, nil
600 }
601 if err != nil {
602 return nil, err
603 }
604 content, err := ignoreFile.Contents()
605 if err != nil {
606 return nil, err
607 }
608 return ignore.ParseIgnoreFile(strings.NewReader(content))
609}
610
611// prepareDeltaBuildFunc is a function that calculates the necessary metadata for preparing
612// a build.Builder instance for generating a delta build.
613type prepareDeltaBuildFunc func(options Options, repository *git.Repository) (repos map[fileKey]BlobLocation, branchVersions map[string]map[string]plumbing.Hash, changedOrDeletedPaths []string, err error)
614
615// prepareNormalBuildFunc is a function that calculates the necessary metadata for preparing
616// a build.Builder instance for generating a normal build.
617type prepareNormalBuildFunc func(options Options, repository *git.Repository) (repos map[fileKey]BlobLocation, branchVersions map[string]map[string]plumbing.Hash, err error)
618
619type gitIndexConfig struct {
620 // prepareDeltaBuild, if not nil, is the function that is used to calculate the metadata that will be used to
621 // prepare the build.Builder instance for generating a delta build.
622 //
623 // If prepareDeltaBuild is nil, gitindex.prepareDeltaBuild will be used instead.
624 prepareDeltaBuild prepareDeltaBuildFunc
625
626 // prepareNormalBuild, if not nil, is the function that is used to calculate the metadata that will be used to
627 // prepare the build.Builder instance for generating a normal build.
628 //
629 // If prepareNormalBuild is nil, gitindex.prepareNormalBuild will be used instead.
630 prepareNormalBuild prepareNormalBuildFunc
631}
632
633func prepareDeltaBuild(options Options, repository *git.Repository) (repos map[fileKey]BlobLocation, branchVersions map[string]map[string]plumbing.Hash, changedOrDeletedPaths []string, err error) {
634 if options.Submodules {
635 return nil, nil, nil, fmt.Errorf("delta builds currently don't support submodule indexing")
636 }
637
638 // discover what commits we indexed during our last build
639 existingRepository, _, ok, err := options.BuildOptions.FindRepositoryMetadata()
640 if err != nil {
641 return nil, nil, nil, fmt.Errorf("failed to get repository metadata: %w", err)
642 }
643
644 if !ok {
645 return nil, nil, nil, fmt.Errorf("no existing shards found for repository")
646 }
647
648 if options.DeltaShardNumberFallbackThreshold > 0 {
649 // HACK: For our interim compaction strategy, we force a full normal index once
650 // the number of shards on disk for this repository exceeds the provided threshold.
651 //
652 // This strategy obviously isn't optimal (as an example: we currently can't differentiate
653 // between "normal" and "delta" shards, so repositories like the gigarepo that generate a large number of shards per
654 // build would be disproportionately affected by this), but it'll allow us to continue experimenting on real workloads
655 // while we create a better compaction strategy).
656
657 oldShards := options.BuildOptions.FindAllShards()
658 if uint64(len(oldShards)) > options.DeltaShardNumberFallbackThreshold {
659 return nil, nil, nil, fmt.Errorf("number of existing shards (%d) > requested shard threshold (%d)", len(oldShards), options.DeltaShardNumberFallbackThreshold)
660 }
661 }
662
663 // Check to see if the set of branch names is consistent with what we last indexed.
664 // If it isn't consistent, that we can't proceed with a delta build (and the caller should fall back to a
665 // normal one).
666
667 if !index.BranchNamesEqual(existingRepository.Branches, options.BuildOptions.RepositoryDescription.Branches) {
668 var existingBranchNames []string
669 for _, b := range existingRepository.Branches {
670 existingBranchNames = append(existingBranchNames, b.Name)
671 }
672
673 var optionsBranchNames []string
674 for _, b := range options.BuildOptions.RepositoryDescription.Branches {
675 optionsBranchNames = append(optionsBranchNames, b.Name)
676 }
677
678 existingBranchList := strings.Join(existingBranchNames, ", ")
679 optionsBranchList := strings.Join(optionsBranchNames, ", ")
680
681 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)
682 }
683
684 // Check if the build options hash does not match the repository metadata's hash
685 // If it does not index then one or more index options has changed and will require a normal build instead of a delta build
686 if options.BuildOptions.GetHash() != existingRepository.IndexOptions {
687 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())
688 }
689
690 // branch => (path, sha1) => repo.
691 repos = map[fileKey]BlobLocation{}
692
693 branches, err := expandBranches(repository, options.Branches, options.BranchPrefix)
694 if err != nil {
695 return nil, nil, nil, fmt.Errorf("expandBranches: %w", err)
696 }
697
698 // branch name -> git worktree at most current commit
699 branchToCurrentTree := make(map[string]*object.Tree, len(branches))
700
701 for _, b := range branches {
702 commit, err := getCommit(repository, options.BranchPrefix, b)
703 if err != nil {
704 return nil, nil, nil, fmt.Errorf("getting last current commit for branch %q: %w", b, err)
705 }
706
707 tree, err := commit.Tree()
708 if err != nil {
709 return nil, nil, nil, fmt.Errorf("getting current git tree for branch %q: %w", b, err)
710 }
711
712 branchToCurrentTree[b] = tree
713 }
714
715 rawURL := options.BuildOptions.RepositoryDescription.URL
716 u, err := url.Parse(rawURL)
717 if err != nil {
718 return nil, nil, nil, fmt.Errorf("parsing repository URL %q: %w", rawURL, err)
719 }
720
721 // TODO: Support repository submodules for delta builds
722
723 // loop over all branches, calculate the diff between our
724 // last indexed commit and the current commit, and add files mentioned in the diff
725 for _, branch := range existingRepository.Branches {
726 lastIndexedCommit, err := getCommit(repository, "", branch.Version)
727 if err != nil {
728 return nil, nil, nil, fmt.Errorf("getting last indexed commit for branch %q: %w", branch.Name, err)
729 }
730
731 lastIndexedTree, err := lastIndexedCommit.Tree()
732 if err != nil {
733 return nil, nil, nil, fmt.Errorf("getting lasted indexed git tree for branch %q: %w", branch.Name, err)
734 }
735
736 changes, err := object.DiffTreeWithOptions(context.Background(), lastIndexedTree, branchToCurrentTree[branch.Name], &object.DiffTreeOptions{DetectRenames: false})
737 if err != nil {
738 return nil, nil, nil, fmt.Errorf("generating changeset for branch %q: %w", branch.Name, err)
739 }
740
741 for i, c := range changes {
742 oldFile, newFile, err := c.Files()
743 if err != nil {
744 return nil, nil, nil, fmt.Errorf("change #%d: getting files before and after change: %w", i, err)
745 }
746
747 if newFile != nil {
748 // note: newFile.Name could be a path that isn't relative to the repository root - using the
749 // change's Name field is the only way that @ggilmore saw to get the full path relative to the root
750 newFileRelativeRootPath := c.To.Name
751
752 // TODO@ggilmore: HACK - remove once ignore files are supported in delta builds
753 if newFileRelativeRootPath == ignore.IgnoreFile {
754 return nil, nil, nil, fmt.Errorf("%q file is not yet supported in delta builds", ignore.IgnoreFile)
755 }
756
757 // either file is added or renamed, so we need to add the new version to the build
758 file := fileKey{Path: newFileRelativeRootPath, ID: newFile.Hash}
759 if existing, ok := repos[file]; ok {
760 existing.Branches = append(existing.Branches, branch.Name)
761 repos[file] = existing
762 } else {
763 repos[file] = BlobLocation{
764 GitRepo: repository,
765 URL: u,
766 Branches: []string{branch.Name},
767 }
768 }
769 }
770
771 if oldFile == nil {
772 // file added - nothing more to do
773 continue
774 }
775
776 // Note: oldFile.Name could be a path that isn't relative to the repository root - using the
777 // change's "Name" field is the only way that ggilmore saw to get the full path relative to the root
778 oldFileRelativeRootPath := c.From.Name
779
780 if oldFileRelativeRootPath == ignore.IgnoreFile {
781 return nil, nil, nil, fmt.Errorf("%q file is not yet supported in delta builds", ignore.IgnoreFile)
782 }
783
784 // The file is either modified or deleted. So, we need to add ALL versions
785 // of the old file (across all branches) to the build.
786 for b, currentTree := range branchToCurrentTree {
787 f, err := currentTree.File(oldFileRelativeRootPath)
788 if err != nil {
789 // the file doesn't exist in this branch
790 if errors.Is(err, object.ErrFileNotFound) {
791 continue
792 }
793
794 return nil, nil, nil, fmt.Errorf("getting hash for file %q in branch %q: %w", oldFile.Name, b, err)
795 }
796
797 file := fileKey{Path: oldFileRelativeRootPath, ID: f.ID()}
798 if existing, ok := repos[file]; ok {
799 existing.Branches = append(existing.Branches, b)
800 repos[file] = existing
801 } else {
802 repos[file] = BlobLocation{
803 GitRepo: repository,
804 URL: u,
805 Branches: []string{b},
806 }
807 }
808 }
809
810 changedOrDeletedPaths = append(changedOrDeletedPaths, oldFileRelativeRootPath)
811 }
812 }
813
814 // we need to de-duplicate the branch map before returning it - it's possible for the same
815 // branch to have been added multiple times if a file has been modified across multiple commits
816 for _, info := range repos {
817 sort.Strings(info.Branches)
818 info.Branches = uniq(info.Branches)
819 }
820
821 // we also need to de-duplicate the list of changed or deleted file paths, it's also possible to have duplicates
822 // for the same reasoning as above
823 sort.Strings(changedOrDeletedPaths)
824 changedOrDeletedPaths = uniq(changedOrDeletedPaths)
825
826 return repos, nil, changedOrDeletedPaths, nil
827}
828
829func prepareNormalBuild(options Options, repository *git.Repository) (repos map[fileKey]BlobLocation, branchVersions map[string]map[string]plumbing.Hash, err error) {
830 var repoCache *RepoCache
831 if options.Submodules {
832 repoCache = NewRepoCache(options.RepoCacheDir)
833 }
834
835 // Branch => Repo => SHA1
836 branchVersions = map[string]map[string]plumbing.Hash{}
837
838 branches, err := expandBranches(repository, options.Branches, options.BranchPrefix)
839 if err != nil {
840 return nil, nil, fmt.Errorf("expandBranches: %w", err)
841 }
842
843 rw := NewRepoWalker(repository, options.BuildOptions.RepositoryDescription.URL, repoCache)
844 for _, b := range branches {
845 commit, err := getCommit(repository, options.BranchPrefix, b)
846 if err != nil {
847 if options.AllowMissingBranch && err.Error() == "reference not found" {
848 continue
849 }
850
851 return nil, nil, fmt.Errorf("getCommit: %w", err)
852 }
853
854 tree, err := commit.Tree()
855 if err != nil {
856 return nil, nil, fmt.Errorf("commit.Tree: %w", err)
857 }
858
859 ig, err := newIgnoreMatcher(tree)
860 if err != nil {
861 return nil, nil, fmt.Errorf("newIgnoreMatcher: %w", err)
862 }
863
864 subVersions, err := rw.CollectFiles(tree, b, ig)
865 if err != nil {
866 return nil, nil, fmt.Errorf("CollectFiles: %w", err)
867 }
868
869 branchVersions[b] = subVersions
870 }
871
872 return rw.Files, branchVersions, nil
873}
874
875func createDocument(key fileKey,
876 repos map[fileKey]BlobLocation,
877 opts index.Options,
878) (index.Document, error) {
879 repo := repos[key]
880 blob, err := repo.GitRepo.BlobObject(key.ID)
881 branches := repos[key].Branches
882
883 // We filter out large documents when fetching the repo. So if an object is too large, it will not be found.
884 if errors.Is(err, plumbing.ErrObjectNotFound) {
885 return skippedLargeDoc(key, branches), nil
886 }
887
888 if err != nil {
889 return index.Document{}, err
890 }
891
892 keyFullPath := key.FullPath()
893 if blob.Size > int64(opts.SizeMax) && !opts.IgnoreSizeMax(keyFullPath) {
894 return skippedLargeDoc(key, branches), nil
895 }
896
897 contents, err := blobContents(blob)
898 if err != nil {
899 return index.Document{}, err
900 }
901
902 return index.Document{
903 SubRepositoryPath: key.SubRepoPath,
904 Name: keyFullPath,
905 Content: contents,
906 Branches: branches,
907 }, nil
908}
909
910func skippedLargeDoc(key fileKey, branches []string) index.Document {
911 return index.Document{
912 SkipReason: index.SkipReasonTooLarge,
913 Name: key.FullPath(),
914 Branches: branches,
915 SubRepositoryPath: key.SubRepoPath,
916 }
917}
918
919func blobContents(blob *object.Blob) ([]byte, error) {
920 r, err := blob.Reader()
921 if err != nil {
922 return nil, err
923 }
924 defer r.Close()
925
926 var buf bytes.Buffer
927 buf.Grow(int(blob.Size))
928 _, err = buf.ReadFrom(r)
929 if err != nil {
930 return nil, err
931 }
932 return buf.Bytes(), nil
933}
934
935func uniq(ss []string) []string {
936 result := ss[:0]
937 var last string
938 for i, s := range ss {
939 if i == 0 || s != last {
940 result = append(result, s)
941 }
942 last = s
943 }
944 return result
945}