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
15package search
16
17import (
18 "bytes"
19 "context"
20 "flag"
21 "fmt"
22 "hash/fnv"
23 "io"
24 "log"
25 "math"
26 "os"
27 "reflect"
28 "runtime"
29 "sort"
30 "strconv"
31 "testing"
32 "testing/quick"
33 "time"
34
35 "github.com/RoaringBitmap/roaring"
36 "github.com/google/go-cmp/cmp"
37 "github.com/google/go-cmp/cmp/cmpopts"
38 "github.com/grafana/regexp"
39
40 "github.com/sourcegraph/zoekt/index"
41
42 "github.com/sourcegraph/zoekt"
43 "github.com/sourcegraph/zoekt/query"
44)
45
46func TestMain(m *testing.M) {
47 flag.Parse()
48 if !testing.Verbose() {
49 log.SetOutput(io.Discard)
50 }
51 os.Exit(m.Run())
52}
53
54type crashSearcher struct{}
55
56func (s *crashSearcher) Search(ctx context.Context, q query.Q, opts *zoekt.SearchOptions) (*zoekt.SearchResult, error) {
57 panic("search")
58}
59
60func (s *crashSearcher) List(ctx context.Context, q query.Q, opts *zoekt.ListOptions) (*zoekt.RepoList, error) {
61 panic("list")
62}
63
64func (s *crashSearcher) Stats() (*zoekt.RepoStats, error) {
65 return &zoekt.RepoStats{}, nil
66}
67
68func (s *crashSearcher) Close() {}
69
70func (s *crashSearcher) String() string { return "crashSearcher" }
71
72func TestCrashResilience(t *testing.T) {
73 out := &bytes.Buffer{}
74 oldOut := log.Writer()
75 log.SetOutput(out)
76 defer log.SetOutput(oldOut)
77
78 ss := newShardedSearcher(2)
79 ss.ranked.Store([]*rankedShard{{Searcher: &crashSearcher{}}})
80
81 var wantCrashes int
82 test := func(t *testing.T) {
83 q := &query.Substring{Pattern: "hoi"}
84 opts := &zoekt.SearchOptions{}
85 if res, err := ss.Search(context.Background(), q, opts); err != nil {
86 t.Fatalf("Search: %v", err)
87 } else if res.Stats.Crashes != wantCrashes {
88 t.Errorf("got stats %#v, want crashes = %d", res.Stats, wantCrashes)
89 }
90
91 if res, err := ss.List(context.Background(), q, nil); err != nil {
92 t.Fatalf("List: %v", err)
93 } else if res.Crashes != wantCrashes {
94 t.Errorf("got result %#v, want crashes = %d", res, wantCrashes)
95 }
96 }
97
98 // Before we are marked as ready we have one extra crash
99 wantCrashes = 2
100 t.Run("loading", test)
101
102 // After marking as ready we should only have the crashSearcher
103 // contributing.
104 ss.markReady()
105 wantCrashes = 1
106 t.Run("ready", test)
107}
108
109type rankSearcher struct {
110 rank uint16
111 repo *zoekt.Repository
112}
113
114func (s *rankSearcher) Close() {
115}
116
117func (s *rankSearcher) String() string {
118 return ""
119}
120
121func (s *rankSearcher) Search(ctx context.Context, q query.Q, opts *zoekt.SearchOptions) (*zoekt.SearchResult, error) {
122 select {
123 case <-ctx.Done():
124 return &zoekt.SearchResult{}, nil
125 default:
126 }
127
128 // Ugly, but without sleep it's too fast, and we can't
129 // simulate the cutoff.
130 time.Sleep(time.Millisecond)
131 return &zoekt.SearchResult{
132 Files: []zoekt.FileMatch{
133 {
134 FileName: fmt.Sprintf("f%d", s.rank),
135 Score: float64(s.rank),
136 },
137 },
138 Stats: zoekt.Stats{
139 MatchCount: 1,
140 },
141 }, nil
142}
143
144func (s *rankSearcher) List(ctx context.Context, q query.Q, opts *zoekt.ListOptions) (*zoekt.RepoList, error) {
145 r := zoekt.Repository{}
146 if s.repo != nil {
147 r = *s.repo
148 }
149 r.Rank = s.rank
150 return &zoekt.RepoList{
151 Repos: []*zoekt.RepoListEntry{
152 {Repository: r},
153 },
154 }, nil
155}
156
157func (s *rankSearcher) Repository() *zoekt.Repository { return s.repo }
158
159func TestOrderByShard(t *testing.T) {
160 ss := newShardedSearcher(1)
161
162 n := 10 * runtime.GOMAXPROCS(0)
163 for i := range n {
164 ss.replace(map[string]zoekt.Searcher{
165 fmt.Sprintf("shard%d", i): &rankSearcher{rank: uint16(i)},
166 })
167 }
168
169 if res, err := ss.Search(context.Background(), &query.Substring{Pattern: "bla"}, &zoekt.SearchOptions{}); err != nil {
170 t.Errorf("Search: %v", err)
171 } else if len(res.Files) != n {
172 t.Fatalf("empty options: got %d results, want %d", len(res.Files), n)
173 }
174
175 opts := zoekt.SearchOptions{
176 TotalMaxMatchCount: 3,
177 }
178 res, err := ss.Search(context.Background(), &query.Substring{Pattern: "bla"}, &opts)
179 if err != nil {
180 t.Errorf("Search: %v", err)
181 }
182
183 if len(res.Files) < opts.TotalMaxMatchCount {
184 t.Errorf("got %d results, want %d", len(res.Files), opts.TotalMaxMatchCount)
185 }
186 if len(res.Files) == n {
187 t.Errorf("got %d results, want < %d", len(res.Files), n)
188 }
189 for i, f := range res.Files {
190 rev := n - 1 - i
191 want := fmt.Sprintf("f%d", rev)
192 got := f.FileName
193
194 if got != want {
195 t.Logf("%d: got %q, want %q", i, got, want)
196 }
197 }
198}
199
200func TestShardedSearcher_Ranking(t *testing.T) {
201 ss := newShardedSearcher(1)
202
203 var nextShardNum int
204 addShard := func(repo string, priority float64, docs ...index.Document) {
205 r := &zoekt.Repository{ID: hash(repo), Name: repo}
206 r.RawConfig = map[string]string{
207 "public": "1",
208 "priority": strconv.FormatFloat(priority, 'f', 2, 64),
209 }
210 b := testShardBuilder(t, r, docs...)
211 shard := searcherForTest(t, b)
212 ss.replace(map[string]zoekt.Searcher{
213 fmt.Sprintf("key-%d", nextShardNum): shard,
214 })
215 nextShardNum++
216 }
217
218 addShard("weekend-project", 20, index.Document{Name: "f2", Content: []byte("foo bas")})
219 addShard("moderately-popular", 500, index.Document{Name: "f3", Content: []byte("foo bar")})
220 addShard("weekend-project-2", 20, index.Document{Name: "f2", Content: []byte("foo bas")})
221 addShard("super-star", 5000, index.Document{Name: "f1", Content: []byte("foo bar bas")})
222
223 want := []string{
224 "super-star",
225 "moderately-popular",
226 "weekend-project",
227 "weekend-project-2",
228 }
229
230 var have []string
231 for _, s := range ss.getLoaded().shards {
232 for _, r := range s.repos {
233 have = append(have, r.Name)
234 }
235 }
236
237 if !reflect.DeepEqual(want, have) {
238 t.Fatalf("\nwant: %s\nhave: %s", want, have)
239 }
240}
241
242func TestShardedSearcher_DocumentRanking(t *testing.T) {
243 ss := newShardedSearcher(1)
244
245 var nextShardNum int
246 addShard := func(repo string, rank uint16, docs ...index.Document) {
247 r := &zoekt.Repository{ID: hash(repo), Name: repo}
248 r.RawConfig = map[string]string{
249 "public": "1",
250 }
251 r.Rank = rank
252 b := testShardBuilder(t, r, docs...)
253 shard := searcherForTest(t, b)
254 ss.replace(map[string]zoekt.Searcher{
255 fmt.Sprintf("key-%d", nextShardNum): shard,
256 })
257 nextShardNum++
258 }
259
260 addShard("old-project", 1, index.Document{Name: "f1", Content: []byte("foobar")})
261 addShard("recent", 2, index.Document{Name: "f2", Content: []byte("foobaz")})
262 addShard("old-project-2", 1, index.Document{Name: "f3", Content: []byte("foo bar")})
263 addShard("new", 3, index.Document{Name: "f4", Content: []byte("foo baz")},
264 index.Document{Name: "f5", Content: []byte("fooooo")})
265
266 // Run a stream search and gather the results
267 var results []*zoekt.SearchResult
268 opts := &zoekt.SearchOptions{
269 FlushWallTime: 100 * time.Millisecond,
270 }
271
272 err := ss.StreamSearch(context.Background(), &query.Substring{Pattern: "foo"}, opts,
273 zoekt.SenderFunc(func(event *zoekt.SearchResult) {
274 results = append(results, event)
275 }))
276 if err != nil {
277 t.Fatal(err)
278 }
279
280 // There should always be two stream results, first progress-only, then the file results
281 if len(results) != 2 {
282 t.Fatalf("expected 2 streamed results, but got %d", len(results))
283 }
284
285 // The ranking should be determined by whether it's an exact word match,
286 // followed by repository priority
287 want := []string{"f4", "f3", "f5", "f2", "f1"}
288
289 files := results[1].Files
290 got := make([]string, len(files))
291 for i := range files {
292 got[i] = files[i].FileName
293 }
294
295 if !reflect.DeepEqual(got, want) {
296 t.Errorf("got %v, want %v", got, want)
297 }
298}
299
300func TestFilteringShardsByRepoSetOrBranchesReposOrRepoIDs(t *testing.T) {
301 ss := newShardedSearcher(1)
302
303 // namePrefix is so we can create a repo:foo filter and match the same set
304 // of repos.
305 namePrefix := [3]string{"foo", "bar", "baz"}
306
307 repoSetNames := []string{}
308 repoIDs := []uint32{}
309 n := 10 * runtime.GOMAXPROCS(0)
310 for i := range n {
311 shardName := fmt.Sprintf("shard%d", i)
312 repoName := fmt.Sprintf("%s-repository%.3d", namePrefix[i%3], i)
313 repoID := hash(repoName)
314
315 if i%3 == 0 {
316 repoSetNames = append(repoSetNames, repoName)
317 repoIDs = append(repoIDs, repoID)
318 }
319
320 ss.replace(map[string]zoekt.Searcher{
321 shardName: &rankSearcher{
322 repo: &zoekt.Repository{ID: repoID, Name: repoName},
323 rank: uint16(n - i),
324 },
325 })
326 }
327
328 res, err := ss.Search(context.Background(), &query.Substring{Pattern: "bla"}, &zoekt.SearchOptions{})
329 if err != nil {
330 t.Errorf("Search: %v", err)
331 }
332 if len(res.Files) != n {
333 t.Fatalf("no reposet: got %d results, want %d", len(res.Files), n)
334 }
335
336 branchesRepos := &query.BranchesRepos{List: []query.BranchRepos{
337 {Branch: "HEAD", Repos: roaring.New()},
338 }}
339
340 for _, name := range repoSetNames {
341 branchesRepos.List[0].Repos.Add(hash(name))
342 }
343
344 set := query.NewRepoSet(repoSetNames...)
345 sub := &query.Substring{Pattern: "bla"}
346
347 repoIDsQuery := query.NewRepoIDs(repoIDs...)
348 repoQuery := &query.Repo{Regexp: regexp.MustCompile("^foo-.*")}
349
350 queries := []query.Q{
351 query.NewAnd(set, sub),
352 // Test with the same reposet again
353 query.NewAnd(set, sub),
354
355 query.NewAnd(branchesRepos, sub),
356 // Test with the same repoBranches with IDs again
357 query.NewAnd(branchesRepos, sub),
358
359 query.NewAnd(repoIDsQuery, sub),
360 // Test with the same repoIDs again
361 query.NewAnd(repoIDsQuery, sub),
362
363 query.NewAnd(repoQuery, sub),
364 query.NewAnd(repoQuery, sub),
365
366 // List has queries which are just the reposet atoms. We also test twice.
367 set,
368 set,
369 branchesRepos,
370 branchesRepos,
371 repoIDsQuery,
372 repoIDsQuery,
373 repoQuery,
374 repoQuery,
375 }
376
377 for _, q := range queries {
378 res, err = ss.Search(context.Background(), q, &zoekt.SearchOptions{})
379 if err != nil {
380 t.Errorf("Search(%s): %v", q, err)
381 }
382 // Note: Assertion is based on fact that `rankSearcher` always returns a
383 // result and using repoSet will half the number of results
384 if len(res.Files) != len(repoSetNames) {
385 t.Fatalf("%s: got %d results, want %d", q, len(res.Files), len(repoSetNames))
386 }
387 }
388}
389
390func hash(name string) uint32 {
391 h := fnv.New32()
392 h.Write([]byte(name))
393 return h.Sum32()
394}
395
396type memSeeker struct {
397 data []byte
398}
399
400func (s *memSeeker) Name() string {
401 return "memseeker"
402}
403
404func (s *memSeeker) Close() {}
405func (s *memSeeker) Read(off, sz uint32) ([]byte, error) {
406 return s.data[off : off+sz], nil
407}
408
409func (s *memSeeker) Size() (uint32, error) {
410 return uint32(len(s.data)), nil
411}
412
413func TestUnloadIndex(t *testing.T) {
414 b := testShardBuilder(t, nil, index.Document{
415 Name: "filename",
416 Content: []byte("needle needle needle"),
417 })
418
419 var buf bytes.Buffer
420 if err := b.Write(&buf); err != nil {
421 t.Fatal(err)
422 }
423 indexBytes := buf.Bytes()
424 indexFile := &memSeeker{indexBytes}
425 searcher, err := index.NewSearcher(indexFile)
426 if err != nil {
427 t.Fatalf("NewSearcher: %v", err)
428 }
429
430 ss := newShardedSearcher(2)
431 ss.replace(map[string]zoekt.Searcher{"key": searcher})
432
433 var opts zoekt.SearchOptions
434 q := &query.Substring{Pattern: "needle"}
435 res, err := ss.Search(context.Background(), q, &opts)
436 if err != nil {
437 t.Fatalf("Search(%s): %v", q, err)
438 }
439
440 forbidden := byte(29)
441 for i := range indexBytes {
442 // non-ASCII
443 indexBytes[i] = forbidden
444 }
445
446 for _, f := range res.Files {
447 if bytes.Contains(f.Content, []byte{forbidden}) {
448 t.Errorf("found %d in content %q", forbidden, f.Content)
449 }
450 if bytes.Contains(f.Checksum, []byte{forbidden}) {
451 t.Errorf("found %d in checksum %q", forbidden, f.Checksum)
452 }
453
454 for _, l := range f.LineMatches {
455 if bytes.Contains(l.Line, []byte{forbidden}) {
456 t.Errorf("found %d in line %q", forbidden, l.Line)
457 }
458 }
459 }
460}
461
462func TestShardedSearcher_List(t *testing.T) {
463 repos := []*zoekt.Repository{
464 {
465 ID: 1234,
466 Name: "repo-a",
467 Branches: []zoekt.RepositoryBranch{{Name: "main"}, {Name: "dev"}},
468 RawConfig: map[string]string{"repoid": "1234"},
469 },
470 {
471 Name: "repo-b",
472 Branches: []zoekt.RepositoryBranch{{Name: "main"}, {Name: "dev"}},
473 },
474 }
475
476 doc := index.Document{
477 Name: "foo.go",
478 Content: []byte("bar\nbaz"),
479 Branches: []string{"main", "dev"},
480 }
481
482 // Test duplicate removal when ListOptions.Minimal is true and false
483 ss := newShardedSearcher(4)
484 ss.replace(map[string]zoekt.Searcher{
485 "1": searcherForTest(t, testShardBuilder(t, repos[0], doc)),
486 "2": searcherForTest(t, testShardBuilder(t, repos[0])),
487 "3": searcherForTest(t, testShardBuilder(t, repos[1], doc)),
488 "4": searcherForTest(t, testShardBuilder(t, repos[1])),
489 })
490 ss.markReady()
491
492 stats := zoekt.RepoStats{
493 Shards: 2,
494 Documents: 1,
495 IndexBytes: 196,
496 ContentBytes: 13,
497 NewLinesCount: 1,
498 DefaultBranchNewLinesCount: 1,
499 OtherBranchesNewLinesCount: 1,
500 }
501
502 aggStats := stats
503 aggStats.Add(&aggStats) // since both repos have the exact same stats, this works
504 aggStats.Repos = 2 // Add doesn't populate Repos, this is done in Shards based on the response sizes.
505
506 for _, tc := range []struct {
507 name string
508 opts *zoekt.ListOptions
509 want *zoekt.RepoList
510 }{
511 {
512 name: "nil opts",
513 opts: nil,
514 want: &zoekt.RepoList{
515 Repos: []*zoekt.RepoListEntry{
516 {
517 Repository: *repos[0],
518 Stats: stats,
519 },
520 {
521 Repository: *repos[1],
522 Stats: stats,
523 },
524 },
525 Stats: aggStats,
526 },
527 },
528 {
529 name: "default",
530 opts: &zoekt.ListOptions{},
531 want: &zoekt.RepoList{
532 Repos: []*zoekt.RepoListEntry{
533 {
534 Repository: *repos[0],
535 Stats: stats,
536 },
537 {
538 Repository: *repos[1],
539 Stats: stats,
540 },
541 },
542 Stats: aggStats,
543 },
544 },
545 {
546 name: "field=repos",
547 opts: &zoekt.ListOptions{Field: zoekt.RepoListFieldRepos},
548 want: &zoekt.RepoList{
549 Repos: []*zoekt.RepoListEntry{
550 {
551 Repository: *repos[0],
552 Stats: stats,
553 },
554 {
555 Repository: *repos[1],
556 Stats: stats,
557 },
558 },
559 Stats: aggStats,
560 },
561 },
562 {
563 name: "field=reposmap",
564 opts: &zoekt.ListOptions{Field: zoekt.RepoListFieldReposMap},
565 want: &zoekt.RepoList{
566 Repos: []*zoekt.RepoListEntry{
567 {
568 Repository: *repos[1],
569 Stats: stats,
570 },
571 },
572 ReposMap: zoekt.ReposMap{
573 repos[0].ID: {
574 HasSymbols: repos[0].HasSymbols,
575 Branches: repos[0].Branches,
576 },
577 },
578 Stats: aggStats,
579 },
580 },
581 } {
582 tc := tc
583 t.Run(tc.name, func(t *testing.T) {
584 t.Parallel()
585
586 q := &query.Repo{Regexp: regexp.MustCompile("repo")}
587
588 res, err := ss.List(context.Background(), q, tc.opts)
589 if err != nil {
590 t.Fatalf("List(%v, %s): %v", q, tc.opts, err)
591 }
592
593 sort.Slice(res.Repos, func(i, j int) bool {
594 return res.Repos[i].Repository.Name < res.Repos[j].Repository.Name
595 })
596
597 ignored := []cmp.Option{
598 cmpopts.EquateEmpty(),
599 cmpopts.IgnoreFields(zoekt.MinimalRepoListEntry{}, "IndexTimeUnix"),
600 cmpopts.IgnoreFields(zoekt.RepoListEntry{}, "IndexMetadata"),
601 cmpopts.IgnoreFields(zoekt.RepoStats{}, "IndexBytes"),
602 cmpopts.IgnoreFields(zoekt.Repository{}, "SubRepoMap"),
603 cmpopts.IgnoreFields(zoekt.Repository{}, "priority"),
604 }
605
606 if diff := cmp.Diff(tc.want, res, ignored...); diff != "" {
607 t.Fatalf("mismatch (-want +got):\n%s", diff)
608 }
609 })
610 }
611}
612
613func testShardBuilder(t testing.TB, repo *zoekt.Repository, docs ...index.Document) *index.ShardBuilder {
614 b, err := index.NewShardBuilder(repo)
615 if err != nil {
616 t.Fatalf("NewShardBuilder: %v", err)
617 }
618
619 for i, d := range docs {
620 if err := b.Add(d); err != nil {
621 t.Fatalf("Add %d: %v", i, err)
622 }
623 }
624 return b
625}
626
627func searcherForTest(t testing.TB, b *index.ShardBuilder) zoekt.Searcher {
628 var buf bytes.Buffer
629 if err := b.Write(&buf); err != nil {
630 t.Fatal(err)
631 }
632 f := &memSeeker{buf.Bytes()}
633
634 searcher, err := index.NewSearcher(f)
635 if err != nil {
636 t.Fatalf("NewSearcher: %v", err)
637 }
638
639 return searcher
640}
641
642func reposForTest(n int) (result []*zoekt.Repository) {
643 for i := range n {
644 result = append(result, &zoekt.Repository{
645 ID: uint32(i + 1),
646 Name: fmt.Sprintf("test-repository-%d", i),
647 })
648 }
649 return result
650}
651
652func testSearcherForRepo(b testing.TB, r *zoekt.Repository, numFiles int) zoekt.Searcher {
653 builder := testShardBuilder(b, r)
654
655 if err := builder.Add(index.Document{
656 Name: fmt.Sprintf("%s/filename-%d.go", r.Name, 0),
657 Content: []byte("needle needle needle haystack"),
658 }); err != nil {
659 b.Fatal(err)
660 }
661
662 for i := 1; i < numFiles; i++ {
663 if err := builder.Add(index.Document{
664 Name: fmt.Sprintf("%s/filename-%d.go", r.Name, i),
665 Content: []byte("haystack haystack haystack"),
666 }); err != nil {
667 b.Fatal(err)
668 }
669 }
670
671 return searcherForTest(b, builder)
672}
673
674func BenchmarkShardedSearch(b *testing.B) {
675 ss := newShardedSearcher(int64(runtime.GOMAXPROCS(0)))
676
677 filesPerRepo := 300
678 repos := reposForTest(3000)
679 var repoSetIDs []uint32
680
681 shards := make(map[string]zoekt.Searcher, len(repos))
682 for i, r := range repos {
683 shards[r.Name] = testSearcherForRepo(b, r, filesPerRepo)
684 if i%2 == 0 {
685 repoSetIDs = append(repoSetIDs, r.ID)
686 }
687 }
688
689 ss.replace(shards)
690
691 ctx := context.Background()
692 opts := &zoekt.SearchOptions{}
693
694 needleSub := &query.Substring{Pattern: "needle"}
695 haystackSub := &query.Substring{Pattern: "haystack"}
696 helloworldSub := &query.Substring{Pattern: "helloworld"}
697 haystackCap, err := query.Parse("hay(s(t))(a)ck")
698 if err != nil {
699 b.Fatal(err)
700 }
701
702 haystackNonCap, err := query.Parse("hay(?:s(?:t))(?:a)ck")
703 if err != nil {
704 b.Fatal(err)
705 }
706
707 setAnd := func(q query.Q) func() query.Q {
708 return func() query.Q {
709 return query.NewAnd(query.NewSingleBranchesRepos("head", repoSetIDs...), q)
710 }
711 }
712
713 search := func(b *testing.B, q query.Q, wantFiles int) {
714 b.Helper()
715
716 res, err := ss.Search(ctx, q, opts)
717 if err != nil {
718 b.Fatalf("Search(%s): %v", q, err)
719 }
720 if have := len(res.Files); have != wantFiles {
721 b.Fatalf("wrong number of file results. have=%d, want=%d", have, wantFiles)
722 }
723 }
724
725 benchmarks := []struct {
726 name string
727 q func() query.Q
728 wantFiles int
729 }{
730 {"substring all results", func() query.Q { return haystackSub }, len(repos) * filesPerRepo},
731 {"substring no results", func() query.Q { return helloworldSub }, 0},
732 {"substring some results", func() query.Q { return needleSub }, len(repos)},
733
734 {"regexp all results capture", func() query.Q { return haystackCap }, len(repos) * filesPerRepo},
735 {"regexp all results non-capture", func() query.Q { return haystackNonCap }, len(repos) * filesPerRepo},
736
737 {"substring all results and repo set", setAnd(haystackSub), len(repoSetIDs) * filesPerRepo},
738 {"substring some results and repo set", setAnd(needleSub), len(repoSetIDs)},
739 {"substring no results and repo set", setAnd(helloworldSub), 0},
740 }
741
742 for _, bb := range benchmarks {
743 b.Run(bb.name, func(b *testing.B) {
744 b.ReportAllocs()
745
746 for n := 0; n < b.N; n++ {
747 q := bb.q()
748
749 search(b, q, bb.wantFiles)
750 }
751 })
752 }
753}
754
755func TestRawQuerySearch(t *testing.T) {
756 ss := newShardedSearcher(1)
757
758 var nextShardNum int
759 addShard := func(repo string, rawConfig map[string]string, docs ...index.Document) {
760 r := &zoekt.Repository{Name: repo}
761 r.RawConfig = rawConfig
762 b := testShardBuilder(t, r, docs...)
763 shard := searcherForTest(t, b)
764 ss.replace(map[string]zoekt.Searcher{fmt.Sprintf("key-%d", nextShardNum): shard})
765 nextShardNum++
766 }
767 addShard("public", map[string]string{"public": "1"}, index.Document{Name: "f1", Content: []byte("foo bar bas")})
768 addShard("private_archived", map[string]string{"archived": "1"}, index.Document{Name: "f2", Content: []byte("foo bas")})
769 addShard("public_fork", map[string]string{"public": "1", "fork": "1"}, index.Document{Name: "f3", Content: []byte("foo bar")})
770
771 cases := []struct {
772 pattern string
773 flags query.RawConfig
774 wantRepos []string
775 wantFiles int
776 }{
777 {
778 pattern: "bas",
779 flags: query.RcOnlyPublic,
780 wantRepos: []string{"public"},
781 wantFiles: 1,
782 },
783 {
784 pattern: "foo",
785 flags: query.RcOnlyPublic,
786 wantRepos: []string{"public", "public_fork"},
787 wantFiles: 2,
788 },
789 {
790 pattern: "foo",
791 flags: query.RcOnlyPublic | query.RcNoForks,
792 wantRepos: []string{"public"},
793 wantFiles: 1,
794 },
795 {
796 pattern: "bar",
797 flags: query.RcOnlyForks,
798 wantRepos: []string{"public_fork"},
799 wantFiles: 1,
800 },
801 {
802 pattern: "bas",
803 flags: query.RcNoArchived,
804 wantRepos: []string{"public"},
805 wantFiles: 1,
806 },
807 {
808 pattern: "foo",
809 flags: query.RcNoForks,
810 wantRepos: []string{"public", "private_archived"},
811 wantFiles: 2,
812 },
813 {
814 pattern: "bas",
815 flags: query.RcOnlyArchived,
816 wantRepos: []string{"private_archived"},
817 wantFiles: 1,
818 },
819 {
820 pattern: "foo",
821 flags: query.RcOnlyPrivate,
822 wantRepos: []string{"private_archived"},
823 wantFiles: 1,
824 },
825 {
826 pattern: "foo",
827 flags: query.RcOnlyPrivate | query.RcNoArchived,
828 wantRepos: []string{},
829 wantFiles: 0,
830 },
831 }
832 for _, c := range cases {
833 t.Run(fmt.Sprintf("pattern:%s", c.pattern), func(t *testing.T) {
834 q := query.NewAnd(&query.Substring{Pattern: c.pattern}, c.flags)
835
836 sr, err := ss.Search(context.Background(), q, &zoekt.SearchOptions{})
837 if err != nil {
838 t.Fatal(err)
839 }
840
841 if got := len(sr.Files); got != c.wantFiles {
842 t.Fatalf("wanted %d, got %d", c.wantFiles, got)
843 }
844
845 if c.wantFiles == 0 {
846 return
847 }
848
849 gotRepos := make([]string, 0, len(sr.RepoURLs))
850 for k := range sr.RepoURLs {
851 gotRepos = append(gotRepos, k)
852 }
853 sort.Strings(gotRepos)
854 sort.Strings(c.wantRepos)
855 if d := cmp.Diff(c.wantRepos, gotRepos); d != "" {
856 t.Fatalf("(-want, +got):\n%s", d)
857 }
858 })
859 }
860}
861
862func TestPrioritySlice(t *testing.T) {
863 p := &prioritySlice{}
864 for step, oper := range []struct {
865 isAppend bool
866 value float64
867 expectedMax float64
868 }{
869 {true, 1, 1},
870 {true, 3, 3},
871 {true, 2, 3},
872 {false, 1, 3},
873 {false, 3, 2},
874 {false, 2, math.Inf(-1)},
875 } {
876 if oper.isAppend {
877 p.append(oper.value)
878 } else {
879 p.remove(oper.value)
880 }
881 max := p.max()
882 if max != oper.expectedMax {
883 t.Errorf("%d: got %f, want %f", step, max, oper.expectedMax)
884 }
885 }
886}
887
888func TestSendByRepository(t *testing.T) {
889 wantStats := zoekt.Stats{}
890 wantStats.ShardsScanned = 1
891
892 // n1, n2, n3 are the number of file matches for each of the 3 repositories in this
893 // test.
894 f := func(n1, n2, n3 uint8) bool {
895 sr := createMockSearchResult(n1, n2, n3, wantStats)
896
897 mock := &mockSender{}
898 sendByRepository(sr, &zoekt.SearchOptions{}, mock)
899
900 if diff := cmp.Diff(wantStats, mock.stats); diff != "" {
901 t.Logf("-want,+got\n%s", diff)
902 return false
903 }
904
905 nonZero := 0
906 for _, l := range []uint8{n1, n2, n3} {
907 if l > 0 {
908 nonZero++
909 }
910 }
911 if l := len(mock.files); l != nonZero {
912 t.Logf("wanted results from %d repositores, got %d", nonZero, l)
913 return false
914 }
915
916 gotTotal := 0
917 for _, fs := range mock.files {
918 gotTotal += len(fs)
919 }
920 wantTotal := int(n1) + int(n2) + int(n3)
921 if gotTotal != wantTotal {
922 t.Logf("wanted %d file matches, got %d", wantTotal, gotTotal)
923 return false
924 }
925
926 for _, fs := range mock.files {
927 if len(fs) == 0 {
928 t.Logf("got search result with 0 file matches after split")
929 return false
930 }
931 }
932 return true
933 }
934
935 if err := quick.Check(f, nil); err != nil {
936 t.Error(err)
937 }
938}
939
940type mockSender struct {
941 stats zoekt.Stats
942 files [][]zoekt.FileMatch
943}
944
945func (s *mockSender) Send(sr *zoekt.SearchResult) {
946 s.stats.Add(sr.Stats)
947 if len(sr.Files) == 0 {
948 return
949 }
950 s.files = append(s.files, sr.Files)
951}
952
953func createMockSearchResult(n1, n2, n3 uint8, stats zoekt.Stats) *zoekt.SearchResult {
954 sr := &zoekt.SearchResult{RepoURLs: make(map[string]string)}
955 for i, n := range []uint8{n1, n2, n3} {
956 if n == 0 {
957 continue
958 }
959 tmp := mkSearchResult(int(n), uint32(i))
960 sr.Files = append(sr.Files, tmp.Files...)
961 for k := range tmp.RepoURLs {
962 sr.RepoURLs[k] = ""
963 }
964 }
965 sr.Stats = stats
966 return sr
967}
968
969func mkSearchResult(n int, repoID uint32) *zoekt.SearchResult {
970 if n == 0 {
971 return &zoekt.SearchResult{}
972 }
973 fm := make([]zoekt.FileMatch, 0, n)
974 for range n {
975 fm = append(fm, zoekt.FileMatch{Repository: fmt.Sprintf("repo%d", repoID), RepositoryID: repoID})
976 }
977
978 return &zoekt.SearchResult{Files: fm, RepoURLs: map[string]string{fmt.Sprintf("repo%d", repoID): ""}}
979}
980
981func TestFileBasedSearch(t *testing.T) {
982 cases := []struct {
983 name string
984 testShardedSearch func(t *testing.T, q query.Q, ib *index.ShardBuilder, useDocumentRanks bool) []zoekt.FileMatch
985 }{
986 {"Search", testShardedSearch},
987 {"StreamSearch", testShardedStreamSearch},
988 }
989
990 c1 := []byte("I love bananas without skin")
991 // -----------0123456789012345678901234567890123456789
992 c2 := []byte("In Dutch, ananas means pineapple")
993 // -----------0123456789012345678901234567890123456789
994 b := testShardBuilder(t, nil,
995 index.Document{Name: "f1", Content: c1},
996 index.Document{Name: "f2", Content: c2},
997 )
998
999 for _, tt := range cases {
1000 for _, useDocumentRanks := range []bool{false, true} {
1001 t.Run(tt.name, func(t *testing.T) {
1002 matches := tt.testShardedSearch(t, &query.Substring{
1003 CaseSensitive: false,
1004 Pattern: "ananas",
1005 }, b, useDocumentRanks)
1006
1007 if len(matches) != 2 {
1008 t.Fatalf("got %v, want 2 matches", matches)
1009 }
1010 if matches[0].FileName != "f2" || matches[1].FileName != "f1" {
1011 t.Fatalf("got %v, want matches {f1,f2}", matches)
1012 }
1013 if matches[0].LineMatches[0].LineFragments[0].Offset != 10 || matches[1].LineMatches[0].LineFragments[0].Offset != 8 {
1014 t.Fatalf("got %#v, want offsets 10,8", matches)
1015 }
1016 })
1017 }
1018 }
1019}
1020
1021func TestWordBoundaryRanking(t *testing.T) {
1022 cases := []struct {
1023 name string
1024 testShardedSearch func(t *testing.T, q query.Q, ib *index.ShardBuilder, useDocumentRanks bool) []zoekt.FileMatch
1025 }{
1026 {"Search", testShardedSearch},
1027 {"StreamSearch", testShardedStreamSearch},
1028 }
1029
1030 b := testShardBuilder(t, nil,
1031 index.Document{Name: "f1", Content: []byte("xbytex xbytex")},
1032 index.Document{Name: "f2", Content: []byte("xbytex\nbytex\nbyte bla")},
1033 // -----------------------------------------0123456 789012 34567890
1034 index.Document{Name: "f3", Content: []byte("xbytex ybytex")})
1035
1036 for _, tt := range cases {
1037 for _, useDocumentRanks := range []bool{false, true} {
1038 t.Run(tt.name, func(t *testing.T) {
1039 files := tt.testShardedSearch(t, &query.Substring{Pattern: "byte"}, b, useDocumentRanks)
1040
1041 if len(files) != 3 {
1042 t.Fatalf("got %#v, want 3 files", files)
1043 }
1044
1045 file0 := files[0]
1046 if file0.FileName != "f2" || len(file0.LineMatches) != 3 {
1047 t.Fatalf("got file %s, num matches %d (%#v), want 3 matches in file f2", file0.FileName, len(file0.LineMatches), file0)
1048 }
1049
1050 if file0.LineMatches[0].LineFragments[0].Offset != 13 {
1051 t.Fatalf("got first match %#v, want full word match", files[0].LineMatches[0])
1052 }
1053 if file0.LineMatches[1].LineFragments[0].Offset != 7 {
1054 t.Fatalf("got second match %#v, want partial word match", files[0].LineMatches[1])
1055 }
1056 })
1057 }
1058 }
1059}
1060
1061func TestAtomCountScore(t *testing.T) {
1062 cases := []struct {
1063 name string
1064 testShardedSearch func(t *testing.T, q query.Q, ib *index.ShardBuilder, useDocumentRanks bool) []zoekt.FileMatch
1065 }{
1066 {"Search", testShardedSearch},
1067 {"StreamSearch", testShardedStreamSearch},
1068 }
1069
1070 b := testShardBuilder(t,
1071 &zoekt.Repository{
1072 Branches: []zoekt.RepositoryBranch{
1073 {Name: "branches", Version: "v1"},
1074 {Name: "needle", Version: "v2"},
1075 },
1076 },
1077 index.Document{Name: "f1", Content: []byte("needle the bla"), Branches: []string{"branches"}},
1078 index.Document{Name: "needle-file-branch", Content: []byte("needle content"), Branches: []string{"needle"}},
1079 index.Document{Name: "needle-file", Content: []byte("needle content"), Branches: []string{"branches"}})
1080
1081 for _, tt := range cases {
1082 for _, useDocumentRanks := range []bool{false, true} {
1083 t.Run(tt.name, func(t *testing.T) {
1084 files := tt.testShardedSearch(t,
1085 query.NewOr(
1086 &query.Substring{Pattern: "needle"},
1087 &query.Substring{Pattern: "needle", FileName: true},
1088 &query.Branch{Pattern: "needle"},
1089 ), b, useDocumentRanks)
1090 var got []string
1091 for _, f := range files {
1092 got = append(got, f.FileName)
1093 }
1094 want := []string{"needle-file-branch", "needle-file", "f1"}
1095 if !reflect.DeepEqual(got, want) {
1096 t.Errorf("got %v, want %v", got, want)
1097 }
1098 })
1099 }
1100 }
1101}
1102
1103func TestUseBM25Scoring(t *testing.T) {
1104 b := testShardBuilder(t,
1105 &zoekt.Repository{},
1106 index.Document{Name: "f1", Content: []byte("one two two three")},
1107 index.Document{Name: "f2", Content: []byte("one two one two")},
1108 index.Document{Name: "f3", Content: []byte("one three three three")})
1109
1110 ss := newShardedSearcher(1)
1111 searcher := searcherForTest(t, b)
1112 ss.replace(map[string]zoekt.Searcher{"r1": searcher})
1113
1114 q := query.NewOr(
1115 &query.Substring{Pattern: "one"},
1116 &query.Substring{Pattern: "three"})
1117
1118 opts := zoekt.SearchOptions{
1119 UseBM25Scoring: true,
1120 }
1121
1122 results, err := ss.Search(context.Background(), q, &opts)
1123 if err != nil {
1124 t.Fatal(err)
1125 }
1126
1127 var got []string
1128 for _, f := range results.Files {
1129 got = append(got, f.FileName)
1130 }
1131
1132 want := []string{"f3", "f1", "f2"}
1133 if !reflect.DeepEqual(got, want) {
1134 t.Errorf("got %v, want %v", got, want)
1135 }
1136}
1137
1138func testShardedStreamSearch(t *testing.T, q query.Q, ib *index.ShardBuilder, useDocumentRanks bool) []zoekt.FileMatch {
1139 ss := newShardedSearcher(1)
1140 searcher := searcherForTest(t, ib)
1141 ss.replace(map[string]zoekt.Searcher{"r1": searcher})
1142
1143 var files []zoekt.FileMatch
1144 sender := zoekt.SenderFunc(func(result *zoekt.SearchResult) {
1145 files = append(files, result.Files...)
1146 })
1147
1148 opts := zoekt.SearchOptions{}
1149 if useDocumentRanks {
1150 opts.FlushWallTime = 10 * time.Millisecond
1151 }
1152 if err := ss.StreamSearch(context.Background(), q, &opts, sender); err != nil {
1153 t.Fatal(err)
1154 }
1155 return files
1156}
1157
1158func testShardedSearch(t *testing.T, q query.Q, ib *index.ShardBuilder, useDocumentRanks bool) []zoekt.FileMatch {
1159 ss := newShardedSearcher(1)
1160 searcher := searcherForTest(t, ib)
1161 ss.replace(map[string]zoekt.Searcher{"r1": searcher})
1162
1163 opts := zoekt.SearchOptions{}
1164 if useDocumentRanks {
1165 opts.FlushWallTime = 50 * time.Millisecond
1166 }
1167 sres, _ := ss.Search(context.Background(), q, &opts)
1168 return sres.Files
1169}
1170
1171// Ensure we work on empty shard directories.
1172func TestNewDirectorySearcher_empty(t *testing.T) {
1173 ctx := context.Background()
1174
1175 test := func(t *testing.T, ss zoekt.Streamer) {
1176 res, err := ss.Search(ctx, &query.Const{Value: true}, nil)
1177 if err != nil {
1178 t.Fatal("Search non-nil error", err)
1179 }
1180
1181 if diff := cmp.Diff(&zoekt.SearchResult{}, res, cmpopts.IgnoreFields(zoekt.Stats{}, "Duration", "Wait"), cmpopts.EquateEmpty()); diff != "" {
1182 t.Fatalf("Search had non empty results (-want, +got):\n%s", diff)
1183 }
1184
1185 rl, err := ss.List(ctx, &query.Const{Value: true}, nil)
1186 if err != nil {
1187 t.Fatal("List non-nil error", err)
1188 }
1189 if diff := cmp.Diff(&zoekt.RepoList{}, rl, cmpopts.EquateEmpty()); diff != "" {
1190 t.Fatalf("List had non empty results (-want, +got):\n%s", diff)
1191 }
1192 }
1193
1194 dir := t.TempDir()
1195 t.Run("blocking", func(t *testing.T) {
1196 ss, err := NewDirectorySearcher(dir)
1197 if err != nil {
1198 t.Fatal(err)
1199 }
1200 t.Cleanup(ss.Close)
1201
1202 // We expect crashes to be empty as soon as NewDirectorySearcher returns
1203 // so we can validate straight away.
1204 test(t, ss)
1205 })
1206
1207 t.Run("fast", func(t *testing.T) {
1208 ss, err := NewDirectorySearcherFast(dir)
1209 if err != nil {
1210 t.Fatal(err)
1211 }
1212 t.Cleanup(ss.Close)
1213
1214 deadline := testDeadline(t, 10*time.Second)
1215
1216 // Wait for scanning of directory to be done. We should be returning
1217 // non-zero crashes until then.
1218 waitForPredicate(deadline, 10*time.Millisecond, func() bool {
1219 res, err := ss.Search(ctx, &query.Const{Value: true}, nil)
1220 if err != nil {
1221 t.Fatal(err)
1222 }
1223 return res.Stats.Crashes == 0
1224 })
1225
1226 test(t, ss)
1227 })
1228}
1229
1230// testDeadline returns the deadline for t, but ensures it is no longer than
1231// maxTimeout away.
1232func testDeadline(t *testing.T, maxTimeout time.Duration) time.Time {
1233 deadline := time.Now().Add(maxTimeout)
1234 if d, ok := t.Deadline(); ok && d.Before(deadline) {
1235 // give 1s for us to do a final test run
1236 deadline = d.Add(-time.Second)
1237 }
1238 return deadline
1239}
1240
1241func waitForPredicate(deadline time.Time, tick time.Duration, pred func() bool) bool {
1242 for time.Now().Before(deadline) {
1243 if pred() {
1244 return true
1245 }
1246
1247 time.Sleep(tick)
1248 }
1249
1250 return pred()
1251}