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