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