fork of https://github.com/sourcegraph/zoekt
eval: fix gatherBranches (#562)
fixes a bug where we wouldn't display branches correctly if the query contained one or more branch filters.
author
Stefan Hengl
committer
GitHub
date
(Mar 23, 2023, 11:02 AM +0100)
commit
cf9500b1
cf9500b19ea3158f7b11a09fd81bf2174cacfd95
parent
27effbad
27effbadc957cb77a48098ce69455f4caeaaa7e2
+74
-17
+24
-16
eval.go
+24
-16
eval.go
···
563
563
return -1
564
564
}
565
565
566
-
// gatherBranches returns a list of branch names.
566
+
// gatherBranches returns a list of branch names taking into account any branch
567
+
// filters in the query. If the query contains a branch filter, it returns all
568
+
// branches containing the docID and matching the branch filter. Otherwise, it
569
+
// returns all branches containing docID.
567
570
func (d *indexData) gatherBranches(docID uint32, mt matchTree, known map[matchTree]bool) []string {
568
-
foundBranchQuery := false
569
-
var branches []string
570
571
repoIdx := d.repos[docID]
572
+
573
+
var mask uint64
571
574
visitMatches(mt, known, func(mt matchTree) {
572
575
bq, ok := mt.(*branchQueryMatchTree)
573
-
if ok {
574
-
foundBranchQuery = true
575
-
branches = append(branches,
576
-
d.branchNames[repoIdx][uint(bq.masks[repoIdx])])
576
+
if !ok {
577
+
return
577
578
}
579
+
580
+
// bq.masks[repoIdx]: the branches we are filtering on
581
+
// bq.branchMask(): the branches of the current file
582
+
mask = mask | (bq.masks[repoIdx] & bq.branchMask())
578
583
})
579
584
580
-
if !foundBranchQuery {
581
-
mask := d.fileBranchMasks[docID]
582
-
id := uint32(1)
583
-
for mask != 0 {
584
-
if mask&0x1 != 0 {
585
-
branches = append(branches, d.branchNames[repoIdx][uint(id)])
586
-
}
587
-
id <<= 1
588
-
mask >>= 1
585
+
if mask == 0 {
586
+
mask = d.fileBranchMasks[docID]
587
+
}
588
+
589
+
var branches []string
590
+
id := uint32(1)
591
+
for mask != 0 {
592
+
if mask&0x1 != 0 {
593
+
branches = append(branches, d.branchNames[repoIdx][uint(id)])
589
594
}
595
+
id <<= 1
596
+
mask >>= 1
590
597
}
598
+
591
599
return branches
592
600
}
593
601
+45
eval_test.go
+45
eval_test.go
···
25
25
"github.com/RoaringBitmap/roaring"
26
26
"github.com/google/go-cmp/cmp"
27
27
"github.com/grafana/regexp"
28
+
28
29
"github.com/sourcegraph/zoekt/query"
29
30
)
30
31
···
339
340
h.Write([]byte(name))
340
341
return h.Sum32()
341
342
}
343
+
344
+
func TestGatherBranches(t *testing.T) {
345
+
content := []byte("dummy")
346
+
b := testIndexBuilder(t, &Repository{
347
+
Branches: []RepositoryBranch{
348
+
{"foo", "v1"},
349
+
{"foo-2", "v1"},
350
+
{"main", "v1"},
351
+
{"bar", "v1"},
352
+
{"quz", "v1"},
353
+
}},
354
+
Document{Name: "f1", Content: content, Branches: []string{"foo", "bar", "quz"}},
355
+
Document{Name: "f2", Content: content, Branches: []string{"foo", "foo-2"}},
356
+
Document{Name: "f3", Content: content, Branches: []string{"main"}})
357
+
358
+
d := searcherForTest(t, b).(*indexData)
359
+
360
+
sr, err := d.Search(
361
+
context.Background(),
362
+
&query.Or{Children: []query.Q{
363
+
&query.Branch{Pattern: "foo"},
364
+
&query.Branch{Pattern: "quz"},
365
+
}},
366
+
&SearchOptions{},
367
+
)
368
+
if err != nil {
369
+
t.Fatal(err)
370
+
}
371
+
372
+
want := map[string][]string{
373
+
"f1": []string{"foo", "quz"},
374
+
"f2": []string{"foo", "foo-2"},
375
+
}
376
+
377
+
if len(sr.Files) != 2 {
378
+
t.Fatalf("len(sr.Files): want %d, got %d", 2, len(sr.Files))
379
+
}
380
+
381
+
for _, f := range sr.Files {
382
+
if d := cmp.Diff(want[f.FileName], f.Branches); d != "" {
383
+
t.Fatalf("-want,+got:\n%s", d)
384
+
}
385
+
}
386
+
}
+5
-1
matchtree.go
+5
-1
matchtree.go
···
181
181
docID uint32
182
182
}
183
183
184
+
func (t *branchQueryMatchTree) branchMask() uint64 {
185
+
return t.fileMasks[t.docID]
186
+
}
187
+
184
188
type symbolRegexpMatchTree struct {
185
189
matchTree
186
190
regexp *regexp.Regexp
···
672
676
}
673
677
674
678
func (t *branchQueryMatchTree) matches(cp *contentProvider, cost int, known map[matchTree]bool) (bool, bool) {
675
-
return t.fileMasks[t.docID]&t.masks[t.repos[t.docID]] != 0, true
679
+
return t.branchMask()&t.masks[t.repos[t.docID]] != 0, true
676
680
}
677
681
678
682
func (t *regexpMatchTree) matches(cp *contentProvider, cost int, known map[matchTree]bool) (bool, bool) {