fork of https://github.com/sourcegraph/zoekt
ranking: add test for scoring based on ctags (#313)
Now that we have ctags available on CI we can add tests that check
regressions of our scoring.
author
Stefan Hengl
committer
GitHub
date
(Mar 30, 2022, 4:59 PM +0200)
commit
da2ce8a9
da2ce8a9b897fc59faebf9a73083f53da914163a
parent
10d20196
10d20196f75e5d62f3c7b01d7b29b316109017b0
+34
-20
+14
-15
build/builder.go
+14
-15
build/builder.go
···
230
230
temp, final string
231
231
}
232
232
233
-
// SetDefaults sets reasonable default options.
234
-
func (o *Options) SetDefaults() {
235
-
if o.CTags == "" {
236
-
if ctags := os.Getenv("CTAGS_COMMAND"); ctags != "" {
237
-
o.CTags = ctags
238
-
}
233
+
func checkCTags() string {
234
+
if ctags := os.Getenv("CTAGS_COMMAND"); ctags != "" {
235
+
return ctags
236
+
}
237
+
238
+
if ctags, err := exec.LookPath("universal-ctags"); err == nil {
239
+
return ctags
239
240
}
240
241
241
-
if o.CTags == "" {
242
-
ctags, err := exec.LookPath("universal-ctags")
243
-
if err == nil {
244
-
o.CTags = ctags
245
-
}
242
+
if ctags, err := exec.LookPath("ctags-exuberant"); err == nil {
243
+
return ctags
246
244
}
245
+
return ""
246
+
}
247
247
248
+
// SetDefaults sets reasonable default options.
249
+
func (o *Options) SetDefaults() {
248
250
if o.CTags == "" {
249
-
ctags, err := exec.LookPath("ctags-exuberant")
250
-
if err == nil {
251
-
o.CTags = ctags
252
-
}
251
+
o.CTags = checkCTags()
253
252
}
254
253
255
254
if o.Parallelism == 0 {
+16
-4
build/e2e_test.go
+16
-4
build/e2e_test.go
···
805
805
}
806
806
807
807
// With this test we want to capture regressions in the names returned by our
808
-
// language detection. We rely on the detected language and its spelling, for
809
-
// example, in scoring (see scoreKind).
810
-
func TestDetectLanguage(t *testing.T) {
808
+
// language detection and the scores assigned to file matches. We rely on the
809
+
// detected language and its spelling, for example, in scoring (see scoreKind).
810
+
func TestScoring(t *testing.T) {
811
+
if os.Getenv("CI") == "" && checkCTags() == "" {
812
+
t.Skip("ctags not available")
813
+
}
811
814
dir := t.TempDir()
812
815
813
816
opts := Options{
···
820
823
cases := []struct {
821
824
fileName string
822
825
content []byte
826
+
query query.Q
823
827
wantLanguage string
828
+
wantScore float64
824
829
}{
825
830
{
826
831
fileName: "hw.java",
···
833
838
}
834
839
}
835
840
`),
841
+
query: &query.Substring{Content: true, Pattern: "lloWorld"},
836
842
wantLanguage: "Java",
843
+
// 5500 (partial symbol at boundary) + 1000 (Java class) + 50 (partial word) + 400 (atom) + 10 (file order)
844
+
wantScore: 6960,
837
845
},
838
846
}
839
847
···
856
864
}
857
865
defer ss.Close()
858
866
859
-
srs, err := ss.Search(context.Background(), &query.Const{true}, new(zoekt.SearchOptions))
867
+
srs, err := ss.Search(context.Background(), c.query, new(zoekt.SearchOptions))
860
868
if err != nil {
861
869
t.Fatal(err)
862
870
}
863
871
864
872
if got, want := len(srs.Files), 1; got != want {
865
873
t.Fatalf("file matches: want %d, got %d", want, got)
874
+
}
875
+
876
+
if got := srs.Files[0].Score; got != c.wantScore {
877
+
t.Fatalf("score: want %f, got %f", c.wantScore, got)
866
878
}
867
879
868
880
if got := srs.Files[0].Language; got != c.wantLanguage {
+4
-1
contentprovider.go
+4
-1
contentprovider.go
···
348
348
switch language {
349
349
case "Java":
350
350
switch kind {
351
-
case "c": // classes
351
+
// 2022-03-30: go-ctags contains a regex rule for Java classes that sets "kind"
352
+
// to "classes" instead of "c". We have to cover both cases to support existing
353
+
// indexes.
354
+
case "c", "classes":
352
355
return scoreKindMatch
353
356
}
354
357
}