fork of https://github.com/sourcegraph/zoekt
Ranking: test sharded search with document ranks (#545)
This adds an integration test for sharded search when document ranking is
enabled. It checks that results are combined correctly across multiple shards,
and that the ranked combined results are streamed out.
author
Julie Tibshirani
committer
GitHub
date
(Feb 23, 2023, 9:39 AM -0800)
commit
25607737
2560773778893ff9f43ebb403ca97f85ffdae8d3
parent
a8188ae8
a8188ae891179adf76a183376b0522fd1069ba8b
+60
+60
shards/shards_test.go
+60
shards/shards_test.go
···
226
226
}
227
227
}
228
228
229
+
func 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", 0.25, zoekt.Document{Name: "f1", Content: []byte("foobar")})
248
+
addShard("moderately-popular", 0.4, zoekt.Document{Name: "f2", Content: []byte("foobaz")})
249
+
addShard("weekend-project-2", 0.25, zoekt.Document{Name: "f3", Content: []byte("foo bar")})
250
+
addShard("super-star", 0.9, 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
+
229
289
func TestFilteringShardsByRepoSetOrBranchesReposOrRepoIDs(t *testing.T) {
230
290
ss := newShardedSearcher(1)
231
291