fork of https://github.com/sourcegraph/zoekt
1package e2e
2
3import (
4 "bytes"
5 "context"
6 "flag"
7 "fmt"
8 "io"
9 "net/url"
10 "os"
11 "os/exec"
12 "path/filepath"
13 "strings"
14 "testing"
15 "time"
16
17 "github.com/google/go-cmp/cmp"
18 "github.com/sourcegraph/zoekt"
19 "github.com/sourcegraph/zoekt/build"
20 "github.com/sourcegraph/zoekt/internal/archive"
21 "github.com/sourcegraph/zoekt/query"
22 "github.com/sourcegraph/zoekt/shards"
23)
24
25var update = flag.Bool("update", false, "update golden file")
26
27var useShardCache = flag.Bool("shard_cache", false, "cache computed shards for faster test runs")
28
29// debugScore can be set to include much more output. Do not commit the
30// updated golden files, this is purely used for debugging in a local
31// environment.
32var debugScore = flag.Bool("debug_score", false, "include debug output in golden files.")
33
34func TestRanking(t *testing.T) {
35 if testing.Short() {
36 t.Skip("skipping due to short flag")
37 }
38
39 requireCTags(t)
40
41 archiveURLs := []string{
42 "https://github.com/sourcegraph/sourcegraph-public-snapshot/tree/v5.2.2",
43 "https://github.com/golang/go/tree/go1.21.4",
44 "https://github.com/sourcegraph/cody/tree/vscode-v0.14.5",
45 // The commit before ranking e2e tests were added to avoid matching
46 // content inside our golden files.
47 "https://github.com/sourcegraph/zoekt/commit/ef907c2371176aa3f97713d5bf182983ef090c6a",
48 }
49 q := func(query, target string) rankingQuery {
50 return rankingQuery{Query: query, Target: target}
51 }
52 queries := []rankingQuery{
53 // golang/go
54 q("test server", "github.com/golang/go/src/net/http/httptest/server.go"),
55 q("bytes buffer", "github.com/golang/go/src/bytes/buffer.go"),
56 q("bufio buffer", "github.com/golang/go/src/bufio/scan.go"),
57 q("time compare\\(", "github.com/golang/go/src/time/time.go"),
58
59 // sourcegraph/sourcegraph
60 q("graphql type User", "github.com/sourcegraph/sourcegraph-public-snapshot/cmd/frontend/graphqlbackend/schema.graphql"),
61 q("Get database/user", "github.com/sourcegraph/sourcegraph-public-snapshot/internal/database/users.go"),
62 q("InternalDoer", "github.com/sourcegraph/sourcegraph-public-snapshot/internal/httpcli/client.go"),
63 q("Repository metadata Write rbac", "github.com/sourcegraph/sourcegraph-public-snapshot/internal/rbac/constants.go"), // unsure if this is the best doc?
64
65 // cody
66 q("generate unit test", "github.com/sourcegraph/cody/lib/shared/src/chat/recipes/generate-test.ts"),
67 q("r:cody sourcegraph url", "github.com/sourcegraph/cody/lib/shared/src/sourcegraph-api/graphql/client.ts"),
68
69 // zoekt
70 q("zoekt searcher", "github.com/sourcegraph/zoekt/api.go"),
71
72 // exact phrases
73 q("assets are not configured for this binary", "github.com/sourcegraph/sourcegraph-public-snapshot/ui/assets/assets.go"),
74 q("sourcegraph/server docker image build", "github.com/sourcegraph/sourcegraph-public-snapshot/dev/tools.go"),
75
76 // symbols split up
77 q("bufio flush writer", "github.com/golang/go/src/net/http/transfer.go"), // bufioFlushWriter
78 q("coverage data writer", "github.com/golang/go/src/internal/coverage/encodecounter/encode.go"), // CoverageDataWriter
79 }
80
81 var indexDir string
82 if *useShardCache {
83 t.Logf("reusing index dir to speed up testing. If you have unexpected results remove %s", shardCache)
84 indexDir = shardCache
85 } else {
86 indexDir = t.TempDir()
87 }
88
89 for _, u := range archiveURLs {
90 if err := indexURL(indexDir, u); err != nil {
91 t.Fatal(err)
92 }
93 }
94
95 ss, err := shards.NewDirectorySearcher(indexDir)
96 if err != nil {
97 t.Fatalf("NewDirectorySearcher(%s): %v", indexDir, err)
98 }
99 defer ss.Close()
100
101 var ranks []int
102 for _, rq := range queries {
103 // normalise queryStr for writing to fs
104 name := strings.Map(func(r rune) rune {
105 if strings.ContainsRune(" :", r) {
106 return '_'
107 }
108 if '0' <= r && r <= '9' ||
109 'a' <= r && r <= 'z' ||
110 'A' <= r && r <= 'Z' {
111 return r
112 }
113 return -1
114 }, rq.Query)
115
116 t.Run(name, func(t *testing.T) {
117 q, err := query.Parse(rq.Query)
118 if err != nil {
119 t.Fatal(err)
120 }
121
122 sOpts := zoekt.SearchOptions{
123 // Use the same options sourcegraph has by default
124 ChunkMatches: true,
125 MaxWallTime: 20 * time.Second,
126 ShardMaxMatchCount: 10_000 * 10,
127 TotalMaxMatchCount: 100_000 * 10,
128 MaxDocDisplayCount: 500,
129
130 DebugScore: *debugScore,
131 }
132 result, err := ss.Search(context.Background(), q, &sOpts)
133 if err != nil {
134 t.Fatal(err)
135 }
136
137 ranks = append(ranks, targetRank(rq, result.Files))
138
139 var gotBuf bytes.Buffer
140 marshalMatches(&gotBuf, rq, q, result.Files)
141 assertGolden(t, name, gotBuf.Bytes())
142 })
143 }
144
145 t.Run("rank_stats", func(t *testing.T) {
146 if len(ranks) != len(queries) {
147 t.Skip("not computing rank stats since not all query cases ran")
148 }
149
150 var gotBuf bytes.Buffer
151 printf := func(format string, a ...any) {
152 _, _ = fmt.Fprintf(&gotBuf, format, a...)
153 }
154
155 printf("queries: %d\n", len(ranks))
156
157 for _, recallThreshold := range []int{1, 5} {
158 count := 0
159 for _, rank := range ranks {
160 if rank <= recallThreshold && rank > 0 {
161 count++
162 }
163 }
164 countp := float64(count) * 100 / float64(len(ranks))
165 printf("recall@%d: %d (%.0f%%)\n", recallThreshold, count, countp)
166 }
167
168 // Mean reciprocal rank
169 mrr := float64(0)
170 for _, rank := range ranks {
171 if rank > 0 {
172 mrr += 1 / float64(rank)
173 }
174 }
175 mrr /= float64(len(ranks))
176 printf("mrr: %f\n", mrr)
177
178 assertGolden(t, "rank_stats", gotBuf.Bytes())
179 })
180}
181
182func assertGolden(t *testing.T, name string, got []byte) {
183 t.Helper()
184
185 wantPath := filepath.Join("testdata", name+".txt")
186 if *update {
187 if err := os.WriteFile(wantPath, got, 0o600); err != nil {
188 t.Fatal(err)
189 }
190 }
191 want, err := os.ReadFile(wantPath)
192 if err != nil {
193 t.Fatal(err)
194 }
195
196 if d := cmp.Diff(string(want), string(got)); d != "" {
197 t.Fatalf("unexpected (-want, +got):\n%s", d)
198 }
199}
200
201type rankingQuery struct {
202 Query string
203 Target string
204}
205
206var (
207 tarballCache = "/tmp/zoekt-test-ranking-tarballs-" + os.Getenv("USER")
208 shardCache = "/tmp/zoekt-test-ranking-shards-" + os.Getenv("USER")
209)
210
211func indexURL(indexDir, u string) error {
212 if err := os.MkdirAll(tarballCache, 0o700); err != nil {
213 return err
214 }
215
216 opts := archive.Options{
217 Archive: u,
218 Incremental: true,
219 }
220 opts.SetDefaults() // sets metadata like Name and the codeload URL
221 u = opts.Archive
222
223 // if opts.Commit is set but opts.Branch is not, then we just need to give
224 // the commit a name for testing.
225 if opts.Commit != "" && opts.Branch == "" {
226 opts.Branch = "test"
227 }
228
229 // update Archive location to cached location
230 cacheBase := fmt.Sprintf("%s-%s%s.tar.gz", url.QueryEscape(opts.Name), opts.Branch, opts.Commit) // assume .tar.gz
231 path := filepath.Join(tarballCache, cacheBase)
232 opts.Archive = path
233
234 if _, err := os.Stat(path); os.IsNotExist(err) {
235 if err := download(u, path); err != nil {
236 return err
237 }
238 }
239
240 // TODO scip
241 // languageMap := make(ctags.LanguageMap)
242 // for _, lang := range []string{"kotlin", "rust", "ruby", "go", "python", "javascript", "c_sharp", "scala", "typescript", "zig"} {
243 // languageMap[lang] = ctags.ScipCTags
244 // }
245
246 err := archive.Index(opts, build.Options{
247 IndexDir: indexDir,
248 CTagsMustSucceed: true,
249 })
250 if err != nil {
251 return fmt.Errorf("failed to index %s: %w", opts.Archive, err)
252 }
253
254 return nil
255}
256
257func download(url, dst string) error {
258 tmpPath := dst + ".part"
259
260 rc, err := archive.OpenReader(url)
261 if err != nil {
262 return err
263 }
264 defer rc.Close()
265
266 f, err := os.OpenFile(tmpPath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0o600)
267 if err != nil {
268 return err
269 }
270 defer f.Close()
271
272 _, err = io.Copy(f, rc)
273 if err != nil {
274 return err
275 }
276
277 err = f.Close()
278 if err != nil {
279 return err
280 }
281
282 return os.Rename(tmpPath, dst)
283}
284
285const (
286 chunkMatchesPerFile = 3
287 fileMatchesPerSearch = 6
288)
289
290func docName(f zoekt.FileMatch) string {
291 return f.Repository + "/" + f.FileName
292}
293
294func marshalMatches(w io.Writer, rq rankingQuery, q query.Q, files []zoekt.FileMatch) {
295 _, _ = fmt.Fprintf(w, "queryString: %s\n", rq.Query)
296 _, _ = fmt.Fprintf(w, "query: %s\n", q)
297 _, _ = fmt.Fprintf(w, "targetRank: %d\n\n", targetRank(rq, files))
298
299 files, hiddenFiles := splitAtIndex(files, fileMatchesPerSearch)
300 for _, f := range files {
301 doc := docName(f)
302 if doc == rq.Target {
303 doc = "**" + doc + "**"
304 }
305 _, _ = fmt.Fprintf(w, "%s%s\n", doc, addTabIfNonEmpty(f.Debug))
306
307 chunks, hidden := splitAtIndex(f.ChunkMatches, chunkMatchesPerFile)
308
309 for _, m := range chunks {
310 _, _ = fmt.Fprintf(w, "%d:%s%s\n", m.ContentStart.LineNumber, strings.TrimRight(string(m.Content), "\n"), addTabIfNonEmpty(m.DebugScore))
311 }
312
313 if len(hidden) > 0 {
314 _, _ = fmt.Fprintf(w, "hidden %d more line matches\n", len(hidden))
315 }
316 _, _ = fmt.Fprintln(w)
317 }
318
319 if len(hiddenFiles) > 0 {
320 fmt.Fprintf(w, "hidden %d more file matches\n", len(hiddenFiles))
321 }
322}
323
324func targetRank(rq rankingQuery, files []zoekt.FileMatch) int {
325 for i, f := range files {
326 if docName(f) == rq.Target {
327 return i + 1
328 }
329 }
330 return -1
331}
332
333func splitAtIndex[E any](s []E, idx int) ([]E, []E) {
334 if idx < len(s) {
335 return s[:idx], s[idx:]
336 }
337 return s, nil
338}
339
340func addTabIfNonEmpty(s string) string {
341 if s != "" {
342 return "\t" + s
343 }
344 return s
345}
346
347func requireCTags(tb testing.TB) {
348 tb.Helper()
349
350 if os.Getenv("CTAGS_COMMAND") != "" {
351 return
352 }
353 if _, err := exec.LookPath("universal-ctags"); err == nil {
354 return
355 }
356
357 // On CI we require ctags to be available. Otherwise we skip
358 if os.Getenv("CI") != "" {
359 tb.Fatal("universal-ctags is missing")
360 } else {
361 tb.Skip("universal-ctags is missing")
362 }
363}