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/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
58 // sourcegraph/sourcegraph
59 q("graphql type User", "github.com/sourcegraph/sourcegraph/cmd/frontend/graphqlbackend/schema.graphql"),
60 q("Get database/user", "github.com/sourcegraph/sourcegraph/internal/database/users.go"),
61 q("InternalDoer", "github.com/sourcegraph/sourcegraph/internal/httpcli/client.go"),
62 q("Repository metadata Write rbac", "github.com/sourcegraph/sourcegraph/internal/rbac/constants.go"), // unsure if this is the best doc?
63
64 // cody
65 q("generate unit test", "github.com/sourcegraph/cody/lib/shared/src/chat/recipes/generate-test.ts"),
66 q("r:cody sourcegraph url", "github.com/sourcegraph/cody/lib/shared/src/sourcegraph-api/graphql/client.ts"),
67
68 // zoekt
69 q("zoekt searcher", "github.com/sourcegraph/zoekt/api.go"),
70
71 // exact phrases
72 q("assets are not configured for this binary", "github.com/sourcegraph/sourcegraph/ui/assets/assets.go"),
73 q("sourcegraph/server docker image build", "github.com/sourcegraph/sourcegraph/dev/tools.go"),
74
75 // symbols split up
76 q("bufio flush writer", "github.com/golang/go/src/net/http/transfer.go"), // bufioFlushWriter
77 q("coverage data writer", "github.com/golang/go/src/internal/coverage/encodecounter/encode.go"), // CoverageDataWriter
78 }
79
80 var indexDir string
81 if *useShardCache {
82 t.Logf("reusing index dir to speed up testing. If you have unexpected results remove %s", shardCache)
83 indexDir = shardCache
84 } else {
85 indexDir = t.TempDir()
86 }
87
88 for _, u := range archiveURLs {
89 if err := indexURL(indexDir, u); err != nil {
90 t.Fatal(err)
91 }
92 }
93
94 ss, err := shards.NewDirectorySearcher(indexDir)
95 if err != nil {
96 t.Fatalf("NewDirectorySearcher(%s): %v", indexDir, err)
97 }
98 defer ss.Close()
99
100 var ranks []int
101 for _, rq := range queries {
102 // normalise queryStr for writing to fs
103 name := strings.Map(func(r rune) rune {
104 if strings.ContainsRune(" :", r) {
105 return '_'
106 }
107 if '0' <= r && r <= '9' ||
108 'a' <= r && r <= 'z' ||
109 'A' <= r && r <= 'Z' {
110 return r
111 }
112 return -1
113 }, rq.Query)
114
115 t.Run(name, func(t *testing.T) {
116 q, err := query.Parse(rq.Query)
117 if err != nil {
118 t.Fatal(err)
119 }
120
121 sOpts := zoekt.SearchOptions{
122 // Use the same options sourcegraph has by default
123 ChunkMatches: true,
124 MaxWallTime: 20 * time.Second,
125 ShardMaxMatchCount: 10_000 * 10,
126 TotalMaxMatchCount: 100_000 * 10,
127 MaxDocDisplayCount: 500,
128
129 DebugScore: *debugScore,
130 }
131 result, err := ss.Search(context.Background(), q, &sOpts)
132 if err != nil {
133 t.Fatal(err)
134 }
135
136 ranks = append(ranks, targetRank(rq, result.Files))
137
138 var gotBuf bytes.Buffer
139 marshalMatches(&gotBuf, rq, q, result.Files)
140 assertGolden(t, name, gotBuf.Bytes())
141 })
142 }
143
144 t.Run("rank_stats", func(t *testing.T) {
145 if len(ranks) != len(queries) {
146 t.Skip("not computing rank stats since not all query cases ran")
147 }
148
149 var gotBuf bytes.Buffer
150 printf := func(format string, a ...any) {
151 _, _ = fmt.Fprintf(&gotBuf, format, a...)
152 }
153
154 printf("queries: %d\n", len(ranks))
155
156 for _, recallThreshold := range []int{1, 5} {
157 count := 0
158 for _, rank := range ranks {
159 if rank <= recallThreshold && rank > 0 {
160 count++
161 }
162 }
163 countp := float64(count) * 100 / float64(len(ranks))
164 printf("recall@%d: %d (%.0f%%)\n", recallThreshold, count, countp)
165 }
166
167 // Mean reciprocal rank
168 mrr := float64(0)
169 for _, rank := range ranks {
170 if rank > 0 {
171 mrr += 1 / float64(rank)
172 }
173 }
174 mrr /= float64(len(ranks))
175 printf("mrr: %f\n", mrr)
176
177 assertGolden(t, "rank_stats", gotBuf.Bytes())
178 })
179}
180
181func assertGolden(t *testing.T, name string, got []byte) {
182 t.Helper()
183
184 wantPath := filepath.Join("testdata", name+".txt")
185 if *update {
186 if err := os.WriteFile(wantPath, got, 0600); err != nil {
187 t.Fatal(err)
188 }
189 }
190 want, err := os.ReadFile(wantPath)
191 if err != nil {
192 t.Fatal(err)
193 }
194
195 if d := cmp.Diff(string(want), string(got)); d != "" {
196 t.Fatalf("unexpected (-want, +got):\n%s", d)
197 }
198}
199
200type rankingQuery struct {
201 Query string
202 Target string
203}
204
205var tarballCache = "/tmp/zoekt-test-ranking-tarballs-" + os.Getenv("USER")
206var shardCache = "/tmp/zoekt-test-ranking-shards-" + os.Getenv("USER")
207
208func indexURL(indexDir, u string) error {
209 if err := os.MkdirAll(tarballCache, 0700); err != nil {
210 return err
211 }
212
213 opts := archive.Options{
214 Archive: u,
215 Incremental: true,
216 }
217 opts.SetDefaults() // sets metadata like Name and the codeload URL
218 u = opts.Archive
219
220 // if opts.Commit is set but opts.Branch is not, then we just need to give
221 // the commit a name for testing.
222 if opts.Commit != "" && opts.Branch == "" {
223 opts.Branch = "test"
224 }
225
226 // update Archive location to cached location
227 cacheBase := fmt.Sprintf("%s-%s%s.tar.gz", url.QueryEscape(opts.Name), opts.Branch, opts.Commit) // assume .tar.gz
228 path := filepath.Join(tarballCache, cacheBase)
229 opts.Archive = path
230
231 if _, err := os.Stat(path); os.IsNotExist(err) {
232 if err := download(u, path); err != nil {
233 return err
234 }
235 }
236
237 // TODO scip
238 // languageMap := make(ctags.LanguageMap)
239 // for _, lang := range []string{"kotlin", "rust", "ruby", "go", "python", "javascript", "c_sharp", "scala", "typescript", "zig"} {
240 // languageMap[lang] = ctags.ScipCTags
241 // }
242
243 err := archive.Index(opts, build.Options{
244 IndexDir: indexDir,
245 CTagsMustSucceed: true,
246 })
247 if err != nil {
248 return fmt.Errorf("failed to index %s: %w", opts.Archive, err)
249 }
250
251 return nil
252}
253
254func download(url, dst string) error {
255 tmpPath := dst + ".part"
256
257 rc, err := archive.OpenReader(url)
258 if err != nil {
259 return err
260 }
261 defer rc.Close()
262
263 f, err := os.OpenFile(tmpPath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0600)
264 if err != nil {
265 return err
266 }
267 defer f.Close()
268
269 _, err = io.Copy(f, rc)
270 if err != nil {
271 return err
272 }
273
274 err = f.Close()
275 if err != nil {
276 return err
277 }
278
279 return os.Rename(tmpPath, dst)
280}
281
282const (
283 chunkMatchesPerFile = 3
284 fileMatchesPerSearch = 6
285)
286
287func docName(f zoekt.FileMatch) string {
288 return f.Repository + "/" + f.FileName
289}
290
291func marshalMatches(w io.Writer, rq rankingQuery, q query.Q, files []zoekt.FileMatch) {
292 _, _ = fmt.Fprintf(w, "queryString: %s\n", rq.Query)
293 _, _ = fmt.Fprintf(w, "query: %s\n", q)
294 _, _ = fmt.Fprintf(w, "targetRank: %d\n\n", targetRank(rq, files))
295
296 files, hiddenFiles := splitAtIndex(files, fileMatchesPerSearch)
297 for _, f := range files {
298 doc := docName(f)
299 if doc == rq.Target {
300 doc = "**" + doc + "**"
301 }
302 _, _ = fmt.Fprintf(w, "%s%s\n", doc, addTabIfNonEmpty(f.Debug))
303
304 chunks, hidden := splitAtIndex(f.ChunkMatches, chunkMatchesPerFile)
305
306 for _, m := range chunks {
307 _, _ = fmt.Fprintf(w, "%d:%s%s\n", m.ContentStart.LineNumber, string(m.Content), addTabIfNonEmpty(m.DebugScore))
308 }
309
310 if len(hidden) > 0 {
311 _, _ = fmt.Fprintf(w, "hidden %d more line matches\n", len(hidden))
312 }
313 _, _ = fmt.Fprintln(w)
314 }
315
316 if len(hiddenFiles) > 0 {
317 fmt.Fprintf(w, "hidden %d more file matches\n", len(hiddenFiles))
318 }
319}
320
321func targetRank(rq rankingQuery, files []zoekt.FileMatch) int {
322 for i, f := range files {
323 if docName(f) == rq.Target {
324 return i + 1
325 }
326 }
327 return -1
328}
329
330func splitAtIndex[E any](s []E, idx int) ([]E, []E) {
331 if idx < len(s) {
332 return s[:idx], s[idx:]
333 }
334 return s, nil
335}
336
337func addTabIfNonEmpty(s string) string {
338 if s != "" {
339 return "\t" + s
340 }
341 return s
342}
343
344func requireCTags(tb testing.TB) {
345 tb.Helper()
346
347 if os.Getenv("CTAGS_COMMAND") != "" {
348 return
349 }
350 if _, err := exec.LookPath("universal-ctags"); err == nil {
351 return
352 }
353
354 // On CI we require ctags to be available. Otherwise we skip
355 if os.Getenv("CI") != "" {
356 tb.Fatal("universal-ctags is missing")
357 } else {
358 tb.Skip("universal-ctags is missing")
359 }
360}