archive: add cody and golang repo to e2e rank corpus

Additionally we add many queries we are tracking.

Test Plan: go test

cmd/zoekt-archive-index/e2e_rank_test.go

@@ -37,9 +37,24 @@ func TestRanking(t *testing.T) {
 
 	archiveURLs := []string{
 		"https://github.com/sourcegraph/sourcegraph/tree/v5.2.2",
+		"https://github.com/golang/go/tree/go1.21.4",
+		"https://github.com/sourcegraph/cody/tree/vscode-v0.14.5",
 	}
 	queries := []string{
+		// golang/go
+		"test server",
+		"bytes buffer",
+		"bufio buffer",
+
+		// sourcegraph/sourcegraph
 		"graphql type User",
+		"Get database/user",
+		"InternalDoer",
+		"Repository metadata Write rbac",
+
+		// cody
+		"generate unit test",
+		"r:cody sourcegraph url",
 	}
 
 	indexDir := t.TempDir()

cmd/zoekt-archive-index/testdata/Get_databaseuser.txt

@@ -0,0 +1,40 @@
+queryString: Get database/user
+query: (and case_substr:"Get" substr:"database/user")
+
+github.com/sourcegraph/sourcegraph/internal/database/user_emails.go
+161:func (s *userEmailsStore) Get(ctx context.Context, userID int32, email string) (emailCanonicalCase string, verified bool, err error) {
+50:	Get(ctx context.Context, userID int32, email string) (emailCanonicalCase string, verified bool, err error)
+91:func (s *userEmailsStore) GetInitialSiteAdminInfo(ctx context.Context) (email string, tosAccepted bool, err error) {
+hidden 14 more line matches
+
+github.com/sourcegraph/sourcegraph/internal/database/user_roles.go
+35:	GetUserRoleOpts    UserRoleOpts
+358:func (r *userRoleStore) GetByUserID(ctx context.Context, opts GetUserRoleOpts) ([]*types.UserRole, error) {
+365:func (r *userRoleStore) GetByRoleID(ctx context.Context, opts GetUserRoleOpts) ([]*types.UserRole, error) {
+hidden 8 more line matches
+
+github.com/sourcegraph/sourcegraph/internal/database/users.go
+940:func (u *userStore) GetByID(ctx context.Context, id int32) (*types.User, error) {
+947:func (u *userStore) GetByVerifiedEmail(ctx context.Context, email string) (*types.User, error) {
+951:func (u *userStore) GetByUsername(ctx context.Context, username string) (*types.User, error) {
+hidden 17 more line matches
+
+github.com/sourcegraph/sourcegraph/internal/database/user_credentials.go
+248:func (s *userCredentialsStore) GetByID(ctx context.Context, id int64) (*UserCredential, error) {
+271:func (s *userCredentialsStore) GetByScope(ctx context.Context, scope UserCredentialScope) (*UserCredential, error) {
+108:	GetByID(ctx context.Context, id int64) (*UserCredential, error)
+hidden 8 more line matches
+
+github.com/sourcegraph/sourcegraph/internal/database/user_emails_test.go
+56:func TestUserEmails_Get(t *testing.T) {
+106:func TestUserEmails_GetPrimary(t *testing.T) {
+585:func TestUserEmails_GetLatestVerificationSentEmail(t *testing.T) {
+hidden 10 more line matches
+
+github.com/sourcegraph/sourcegraph/internal/database/users_test.go
+628:func TestUsers_GetByVerifiedEmail(t *testing.T) {
+664:func TestUsers_GetByUsername(t *testing.T) {
+711:func TestUsers_GetByUsernames(t *testing.T) {
+hidden 32 more line matches
+
+hidden 3 more file matches

cmd/zoekt-archive-index/testdata/InternalDoer.txt

@@ -0,0 +1,27 @@
+queryString: InternalDoer
+query: case_substr:"InternalDoer"
+
+github.com/sourcegraph/sourcegraph/internal/httpcli/client.go
+217:var InternalDoer, _ = InternalClientFactory.Doer()
+215:// InternalDoer is a shared client for internal communication. This is a
+
+github.com/sourcegraph/sourcegraph/internal/api/internalapi/client.go
+144:	resp, err := httpcli.InternalDoer.Do(req.WithContext(ctx))
+
+github.com/sourcegraph/sourcegraph/enterprise/cmd/embeddings/qa/context_data.tsv
+3:In the sourcegraph repository, what does InternalDoer do?	internal/httpcli/client.go
+4:In my codebase, what does InternalDoer do?	internal/httpcli/client.go
+
+github.com/sourcegraph/sourcegraph/cmd/frontend/internal/app/badge.go
+23:	totalRefs, err := backend.CountGoImporters(r.Context(), httpcli.InternalDoer, routevar.ToRepo(mux.Vars(r)))
+
+github.com/sourcegraph/sourcegraph/internal/batches/webhooks/webhooks.go
+67:	Enqueue(ctx, logger, db, eventType, marshalBatchChange, id, httpcli.InternalDoer)
+74:	Enqueue(ctx, logger, db, eventType, marshalChangeset, id, httpcli.InternalDoer)
+
+github.com/sourcegraph/sourcegraph/cmd/frontend/internal/app/resolvers/app.go
+53:		doer:      httpcli.InternalDoer,
+354:	cli := httpcli.InternalDoer
+424:	cli := httpcli.InternalDoer
+
+hidden 9 more file matches

cmd/zoekt-archive-index/testdata/Repository_metadata_Write_rbac.txt

@@ -0,0 +1,33 @@
+queryString: Repository metadata Write rbac
+query: (and case_substr:"Repository" substr:"metadata" case_substr:"Write" substr:"rbac")
+
+github.com/sourcegraph/sourcegraph/cmd/frontend/graphqlbackend/repository_metadata.go
+54:func (r *schemaResolver) AddRepoMetadata(ctx context.Context, args struct {
+95:func (r *schemaResolver) UpdateRepoMetadata(ctx context.Context, args struct {
+134:func (r *schemaResolver) DeleteRepoMetadata(ctx context.Context, args struct {
+hidden 30 more line matches
+
+github.com/sourcegraph/sourcegraph/client/web/src/repo/tree/TreePageContent.tsx
+666:interface RepositoryContributorNodeProps extends QuerySpec {
+10:import { RepoMetadata } from '@sourcegraph/branded'
+16:import { RepositoryType, SearchPatternType, type TreeFields } from '@sourcegraph/shared/src/graphql-operations'
+hidden 46 more line matches
+
+github.com/sourcegraph/sourcegraph/doc/admin/repo/metadata.md
+1:# Custom repository metadata
+18:## Adding metadata
+8:### Repository owners
+hidden 14 more line matches
+
+github.com/sourcegraph/sourcegraph/cmd/frontend/graphqlbackend/repository_metadata_test.go
+26:func TestRepositoryMetadata(t *testing.T) {
+17:	"github.com/sourcegraph/sourcegraph/internal/rbac"
+23:	rtypes "github.com/sourcegraph/sourcegraph/internal/rbac/types"
+hidden 25 more line matches
+
+github.com/sourcegraph/sourcegraph/client/web/src/repo/repoContainerRoutes.tsx
+3:import { canWriteRepoMetadata } from '../util/rbac'
+5:import { RepositoryChangelistPage } from './commit/RepositoryCommitPage'
+9:const RepositoryCommitPage = lazyComponent(() => import('./commit/RepositoryCommitPage'), 'RepositoryCommitPage')
+hidden 19 more line matches
+

cmd/zoekt-archive-index/testdata/bufio_buffer.txt

@@ -0,0 +1,40 @@
+queryString: bufio buffer
+query: (and substr:"bufio" substr:"buffer")
+
+github.com/golang/go/src/bytes/buffer.go
+20:type Buffer struct {
+60:func (b *Buffer) AvailableBuffer() []byte { return b.buf[len(b.buf):] }
+472:func NewBuffer(buf []byte) *Buffer { return &Buffer{buf: buf} }
+hidden 108 more line matches
+
+github.com/golang/go/src/bufio/scan.go
+267:func (s *Scanner) Buffer(buf []byte, max int) {
+5:package bufio
+25:// large to fit in the buffer. When a scan stops, the reader may have
+hidden 21 more line matches
+
+github.com/golang/go/src/bufio/bufio.go
+8:package bufio
+665:func (b *Writer) AvailableBuffer() []byte {
+338:func (b *Reader) Buffered() int { return b.w - b.r }
+hidden 89 more line matches
+
+github.com/golang/go/src/cmd/doc/pkg.go
+59:	bytes.Buffer
+56:type pkgBuffer struct {
+8:	"bufio"
+hidden 8 more line matches
+
+github.com/golang/go/src/net/http/h2_bundle.go
+3716:type http2pipeBuffer interface {
+1086:type http2dataBuffer struct {
+3724:func (p *http2pipe) setBuffer(b http2pipeBuffer) {
+hidden 116 more line matches
+
+github.com/golang/go/src/image/png/writer.go
+36:type EncoderBuffer encoder
+24:	BufferPool EncoderBufferPool
+30:type EncoderBufferPool interface {
+hidden 18 more line matches
+
+hidden 112 more file matches

cmd/zoekt-archive-index/testdata/bytes_buffer.txt

@@ -0,0 +1,40 @@
+queryString: bytes buffer
+query: (and substr:"bytes" substr:"buffer")
+
+github.com/golang/go/src/bytes/buffer.go
+20:type Buffer struct {
+54:func (b *Buffer) Bytes() []byte { return b.buf[b.off:] }
+5:package bytes
+hidden 126 more line matches
+
+github.com/golang/go/src/cmd/internal/edit/edit.go
+14:type Buffer struct {
+68:func (b *Buffer) Bytes() []byte {
+41:func NewBuffer(data []byte) *Buffer {
+hidden 13 more line matches
+
+github.com/golang/go/src/hash/crc32/crc32_ppc64le.s
+122:	SLD     $2,R8           // convert index-> bytes
+59:	MOVWZ	0(R5),R8	// 0-3 bytes of p ?Endian?
+60:	MOVWZ	4(R5),R9	// 4-7 bytes of p
+hidden 35 more line matches
+
+github.com/golang/go/src/fmt/print.go
+101:type buffer []byte
+509:func (p *pp) fmtBytes(v []byte, verb rune, typeString string) {
+17:// Strings for use with buffer.WriteString.
+hidden 28 more line matches
+
+github.com/golang/go/src/bufio/scan.go
+106:func (s *Scanner) Bytes() []byte {
+267:func (s *Scanner) Buffer(buf []byte, max int) {
+289:func ScanBytes(data []byte, atEOF bool) (advance int, token []byte, err error) {
+hidden 26 more line matches
+
+github.com/golang/go/src/os/exec/exec.go
+1134:func (w *prefixSuffixSaver) Bytes() []byte {
+94:	"bytes"
+396:			if i := bytes.Index(stack, []byte("\nos/exec.Command(")); i >= 0 {
+hidden 17 more line matches
+
+hidden 494 more file matches

cmd/zoekt-archive-index/testdata/generate_unit_test.txt

@@ -0,0 +1,40 @@
+queryString: generate unit test
+query: (and substr:"generate" substr:"unit" substr:"test")
+
+github.com/sourcegraph/sourcegraph/cmd/frontend/internal/insights/resolvers/insight_series_resolver.go
+300:func (j *seriesResolverGenerator) Generate(ctx context.Context, series types.InsightViewSeries, baseResolver baseInsightResolver, filters types.InsightViewFilters, options types.SeriesDisplayOptions) ([]graphqlbackend.InsightSeriesResolver, error) {
+275:	Generate(ctx context.Context, series types.InsightViewSeries, baseResolver baseInsightResolver, filters types.InsightViewFilters, options types.SeriesDisplayOptions) ([]graphqlbackend.InsightSeriesResolver, error)
+286:	generateResolver resolverGenerator
+hidden 16 more line matches
+
+github.com/golang/go/src/cmd/vendor/github.com/google/pprof/internal/report/report.go
+87:func Generate(w io.Writer, rpt *Report, obj plugin.ObjTool) error {
+187:func (rpt *Report) selectOutputUnit(g *graph.Graph) {
+75:	SampleUnit        string // Unit for the sample data from the profile.
+hidden 48 more line matches
+
+github.com/sourcegraph/sourcegraph/internal/codeintel/autoindexing/internal/inference/lua/test.lua
+9:  generate = function(_, paths)
+6:  patterns = { pattern.new_path_basename "sg-test" },
+8:  -- Invoked as part of unit tests for the autoindexing service
+hidden 1 more line matches
+
+github.com/golang/go/src/cmd/internal/testdir/testdir_test.go
+273:type test struct {
+74:func Test(t *testing.T) {
+263:type testCommon struct {
+hidden 120 more line matches
+
+github.com/golang/go/src/cmd/vendor/github.com/google/pprof/profile/profile.go
+65:	Unit string // seconds, nanoseconds, bytes, etc
+77:	NumUnit  map[string][]string
+68:	unitX int64
+hidden 44 more line matches
+
+github.com/golang/go/src/cmd/link/internal/loader/loader.go
+79:	unit         *sym.CompilationUnit
+1544:func (l *Loader) SymUnit(i Sym) *sym.CompilationUnit {
+228:	generatedSyms        Bitmap // symbols that generate their content, indexed by ext sym idx
+hidden 50 more line matches
+
+hidden 244 more file matches

cmd/zoekt-archive-index/testdata/r_cody_sourcegraph_url.txt

@@ -0,0 +1,39 @@
+queryString: r:cody sourcegraph url
+query: (and repo:cody substr:"sourcegraph" substr:"url")
+
+github.com/sourcegraph/cody/lib/shared/src/sourcegraph-api/graphql/client.ts
+611:        const url = buildGraphQLUrl({ request: query, baseUrl: this.config.serverEndpoint })
+626:        const url = buildGraphQLUrl({ request: query, baseUrl: this.dotcomUrl.href })
+641:        const url = 'http://localhost:49300/.api/testLogging'
+hidden 51 more line matches
+
+github.com/sourcegraph/cody/vscode/src/completions/client.ts
+85:        const url = getCodeCompletionsEndpoint()
+1:import { FeatureFlag, featureFlagProvider } from '@sourcegraph/cody-shared/src/experimentation/FeatureFlagProvider'
+5:} from '@sourcegraph/cody-shared/src/sourcegraph-api/completions/client'
+hidden 6 more line matches
+
+github.com/sourcegraph/cody/vscode/scripts/download-wasm-modules.ts
+83:    for (const url of urls) {
+93:function getFilePathFromURL(url: string): string {
+20:const urls = [
+hidden 21 more line matches
+
+github.com/sourcegraph/cody/slack/src/services/local-vector-store.ts
+18:    const { content, url } = codyNotice
+9:        owner: 'sourcegraph',
+24:            fileName: url,
+
+github.com/sourcegraph/cody/lib/shared/src/sourcegraph-api/completions/client.ts
+23:export abstract class SourcegraphCompletionsClient {
+21: * Access the chat based LLM APIs via a Sourcegraph server instance.
+36:        return new URL('/.api/completions/stream', this.config.serverEndpoint).href
+hidden 1 more line matches
+
+github.com/sourcegraph/cody/lib/shared/src/sourcegraph-api/completions/browserClient.ts
+8:export class SourcegraphBrowserCompletionsClient extends SourcegraphCompletionsClient {
+5:import { SourcegraphCompletionsClient } from './client'
+20:            headersInstance.set('X-Sourcegraph-Should-Trace', 'true')
+hidden 1 more line matches
+
+hidden 71 more file matches

cmd/zoekt-archive-index/testdata/test_server.txt

@@ -0,0 +1,40 @@
+queryString: test server
+query: (and substr:"test" substr:"server")
+
+github.com/golang/go/src/net/http/httptest/server.go
+26:type Server struct {
+105:func NewServer(handler http.Handler) *Server {
+117:func NewUnstartedServer(handler http.Handler) *Server {
+hidden 62 more line matches
+
+github.com/golang/go/src/net/rpc/server.go
+188:type Server struct {
+656:type ServerCodec interface {
+197:func NewServer() *Server {
+hidden 104 more line matches
+
+github.com/sourcegraph/cody/vscode/test/fixtures/mock-server.ts
+126:    const server = app.listen(SERVER_PORT, () => {
+19:const SERVER_PORT = 49300
+21:export const SERVER_URL = 'http://localhost:49300'
+hidden 24 more line matches
+
+github.com/golang/go/src/net/http/server.go
+2617:type Server struct {
+256:	server *Server
+2925:type serverHandler struct {
+hidden 180 more line matches
+
+github.com/sourcegraph/sourcegraph/cmd/gitserver/server/server.go
+132:type Server struct {
+2:package server
+741:func (s *Server) serverContext() (context.Context, context.CancelFunc) {
+hidden 166 more line matches
+
+github.com/golang/go/src/cmd/go/internal/vcweb/vcstest/vcstest.go
+32:type Server struct {
+42:func NewServer() (srv *Server, err error) {
+7:package vcstest
+hidden 31 more line matches
+
+hidden 494 more file matches