Refactor code_indexer to use an SearchOptions struct for PerformSearch (#29724)

similar to how it's already done for the issue_indexer


---
*Sponsored by Kithara Software GmbH*
This commit is contained in:
6543 2024-03-16 11:32:45 +01:00 committed by GitHub
parent e0ea3811c4
commit 1262ff6734
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
9 changed files with 86 additions and 39 deletions

View File

@ -142,7 +142,7 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
return err
}
if size, err = strconv.ParseInt(strings.TrimSpace(stdout), 10, 64); err != nil {
return fmt.Errorf("Misformatted git cat-file output: %w", err)
return fmt.Errorf("misformatted git cat-file output: %w", err)
}
}
@ -233,26 +233,26 @@ func (b *Indexer) Delete(_ context.Context, repoID int64) error {
// Search searches for files in the specified repo.
// Returns the matching file-paths
func (b *Indexer) Search(ctx context.Context, repoIDs []int64, language, keyword string, page, pageSize int, isFuzzy bool) (int64, []*internal.SearchResult, []*internal.SearchResultLanguages, error) {
func (b *Indexer) Search(ctx context.Context, opts *internal.SearchOptions) (int64, []*internal.SearchResult, []*internal.SearchResultLanguages, error) {
var (
indexerQuery query.Query
keywordQuery query.Query
)
if isFuzzy {
phraseQuery := bleve.NewMatchPhraseQuery(keyword)
if opts.IsKeywordFuzzy {
phraseQuery := bleve.NewMatchPhraseQuery(opts.Keyword)
phraseQuery.FieldVal = "Content"
phraseQuery.Analyzer = repoIndexerAnalyzer
keywordQuery = phraseQuery
} else {
prefixQuery := bleve.NewPrefixQuery(keyword)
prefixQuery := bleve.NewPrefixQuery(opts.Keyword)
prefixQuery.FieldVal = "Content"
keywordQuery = prefixQuery
}
if len(repoIDs) > 0 {
repoQueries := make([]query.Query, 0, len(repoIDs))
for _, repoID := range repoIDs {
if len(opts.RepoIDs) > 0 {
repoQueries := make([]query.Query, 0, len(opts.RepoIDs))
for _, repoID := range opts.RepoIDs {
repoQueries = append(repoQueries, inner_bleve.NumericEqualityQuery(repoID, "RepoID"))
}
@ -266,8 +266,8 @@ func (b *Indexer) Search(ctx context.Context, repoIDs []int64, language, keyword
// Save for reuse without language filter
facetQuery := indexerQuery
if len(language) > 0 {
languageQuery := bleve.NewMatchQuery(language)
if len(opts.Language) > 0 {
languageQuery := bleve.NewMatchQuery(opts.Language)
languageQuery.FieldVal = "Language"
languageQuery.Analyzer = analyzer_keyword.Name
@ -277,12 +277,12 @@ func (b *Indexer) Search(ctx context.Context, repoIDs []int64, language, keyword
)
}
from := (page - 1) * pageSize
from, pageSize := opts.GetSkipTake()
searchRequest := bleve.NewSearchRequestOptions(indexerQuery, pageSize, from, false)
searchRequest.Fields = []string{"Content", "RepoID", "Language", "CommitID", "UpdatedAt"}
searchRequest.IncludeLocations = true
if len(language) == 0 {
if len(opts.Language) == 0 {
searchRequest.AddFacet("languages", bleve.NewFacetRequest("Language", 10))
}
@ -326,7 +326,7 @@ func (b *Indexer) Search(ctx context.Context, repoIDs []int64, language, keyword
}
searchResultLanguages := make([]*internal.SearchResultLanguages, 0, 10)
if len(language) > 0 {
if len(opts.Language) > 0 {
// Use separate query to go get all language counts
facetRequest := bleve.NewSearchRequestOptions(facetQuery, 1, 0, false)
facetRequest.Fields = []string{"Content", "RepoID", "Language", "CommitID", "UpdatedAt"}

View File

@ -281,18 +281,18 @@ func extractAggs(searchResult *elastic.SearchResult) []*internal.SearchResultLan
}
// Search searches for codes and language stats by given conditions.
func (b *Indexer) Search(ctx context.Context, repoIDs []int64, language, keyword string, page, pageSize int, isFuzzy bool) (int64, []*internal.SearchResult, []*internal.SearchResultLanguages, error) {
func (b *Indexer) Search(ctx context.Context, opts *internal.SearchOptions) (int64, []*internal.SearchResult, []*internal.SearchResultLanguages, error) {
searchType := esMultiMatchTypePhrasePrefix
if isFuzzy {
if opts.IsKeywordFuzzy {
searchType = esMultiMatchTypeBestFields
}
kwQuery := elastic.NewMultiMatchQuery(keyword, "content").Type(searchType)
kwQuery := elastic.NewMultiMatchQuery(opts.Keyword, "content").Type(searchType)
query := elastic.NewBoolQuery()
query = query.Must(kwQuery)
if len(repoIDs) > 0 {
repoStrs := make([]any, 0, len(repoIDs))
for _, repoID := range repoIDs {
if len(opts.RepoIDs) > 0 {
repoStrs := make([]any, 0, len(opts.RepoIDs))
for _, repoID := range opts.RepoIDs {
repoStrs = append(repoStrs, repoID)
}
repoQuery := elastic.NewTermsQuery("repo_id", repoStrs...)
@ -300,16 +300,12 @@ func (b *Indexer) Search(ctx context.Context, repoIDs []int64, language, keyword
}
var (
start int
kw = "<em>" + keyword + "</em>"
start, pageSize = opts.GetSkipTake()
kw = "<em>" + opts.Keyword + "</em>"
aggregation = elastic.NewTermsAggregation().Field("language").Size(10).OrderByCountDesc()
)
if page > 0 {
start = (page - 1) * pageSize
}
if len(language) == 0 {
if len(opts.Language) == 0 {
searchResult, err := b.inner.Client.Search().
Index(b.inner.VersionedIndexName()).
Aggregation("language", aggregation).
@ -330,7 +326,7 @@ func (b *Indexer) Search(ctx context.Context, repoIDs []int64, language, keyword
return convertResult(searchResult, kw, pageSize)
}
langQuery := elastic.NewMatchQuery("language", language)
langQuery := elastic.NewMatchQuery("language", opts.Language)
countResult, err := b.inner.Client.Search().
Index(b.inner.VersionedIndexName()).
Aggregation("language", aggregation).

View File

@ -32,7 +32,7 @@ func getRepoChanges(ctx context.Context, repo *repo_model.Repository, revision s
needGenesis := len(status.CommitSha) == 0
if !needGenesis {
hasAncestorCmd := git.NewCommand(ctx, "merge-base").AddDynamicArguments(repo.CodeIndexerStatus.CommitSha, revision)
hasAncestorCmd := git.NewCommand(ctx, "merge-base").AddDynamicArguments(status.CommitSha, revision)
stdout, _, _ := hasAncestorCmd.RunStdString(&git.RunOpts{Dir: repo.RepoPath()})
needGenesis = len(stdout) == 0
}

View File

@ -8,6 +8,7 @@ import (
"os"
"testing"
"code.gitea.io/gitea/models/db"
"code.gitea.io/gitea/models/unittest"
"code.gitea.io/gitea/modules/git"
"code.gitea.io/gitea/modules/indexer/code/bleve"
@ -70,7 +71,15 @@ func testIndexer(name string, t *testing.T, indexer internal.Indexer) {
for _, kw := range keywords {
t.Run(kw.Keyword, func(t *testing.T) {
total, res, langs, err := indexer.Search(context.TODO(), kw.RepoIDs, "", kw.Keyword, 1, 10, true)
total, res, langs, err := indexer.Search(context.TODO(), &internal.SearchOptions{
RepoIDs: kw.RepoIDs,
Keyword: kw.Keyword,
Paginator: &db.ListOptions{
Page: 1,
PageSize: 10,
},
IsKeywordFuzzy: true,
})
assert.NoError(t, err)
assert.Len(t, kw.IDs, int(total))
assert.Len(t, langs, kw.Langs)

View File

@ -7,6 +7,7 @@ import (
"context"
"fmt"
"code.gitea.io/gitea/models/db"
repo_model "code.gitea.io/gitea/models/repo"
"code.gitea.io/gitea/modules/indexer/internal"
)
@ -16,7 +17,17 @@ type Indexer interface {
internal.Indexer
Index(ctx context.Context, repo *repo_model.Repository, sha string, changes *RepoChanges) error
Delete(ctx context.Context, repoID int64) error
Search(ctx context.Context, repoIDs []int64, language, keyword string, page, pageSize int, isFuzzy bool) (int64, []*SearchResult, []*SearchResultLanguages, error)
Search(ctx context.Context, opts *SearchOptions) (int64, []*SearchResult, []*SearchResultLanguages, error)
}
type SearchOptions struct {
RepoIDs []int64
Keyword string
Language string
IsKeywordFuzzy bool
db.Paginator
}
// NewDummyIndexer returns a dummy indexer
@ -38,6 +49,6 @@ func (d *dummyIndexer) Delete(ctx context.Context, repoID int64) error {
return fmt.Errorf("indexer is not ready")
}
func (d *dummyIndexer) Search(ctx context.Context, repoIDs []int64, language, keyword string, page, pageSize int, isFuzzy bool) (int64, []*SearchResult, []*SearchResultLanguages, error) {
func (d *dummyIndexer) Search(ctx context.Context, opts *SearchOptions) (int64, []*SearchResult, []*SearchResultLanguages, error) {
return 0, nil, nil, fmt.Errorf("indexer is not ready")
}

View File

@ -32,6 +32,8 @@ type ResultLine struct {
type SearchResultLanguages = internal.SearchResultLanguages
type SearchOptions = internal.SearchOptions
func indices(content string, selectionStartIndex, selectionEndIndex int) (int, int) {
startIndex := selectionStartIndex
numLinesBefore := 0
@ -125,12 +127,12 @@ func searchResult(result *internal.SearchResult, startIndex, endIndex int) (*Res
// PerformSearch perform a search on a repository
// if isFuzzy is true set the Damerau-Levenshtein distance from 0 to 2
func PerformSearch(ctx context.Context, repoIDs []int64, language, keyword string, page, pageSize int, isFuzzy bool) (int, []*Result, []*internal.SearchResultLanguages, error) {
if len(keyword) == 0 {
func PerformSearch(ctx context.Context, opts *SearchOptions) (int, []*Result, []*SearchResultLanguages, error) {
if opts == nil || len(opts.Keyword) == 0 {
return 0, nil, nil, nil
}
total, results, resultLanguages, err := (*globalIndexer.Load()).Search(ctx, repoIDs, language, keyword, page, pageSize, isFuzzy)
total, results, resultLanguages, err := (*globalIndexer.Load()).Search(ctx, opts)
if err != nil {
return 0, nil, nil, err
}

View File

@ -6,6 +6,7 @@ package explore
import (
"net/http"
"code.gitea.io/gitea/models/db"
repo_model "code.gitea.io/gitea/models/repo"
"code.gitea.io/gitea/modules/base"
code_indexer "code.gitea.io/gitea/modules/indexer/code"
@ -76,7 +77,16 @@ func Code(ctx *context.Context) {
)
if (len(repoIDs) > 0) || isAdmin {
total, searchResults, searchResultLanguages, err = code_indexer.PerformSearch(ctx, repoIDs, language, keyword, page, setting.UI.RepoSearchPagingNum, isFuzzy)
total, searchResults, searchResultLanguages, err = code_indexer.PerformSearch(ctx, &code_indexer.SearchOptions{
RepoIDs: repoIDs,
Keyword: keyword,
IsKeywordFuzzy: isFuzzy,
Language: language,
Paginator: &db.ListOptions{
Page: page,
PageSize: setting.UI.RepoSearchPagingNum,
},
})
if err != nil {
if code_indexer.IsAvailable(ctx) {
ctx.ServerError("SearchResults", err)

View File

@ -6,6 +6,7 @@ package repo
import (
"net/http"
"code.gitea.io/gitea/models/db"
"code.gitea.io/gitea/modules/base"
code_indexer "code.gitea.io/gitea/modules/indexer/code"
"code.gitea.io/gitea/modules/setting"
@ -41,8 +42,16 @@ func Search(ctx *context.Context) {
page = 1
}
total, searchResults, searchResultLanguages, err := code_indexer.PerformSearch(ctx, []int64{ctx.Repo.Repository.ID},
language, keyword, page, setting.UI.RepoSearchPagingNum, isFuzzy)
total, searchResults, searchResultLanguages, err := code_indexer.PerformSearch(ctx, &code_indexer.SearchOptions{
RepoIDs: []int64{ctx.Repo.Repository.ID},
Keyword: keyword,
IsKeywordFuzzy: isFuzzy,
Language: language,
Paginator: &db.ListOptions{
Page: page,
PageSize: setting.UI.RepoSearchPagingNum,
},
})
if err != nil {
if code_indexer.IsAvailable(ctx) {
ctx.ServerError("SearchResults", err)

View File

@ -6,6 +6,7 @@ package user
import (
"net/http"
"code.gitea.io/gitea/models/db"
repo_model "code.gitea.io/gitea/models/repo"
"code.gitea.io/gitea/modules/base"
code_indexer "code.gitea.io/gitea/modules/indexer/code"
@ -74,7 +75,16 @@ func CodeSearch(ctx *context.Context) {
)
if len(repoIDs) > 0 {
total, searchResults, searchResultLanguages, err = code_indexer.PerformSearch(ctx, repoIDs, language, keyword, page, setting.UI.RepoSearchPagingNum, isFuzzy)
total, searchResults, searchResultLanguages, err = code_indexer.PerformSearch(ctx, &code_indexer.SearchOptions{
RepoIDs: repoIDs,
Keyword: keyword,
IsKeywordFuzzy: isFuzzy,
Language: language,
Paginator: &db.ListOptions{
Page: page,
PageSize: setting.UI.RepoSearchPagingNum,
},
})
if err != nil {
if code_indexer.IsAvailable(ctx) {
ctx.ServerError("SearchResults", err)