gitforge/modules/setting/indexer.go
Alex619829 e35880e7ac Add code search with zoekt support (#8827)
This PR adds zoekt as a code search engine for forgejo. This Pull Request is a continuation of the discussion #8302.
The meilisearch search engine was not suitable, as it is not designed for searching by code. The zoekt project was proposed instead. Zoekt copes well with code indexing, but its operating principle differs from such search engines as elasticsearch.
While elasticsearch can return a result in a ready-made form (with pagination, ready-made snippets, etc.) and forgejo only needs to show this result in the interface with a little work with the data, zoekt works completely differently.

Zoekt finds matches in the repository index and returns a response. The response contains a line with the search word, its number from the file, and also a context, if specified in the request. This response is not suitable for Forgejo, so you need to assemble it yourself. To assemble the response from Zoekt into a form acceptable for Forgejo, I had to write some code and create a new function `searchZoektResult`, since the existing `searchResult` function is completely unsuitable for this search engine. I also had to write logic for pagination, highlighting, and correct display of lines in found snippets with a match, but this is a feature of Zoekt.
At the moment, Zoekt does not support deleting a repository index by repo_id, it only supports complete deletion of all repositories. But I still implemented the Delete function, which deletes a specific repository by its ID.

Co-authored-by: Aleksandr Gamzin <gamzin@altlinux.org>
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/8827
Reviewed-by: Shiny Nematoda <snematoda@noreply.codeberg.org>
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
2026-05-28 20:52:34 +02:00

122 lines
4.1 KiB
Go

// Copyright 2019 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package setting
import (
"net/url"
"path/filepath"
"strings"
"time"
"forgejo.org/modules/log"
"github.com/gobwas/glob"
)
// Indexer settings
var Indexer = struct {
IssueType string
IssuePath string
IssueConnStr string
IssueConnAuth string
IssueIndexerName string
StartupTimeout time.Duration
RepoIndexerEnabled bool
RepoIndexerRepoTypes []string
RepoIndexerEnableFuzzy bool
RepoType string
RepoPath string
RepoConnStr string
RepoIndexerName string
MaxIndexerFileSize int64
IncludePatterns []Glob
ExcludePatterns []Glob
ExcludeVendored bool
}{
IssueType: "bleve",
IssuePath: "indexers/issues.bleve",
IssueConnStr: "",
IssueConnAuth: "",
IssueIndexerName: "gitea_issues",
RepoIndexerEnabled: false,
RepoIndexerRepoTypes: []string{"sources", "forks", "mirrors", "templates"},
RepoIndexerEnableFuzzy: false,
RepoType: "bleve",
RepoPath: "indexers/repos.bleve",
RepoConnStr: "",
RepoIndexerName: "gitea_codes",
MaxIndexerFileSize: 1024 * 1024,
ExcludeVendored: true,
}
type Glob struct {
glob glob.Glob
pattern string
}
func (g *Glob) Match(s string) bool {
return g.glob.Match(s)
}
func (g *Glob) Pattern() string {
return g.pattern
}
func loadIndexerFrom(rootCfg ConfigProvider) {
sec := rootCfg.Section("indexer")
Indexer.IssueType = sec.Key("ISSUE_INDEXER_TYPE").MustString("bleve")
Indexer.IssuePath = filepath.ToSlash(sec.Key("ISSUE_INDEXER_PATH").MustString(filepath.ToSlash(filepath.Join(AppDataPath, "indexers/issues.bleve"))))
if !filepath.IsAbs(Indexer.IssuePath) {
Indexer.IssuePath = filepath.ToSlash(filepath.Join(AppWorkPath, Indexer.IssuePath))
}
Indexer.IssueConnStr = sec.Key("ISSUE_INDEXER_CONN_STR").MustString(Indexer.IssueConnStr)
if Indexer.IssueType == "meilisearch" {
u, err := url.Parse(Indexer.IssueConnStr)
if err != nil {
log.Warn("Failed to parse ISSUE_INDEXER_CONN_STR: %v", err)
u = &url.URL{}
}
Indexer.IssueConnAuth, _ = u.User.Password()
u.User = nil
Indexer.IssueConnStr = u.String()
}
Indexer.IssueIndexerName = sec.Key("ISSUE_INDEXER_NAME").MustString(Indexer.IssueIndexerName)
Indexer.RepoIndexerEnabled = sec.Key("REPO_INDEXER_ENABLED").MustBool(false)
Indexer.RepoIndexerRepoTypes = strings.Split(sec.Key("REPO_INDEXER_REPO_TYPES").MustString("sources,forks,mirrors,templates"), ",")
Indexer.RepoIndexerEnableFuzzy = sec.Key("REPO_INDEXER_FUZZY_ENABLED").MustBool(false)
Indexer.RepoType = sec.Key("REPO_INDEXER_TYPE").MustString("bleve")
Indexer.RepoPath = filepath.ToSlash(sec.Key("REPO_INDEXER_PATH").MustString(filepath.ToSlash(filepath.Join(AppDataPath, "indexers", "repos."+Indexer.RepoType))))
if !filepath.IsAbs(Indexer.RepoPath) {
Indexer.RepoPath = filepath.ToSlash(filepath.Join(AppWorkPath, Indexer.RepoPath))
}
Indexer.RepoConnStr = sec.Key("REPO_INDEXER_CONN_STR").MustString("")
Indexer.RepoIndexerName = sec.Key("REPO_INDEXER_NAME").MustString("gitea_codes")
Indexer.IncludePatterns = IndexerGlobFromString(sec.Key("REPO_INDEXER_INCLUDE").MustString(""))
Indexer.ExcludePatterns = IndexerGlobFromString(sec.Key("REPO_INDEXER_EXCLUDE").MustString(""))
Indexer.ExcludeVendored = sec.Key("REPO_INDEXER_EXCLUDE_VENDORED").MustBool(true)
Indexer.MaxIndexerFileSize = sec.Key("MAX_FILE_SIZE").MustInt64(1024 * 1024)
Indexer.StartupTimeout = sec.Key("STARTUP_TIMEOUT").MustDuration(30 * time.Second)
}
// IndexerGlobFromString parses a comma separated list of patterns and returns a glob.Glob slice suited for repo indexing
func IndexerGlobFromString(globstr string) []Glob {
extarr := make([]Glob, 0, 10)
for expr := range strings.SplitSeq(strings.ToLower(globstr), ",") {
expr = strings.TrimSpace(expr)
if expr != "" {
if g, err := glob.Compile(expr, '.', '/'); err != nil {
log.Warn("Invalid glob expression '%s' (skipped): %v", expr, err)
} else {
extarr = append(extarr, Glob{glob: g, pattern: expr})
}
}
}
return extarr
}