Make repository indexer tokenize by camel case selectable

guillep2k · Aug 3, 2019 · a8d4e40 · a8d4e40
1 parent 6c0c5c5
commit a8d4e40
Show file tree

Hide file tree

Showing 4 changed files with 28 additions and 12 deletions.
diff --git a/custom/conf/app.ini.sample b/custom/conf/app.ini.sample
@@ -296,6 +296,9 @@ REPO_INDEXER_ENABLED = false
 REPO_INDEXER_PATH = indexers/repos.bleve
 UPDATE_BUFFER_LEN = 20
 MAX_FILE_SIZE = 1048576
+; Break camel case names into separate words for indexing.
+; It's imperative to delete any previous indexes from REPO_INDEXER_PATH after changing this setting.
+REPO_INDEXER_CAMEL_CASE = true
 
 [admin]
 ; Disallow regular (non-admin) users from creating organizations.

diff --git a/docs/content/doc/advanced/config-cheat-sheet.en-us.md b/docs/content/doc/advanced/config-cheat-sheet.en-us.md
@@ -179,6 +179,8 @@ Values containing `#` or `;` must be quoted using `` ` `` or `"""`.
 - `REPO_INDEXER_PATH`: **indexers/repos.bleve**: Index file used for code search.
 - `UPDATE_BUFFER_LEN`: **20**: Buffer length of index request.
 - `MAX_FILE_SIZE`: **1048576**: Maximum size in bytes of files to be indexed.
+- `REPO_INDEXER_CAMEL_CASE`: **true**: When `REPO_INDEXER_CAMEL_CASE` is true, repository indexer will break camel case into words, so thisCameCaseName will be indexed as this, camel, case, name. It's imperative to delete any previous indexes from REPO_INDEXER_PATH after changing this setting.
+
 
 ## Security (`security`)
 

diff --git a/modules/indexer/repo.go b/modules/indexer/repo.go
@@ -107,11 +107,20 @@ func createRepoIndexer(path string, latestVersion int) error {
 	mapping := bleve.NewIndexMapping()
 	if err = addUnicodeNormalizeTokenFilter(mapping); err != nil {
 		return err
-	} else if err = mapping.AddCustomAnalyzer(repoIndexerAnalyzer, map[string]interface{}{
+	}
+
+	var tokenFilters []string
+	if setting.Indexer.RepoUseCamelCaseTokenizer {
+		tokenFilters = []string{unicodeNormalizeName, camelcase.Name, lowercase.Name, unique.Name}
+	} else {
+		tokenFilters = []string{unicodeNormalizeName, lowercase.Name, unique.Name}
+	}
+
+	if err = mapping.AddCustomAnalyzer(repoIndexerAnalyzer, map[string]interface{}{
 		"type":          custom.Name,
 		"char_filters":  []string{},
 		"tokenizer":     unicode.Name,
-		"token_filters": []string{unicodeNormalizeName, camelcase.Name, lowercase.Name, unique.Name},
+		"token_filters": tokenFilters,
 	}); err != nil {
 		return err
 	}

diff --git a/modules/setting/indexer.go b/modules/setting/indexer.go
@@ -19,16 +19,17 @@ const (
 var (
 	// Indexer settings
 	Indexer = struct {
-		IssueType             string
-		IssuePath             string
-		RepoIndexerEnabled    bool
-		RepoPath              string
-		UpdateQueueLength     int
-		MaxIndexerFileSize    int64
-		IssueQueueType        string
-		IssueQueueDir         string
-		IssueQueueConnStr     string
-		IssueQueueBatchNumber int
+		IssueType                 string
+		IssuePath                 string
+		RepoIndexerEnabled        bool
+		RepoPath                  string
+		UpdateQueueLength         int
+		MaxIndexerFileSize        int64
+		IssueQueueType            string
+		IssueQueueDir             string
+		IssueQueueConnStr         string
+		IssueQueueBatchNumber     int
+		RepoUseCamelCaseTokenizer bool
 	}{
 		IssueType:             "bleve",
 		IssuePath:             "indexers/issues.bleve",
@@ -53,6 +54,7 @@ func newIndexerService() {
 	}
 	Indexer.UpdateQueueLength = sec.Key("UPDATE_BUFFER_LEN").MustInt(20)
 	Indexer.MaxIndexerFileSize = sec.Key("MAX_FILE_SIZE").MustInt64(1024 * 1024)
+	Indexer.RepoUseCamelCaseTokenizer = sec.Key("REPO_INDEXER_CAMEL_CASE").MustBool(true)
 	Indexer.IssueQueueType = sec.Key("ISSUE_INDEXER_QUEUE_TYPE").MustString(LevelQueueType)
 	Indexer.IssueQueueDir = sec.Key("ISSUE_INDEXER_QUEUE_DIR").MustString(path.Join(AppDataPath, "indexers/issues.queue"))
 	Indexer.IssueQueueConnStr = sec.Key("ISSUE_INDEXER_QUEUE_CONN_STR").MustString(path.Join(AppDataPath, ""))