mirror of
https://github.com/hanzoai/forge.git
synced 2026-10-30 15:00:39 +00:00
zeekayandHanzo Dev
9a07b4d089
2978 files. The fork is 2026-07-15 (rebrand: gitea.dev -> github.com/hanzoai/git), so 2026 is the year, and "Copyright 2026 Hanzo AI, Inc." was already the form used by the handful of files that had ours. Shape is now: // Copyright 2026 Hanzo AI, Inc. All rights reserved. // Copyright 2015 The Gogs Authors. All rights reserved. // Copyright 2019 The Gitea Authors. All rights reserved. // SPDX-License-Identifier: MIT Ours on top, theirs retained verbatim. Retained because MIT requires it, and separately because they are true — those authors did write that code. Not renamed to "The Git Authors": Git is a different project under a different license, and putting its name here would attribute this code to people who have nothing to do with it. Adding our line is the fix; overwriting theirs would have been both a licence violation and a false statement. gofmt still reports import-order drift in some of these files. It predates this change (verified against a stash) and is left alone rather than burying a one-line-per-file legal change under 3000 files of unrelated reordering. Build clean. Co-authored-by: Hanzo Dev <[email protected]>
92 lines
2.7 KiB
Go
92 lines
2.7 KiB
Go
// Copyright 2026 Hanzo AI, Inc. All rights reserved.
|
|
// Copyright 2023 The Gitea Authors. All rights reserved.
|
|
// SPDX-License-Identifier: MIT
|
|
|
|
package bleve
|
|
|
|
import (
|
|
"errors"
|
|
"os"
|
|
"unicode"
|
|
|
|
"github.com/hanzoai/git/modules/log"
|
|
"github.com/hanzoai/git/modules/setting"
|
|
"github.com/hanzoai/git/modules/util"
|
|
|
|
"github.com/blevesearch/bleve/v2"
|
|
unicode_tokenizer "github.com/blevesearch/bleve/v2/analysis/tokenizer/unicode"
|
|
"github.com/blevesearch/bleve/v2/index/upsidedown"
|
|
"github.com/ethantkoenig/rupture"
|
|
)
|
|
|
|
const (
|
|
maxFuzziness = 2
|
|
)
|
|
|
|
// openIndexer open the index at the specified path, checking for metadata
|
|
// updates and bleve version updates. If index needs to be created (or
|
|
// re-created), returns (nil, nil)
|
|
func openIndexer(path string, latestVersion int) (bleve.Index, int, error) {
|
|
_, err := os.Stat(path)
|
|
if err != nil && os.IsNotExist(err) {
|
|
return nil, 0, nil
|
|
} else if err != nil {
|
|
return nil, 0, err
|
|
}
|
|
|
|
metadata, err := rupture.ReadIndexMetadata(path)
|
|
if err != nil {
|
|
return nil, 0, err
|
|
}
|
|
if metadata.Version < latestVersion {
|
|
// the indexer is using a previous version, so we should delete it and
|
|
// re-populate
|
|
return nil, metadata.Version, util.RemoveAll(path)
|
|
}
|
|
|
|
index, err := bleve.Open(path)
|
|
if err != nil {
|
|
if errors.Is(err, upsidedown.IncompatibleVersion) {
|
|
log.Warn("Indexer was built with a previous version of bleve, deleting and rebuilding")
|
|
return nil, 0, util.RemoveAll(path)
|
|
}
|
|
return nil, 0, err
|
|
}
|
|
|
|
return index, 0, nil
|
|
}
|
|
|
|
// GuessFuzzinessByKeyword guesses fuzziness based on the levenshtein distance and determines how many chars
|
|
// may be different on two string, and they still be considered equivalent.
|
|
// Given a phrase, its shortest word determines its fuzziness. If a phrase uses CJK (eg: `갃갃갃` `啊啊啊`), the fuzziness is zero.
|
|
func GuessFuzzinessByKeyword(s string) int {
|
|
tokenizer := unicode_tokenizer.NewUnicodeTokenizer()
|
|
tokens := tokenizer.Tokenize([]byte(s))
|
|
|
|
if len(tokens) > 0 {
|
|
fuzziness := maxFuzziness
|
|
|
|
for _, token := range tokens {
|
|
fuzziness = min(fuzziness, guessFuzzinessByKeyword(string(token.Term)))
|
|
}
|
|
|
|
return fuzziness
|
|
}
|
|
|
|
return 0
|
|
}
|
|
|
|
func guessFuzzinessByKeyword(s string) int {
|
|
// according to https://github.com/blevesearch/bleve/issues/1563, the supported max fuzziness is 2
|
|
// magic number 4 was chosen to determine the levenshtein distance per each character of a keyword
|
|
// BUT, when using CJK (eg: `갃갃갃` `啊啊啊`), it mismatches a lot.
|
|
// Likewise, queries whose terms contains characters that are *not* letters should not use fuzziness
|
|
|
|
for _, r := range s {
|
|
if r >= 128 || !unicode.IsLetter(r) {
|
|
return 0
|
|
}
|
|
}
|
|
return min(min(setting.Indexer.TypeBleveMaxFuzzniess, maxFuzziness), len(s)/4)
|
|
}
|