renil/um-cli-artifact
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(qmc): use unicode normalize to match filename

Browse Source
This commit is contained in:
Unlock Music Dev
2022-12-05 11:04:57 +08:00
parent f6149c9109
commit 79d00b356f
3 changed files with 28 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

54
algo/qmc/key_mmkv.go
View File

@ -6,12 +6,11 @@ import (
"os"
"path/filepath"
"runtime"
"strings"
"github.com/hbollon/go-edlib"
"github.com/samber/lo"
"go.uber.org/zap"
"golang.org/x/exp/slices"
"golang.org/x/text/unicode/norm"
"unlock-music.dev/mmkv"
)
@ -51,45 +50,34 @@ func readKeyFromMMKV(file string, logger *zap.Logger) ([]byte, error) {
}
_, partName := filepath.Split(file)
buf, err := streamKeyVault.GetBytes(file)
partName = normalizeUnicode(partName)
buf, err := streamKeyVault.GetBytes(file)
if buf == nil {
filePaths := streamKeyVault.Keys()
fileNames := lo.Map(filePaths, func(filePath string, _ int) string {
_, name := filepath.Split(filePath)
return normalizeUnicode(name)
})
for _, key := range filePaths { // fallback 1: match filename only
if !strings.HasSuffix(key, partName) {
logger.Debug("filename",
zap.String("partName", partName),
zap.Int("partName", len(partName)),
zap.String("fileNames", fileNames[0]),
zap.Int("fileNames", len(fileNames[0])),
)
for _, key := range fileNames { // fallback: match filename only
if key != partName {
continue
}
buf, err = streamKeyVault.GetBytes(key)
idx := slices.Index(fileNames, key)
buf, err = streamKeyVault.GetBytes(filePaths[idx])
if err != nil {
logger.Warn("read key from mmkv", zap.String("key", key), zap.Error(err))
logger.Warn("read key from mmkv", zap.String("key", filePaths[idx]), zap.Error(err))
}
}
if buf == nil { // fallback 2: match filename with edit distance
// use editorial judgement to select the best match
// since macOS may change some characters in the file name.
// e.g. "ぜ"(e3 81 9c) -> "ぜ"(e3 81 9b e3 82 99)
fileNames := lo.Map(filePaths, func(filePath string, _ int) string {
_, name := filepath.Split(filePath)
return name
})
minDisStr, err := edlib.FuzzySearch(partName, fileNames, edlib.Levenshtein)
if err != nil {
logger.Warn("fuzzy search failed", zap.Error(err))
}
// TODO: make distance configurable
// for now, assume only 1 character changed to 2 characters
if edlib.LevenshteinDistance(partName, minDisStr) < 3 {
idx := slices.Index(fileNames, minDisStr)
buf, err = streamKeyVault.GetBytes(filePaths[idx])
if err != nil {
logger.Warn("read key from mmkv", zap.String("key", minDisStr), zap.Error(err))
}
}
}
}
if len(buf) == 0 {
@ -135,3 +123,7 @@ func getDefaultMMKVDir() (string, error) {
return mmkvDir, nil
}
func normalizeUnicode(str string) string {
return norm.NFC.String(str)
}