mirror of
https://github.com/Psychotoxical/psysonic.git
synced 2026-07-22 15:25:46 +00:00
refactor(playlist): G.62 — extract CSV match helpers + import orchestrator (cluster) (#629)
Two-cut cluster on PlaylistDetail.tsx. 1761 → 1400 LOC (−361). The page keeps the same handleImportCsv arrow function (now a thin guard + delegate), but the matching algorithm and the CSV import pipeline live in their own modules. spotifyCsvMatch — six pure helpers: normalizeForMatching, cleanTrackTitle (the 100-LOC suffix-regex array), levenshtein, similarityScore, calculateDynamicThreshold, processBatch. All exports. No state, no side effects, no React deps. runPlaylistCsvImport — full Spotify CSV import pipeline: openDialog + readTextFile, parseSpotifyCsv parse step, batched search with 2-attempt retry, dynamic-threshold scored matching with ISRC fast path, dedupe against existing + already-queued, savePlaylist commit, auto-show report modal on issues, toast variant by outcome (success / warning / error). Takes a deps object with songs / t / savePlaylist + setSongs / setCsvImporting / setCsvImportReport setters. PlaylistDetail keeps the `if (!id || csvImporting) return;` guard on the caller side; id is no longer needed inside the runner (savePlaylist captures it). PlaylistDetail drops three direct imports (search, openDialog, readTextFile, parseSpotifyCsv) and picks up runPlaylistCsvImport. search comes back as a top-level import because the search-add overlay useEffect on the same page still hits the Subsonic search endpoint directly. SpotifyCsvTrack stays as a type-only import for the csvImportReport state shape. Pure code move otherwise — no behaviour change.
This commit is contained in:
Frank Stellmacher
committed by
GitHub
GitHub
parent
84ceb5f423
commit
1b1122f086
@@ -0,0 +1,177 @@
|
||||
// Normalize strings for matching: remove accents, special chars, lowercase, trim
|
||||
export function normalizeForMatching(s: string): string {
|
||||
return s.toLowerCase()
|
||||
.normalize('NFD')
|
||||
.replace(/[̀-ͯ]/g, '')
|
||||
.replace(/[ø]/gi, 'o')
|
||||
.replace(/[æ]/gi, 'ae')
|
||||
.trim();
|
||||
}
|
||||
|
||||
// Clean common title suffixes (remastered, live, editions, etc.)
|
||||
export function cleanTrackTitle(title: string): string {
|
||||
const suffixes = [
|
||||
// Remastered variants
|
||||
/\s*-\s*remasterizado$/i,
|
||||
/\s*-\s*remaster$/i,
|
||||
/\s*-\s*remastered$/i,
|
||||
/\s*\(remasterizado\)$/i,
|
||||
/\s*\(remaster\)$/i,
|
||||
/\s*\(remastered\)$/i,
|
||||
/\s*\[remasterizado\]$/i,
|
||||
/\s*\[remaster\]$/i,
|
||||
/\s*\[remastered\]$/i,
|
||||
/\s*-\s*remasterizado\s+\d{4}$/i,
|
||||
/\s*-\s*remastered\s+\d{4}$/i,
|
||||
/\s*\(\d{4}\s+remaster\)$/i,
|
||||
/\s*\(\d{4}\s+remastered\)$/i,
|
||||
// Live variants
|
||||
/\s*-\s*en vivo$/i,
|
||||
/\s*-\s*live$/i,
|
||||
/\s*-\s*version en vivo$/i,
|
||||
/\s*-\s*studio version$/i,
|
||||
/\s*-\s*version de estudio$/i,
|
||||
/\s*\(en vivo\)$/i,
|
||||
/\s*\(live\)$/i,
|
||||
/\s*\(live .*\)$/i,
|
||||
/\s*\[en vivo\]$/i,
|
||||
/\s*\[live\]$/i,
|
||||
/\s*\[live .*\]$/i,
|
||||
/\s*-\s*live at.*$/i,
|
||||
/\s*\(live at.*\)$/i,
|
||||
// Version/Edition variants
|
||||
/\s*-\s*version$/i,
|
||||
/\s*-\s*versión$/i,
|
||||
/\s*\(version\)$/i,
|
||||
/\s*\(versión\)$/i,
|
||||
/\s*\[version\]$/i,
|
||||
/\s*\[versión\]$/i,
|
||||
/\s*-\s*album version$/i,
|
||||
/\s*\(album version\)$/i,
|
||||
/\s*\[album version\]$/i,
|
||||
// Radio/Edit variants
|
||||
/\s*-\s*radio edit$/i,
|
||||
/\s*-\s*radio version$/i,
|
||||
/\s*\(radio edit\)$/i,
|
||||
/\s*\(radio version\)$/i,
|
||||
/\s*\[radio edit\]$/i,
|
||||
/\s*\[radio version\]$/i,
|
||||
/\s*-\s*edit$/i,
|
||||
/\s*\(edit\)$/i,
|
||||
/\s*\[edit\]$/i,
|
||||
// Acoustic/Instrumental variants
|
||||
/\s*-\s*acoustic$/i,
|
||||
/\s*-\s*acústico$/i,
|
||||
/\s*\(acoustic\)$/i,
|
||||
/\s*\(acústico\)$/i,
|
||||
/\s*\[acoustic\]$/i,
|
||||
/\s*\[acústico\]$/i,
|
||||
/\s*-\s*instrumental$/i,
|
||||
/\s*\(instrumental\)$/i,
|
||||
/\s*\[instrumental\]$/i,
|
||||
// Featuring/Feat/Ft/With variants
|
||||
/\s*\(feat\.?\s+.*\)$/i,
|
||||
/\s*\[feat\.?\s+.*\]$/i,
|
||||
/\s*-\s*feat\.?\s+.*$/i,
|
||||
/\s*\(featuring\s+.*\)$/i,
|
||||
/\s*\[featuring\s+.*\]$/i,
|
||||
/\s*\(ft\.?\s+.*\)$/i,
|
||||
/\s*\[ft\.?\s+.*\]$/i,
|
||||
/\s*-\s*ft\.?\s+.*$/i,
|
||||
/\s*\(with\s+.*\)$/i,
|
||||
/\s*\[with\s+.*\]$/i,
|
||||
/\s*-\s*with\s+.*$/i,
|
||||
/\s*ft\.?\s+.*$/i,
|
||||
// Explicit/Clean tags
|
||||
/\s*\(explicit\)$/i,
|
||||
/\s*\[explicit\]$/i,
|
||||
/\s*\(clean\)$/i,
|
||||
/\s*\[clean\]$/i,
|
||||
// Mono/Stereo
|
||||
/\s*\(mono\)$/i,
|
||||
/\s*\[mono\]$/i,
|
||||
/\s*\(stereo\)$/i,
|
||||
/\s*\[stereo\]$/i,
|
||||
// Deluxe/Special editions
|
||||
/\s*-\s*deluxe$/i,
|
||||
/\s*\(deluxe\)$/i,
|
||||
/\s*\[deluxe\]$/i,
|
||||
/\s*-\s*special edition$/i,
|
||||
/\s*\(special edition\)$/i,
|
||||
/\s*\[special edition\]$/i,
|
||||
// Year in parentheses (common in remasters)
|
||||
/\s*\(\d{4}\)$/i,
|
||||
];
|
||||
let cleaned = title.trim();
|
||||
// Apply patterns multiple times for nested cases
|
||||
for (let i = 0; i < 3; i++) {
|
||||
const previous = cleaned;
|
||||
for (const regex of suffixes) {
|
||||
cleaned = cleaned.replace(regex, '');
|
||||
}
|
||||
cleaned = cleaned.trim();
|
||||
if (previous === cleaned) break; // No more changes
|
||||
}
|
||||
return cleaned;
|
||||
}
|
||||
|
||||
// Levenshtein distance for similarity scoring
|
||||
export function levenshtein(a: string, b: string): number {
|
||||
const matrix: number[][] = [];
|
||||
for (let i = 0; i <= b.length; i++) matrix[i] = [i];
|
||||
for (let j = 0; j <= a.length; j++) matrix[0][j] = j;
|
||||
for (let i = 1; i <= b.length; i++) {
|
||||
for (let j = 1; j <= a.length; j++) {
|
||||
matrix[i][j] = b[i - 1] === a[j - 1]
|
||||
? matrix[i - 1][j - 1]
|
||||
: Math.min(matrix[i - 1][j - 1] + 1, matrix[i][j - 1] + 1, matrix[i - 1][j] + 1);
|
||||
}
|
||||
}
|
||||
return matrix[b.length][a.length];
|
||||
}
|
||||
|
||||
export function similarityScore(a: string, b: string): number {
|
||||
const maxLen = Math.max(a.length, b.length);
|
||||
if (maxLen === 0) return 1;
|
||||
// Use normalized strings (without accents) for comparison
|
||||
const dist = levenshtein(normalizeForMatching(a), normalizeForMatching(b));
|
||||
return 1 - dist / maxLen;
|
||||
}
|
||||
|
||||
// Calculate dynamic threshold based on match quality signals
|
||||
export function calculateDynamicThreshold(
|
||||
bestMatch: { score: number; artistScore: number },
|
||||
secondMatch: { score: number } | undefined,
|
||||
titleWords: number
|
||||
): number {
|
||||
const baseThreshold = 0.6; // Minimum acceptable score
|
||||
|
||||
// Bonus if there's a large gap between best and second match (clear winner)
|
||||
const gap = secondMatch ? bestMatch.score - secondMatch.score : 0.3;
|
||||
const gapBonus = gap > 0.15 ? 0.1 : gap > 0.08 ? 0.05 : 0;
|
||||
|
||||
// Short titles (< 3 words) are more ambiguous, need higher threshold
|
||||
// Long titles (> 4 words) are more specific, can use lower threshold
|
||||
const lengthBonus = titleWords > 4 ? 0.05 : titleWords < 3 ? -0.05 : 0;
|
||||
|
||||
// Strong artist match gives confidence to accept lower overall score
|
||||
const artistBonus = bestMatch.artistScore > 0.85 ? 0.08 : bestMatch.artistScore > 0.7 ? 0.04 : 0;
|
||||
|
||||
// Calculate final threshold, clamp between 0.55 and 0.75
|
||||
return Math.max(0.55, Math.min(0.75, baseThreshold - gapBonus - lengthBonus - artistBonus));
|
||||
}
|
||||
|
||||
// Process searches in batches to avoid overloading the server
|
||||
export async function processBatch<T, R>(
|
||||
items: T[],
|
||||
batchSize: number,
|
||||
processor: (item: T) => Promise<R | null>
|
||||
): Promise<(R | null)[]> {
|
||||
const results: (R | null)[] = [];
|
||||
for (let i = 0; i < items.length; i += batchSize) {
|
||||
const batch = items.slice(i, i + batchSize);
|
||||
const batchResults = await Promise.all(batch.map(processor));
|
||||
results.push(...batchResults);
|
||||
}
|
||||
return results;
|
||||
}
|
||||
Reference in New Issue
Block a user