Files
Nodecast/server/services/liveProjectionService.js
blomios 2a1c745a26 feat(live): add language-aware flexible channel search (#23)
Add backend language detection, flexible live channel search, iptv-org catalog caching, and protected built-in EPG source registration.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-10-06 11:42:06 +02:00

470 lines
20 KiB
JavaScript

const crypto = require('crypto');
const { pairKey, scoreCandidate, identity } = require('./liveGroupingScoring');
const { detectLanguage } = require('./channelLanguage');
const { matchesSearch } = require('./flexibleChannelSearch');
const iptvOrgCatalog = require('./iptvOrgCatalogService');
const QUALITY_TOKENS = new Set([
'4k', 'uhd', 'fhd', 'fullhd', 'hd', 'sd', 'hevc', 'h265', 'h264',
'1080p', '1080i', '720p', '576p', '480p'
]);
const REGION_PREFIX = /^(?:(?:\|\s*[a-z]{2,3}\s*\|)|(?:[a-z]{2,3}(?:-[a-z]{2,3})?\s*\|))\s*/i;
const BEIN_SUFFIX_TOKENS = new Set(['fr', 'en', 'thai', 'tod', 'appel', 'app']);
const BEIN_TIMESHIFT_TOKENS = new Set(['2h', '4h', '8h', '12h']);
function words(value) {
return String(value || '')
.normalize('NFKD')
.replace(/[\u0300-\u036f]/g, '')
.toLowerCase()
.replace(REGION_PREFIX, '')
.replace(/[|()[\]{}_.,:;+/\\-]+/g, ' ')
.split(/\s+/)
.filter(Boolean);
}
function normalizeChannelName(name) {
const tokens = words(name).filter(token => !QUALITY_TOKENS.has(token));
const brandIndex = tokens.findIndex((token, index) => token === 'bein' && tokens[index + 1] === 'sports');
if (brandIndex >= 0) {
const branded = tokens.slice(brandIndex);
const numberedBaseLength = branded[2] === 'max' && /^\d+$/.test(branded[3])
? 4
: (/^\d+$/.test(branded[2]) ? 3 : 0);
if (numberedBaseLength) {
const suffix = branded.slice(numberedBaseLength);
const removable = suffix.every(token =>
BEIN_SUFFIX_TOKENS.has(token) || BEIN_TIMESHIFT_TOKENS.has(token)
);
if (removable) return branded.slice(0, numberedBaseLength).join(' ');
}
return branded.join(' ').trim();
}
return tokens.join(' ').trim();
}
function displayNameForFingerprint(fingerprint) {
return fingerprint.split(' ').map(token => {
if (token === 'bein') return 'beIN';
if (/^\d+$/.test(token)) return token;
return token.charAt(0).toUpperCase() + token.slice(1);
}).join(' ');
}
function qualityForName(name) {
const tokens = words(name);
if (tokens.includes('4k') || tokens.includes('uhd')) return { label: '4K', rank: 4 };
if (tokens.includes('fhd') || tokens.includes('fullhd') || tokens.includes('1080p') || tokens.includes('1080i')) {
return { label: 'Full HD', rank: 3 };
}
if (tokens.includes('hd') || tokens.includes('720p')) return { label: 'HD', rank: 2 };
if (tokens.includes('sd') || tokens.includes('576p') || tokens.includes('480p')) return { label: 'SD', rank: 0 };
return { label: 'Standard', rank: 1 };
}
function stableHash(value) {
return crypto.createHash('sha256').update(value).digest('hex');
}
function projectCatalog(items, options = {}) {
const candidateBuckets = new Map();
for (const item of items) {
const fallback = `channel-${item.source_id}:${item.item_id}`;
const fingerprint = normalizeChannelName(item.name) || fallback;
const quality = qualityForName(item.name);
const projected = {
...item,
fingerprint,
groupingInputHash: stableHash(`${item.name || ''}\0${item.category_id || ''}`),
qualityLabel: quality.label,
qualityRank: quality.rank
};
const language = detectLanguage({ ...projected, data: parseData(projected.data) });
projected.languageCode = language.code;
projected.languageConfidence = language.confidence;
projected.languageScores = language.scores;
projected.languageEvidence = language.evidence;
const itemIdentity = identity(fingerprint);
const candidateKey = itemIdentity.family && itemIdentity.number
? `${itemIdentity.family}:${itemIdentity.number}`
: fingerprint;
if (!candidateBuckets.has(candidateKey)) candidateBuckets.set(candidateKey, []);
const groups = candidateBuckets.get(candidateKey);
const matchingGroup = groups.find(group => group.some(member => scoreCandidate(member, projected, {
mode: options.mode || 'algo',
layaHint: options.identityHints?.get(pairKey(member, projected))
}).matched));
if (matchingGroup) matchingGroup.push(projected);
else groups.push([projected]);
}
return Array.from(candidateBuckets.values()).flat().map(variants => {
const fingerprint = variants.map(variant => variant.fingerprint).sort((a, b) => a.length - b.length || a.localeCompare(b))[0];
variants.sort((a, b) => b.qualityRank - a.qualityRank || a.source_id - b.source_id || String(a.item_id).localeCompare(String(b.item_id)));
variants.forEach((variant, index) => { variant.variantRank = index; });
return {
id: `logical:${stableHash(fingerprint).slice(0, 24)}`,
fingerprint,
displayName: displayNameForFingerprint(fingerprint),
preferredVariantId: variants[0].id,
variants
};
}).sort((a, b) => a.displayName.localeCompare(b.displayName));
}
function safeIdentityDescriptor(item) {
return {
name: String(item.name || '').slice(0, 300),
group: String(item.category_name || item.groupTitle || '').slice(0, 200),
source: String(item.source_name || '').slice(0, 120),
tvgId: String(item.tvg_id || '').slice(0, 160)
};
}
function identityCandidatePairs(items) {
const buckets = new Map();
for (const item of items) {
const fingerprint = normalizeChannelName(item.name);
const itemIdentity = identity(fingerprint);
if (!itemIdentity.family || !itemIdentity.number) continue;
const key = `${itemIdentity.family}:${itemIdentity.number}`;
if (!buckets.has(key)) buckets.set(key, []);
buckets.get(key).push({ ...item, fingerprint });
}
const pairs = [];
for (const bucket of buckets.values()) {
for (let left = 0; left < bucket.length; left++) {
for (let right = left + 1; right < bucket.length; right++) {
if (bucket[left].fingerprint === bucket[right].fingerprint) continue;
if (scoreCandidate(bucket[left], bucket[right], { mode: 'algo' }).blocker) continue;
pairs.push({
id: pairKey(bucket[left], bucket[right]),
left: safeIdentityDescriptor(bucket[left]),
right: safeIdentityDescriptor(bucket[right])
});
if (pairs.length >= 256) return pairs;
}
}
}
return pairs;
}
async function collectIdentityHints(items, mode, layaClient, options = {}) {
if (mode === 'algo' || !layaClient) return new Map();
const pairs = identityCandidatePairs(items);
const total = Math.ceil(pairs.length / 64);
let processed = 0;
const hints = new Map();
options.onProgress?.({ processed, total });
for (let index = 0; index < pairs.length; index += 64) {
const results = await layaClient.identityHintBatch(pairs.slice(index, index + 64));
results.forEach(result => hints.set(result.id, result));
processed += 1;
options.onProgress?.({ processed, total });
}
return hints;
}
async function collectOptionalIdentityHints(items, mode, layaClient, options = {}) {
if (mode === 'algo') return new Map();
try {
return await collectIdentityHints(items, mode, layaClient, options);
} catch (error) {
console.warn('[LiveProjection] Laya identity hints unavailable; continuing deterministically');
return new Map();
}
}
function catalogInputHash(items) {
return stableHash(items
.map(item => `${item.id}\0${item.name}\0${item.category_id || ''}`)
.sort()
.join('\n'));
}
function defaultDatabase() {
return require('../db/sqlite').getDb();
}
function projectionState(database = null) {
database ||= defaultDatabase();
return database.prepare('SELECT * FROM catalog_projection_state WHERE projection = ?').get('live') || {
projection: 'live', status: 'missing', started_at: null, completed_at: null,
input_hash: null, projected_count: 0, phase: 'idle', processed_count: 0, total_count: 0,
laya_processed_count: 0, laya_total_count: 0, error: null
};
}
function updateProjectionProgress(database, phase, progress = {}) {
database.prepare(`
UPDATE catalog_projection_state SET phase = ?,
laya_processed_count = COALESCE(?, laya_processed_count),
laya_total_count = COALESCE(?, laya_total_count)
WHERE projection = 'live'
`).run(phase, progress.layaProcessed ?? null, progress.layaTotal ?? null);
}
function markProjectionBuilding(database = null) {
database ||= defaultDatabase();
const total = database.prepare("SELECT COUNT(*) AS count FROM playlist_items WHERE type = 'live'").get().count;
const startedAt = Date.now();
database.prepare(`
INSERT INTO catalog_projection_state
(projection, status, started_at, completed_at, input_hash, projected_count, phase,
processed_count, total_count, laya_processed_count, laya_total_count, error)
VALUES ('live', 'building', ?, NULL, NULL, 0, 'preparing', 0, ?, 0, 0, NULL)
ON CONFLICT(projection) DO UPDATE SET status = 'building', started_at = excluded.started_at,
completed_at = NULL, phase = 'preparing', processed_count = 0,
total_count = excluded.total_count, laya_processed_count = 0,
laya_total_count = 0, error = NULL
`).run(startedAt, total);
return projectionState(database);
}
function loadProjectionItems(database) {
return database.prepare(`
SELECT p.id, p.source_id, p.item_id, p.name, p.category_id, p.data,
c.name AS category_name
FROM playlist_items p
LEFT JOIN categories c ON c.source_id = p.source_id AND c.type = 'live' AND c.category_id = p.category_id
WHERE p.type = 'live'
ORDER BY p.id
`).all().map(item => {
const data = parseData(item.data);
return { ...item, tvg_id: data.tvgId || data.tvg_id || null };
});
}
function rebuild(database = null, options = {}) {
database ||= defaultDatabase();
const items = options.items || loadProjectionItems(database);
if (!options.stateAlreadyBuilding) markProjectionBuilding(database);
updateProjectionProgress(database, 'projecting');
const groups = projectCatalog(items, { mode: options.mode, identityHints: options.identityHints });
const inputHash = catalogInputHash(items);
updateProjectionProgress(database, 'writing');
try {
const apply = database.transaction(() => {
database.prepare('DELETE FROM logical_channels').run();
database.prepare(`
UPDATE playlist_items SET logical_channel_id = NULL, grouping_fingerprint = NULL,
grouping_input_hash = NULL, quality_label = NULL, quality_rank = NULL, variant_rank = NULL,
language_code = NULL, language_confidence = NULL, language_scores = NULL,
language_evidence = NULL, language_detector = NULL, language_input_hash = NULL
WHERE type = 'live'
`).run();
const insertLogical = database.prepare(`
INSERT INTO logical_channels
(id, fingerprint, display_name, preferred_variant_id, variant_count, created_at, updated_at)
VALUES (?, ?, ?, ?, ?, ?, ?)
`);
const updateVariant = database.prepare(`
UPDATE playlist_items SET logical_channel_id = ?, grouping_fingerprint = ?,
grouping_input_hash = ?, quality_label = ?, quality_rank = ?, variant_rank = ?,
language_code = ?, language_confidence = ?, language_scores = ?,
language_evidence = ?, language_detector = 'algorithm', language_input_hash = ?
WHERE id = ? AND type = 'live'
`);
const now = Date.now();
for (const group of groups) {
insertLogical.run(group.id, group.fingerprint, group.displayName, group.preferredVariantId, group.variants.length, now, now);
for (const variant of group.variants) {
updateVariant.run(group.id, group.fingerprint, variant.groupingInputHash,
variant.qualityLabel, variant.qualityRank, variant.variantRank,
variant.languageCode, variant.languageConfidence,
JSON.stringify(variant.languageScores), JSON.stringify(variant.languageEvidence),
stableHash(`${variant.name || ''}\0${variant.category_name || ''}\0${variant.tvg_id || ''}`), variant.id);
}
}
database.prepare(`
UPDATE catalog_projection_state SET status = 'success', phase = 'complete', completed_at = ?,
input_hash = ?, projected_count = ?, processed_count = ?, total_count = ?, error = NULL
WHERE projection = 'live'
`).run(Date.now(), inputHash, groups.length, items.length, items.length);
});
apply();
return projectionState(database);
} catch (error) {
const diagnosticCode = /^SQLITE_[A-Z_]+$/.test(error.code || '') ? error.code : 'INTERNAL';
console.error(`[LiveProjection] Rebuild transaction failed (${diagnosticCode}, ${error.name || 'Error'})`);
database.prepare(`
UPDATE catalog_projection_state SET status = 'error', phase = 'failed', completed_at = ?, error = ?
WHERE projection = 'live'
`).run(Date.now(), `Live catalog projection failed [${diagnosticCode}]`);
throw error;
}
}
async function rebuildWithSignals(database = null, options = {}) {
database ||= defaultDatabase();
const items = loadProjectionItems(database);
const configured = options.mode || (await require('../db').settings.get()).languageDetectionMode || 'algo';
const mode = ['algo', 'hybrid', 'laya'].includes(configured) ? configured : 'algo';
let identityHints = new Map();
if (mode !== 'algo') {
const layaClient = options.layaClient || new (require('./layaService').LayaService)();
identityHints = await collectOptionalIdentityHints(items, mode, layaClient, {
onProgress: ({ processed, total }) => updateProjectionProgress(database, 'laya_identity', {
layaProcessed: processed,
layaTotal: total
})
});
}
return rebuild(database, { ...options, items, mode, identityHints });
}
let activeRebuild = null;
let activeState = null;
function startRebuild(database = null, options = {}) {
if (activeRebuild) return { alreadyRunning: true, state: activeState, promise: activeRebuild };
database ||= defaultDatabase();
const mark = options.mark || markProjectionBuilding;
const runner = options.runner || rebuildWithSignals;
const schedule = options.schedule || setImmediate;
activeState = mark(database);
activeRebuild = new Promise((resolve, reject) => {
schedule(() => {
Promise.resolve()
.then(() => runner(database, { stateAlreadyBuilding: true }))
.then(resolve, reject);
});
}).finally(() => {
activeRebuild = null;
activeState = null;
});
return { alreadyRunning: false, state: activeState, promise: activeRebuild };
}
function queueRebuild(database = null) {
return startRebuild(database).promise;
}
function parseData(value) {
try { return value ? JSON.parse(value) : {}; } catch { return {}; }
}
function getChannels({ sourceId = null, includeHidden = false, userId = null, q = '', languages = [] }, database = null) {
database ||= defaultDatabase();
const params = [];
let sourceFilter = '';
if (sourceId !== null) {
sourceFilter = 'AND p.source_id = ?';
params.push(sourceId);
}
const rows = database.prepare(`
SELECT p.*, c.name AS category_name, COALESCE(c.is_hidden, 0) AS category_hidden,
CASE WHEN f.id IS NULL THEN 0 ELSE 1 END AS user_favorite,
l.display_name, l.preferred_variant_id
FROM playlist_items p
JOIN logical_channels l ON l.id = p.logical_channel_id
LEFT JOIN categories c ON c.source_id = p.source_id AND c.type = 'live' AND c.category_id = p.category_id
LEFT JOIN favorites f ON f.user_id = ? AND f.source_id = p.source_id
AND f.item_id = p.item_id AND f.item_type = 'channel'
WHERE p.type = 'live' ${sourceFilter}
ORDER BY l.display_name, p.variant_rank
`).all(userId ?? -1, ...params);
return serializeChannels(rows, includeHidden).map(channel => {
const match = iptvOrgCatalog.findMatch(database, channel);
if (match) {
channel.standardName = match.name;
channel.standard = { provider: 'iptv-org', channelId: match.channelId, country: match.country, confidence: match.confidence };
}
return channel;
}).filter(channel =>
(!languages.length || languages.includes(channel.language.code)) && matchesSearch(channel, q)
);
}
function serializeChannels(rows, includeHidden = false) {
const groups = new Map();
for (const row of rows) {
const hidden = Boolean(row.is_hidden || row.category_hidden);
if (hidden && !includeHidden) continue;
const raw = parseData(row.data);
const categoryName = raw.groupTitle || row.category_name || 'Uncategorized';
const variant = {
...raw,
id: row.item_id,
sourceId: row.source_id,
name: row.name,
categoryId: row.category_id,
categoryName,
groupTitle: categoryName,
streamIcon: row.stream_icon,
streamUrl: row.stream_url,
qualityLabel: row.quality_label,
qualityRank: row.quality_rank,
variantRank: row.variant_rank,
isHidden: hidden,
isFavorite: Boolean(row.user_favorite)
};
const languageCode = row.language_code || 'und';
const languageKey = `${row.logical_channel_id}:${languageCode}`;
if (!groups.has(languageKey)) {
groups.set(languageKey, {
id: languageKey,
logicalChannelId: row.logical_channel_id,
fingerprint: row.grouping_fingerprint,
displayName: row.display_name,
standardName: row.display_name,
standard: { provider: 'local', channelId: null, confidence: 0 },
language: {
code: languageCode,
short: languageCode === 'und' ? 'UND' : languageCode.toUpperCase(),
label: languageCode === 'und' ? 'Unknown' : new Intl.DisplayNames(['en'], { type: 'language' }).of(languageCode),
confidence: Number(row.language_confidence || 0)
},
name: row.display_name,
variants: [],
isFavorite: false,
isHidden: true
});
}
const group = groups.get(languageKey);
group.variants.push(variant);
group.isFavorite ||= variant.isFavorite;
group.isHidden &&= variant.isHidden;
if (!group.preferredVariant || Number(variant.variantRank) < Number(group.preferredVariant.variantRank)) group.preferredVariant = variant;
}
return Array.from(groups.values()).map(group => {
group.preferredVariant ||= group.variants[0];
group.categoryId = group.preferredVariant?.categoryId || null;
group.categoryName = group.preferredVariant?.categoryName || 'Uncategorized';
group.groupTitle = group.preferredVariant?.groupTitle || group.categoryName;
group.selectedVariantId = group.preferredVariant
? `${group.preferredVariant.sourceId}:${group.preferredVariant.id}`
: null;
const epgChannelId = group.preferredVariant?.tvgId || group.preferredVariant?.epg_channel_id || null;
group.epg = {
sourceId: null,
channelId: epgChannelId,
language: group.language.code,
matchConfidence: epgChannelId ? 100 : 0
};
return group;
});
}
module.exports = {
catalogInputHash,
collectIdentityHints,
collectOptionalIdentityHints,
displayNameForFingerprint,
getChannels,
identityCandidatePairs,
normalizeChannelName,
markProjectionBuilding,
matchesSearch,
projectCatalog,
projectionState,
qualityForName,
queueRebuild,
rebuild,
rebuildWithSignals,
safeIdentityDescriptor,
serializeChannels,
startRebuild
};