From 832d604aedaaf276459a42e0c02e29f2151431cd Mon Sep 17 00:00:00 2001 From: Elegy233 <82034747+wan0ge@users.noreply.github.com> Date: Sun, 22 Mar 2026 08:04:32 +0800 Subject: [PATCH] =?UTF-8?q?=E4=B8=BA=E5=90=88=E5=B9=B6=E6=94=AF=E6=8C=81?= =?UTF-8?q?=E5=88=A4=E6=96=AD3D2D=EF=BC=8C=E4=BF=AE=E5=A4=8D=E5=B8=A6?= =?UTF-8?q?=E6=9C=89=E4=B8=8D=E5=90=8C=E9=85=8D=E9=9F=B3=E7=9A=84=E7=BB=93?= =?UTF-8?q?=E6=9E=9C=E5=90=88=E9=9B=86=E5=88=A4=E6=96=AD=E5=A4=B1=E8=AF=AF?= =?UTF-8?q?=EF=BC=8C=E5=A2=9E=E5=8A=A0=E7=AE=80=E6=98=93=E7=9A=84=E5=86=85?= =?UTF-8?q?=E9=83=A8=E7=89=B9=E6=AE=8A=E6=9D=A1=E7=9B=AE=E6=98=A0=E5=B0=84?= =?UTF-8?q?=E8=A7=84=E5=88=99=20(#252)?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit * animeko.js：从tags中提取3D/2D追加到类型 * iqiyi.js：从metaTags中提取3D/2D追加到类型 * tencent.js：从richTags中提取3D/2D追加到类型 * merge-util.js：支持判断3D2D，修复带有不同配音的结果合集判断失误，增加简易的内部特殊条目映射规则 --- danmu_api/sources/animeko.js | 16 +- danmu_api/sources/iqiyi.js | 23 ++- danmu_api/sources/tencent.js | 28 ++- danmu_api/utils/merge-util.js | 361 ++++++++++++++++++++++++---------- 4 files changed, 317 insertions(+), 111 deletions(-) diff --git a/danmu_api/sources/animeko.js b/danmu_api/sources/animeko.js index 0599a36..8ea449f 100644 --- a/danmu_api/sources/animeko.js +++ b/danmu_api/sources/animeko.js @@ -260,6 +260,18 @@ export default class AnimekoSource extends BaseSource { } } + // 识别 3D 与 2D 标签并追加至类型描述 + let is3D = false; + let is2D = false; + if (item.tags && Array.isArray(item.tags)) { + item.tags.forEach(tag => { + if (tag.name === '3D') is3D = true; + if (tag.name === '2D') is2D = true; + }); + } + if (is3D) typeDesc = "3D" + typeDesc; + else if (is2D) typeDesc = "2D" + typeDesc; + const titleSuffix = item._relation_mark ? ` ${item._relation_mark}` : ""; // 提取别名列表 (用于合并工具进行模糊匹配) @@ -370,7 +382,7 @@ export default class AnimekoSource extends BaseSource { * @param {string} queryTitle 原始查询标题 * @param {Array} curAnimes 当前缓存的番剧列表 */ - async handleAnimes(sourceAnimes, queryTitle, curAnimes, detailStore = null) { + async handleAnimes(sourceAnimes, queryTitle, curAnimes, detailStore = null) { const tmpAnimes = []; if (!sourceAnimes || !Array.isArray(sourceAnimes)) { @@ -426,7 +438,7 @@ export default class AnimekoSource extends BaseSource { }; tmpAnimes.push(transformedAnime); - addAnime({...transformedAnime, links: links}, detailStore); + addAnime({...transformedAnime, links: links}, detailStore); if (globals.animes.length > globals.MAX_ANIMES) removeEarliestAnime(); } diff --git a/danmu_api/sources/iqiyi.js b/danmu_api/sources/iqiyi.js index 0a17e94..11fbe52 100644 --- a/danmu_api/sources/iqiyi.js +++ b/danmu_api/sources/iqiyi.js @@ -148,8 +148,29 @@ export default class IqiyiSource extends BaseSource { return null; } + // 提取 3D 与 2D 属性标签并追加至媒体类型 + let is3D = false; + let is2D = false; + if (album.metaTags && Array.isArray(album.metaTags)) { + album.metaTags.forEach(tag => { + if (tag.name === '3D') is3D = true; + if (tag.name === '2D') is2D = true; + }); + } + if (album.baseTags && Array.isArray(album.baseTags)) { + album.baseTags.forEach(tag => { + if (tag.value === '3D') is3D = true; + if (tag.value === '2D') is2D = true; + }); + } + if (is3D) { + mediaType = "3D" + mediaType; + } else if (is2D) { + mediaType = "2D" + mediaType; + } + // 电影类型：使用 qipuId 作为 mediaId - if (mediaType === "电影") { + if (mediaType.includes("电影")) { const qipuId = album.qipuId || album.playQipuId; if (!qipuId) { log("debug", `[iQiyi] 电影缺少 qipuId: ${album.title}`); diff --git a/danmu_api/sources/tencent.js b/danmu_api/sources/tencent.js index 5a16a10..c5ed55c 100644 --- a/danmu_api/sources/tencent.js +++ b/danmu_api/sources/tencent.js @@ -55,6 +55,32 @@ export default class TencentSource extends BaseSource { return null; } + let mediaType = contentType; + + // 提取 3D 与 2D 属性标签并追加至媒体类型 + let is3D = false; + let is2D = false; + if (videoInfo.coverDoc) { + if (videoInfo.coverDoc.richTags && Array.isArray(videoInfo.coverDoc.richTags)) { + videoInfo.coverDoc.richTags.forEach(tag => { + if (tag.text && tag.text.includes('3D')) is3D = true; + if (tag.text && tag.text.includes('2D')) is2D = true; + }); + } + if (videoInfo.coverDoc.tags && Array.isArray(videoInfo.coverDoc.tags)) { + videoInfo.coverDoc.tags.forEach(tag => { + if (tag && tag.includes('3D')) is3D = true; + if (tag && tag.includes('2D')) is2D = true; + }); + } + } + + if (is3D) { + mediaType = "3D" + mediaType; + } else if (is2D) { + mediaType = "2D" + mediaType; + } + // 过滤非腾讯视频内容 const allSites = (videoInfo.playSites || []).concat(videoInfo.episodeSites || []); if (allSites.length > 0 && !allSites.some(site => site.enName === 'qq')) { @@ -75,7 +101,7 @@ export default class TencentSource extends BaseSource { provider: "tencent", mediaId: mediaId, title: title, - type: contentType, // 使用中文类型,与360/vod保持一致 + type: mediaType, year: videoInfo.year, imageUrl: videoInfo.imgUrl, episodeCount: episodeCount diff --git a/danmu_api/utils/merge-util.js b/danmu_api/utils/merge-util.js index 34070da..674c0b5 100644 --- a/danmu_api/utils/merge-util.js +++ b/danmu_api/utils/merge-util.js @@ -34,10 +34,10 @@ const MergeWeights = Object.freeze({ TITLE_STRUCTURE_CONFLICT: -0.30, // 标题结构冲突（如父子集关系） LANG_MATCH_CN: 0.15, // 双端均为中文时的奖励 LANG_MISMATCH: -0.20, // 语言不一致时的惩罚 - + // 日期 DATE_MATCH: 0.0, // 基础日期匹配（动态计算，此处为占位） - + // 集数对齐 (Alignment) EP_ALIGN: { MOVIE_TYPE_MISMATCH: -5.0, // 电影/TV 类型不符 @@ -216,6 +216,84 @@ const SEASON_PATTERNS = [ { regex: /[^0-9](\d)$/, prefix: 'S', useCleaned: true } ]; +// ========================================== +// 扩展配置: 特殊番剧特征规则库 (Special Series Registry) +// ========================================== +const SpecialSeriesRegistry = [ + { + // 案例 1：经典的带副标题映射（美少女战士） + seriesKeywords: ["美少女战士"], + mappings: [ + // 忽略大小写，填入 "R" 即可自动匹配 "r" + { markers: ["R"], targetStandard: "第二季" }, + { markers: ["S"], targetStandard: "第三季" }, + { markers: ["SuperS", "Super S"], targetStandard: "第四季" }, + { markers: ["Sailor Stars", "最后的星光"], targetStandard: "第五季" } + ] + }, + { + seriesKeywords: ["小林家的龙女仆"], + mappings: [ + { markers: ["S"], targetStandard: "第二季" }, + ] + }, + { + // 案例 2：可能携带副标题的（我们不可能成为恋人） + seriesKeywords: ["可能成为恋人", "不行"], + mappings: [ + { markers: ["NEXT SHINE", "再次闪耀"], targetStandard: "续篇" } + ] + }, + { + // 案例 3：没有任何副标题，主标题本身就是特殊类型 + seriesKeywords: ["红猪", "千与千寻", "龙猫"], + // 当没有匹配到任何 mappings（或根本没写 mappings）时，直接给这个条目打上默认标签 + defaultStandard: "剧场版" + } +]; + +function escapeRegExp(string) { + return string.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); +} + +/** + * 引擎级标题语义转换器 (Semantic Normalizer) + */ +function normalizeTitleForEngine(title) { + if (!title) return ''; + let normTitle = title; + const upperTitle = normTitle.toUpperCase(); + + for (const rule of SpecialSeriesRegistry) { + // 判断是否命中了设定的系列 + const isTargetSeries = rule.seriesKeywords.every(kw => upperTitle.includes(kw.toUpperCase())); + + if (isTargetSeries) { + let hasMapped = false; + + // 1. 尝试匹配并替换副标题 + if (rule.mappings) { + for (const mapping of rule.mappings) { + mapping.markers.forEach(marker => { + // 'gi' 保证了全局且忽略大小写的替换 + const reg = new RegExp(escapeRegExp(marker), 'gi'); + if (reg.test(normTitle)) { + normTitle = normTitle.replace(reg, ` ${mapping.targetStandard} `); + hasMapped = true; + } + }); + } + } + + // 如果没有任何副标题被替换，且配置了 defaultStandard，则强行追加标准类型 + if (!hasMapped && rule.defaultStandard) { + normTitle += ` ${rule.defaultStandard} `; + } + } + } + return normTitle; +} + // ========================================== // 3. 基础文本处理工具 (Utilities) // ========================================== @@ -243,13 +321,13 @@ function getLanguageType(text) { function cleanText(text) { if (!text) return ''; let clean = simplized(text); - + clean = clean.replace(RegexStore.Clean.NA_TAG, ''); clean = clean.replace(RegexStore.Season.PART_NORM, 'part $1'); clean = clean.replace(RegexStore.Season.PART_NORM_2, 'part $1'); clean = clean.replace(RegexStore.Season.FINAL, '最终季'); clean = clean.replace(RegexStore.Season.NORM, '第$1季'); - + const cnNums = {'一':'1', '二':'2', '三':'3', '四':'4', '五':'5', '六':'6', '七':'7', '八':'8', '九':'9', '十':'10'}; clean = clean.replace(RegexStore.Season.CN, (m, num) => `第${cnNums[num]}季`); clean = clean.replace(RegexStore.Season.ROMAN, (match, p1, roman, p2) => { @@ -264,19 +342,20 @@ function cleanText(text) { clean = clean.replace(/(\d+)\.(\d+)/g, '$1{{DOT}}$2'); clean = clean.replace(RegexStore.Clean.PUNCTUATION, ' '); clean = clean.replace(/{{DOT}}/g, '.'); - + return clean.replace(RegexStore.Clean.WHITESPACE, ' ').toLowerCase().trim(); } /** * 相似度计算专用极简清洗 - * 强力移除所有季数、Part、括号内容、语言标识，只保留核心标题 + * 强力移除所有季数、Part、括号内容、语言标识、维数标识等，只保留核心标题 * @param {string} text - 原始标题 * @returns {string} - 用于相似度计算的极简标题 */ function cleanTitleForSimilarity(text) { if (!text) return ''; - let clean = simplized(text); + let clean = normalizeTitleForEngine(text); + clean = simplized(clean); // 标题保护逻辑：如果标题被包裹在【】或[]中且位于开头，尝试提取内容 const startBracketMatch = clean.match(/^(?:【|\[)(.+?)(?:】|\])/); if (startBracketMatch) { @@ -432,12 +511,12 @@ function calculateSimilarity(str1, str2) { if (s1 === s2) return 1.0; const len1 = s1.length, len2 = s2.length; const maxLen = Math.max(len1, len2), minLen = Math.min(len1, len2); - + if (s1.includes(s2) || s2.includes(s1)) { const lenRatio = minLen / maxLen; if (lenRatio > 0.5) return 0.8 + (lenRatio * 0.2); } - + const distance = editDistance(s1, s2); const editScore = maxLen === 0 ? 1.0 : 1.0 - (distance / maxLen); const set1 = new Set(s1.replace(RegexStore.Clean.WHITESPACE, '')); @@ -521,6 +600,7 @@ function checkTitleSubtitleConflict(titleA, titleB, isDateValid = true) { * @returns {Set} */ function extractSeasonMarkers(title, typeDesc = '') { + title = normalizeTitleForEngine(title); const markers = new Set(); const t = cleanText(title); const type = cleanText(typeDesc || ''); @@ -583,6 +663,7 @@ function extractSeasonMarkers(title, typeDesc = '') { * @returns {number|null} */ function getSeasonNumber(title, typeDesc = '') { + title = normalizeTitleForEngine(title); const markers = extractSeasonMarkers(title, typeDesc); let maxSeason = null; for (const m of markers) { @@ -657,13 +738,38 @@ function checkTheatricalExemption(titleA, titleB, typeDescA, typeDescB) { } /** - * 校验媒体类型是否冲突 (真人 vs 动漫, TV vs Movie) + * 校验媒体类型是否冲突 (真人 vs 动漫, TV vs Movie, 3D vs 2D) * @returns {boolean} true=冲突, false=兼容 */ function checkMediaTypeMismatch(titleA, titleB, typeDescA, typeDescB, countA, countB, sourceA = '', sourceB = '') { const catA = getContentCategory(titleA, typeDescA, sourceA); const catB = getContentCategory(titleB, typeDescB, sourceB); if ((catA === 'REAL' && catB === 'ANIME') || (catA === 'ANIME' && catB === 'REAL')) return true; + + // 提取明确的维数属性 + let is3DA = (typeDescA || '').includes('3D'); + let is3DB = (typeDescB || '').includes('3D'); + let is2DA = (typeDescA || '').includes('2D'); + let is2DB = (typeDescB || '').includes('2D'); + + // 维数动态探测逻辑 (Dynamic Dimension Detection): + // 如果有一方在类型中明确是 3D，则被授权从另一方的标题中主动探测 3D 关键词 + if (is3DA && !is3DB) { + if (/3[dD]/.test(titleB)) is3DB = true; + } else if (is3DB && !is3DA) { + if (/3[dD]/.test(titleA)) is3DA = true; + } + + // 同样适用 2D 的动态探测：明确有 2D 标签时，才去对方标题里找 2D + if (is2DA && !is2DB) { + if (/2[dD]/.test(titleB)) is2DB = true; + } else if (is2DB && !is2DA) { + if (/2[dD]/.test(titleA)) is2DA = true; + } + + // 检测 3D 与 2D 的维数属性冲突：任意一方带 3D 且另一方不带 3D（即视为 2D）则判定为冲突 + if (is3DA !== is3DB) return true; + const mediaA = getStrictMediaType(titleA, typeDescA); const mediaB = getStrictMediaType(titleB, typeDescB); if (!mediaA || !mediaB || mediaA === mediaB) return false; @@ -912,7 +1018,7 @@ export function findSecondaryMatches(primaryAnime, secondaryList, collectionAnim const isSecDub = !!(secTitleForSim.match(RegexStore.Lang.CN_DUB_VER)) || RegexStore.Lang.CN.test(secTitleForSim); const isDubRelation = isPrimaryDub || isSecDub; const secCount = secAnime.episodeCount || (secAnime.links ? secAnime.links.length : 0); - + // 之字结构强阻断 if (secTitleForSim.includes('之')) { const parts = secTitleForSim.split('之'); @@ -926,7 +1032,7 @@ export function findSecondaryMatches(primaryAnime, secondaryList, collectionAnim if (checkMediaTypeMismatch(rawPrimaryTitle, rawSecTitle, primaryAnime.typeDescription, secAnime.typeDescription, primaryCount, secCount, primaryAnime.source, secAnime.source)) { const pType = getContentCategory(rawPrimaryTitle, primaryAnime.typeDescription, primaryAnime.source); const sType = getContentCategory(rawSecTitle, secAnime.typeDescription, secAnime.source); - logReason(rawSecTitle, `媒体类型不匹配 (P:${pType}/${getStrictMediaType(rawPrimaryTitle, primaryAnime.typeDescription)} vs S:${sType}/${getStrictMediaType(rawSecTitle, secAnime.typeDescription)})`); + logReason(rawSecTitle, `媒体类型或维数不匹配 (P:${pType}/${getStrictMediaType(rawPrimaryTitle, primaryAnime.typeDescription)} [${primaryAnime.typeDescription}] vs S:${sType}/${getStrictMediaType(rawSecTitle, secAnime.typeDescription)} [${secAnime.typeDescription}])`); continue; } @@ -945,7 +1051,7 @@ export function findSecondaryMatches(primaryAnime, secondaryList, collectionAnim } } } - + if (isPrimaryContextSequel) { if (cleanTitleForSimilarity(secTitleForSim) === cleanTitleForSimilarity(primaryBaseTitleFromContext)) { logReason(rawSecTitle, `上下文阻断: 主源(S2/Sequel) vs 副源(Base/S1) (Base: "${primaryBaseTitleFromContext}")`); @@ -1141,7 +1247,7 @@ function extractEpisodeInfo(title, sourceName = '') { */ function filterEpisodes(links, filterRegex, sourceName = '') { if (!links) return []; - + // 特定源白名单：这些源的集标题通常比较规范，免除正则过滤拦截 const skipFilterSources = ['animeko', 'bilibili', 'bilibili1', 'bahamut', 'dandan']; const shouldSkipFilter = skipFilterSources.includes(sourceName); @@ -1149,7 +1255,7 @@ function filterEpisodes(links, filterRegex, sourceName = '') { if (!filterRegex || shouldSkipFilter) { return links.map((link, index) => ({ link, originalIndex: index })); } - + const validLinks = []; const droppedTitles = []; @@ -1330,7 +1436,7 @@ function findBestAlignmentOffset(primaryLinks, secondaryLinks, seriesLangA = 'Un else sim = MergeWeights.EP_ALIGN.CN_STRICT_MISMATCH; } else sim = calculateSimilarity(dataA.cleanEpText, dataB.cleanEpText); } else sim = calculateSimilarity(dataA.cleanEpText, dataB.cleanEpText); - + pairScore += sim; rawTextScoreSum += sim; if (infoA.num !== null && infoB.num !== null && infoA.num === infoB.num) pairScore += MergeWeights.EP_ALIGN.NUMERIC_MATCH; @@ -1348,7 +1454,7 @@ function findBestAlignmentOffset(primaryLinks, secondaryLinks, seriesLangA = 'Un let finalScore = totalTextScore / matchCount; let maxFrequency = 0; for (const count of numericDiffs.values()) maxFrequency = Math.max(maxFrequency, count); - + const consistencyRatio = maxFrequency / matchCount; const avgRawTextScore = rawTextScoreSum / matchCount; @@ -1431,7 +1537,7 @@ function stitchUnmatchedEpisodes(derivedAnime, orphans, sourceName) { break; } } - + for (const item of orphans) { const relativeIdx = item.relativeIndex; const isStrictSpecial = item.info && item.info.isStrictSpecial; @@ -1540,7 +1646,7 @@ function buildSeasonLengthMap(allGroupAnimes, epFilter, collectionAnimeIds) { } const realAnime = globals.animes.find(a => String(a.animeId) === String(anime.animeId)) || anime; - + // 2. 类型过滤：严格剔除电影和真人剧 // 地图构建是为了给 TV 动画切片，电影(1集)和真人剧(集数不同)会严重干扰统计 const category = getContentCategory(realAnime.animeTitle, realAnime.typeDescription, realAnime.source); @@ -1548,7 +1654,7 @@ function buildSeasonLengthMap(allGroupAnimes, epFilter, collectionAnimeIds) { debugLogs.push(` [剔除] [${realAnime.source}] ${realAnime.animeTitle} (类型: 真人剧/REAL)`); continue; } - + const mediaType = getStrictMediaType(realAnime.animeTitle, realAnime.typeDescription); if (mediaType === 'MOVIE') { debugLogs.push(` [剔除] [${realAnime.source}] ${realAnime.animeTitle} (类型: 电影/MOVIE)`); @@ -1561,7 +1667,7 @@ function buildSeasonLengthMap(allGroupAnimes, epFilter, collectionAnimeIds) { const title = item.link.title || item.link.name || ""; const cleanT = cleanText(title); const rawTemp = cleanT.replace(RegexStore.Clean.SOURCE_TAG, '').replace(RegexStore.Clean.FROM_SUFFIX, '').trim(); - + // 特定源的番外过滤逻辑 if (/^(dandan|animeko)$/i.test(realAnime.source)) { if (RegexStore.Episode.SPECIAL_CHECK.test(rawTemp) || RegexStore.Episode.DANDAN_IGNORE.test(rawTemp)) return false; @@ -1574,10 +1680,10 @@ function buildSeasonLengthMap(allGroupAnimes, epFilter, collectionAnimeIds) { if (count > 0) { if (!seasonStats.has(seasonNum)) seasonStats.set(seasonNum, new Map()); const freqMap = seasonStats.get(seasonNum); - + if (!freqMap.has(count)) freqMap.set(count, []); freqMap.get(count).push(realAnime.source); - + debugLogs.push(` [采纳] [${realAnime.source}] S${seasonNum} = ${count}集 ("${realAnime.animeTitle}")`); } else { debugLogs.push(` [忽略] [${realAnime.source}] S${seasonNum} (有效集数为 0)`); @@ -1615,9 +1721,9 @@ function buildSeasonLengthMap(allGroupAnimes, epFilter, collectionAnimeIds) { } } } - + seasonMap.set(sNum, modeCount); - + // 构建详细的统计日志字符串 const statDetails = Array.from(freqMap.entries()) .map(([cnt, srcs]) => `${cnt}集(x${srcs.length})[${srcs.join(',')}]`) @@ -1654,7 +1760,7 @@ function fastCloneAnime(anime) { /** * 执行单个主源的合并任务 - * 包含：寻找匹配、ID生成、链接映射、集数补全 + * 包含：寻找匹配、ID生成、链接映射、集数补全、跨源合集时序接管 * @param {Object} params - 任务参数对象 * @returns {Promise} 合并后的新对象或 null */ @@ -1679,14 +1785,14 @@ async function processMergeTask(params) { const logTitleA = pAnime.animeTitle.replace(RegexStore.Clean.FROM_SUFFIX, ''); let derivedAnime = fastCloneAnime(cachedPAnime); - + const actualMergedSources = []; const contentSignatureParts = [pAnime.animeId]; let hasMergedAny = false; const seriesLangA = getLanguageType(pAnime.animeTitle); const redundantP = identifyRedundantTitle(derivedAnime.links, pAnime.animeTitle, currentPrimarySource); - + const getTempTitle = (rawTitle, redundantStr) => { if (!rawTitle) return ""; if (redundantStr && rawTitle.includes(redundantStr)) return rawTitle.replace(redundantStr, ''); @@ -1697,7 +1803,7 @@ async function processMergeTask(params) { const pCleanTitle = cleanTitleForSimilarity(pAnime.animeTitle); const peerAnimes = curAnimes.filter(a => cleanTitleForSimilarity(a.animeTitle) === pCleanTitle); const seasonLengthMap = buildSeasonLengthMap(peerAnimes, epFilter, collectionAnimeIds); - + if (seasonLengthMap.size > 0) { const mapDesc = Array.from(seasonLengthMap.entries()).map(([k,v]) => `S${k}=${v}`).join(', '); if (isPrimaryCollection || availableSecondaries.some(s => curAnimes.some(a => a.source === s && collectionAnimeIds.has(a.animeId)))) { @@ -1705,10 +1811,13 @@ async function processMergeTask(params) { } } + // 隔离检索，汇总排序池：避免不同源之间因微小相似度差异在 findSecondaryMatches 中发生内卷排挤 + let allMatches = []; + for (const secSource of availableSecondaries) { let secondaryItems = curAnimes.filter(a => { if (a.source !== secSource) return false; - + // 合集主源特权：当主源为合集时，允许无视当前组的消耗状态，复用已被消耗的副源以拼凑完整季度 if (isPrimaryCollection) return true; @@ -1717,7 +1826,7 @@ async function processMergeTask(params) { if (isConsumed && !isAllowedReuse) return false; return true; }); - + if (limitSecondaryLang) secondaryItems = secondaryItems.filter(a => getLanguageType(a.animeTitle) === limitSecondaryLang); if (secondaryItems.length > 1) { @@ -1730,9 +1839,31 @@ async function processMergeTask(params) { } if (secondaryItems.length === 0) continue; - const matches = findSecondaryMatches(pAnime, secondaryItems, collectionAnimeIds); - - for (const match of matches) { + // 逐源查找匹配，确保每个源的候选者在自身赛道内出线 + const matchesForSource = findSecondaryMatches(pAnime, secondaryItems, collectionAnimeIds); + allMatches.push(...matchesForSource); + } + + if (allMatches.length > 0) { + // 跨源合集时序接管：汇集所有有效匹配后，强制按季数升序排列全局匹配队列 + if (allMatches.length > 1 && (isPrimaryCollection || allMatches.some(m => collectionAnimeIds.has(m.animeId)))) { + allMatches.sort((a, b) => { + const sA = getSeasonNumber(a.animeTitle, a.typeDescription) || 1; + const sB = getSeasonNumber(b.animeTitle, b.typeDescription) || 1; + if (sA !== sB) return sA - sB; // 季数优先 + + // 季数相同时，遵循用户配置的源优先级 + const idxA = availableSecondaries.indexOf(a.source); + const idxB = availableSecondaries.indexOf(b.source); + return idxA - idxB; + }); + const seqLogs = allMatches.map(m => `[${m.source}]S${getSeasonNumber(m.animeTitle, m.typeDescription) || 1}`); + log("info", `${logPrefix} [合集时序] 已将跨源匹配队列按季数升序排列，保障切片推断严格自举: ${seqLogs.join(' -> ')}`); + } + + for (const match of allMatches) { + const secSource = match.source; + // 合集主源特权：如果是合集主源，直接跳过此处的二次消耗校验 if (!isPrimaryCollection) { const isReuse = allowReuseIds && allowReuseIds.has(match.animeId); @@ -1750,9 +1881,9 @@ async function processMergeTask(params) { const decimalsS = getDecimalEpisodes(derivedMatch.links, secSource); const toSinkS = new Set([...decimalsS].filter(x => !decimalsP.has(x))); const toSinkP = new Set([...decimalsP].filter(x => !decimalsS.has(x))); - + if (toSinkP.size > 0) sinkDecimalEpisodes(derivedAnime.links, toSinkP, currentPrimarySource, `主源:${currentPrimarySource}`); - + if (toSinkS.size > 0) { sinkDecimalEpisodes(derivedMatch.links, toSinkS, secSource, `副源:${secSource}`); } @@ -1765,14 +1896,15 @@ async function processMergeTask(params) { let activePLinks = filteredPLinksWithIndex, activeMLinks = filteredMLinksWithIndex; let sliceStartP = 0, sliceStartS = 0; const isSecondaryCollection = collectionAnimeIds.has(match.animeId); - + const performSlicing = (isPrimarySide, collectionLinks, seasonNum) => { let sliceStart = 0, slicedList = collectionLinks; if (seasonNum && seasonNum > 1) { // 1. 优先尝试历史推断 let historyFound = false; - if (!isPrimarySide && collectionProgress && collectionProgress.has(match.animeId)) { - const progress = collectionProgress.get(match.animeId); + const collectionIdToCheck = isPrimarySide ? pAnime.animeId : match.animeId; + if (collectionProgress && collectionProgress.has(collectionIdToCheck)) { + const progress = collectionProgress.get(collectionIdToCheck); const prevSeason = seasonNum - 1; if (progress[`S${prevSeason}`] !== undefined) { const inferredStart = progress[`S${prevSeason}`] + 1; @@ -1789,7 +1921,7 @@ async function processMergeTask(params) { if (!historyFound) { let accumulatedCount = 0; for (let s = 1; s < seasonNum; s++) accumulatedCount += (seasonLengthMap.get(s) || 0); - + // 移除 -2 缓冲。既然使用 Mode 策略得到了较准确的长度，直接信任该长度。 // 这样可以避免 S2 的第 1 集错误地回溯匹配到 S1 的最后几集。 let safeAccumulated = accumulatedCount; @@ -1850,18 +1982,18 @@ async function processMergeTask(params) { const orphanItem = { link: sourceLink, originalIndex: sourceLinkItem.originalIndex, relativeIndex: pIndex, info: null }; const cleanTitleS = getTempTitle(sourceLink.title, redundantS); orphanItem.info = extractEpisodeInfo(cleanTitleS, secSource); - + if (pIndex >= 0 && pIndex < filteredPLinksWithIndex.length) { const originalPIndex = filteredPLinksWithIndex[pIndex].originalIndex; const targetLink = derivedAnime.links[originalPIndex]; const pTitleShort = targetLink.name || targetLink.title || `Index ${originalPIndex}`; - + const cleanTitleP = getTempTitle(targetLink.title, redundantP); const specialP = getSpecialEpisodeType(cleanTitleP); const specialS = getSpecialEpisodeType(cleanTitleS); const infoP = extractEpisodeInfo(cleanTitleP, currentPrimarySource); const infoS = orphanItem.info; - + if (infoS.isPV && !specialP) { mappingEntries.push({ idx: pIndex, text: ` [略过] ${pTitleShort} =/= ${sTitleShort} (PV不匹配正片)` }); orphanedEpisodes.push(orphanItem); @@ -1877,7 +2009,7 @@ async function processMergeTask(params) { orphanedEpisodes.push(orphanItem); continue; } - + const idB = sanitizeUrl(sourceLink.url); let currentUrl = targetLink.url; const secPart = `${secSource}:${idB}`; @@ -1885,7 +2017,7 @@ async function processMergeTask(params) { if (!currentUrl.startsWith(currentPrimarySource + ':')) currentUrl = `${currentPrimarySource}:${currentUrl}`; } const newMergedUrl = `${currentUrl}${MERGE_DELIMITER}${secPart}`; - + let newMergedTitle = targetLink.title; if (newMergedTitle) { let sLabel = secSource; @@ -1905,7 +2037,7 @@ async function processMergeTask(params) { orphanedEpisodes.push(orphanItem); } } - + for (let j = 0; j < filteredPLinksWithIndex.length; j++) { if (!matchedPIndices.has(j)) { const originalPIndex = filteredPLinksWithIndex[j].originalIndex; @@ -1928,33 +2060,52 @@ async function processMergeTask(params) { mappingEntries.sort((a, b) => a.idx - b.idx); log("info", `${logPrefix} [${secSource}] 映射详情:\n${mappingEntries.map(e => e.text).join('\n')}`); } - - // [进度自举] 成功合并后，更新历史进度 - if (collectionProgress && isSecondaryCollection) { + + // 支持双向进度写入：主源为合集与副源为合集的情况都被覆盖，为链式关联铺路 + if (collectionProgress && (isSecondaryCollection || isPrimaryCollection)) { let maxUsedIndex = -1; - for (let k = 0; k < filteredMLinksWithIndex.length; k++) { - const pIndex = k + offset; - if (matchedPIndices.has(pIndex)) { - const item = filteredMLinksWithIndex[k]; - if (item.originalIndex > maxUsedIndex) maxUsedIndex = item.originalIndex; - } - } - - if (maxUsedIndex !== -1) { - const pSeason = getSeasonNumber(pAnime.animeTitle, pAnime.typeDescription) || 1; - if (!collectionProgress.has(match.animeId)) collectionProgress.set(match.animeId, {}); - const progress = collectionProgress.get(match.animeId); - // 只要有合并成功，就记录下来，为后续季度的关联提供基准 - if (mergedCount >= 3) { - progress[`S${pSeason}`] = maxUsedIndex; - // log("info", `[Merge-Check] 更新合集进度 [${match.animeId}]: S${pSeason} -> MaxIndex ${maxUsedIndex}`); + + if (isPrimaryCollection && !isSecondaryCollection) { + // 主源是合集，在主源索引中找最大落点 + for (let k = 0; k < filteredMLinksWithIndex.length; k++) { + const pIndex = k + offset; + if (matchedPIndices.has(pIndex)) { + const originalPIndex = filteredPLinksWithIndex[pIndex].originalIndex; + if (originalPIndex > maxUsedIndex) maxUsedIndex = originalPIndex; + } + } + if (maxUsedIndex !== -1) { + const sSeason = getSeasonNumber(derivedMatch.animeTitle, derivedMatch.typeDescription) || 1; + if (!collectionProgress.has(pAnime.animeId)) collectionProgress.set(pAnime.animeId, {}); + const progress = collectionProgress.get(pAnime.animeId); + if (mergedCount >= 3) { + progress[`S${sSeason}`] = maxUsedIndex; + } + } + } else if (!isPrimaryCollection && isSecondaryCollection) { + // 副源是合集，在副源索引中找最大落点 + for (let k = 0; k < filteredMLinksWithIndex.length; k++) { + const pIndex = k + offset; + if (matchedPIndices.has(pIndex)) { + const item = filteredMLinksWithIndex[k]; + if (item.originalIndex > maxUsedIndex) maxUsedIndex = item.originalIndex; + } + } + if (maxUsedIndex !== -1) { + const pSeason = getSeasonNumber(pAnime.animeTitle, pAnime.typeDescription) || 1; + if (!collectionProgress.has(match.animeId)) collectionProgress.set(match.animeId, {}); + const progress = collectionProgress.get(match.animeId); + if (mergedCount >= 3) { + progress[`S${pSeason}`] = maxUsedIndex; + log("info", `[Merge-Check] 更新合集进度 [${match.animeId}]: S${pSeason} -> MaxIndex ${maxUsedIndex}`); + } } } } if (collectionAnimeIds.has(match.animeId)) log("info", `${logPrefix} [智能补全] 跳过: 副源 [${secSource}] 为合集，为避免混入其他季度集数，不执行补全。`); else stitchUnmatchedEpisodes(derivedAnime, orphanedEpisodes, secSource); - + const normals = [], sinkers = []; derivedAnime.links.forEach(link => { const rawContent = link.title.replace(RegexStore.Clean.SOURCE_TAG, '').trim(); @@ -1980,7 +2131,7 @@ async function processMergeTask(params) { return null; } generatedSignatures.add(signature); - + for (let i = 1; i < contentSignatureParts.length; i++) { const secId = contentSignatureParts[i]; if (!collectionAnimeIds.has(secId)) { @@ -2035,7 +2186,7 @@ function detectCollectionCandidates(curAnimes) { .replace(RegexStore.Clean.YEAR_TAG, '') .replace(RegexStore.Clean.META_SUFFIX, '') .replace(RegexStore.Lang.KEYWORDS_STRONG, ''); - + SUFFIX_SPECIFIC_MAP.forEach(m => clean = clean.replace(m.regex, '')); clean = clean.replace(RegexStore.Season.SUFFIX_AMBIGUOUS, '') @@ -2065,7 +2216,7 @@ function detectCollectionCandidates(curAnimes) { list.forEach(anime => { const realAnime = globals.animes.find(a => String(a.animeId) === String(anime.animeId)) || anime; const markers = extractSeasonMarkers(realAnime.animeTitle, realAnime.typeDescription); - + let seasonNum = 1; for (const m of markers) { if (m.startsWith('S')) { @@ -2095,7 +2246,6 @@ function detectCollectionCandidates(curAnimes) { const t = cleanText(rawTitle); if (RegexStore.Episode.SPECIAL_CHECK.test(t) || RegexStore.Episode.DANDAN_IGNORE.test(t)) return false; if (RegexStore.Episode.MAP_EXCLUDE_KEYWORDS.test(rawContent) || RegexStore.Episode.MAP_EXCLUDE_KEYWORDS.test(rawTitle)) return false; - log("info", `[Merge-Check] [Detail-Ep] [${realAnime.source}] 保留: "${rawTitle}"`); return true; }).length; } else validCount = realAnime.links.length; @@ -2103,9 +2253,12 @@ function detectCollectionCandidates(curAnimes) { if (!sourceStats.has(realAnime.source)) sourceStats.set(realAnime.source, { seasonCounts: {}, maxSeason: 0, s1Candidates: [] }); const stat = sourceStats.get(realAnime.source); - + if (!stat.seasonCounts[seasonNum]) stat.seasonCounts[seasonNum] = 0; - stat.seasonCounts[seasonNum] += validCount; + + // 取单季度的最大集数，避免多语言版本或相同季度的重复条目累加导致集数虚高 + stat.seasonCounts[seasonNum] = Math.max(stat.seasonCounts[seasonNum], validCount); + if (seasonNum > stat.maxSeason) stat.maxSeason = seasonNum; if (seasonNum === 1) stat.s1Candidates.push({ anime: realAnime, originalCount: validCount }); }); @@ -2131,28 +2284,22 @@ function detectCollectionCandidates(curAnimes) { if (!hasOtherSources) continue; - const currentS1Total = stat.seasonCounts[1]; const threshold = maxOtherS1 + Thresholds.COLLECTION_DIFF; - const ratio = maxOtherS1 > 0 ? (currentS1Total / maxOtherS1) : 0; - if (currentS1Total > threshold) { - if (ratio > Thresholds.COLLECTION_RATIO) { - log("info", `[Merge-Check] [合集探测] 拒绝: [${source}] ${baseTitle} (Ratio too high: ${ratio.toFixed(2)} > ${Thresholds.COLLECTION_RATIO})`); - continue; - } - const giants = stat.s1Candidates.filter(c => c.originalCount > threshold); - if (giants.length > 0) { - giants.forEach(cand => { - collectionIds.add(cand.anime.animeId); - log("info", `[Merge-Check] [合集探测] 发现疑似合集(单体): [${source}] ${cand.anime.animeTitle} (Count:${cand.originalCount} > Thr:${threshold}, Ratio:${ratio.toFixed(2)}) -> 标记为合集`); - }); + // 独立评估每个候选条目，确认为大体积合集才进行标记，杜绝聚合误判 + stat.s1Candidates.forEach(cand => { + const ratio = maxOtherS1 > 0 ? (cand.originalCount / maxOtherS1) : 0; + if (cand.originalCount > threshold) { + if (ratio > Thresholds.COLLECTION_RATIO) { + log("info", `[Merge-Check] [合集探测] 拒绝: [${source}] ${cand.anime.animeTitle} (Ratio too high: ${ratio.toFixed(2)} > ${Thresholds.COLLECTION_RATIO})`); + return; + } + collectionIds.add(cand.anime.animeId); + log("info", `[Merge-Check] [合集探测] 发现疑似合集(单体): [${source}] ${cand.anime.animeTitle} (Count:${cand.originalCount} > Thr:${threshold}, Ratio:${ratio.toFixed(2)}) -> 标记为合集`); } else { - stat.s1Candidates.forEach(cand => { - collectionIds.add(cand.anime.animeId); - log("info", `[Merge-Check] [合集探测] 发现疑似合集(聚合): [${source}] ${cand.anime.animeTitle} (TotalS1:${currentS1Total} > Thr:${threshold}, Ratio:${ratio.toFixed(2)}) -> 标记为合集`); - }); + log("info", `[Merge-Check] [合集探测] 未命中: [${source}] ${cand.anime.animeTitle} (Count:${cand.originalCount} <= Threshold:${threshold})`); } - } else log("info", `[Merge-Check] [合集探测] 未命中: [${source}] ${baseTitle} (TotalS1:${currentS1Total} <= Threshold:${threshold})`); + }); } } return collectionIds; @@ -2175,7 +2322,7 @@ function detectCollectionCandidates(curAnimes) { * @param {Array} curAnimes - 待处理的番剧列表（将被原地修改） * @returns {Promise} */ -export async function applyMergeLogic(curAnimes, detailStore = null) { +export async function applyMergeLogic(curAnimes, detailStore = null) { const groups = globals.mergeSourcePairs; if (!groups || groups.length === 0) return; @@ -2188,7 +2335,7 @@ export async function applyMergeLogic(curAnimes, detailStore = null) { // 1. 合集探测 (前置计算) const collectionAnimeIds = detectCollectionCandidates(curAnimes); - + // 用于记录合集的使用进度，辅助切片推理 (Map) // 此进度对象在所有 Phase 间共享，确保 Phase 1 产生的进度能被 Phase 2 利用 const collectionProgress = new Map(); @@ -2201,20 +2348,20 @@ export async function applyMergeLogic(curAnimes, detailStore = null) { for (const group of groups) { if (group.secondaries.length === 0) continue; - + // 构建全局优先级地图，用于排序 (Primary=0, Sec1=1, Sec2=2...) // 优先级数值越小，代表在配置文件中越靠前，优先级越高 const sourcePriorityMap = new Map(); const fullPriorityList = [group.primary, ...group.secondaries]; fullPriorityList.forEach((src, idx) => sourcePriorityMap.set(src, idx)); - + const groupFingerprint = fullPriorityList.join('&'); const groupConsumedIds = new Set(); // 通用排序函数：源优先级 ASC > 媒体类型 (TV>Movie) > 季度编号 ASC (S1->S2) const sortCandidates = (list, phaseName) => { if (!list || list.length < 2) return list; - + log("info", `[Merge-Check] [Sort] ${phaseName} 排序前首个元素: ${list[0].animeTitle}`); list.sort((a, b) => { @@ -2230,11 +2377,11 @@ export async function applyMergeLogic(curAnimes, detailStore = null) { const markers = extractSeasonMarkers(anime.animeTitle, anime.typeDescription); if (markers.has('MOVIE')) return 2; if (markers.has('OVA') || markers.has('SP')) return 2; - + // 补充检查：防止 extractSeasonMarkers 漏网，使用严格类型判断 const strictType = getStrictMediaType(anime.animeTitle, anime.typeDescription); if (strictType === 'MOVIE') return 2; - + return 1; // 默认为 TV 正片季度 }; @@ -2253,7 +2400,7 @@ export async function applyMergeLogic(curAnimes, detailStore = null) { const sNum = getSeasonNumber(a.animeTitle, a.typeDescription) || 1; const typeLabel = (extractSeasonMarkers(a.animeTitle, a.typeDescription).has('MOVIE') || getStrictMediaType(a.animeTitle, a.typeDescription) === 'MOVIE') ? 'Movie' : `S${sNum}`; const pLevel = sourcePriorityMap.get(a.source) ?? '?'; - + return `[P${pLevel}] [${typeLabel}] [${a.source}] ${a.animeTitle}`; }); log("info", `[Merge-Check] [Sort] ${phaseName} 执行顺序:\n ${debugOrder.join('\n ')}`); @@ -2322,7 +2469,7 @@ export async function applyMergeLogic(curAnimes, detailStore = null) { const currentPriorityIdx = sourcePriorityMap.get(tAnime.source); const availableSecondaries = fullPriorityList.slice(currentPriorityIdx + 1); - + if (availableSecondaries.length === 0) continue; const resultAnime = await processMergeTask({ @@ -2391,7 +2538,7 @@ export async function applyMergeLogic(curAnimes, detailStore = null) { // 将所有合法衍生出来的合并对象推入主列表 if (newMergedAnimes.length > 0) { - for (const anime of newMergedAnimes) addAnime(anime, detailStore); + for (const anime of newMergedAnimes) addAnime(anime, detailStore); curAnimes.unshift(...newMergedAnimes); } @@ -2401,13 +2548,13 @@ export async function applyMergeLogic(curAnimes, detailStore = null) { if (globalConsumedIds.has(anime.animeId) && keepSources.has(anime.source)) globalConsumedIds.delete(anime.animeId); } } - + // 最终清理：移除已被任意一组所消费的单源原始资源 for (let i = curAnimes.length - 1; i >= 0; i--) { const item = curAnimes[i]; if (item._isMerged || globalConsumedIds.has(item.animeId)) curAnimes.splice(i, 1); } - + log("info", `[Merge] 合并执行完毕，最终列表数量: ${curAnimes.length}`); } @@ -2456,13 +2603,13 @@ export function alignSourceTimelines(results, sourceNames, realIds, dandanShifts } return targetStr.replace(/^https?:\/\/(www\.)?/, '').split('?')[0]; }; - + // 遍历其他来源进行对齐 for (let idx = 0; idx < results.length; idx++) { const sourceName = sourceNames[idx]; const realId = realIds[idx]; const list = results[idx]; - + if (sourceName === 'dandan' || !Array.isArray(list) || list.length === 0) { continue; } @@ -2486,13 +2633,13 @@ export function alignSourceTimelines(results, sourceNames, realIds, dandanShifts // 仅根据 dandan related 接口提供的精确偏移量进行对齐，未命中则跳过 if (appliedShift !== undefined) { log("info", `[Merge][AlignTimeline] ${sourceName}:${realId} 应用 dandan API 精确偏移量 ${appliedShift}s`); - + for (let i = 0; i < list.length; i++) { const danmu = list[i]; const time = getDanmuTime(danmu); // 原本 time 是基准，shift表示加上该偏差到达 Dandan 时间轴 const targetTime = Math.max(0, time + appliedShift); - + // 替换位移 if (danmu.p && typeof danmu.p === 'string') { const firstComma = danmu.p.indexOf(','); @@ -2511,7 +2658,7 @@ export function alignSourceTimelines(results, sourceNames, realIds, dandanShifts log("info", `[Merge][AlignTimeline] ${sourceName}:${realId} 未找到相关 API 偏移量数据，跳过时间轴对齐`); } } - + // 直接返回修改后的原始数组，不再产生深拷贝开销 return results; }