feat: DSL 磁力聚合 — 搜索/解析/预览/投递 + cilisousuo 源 + 中英双语 + Docker 部署

2026-10-16 18:53:45 +08:00
commit d6999104bf
64 changed files with 16524 additions and 0 deletions
@@ -0,0 +1,62 @@
+import { DEFAULT_TRACKERS } from '../torrent'
+import { DEFAULT_UA, httpJson } from '../http'
+import type { ProviderContext, SearchProvider, SearchResult } from './types'
+
+/** The Pirate Bay 分类编号 → 可读标签（只保留常见项） */
+const TPB_CATEGORIES: Record<string, string> = {
+  '101': '音乐', '102': '有声书', '103': '视频', '104': '电影', '105': '剧集', '106': 'MV', '199': '其他',
+  '201': '软件', '202': '安卓', '203': 'Mac', '204': '游戏', '205': 'PC游戏', '206': 'Mac游戏', '207': '其他',
+  '301': '视频', '302': '视频', '303': '图片', '304': '其他', '399': '其他',
+  '401': '游戏', '402': 'PC游戏', '403': 'Mac游戏', '404': '其他',
+  '501': '其他', '502': '电子书', '503': '漫画', '504': '图片', '505': '其他', '599': '其他',
+  '601': '其他', '602': '电子书', '603': '音频', '604': '视频', '605': '其他', '699': '其他'
+}
+
+function apibayBase() {
+  return (process.env.APIBAY_BASE || 'https://apibay.org').replace(/\/+$/, '')
+}
+
+export const apibayProvider: SearchProvider = {
+  id: 'apibay',
+  label: 'The Pirate Bay',
+  homepage: 'https://thepiratebay.org',
+
+  async search(query: string, ctx: ProviderContext): Promise<SearchResult[]> {
+    const url = `${apibayBase()}/q.php?q=${encodeURIComponent(query)}&cat=0`
+    const raw = await httpJson<any[]>(url, {
+      proxy: ctx.proxy,
+      timeoutMs: ctx.timeoutMs,
+      headers: { accept: 'application/json', 'user-agent': DEFAULT_UA }
+    })
+    if (!Array.isArray(raw)) return []
+
+    return raw
+      .filter((it) => {
+        if (!it || typeof it.info_hash !== 'string') return false
+        if (it.info_hash === '0000000000000000000000000000000000000000') return false
+        if (typeof it.name === 'string' && /no results returned/i.test(it.name)) return false
+        return true
+      })
+      .map((it) => {
+        const infoHash = String(it.info_hash).toLowerCase()
+        const name = String(it.name || infoHash)
+        return {
+          infoHash,
+          name,
+          size: Number(it.size) || 0,
+          seeders: Number(it.seeders) || 0,
+          leechers: Number(it.leechers) || 0,
+          fileCount: Number(it.num_files) || null,
+          category: TPB_CATEGORIES[String(it.category)] || (it.category ? `分类 ${it.category}` : null),
+          source: 'apibay',
+          sourceLabel: 'The Pirate Bay',
+          sourceId: it.id ? String(it.id) : null,
+          sourceUrl: it.id ? `https://thepiratebay.org/description.php?id=${it.id}` : null,
+          addedAt: it.added ? new Date(Number(it.added) * 1000) : null,
+          magnet: `magnet:?xt=urn:btih:${infoHash}&dn=${encodeURIComponent(name)}${DEFAULT_TRACKERS.map(
+            (t) => `&tr=${encodeURIComponent(t)}`
+          ).join('')}`
+        } satisfies SearchResult
+      })
+  }
+}
@@ -0,0 +1,145 @@
+import { buildMagnet } from '../torrent'
+import { DEFAULT_UA, httpText } from '../http'
+import type { ProviderContext, SearchProvider, SearchResult } from './types'
+
+const HOMEPAGE = 'https://cilisousuo.cc'
+
+const SIZE_UNITS: Record<string, number> = {
+  b: 1,
+  kb: 1024,
+  kib: 1024,
+  mb: 1024 ** 2,
+  mib: 1024 ** 2,
+  gb: 1024 ** 3,
+  gib: 1024 ** 3,
+  tb: 1024 ** 4,
+  tib: 1024 ** 4
+}
+
+/** "330.3MB" / "1.1TB" → 字节数 */
+function parseSize(input: unknown): number {
+  const text = String(input || '').trim()
+  const m = text.match(/^([\d.]+)\s*([a-zA-Z]+)?$/)
+  if (!m) return 0
+  const value = Number(m[1]) || 0
+  const unit = (m[2] || 'b').toLowerCase()
+  return Math.round(value * (SIZE_UNITS[unit] ?? 1))
+}
+
+function cilisousuoBase() {
+  return (process.env.CILISOUSUO_BASE || HOMEPAGE).replace(/\/+$/, '')
+}
+
+/** 去掉 HTML 标签并解码常见实体 */
+function stripHtml(input: string): string {
+  return input
+    .replace(/<[^>]+>/g, '')
+    .replace(/&nbsp;/g, ' ')
+    .replace(/&amp;/g, '&')
+    .replace(/&lt;/g, '<')
+    .replace(/&gt;/g, '>')
+    .replace(/&#0?39;|&apos;/g, "'")
+    .replace(/&quot;/g, '"')
+    .replace(/\s+/g, ' ')
+    .trim()
+}
+
+interface ListItem {
+  title: string
+  filename: string
+  size: number
+  code: string
+}
+
+/** 从列表页解析出条目（列表页只含名称/大小与 /magnet/{code}，infoHash 在详情页） */
+function parseList(html: string): ListItem[] {
+  const items: ListItem[] = []
+  const liRe = /<li class="item">([\s\S]*?)<\/li>/g
+  let m: RegExpExecArray | null
+  while ((m = liRe.exec(html))) {
+    const block = m[1]
+    const titleM = block.match(/<div class="result-title">([\s\S]*?)<\/div>/)
+    const fileM = block.match(/<div class="filename">([\s\S]*?)<\/div>/)
+    const sizeM = block.match(/<div class="size">([\s\S]*?)<\/div>/)
+    const codeM = block.match(/href="\/magnet\/([^"]+)"/)
+    if (!codeM) continue
+    const title = titleM ? stripHtml(titleM[1]) : ''
+    const filename = fileM ? stripHtml(fileM[1]) : ''
+    items.push({
+      title,
+      filename,
+      size: sizeM ? parseSize(sizeM[1]) : 0,
+      code: codeM[1]
+    })
+  }
+  return items
+}
+
+/** 并发受限地映射 */
+async function mapWithConcurrency<T, R>(items: T[], concurrency: number, fn: (item: T) => Promise<R>): Promise<R[]> {
+  const out: R[] = new Array(items.length)
+  let cursor = 0
+  async function worker() {
+    while (cursor < items.length) {
+      const i = cursor++
+      out[i] = await fn(items[i])
+    }
+  }
+  await Promise.all(Array.from({ length: Math.min(concurrency, items.length) }, () => worker()))
+  return out
+}
+
+async function fetchDetail(it: ListItem, ctx: ProviderContext): Promise<SearchResult | null> {
+  const url = `${cilisousuoBase()}/magnet/${it.code}`
+  const html = await httpText(url, {
+    proxy: ctx.proxy,
+    timeoutMs: Math.min(ctx.timeoutMs, 8000),
+    headers: { accept: 'text/html,application/xhtml+xml', 'user-agent': DEFAULT_UA }
+  })
+
+  const hashM = html.match(/xt=urn:btih:([a-f0-9]{40})/i)
+  if (!hashM) return null
+  const infoHash = hashM[1].toLowerCase()
+  const fileCountM = html.match(/文件数量[^\d]*<dd>\s*(\d+)/)
+  const fileCount = fileCountM ? Number(fileCountM[1]) : null
+  const name = it.title || it.filename || infoHash
+
+  return {
+    infoHash,
+    name,
+    size: it.size,
+    seeders: null,
+    leechers: null,
+    fileCount,
+    category: null,
+    source: 'cilisousuo',
+    sourceLabel: '磁力搜索',
+    sourceId: it.code,
+    sourceUrl: `${cilisousuoBase()}/magnet/${it.code}`,
+    addedAt: null,
+    magnet: buildMagnet(infoHash, name)
+  } satisfies SearchResult
+}
+
+export const cilisousuoProvider: SearchProvider = {
+  id: 'cilisousuo',
+  label: '磁力搜索',
+  homepage: HOMEPAGE,
+
+  async search(query: string, ctx: ProviderContext): Promise<SearchResult[]> {
+    const url = `${cilisousuoBase()}/search?q=${encodeURIComponent(query)}`
+    const html = await httpText(url, {
+      proxy: ctx.proxy,
+      timeoutMs: ctx.timeoutMs,
+      headers: { accept: 'text/html,application/xhtml+xml', 'user-agent': DEFAULT_UA }
+    })
+
+    const items = parseList(html)
+    if (!items.length) return []
+
+    // 列表页不含 infoHash，逐条抓详情页补全磁力链；最多取前 24 条并发抓取
+    const capped = items.slice(0, 24)
+    const results = await mapWithConcurrency(capped, 5, (it) => fetchDetail(it, ctx).catch(() => null))
+    return results.filter((x): x is SearchResult => x !== null)
+  }
+}
@@ -0,0 +1,86 @@
+import { loadSettings } from '../appconfig'
+import { apibayProvider } from './apibay'
+import { cilisousuoProvider } from './cilisousuo'
+import { nyaaProvider } from './nyaa'
+import { solidtorrentsProvider } from './solidtorrents'
+import type { ProviderReport, SearchProvider, SearchResult } from './types'
+
+export const PROVIDERS: SearchProvider[] = [apibayProvider, nyaaProvider, solidtorrentsProvider, cilisousuoProvider]
+
+export function listProviders() {
+  return PROVIDERS.map((p) => ({ id: p.id, label: p.label, homepage: p.homepage }))
+}
+
+function friendlyError(err: unknown): string {
+  const msg = err instanceof Error ? err.message : String(err)
+  if (/fetch failed|ENOTFOUND|EAI_AGAIN|ECONNREFUSED|ECONNRESET|ETIMEDOUT|socket hang up/i.test(msg)) {
+    return '网络不可达，通常是站点被墙，请在「设置」里配置 HTTP 代理后重试'
+  }
+  if (/aborted|超时/.test(msg)) return msg
+  if (/HTTP 5\d\d/.test(msg)) return `源站异常：${msg}`
+  if (/HTTP 4\d\d/.test(msg)) return `源站拒绝访问：${msg}`
+  return msg
+}
+
+export interface RunSearchOptions {
+  providers?: string[]
+  limit?: number
+}
+
+export interface RunSearchResult {
+  results: SearchResult[]
+  reports: ProviderReport[]
+}
+
+export async function runSearch(query: string, options: RunSearchOptions = {}): Promise<RunSearchResult> {
+  const settings = await loadSettings()
+  const wanted = options.providers?.length ? options.providers : settings.enabledProviders
+  const targets = PROVIDERS.filter((p) => wanted.includes(p.id))
+  const ctx = { proxy: settings.searchProxy || null, timeoutMs: settings.searchTimeout }
+
+  const reports = await Promise.all(
+    targets.map(async (provider): Promise<ProviderReport & { results: SearchResult[] }> => {
+      const started = Date.now()
+      try {
+        const results = await provider.search(query, ctx)
+        return {
+          id: provider.id,
+          label: provider.label,
+          ok: true,
+          count: results.length,
+          ms: Date.now() - started,
+          error: null,
+          results
+        }
+      } catch (err) {
+        return {
+          id: provider.id,
+          label: provider.label,
+          ok: false,
+          count: 0,
+          ms: Date.now() - started,
+          error: friendlyError(err),
+          results: []
+        }
+      }
+    })
+  )
+
+  // 按 infoHash 去重，保留做种数更高的那条
+  const merged = new Map<string, SearchResult>()
+  for (const report of reports) {
+    for (const item of report.results) {
+      const exists = merged.get(item.infoHash)
+      if (!exists || (item.seeders ?? 0) > (exists.seeders ?? 0)) merged.set(item.infoHash, item)
+    }
+  }
+
+  const results = [...merged.values()].sort(
+    (a, b) => (b.seeders ?? 0) - (a.seeders ?? 0) || b.size - a.size
+  )
+
+  return {
+    results: options.limit ? results.slice(0, options.limit) : results,
+    reports: reports.map(({ results: _omit, ...rest }) => rest)
+  }
+}
@@ -0,0 +1,94 @@
+import { XMLParser } from 'fast-xml-parser'
+import { DEFAULT_TRACKERS } from '../torrent'
+import { DEFAULT_UA, httpText } from '../http'
+import type { ProviderContext, SearchProvider, SearchResult } from './types'
+
+const SIZE_UNITS: Record<string, number> = {
+  b: 1,
+  kb: 1024,
+  kib: 1024,
+  mb: 1024 ** 2,
+  mib: 1024 ** 2,
+  gb: 1024 ** 3,
+  gib: 1024 ** 3,
+  tb: 1024 ** 4,
+  tib: 1024 ** 4
+}
+
+/** "1.23 GiB" / "456 MB" → 字节数 */
+function parseSize(input: unknown): number {
+  if (typeof input === 'number') return input
+  const text = String(input || '').trim()
+  const m = text.match(/^([\d.]+)\s*([a-zA-Z]+)?$/)
+  if (!m) return 0
+  const value = Number(m[1]) || 0
+  const unit = (m[2] || 'b').toLowerCase()
+  return Math.round(value * (SIZE_UNITS[unit] ?? 1))
+}
+
+function nyaaBase() {
+  return (process.env.NYAA_BASE || 'https://nyaa.si').replace(/\/+$/, '')
+}
+
+function asArray<T>(value: T | T[] | undefined): T[] {
+  if (value === undefined || value === null) return []
+  return Array.isArray(value) ? value : [value]
+}
+
+function pickString(value: unknown): string {
+  if (value === undefined || value === null) return ''
+  if (typeof value === 'object' && '#text' in (value as Record<string, unknown>)) {
+    return String((value as Record<string, unknown>)['#text'] ?? '')
+  }
+  return String(value)
+}
+
+export const nyaaProvider: SearchProvider = {
+  id: 'nyaa',
+  label: 'Nyaa',
+  homepage: 'https://nyaa.si',
+
+  async search(query: string, ctx: ProviderContext): Promise<SearchResult[]> {
+    // c=0_0 全部分类；f=0 只过滤可信发布者之外的全部结果
+    const url = `${nyaaBase()}/?page=rss&q=${encodeURIComponent(query)}&c=0_0&f=0`
+    const xml = await httpText(url, {
+      proxy: ctx.proxy,
+      timeoutMs: ctx.timeoutMs,
+      headers: { accept: 'application/rss+xml, application/xml, text/xml', 'user-agent': DEFAULT_UA }
+    })
+
+    const parser = new XMLParser({ removeNSPrefix: true, trimValues: true, parseTagValue: false })
+    const doc = parser.parse(xml) as Record<string, any>
+    const items = asArray(doc?.rss?.channel?.item)
+
+    return items
+      .map((item: Record<string, any>) => {
+        const infoHash = pickString(item.infoHash).toLowerCase()
+        if (!/^[a-f0-9]{40}$/.test(infoHash)) return null
+        const name = pickString(item.title) || infoHash
+        const torrentUrl = pickString(item.guid) || pickString(item.link)
+        const category = pickString(item.category)
+        const pubDate = pickString(item.pubDate)
+        const added = pubDate ? new Date(pubDate) : null
+
+        return {
+          infoHash,
+          name,
+          size: parseSize(item.size),
+          seeders: Number(pickString(item.seeders)) || 0,
+          leechers: Number(pickString(item.leechers)) || 0,
+          fileCount: null,
+          category: category || null,
+          source: 'nyaa',
+          sourceLabel: 'Nyaa',
+          sourceId: pickString(item.link) || null,
+          sourceUrl: torrentUrl || null,
+          addedAt: added && !Number.isNaN(added.getTime()) ? added : null,
+          magnet: `magnet:?xt=urn:btih:${infoHash}&dn=${encodeURIComponent(name)}${DEFAULT_TRACKERS.map(
+            (t) => `&tr=${encodeURIComponent(t)}`
+          ).join('')}`
+        } satisfies SearchResult
+      })
+      .filter((item): item is SearchResult => item !== null)
+  }
+}
@@ -0,0 +1,42 @@
+import { DEFAULT_TRACKERS } from '../torrent'
+import { DEFAULT_UA, httpJson } from '../http'
+import type { ProviderContext, SearchProvider, SearchResult } from './types'
+
+export const solidtorrentsProvider: SearchProvider = {
+  id: 'solidtorrents',
+  label: 'SolidTorrents',
+  homepage: 'https://solidtorrents.net',
+
+  async search(query: string, ctx: ProviderContext): Promise<SearchResult[]> {
+    const url = `https://solidtorrents.net/api/v1/search?q=${encodeURIComponent(query)}&category=all&sort=seeders`
+    const raw = await httpJson<any>(url, {
+      proxy: ctx.proxy,
+      timeoutMs: ctx.timeoutMs,
+      headers: { accept: 'application/json', 'user-agent': DEFAULT_UA }
+    })
+
+    if (!raw || !Array.isArray(raw.results)) return []
+
+    return raw.results.map((it: any) => {
+      const infoHash = String(it.infohash).toLowerCase()
+      const name = String(it.title || infoHash)
+      return {
+        infoHash,
+        name,
+        size: Number(it.size) || 0,
+        seeders: Number(it.swarm?.seeders) || 0,
+        leechers: Number(it.swarm?.leechers) || 0,
+        fileCount: null, // API doesn't seem to provide this in search list
+        category: it.category || null,
+        source: 'solidtorrents',
+        sourceLabel: 'SolidTorrents',
+        sourceId: null,
+        sourceUrl: `https://solidtorrents.to/view/${infoHash}`,
+        addedAt: it.createdAt ? new Date(it.createdAt) : null,
+        magnet: `magnet:?xt=urn:btih:${infoHash}&dn=${encodeURIComponent(name)}${DEFAULT_TRACKERS.map(
+          (t) => `&tr=${encodeURIComponent(t)}`
+        ).join('')}`
+      } satisfies SearchResult
+    })
+  }
+}
@@ -0,0 +1,37 @@
+export interface SearchResult {
+  infoHash: string
+  name: string
+  size: number
+  seeders: number | null
+  leechers: number | null
+  fileCount: number | null
+  category: string | null
+  source: string
+  sourceLabel: string
+  sourceId: string | null
+  sourceUrl: string | null
+  addedAt: Date | null
+  magnet: string
+}
+
+export interface ProviderContext {
+  proxy: string | null
+  timeoutMs: number
+}
+
+export interface SearchProvider {
+  id: string
+  label: string
+  /** 该站点所在的站点根地址，便于 UI 上提示 */
+  homepage: string
+  search(query: string, ctx: ProviderContext): Promise<SearchResult[]>
+}
+
+export interface ProviderReport {
+  id: string
+  label: string
+  ok: boolean
+  count: number
+  ms: number
+  error: string | null
+}