@chilfish/gallery-dl-instagram 0.1.0

package/instagram/extractors.ts

@@ -0,0 +1,521 @@
+/**
+ * Instagram sub-extractors — one class per resource type.
+ */
+
+import type { ExtractorClass, MessageIter, Metadata } from '../types'
+import type { InstagramExtractorOptions } from './base'
+import type { InstagramPost, InstagramUser } from './types'
+import { directory, queue } from '../message'
+import { shortcodeFromId } from '../utils/id-codec'
+import { ensureHttpScheme } from '../utils/text'
+import { InstagramExtractor } from './base'
+
+const BASE_RE = /^(?:https?:\/\/)?(?:www\.)?instagram\.com/
+
+function re(base: RegExp, path: RegExp | string): RegExp {
+  const pathSrc = typeof path === 'string' ? path : path.source
+  return new RegExp(base.source + pathSrc, 'i')
+}
+
+export class InstagramPostExtractor extends InstagramExtractor {
+  static readonly subcategory = 'post'
+  static override pattern = re(
+    /^(?:https?:\/\/)?(?:www\.)?instagram\.com\//,
+    /(?:share(?:\/(?:p|tv|reels?))?|(?:[^/?#]+\/)?(?:p|tv|reels?))\/([^/?#]+)/,
+  )
+
+  override readonly subcategory = InstagramPostExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+    if (opts.match[2] != null || opts.match[3] != null) {
+      (this as { subcategory: string }).subcategory = 'reel'
+    }
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramPostExtractor | null {
+    const match = InstagramPostExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramPostExtractor({ ...opts, url, match })
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    const groups = this.groups as string[]
+    let shortcode: string | undefined = groups[0]
+    if (!shortcode)
+      return
+
+    if (groups[1] === '') {
+      // Share link — follow redirect
+      this.log.info(`Resolving share link: ${this.url}`)
+      const resp = await this.request(ensureHttpScheme(this.url), {
+        headers: {
+          'Sec-Fetch-Dest': 'empty',
+          'Sec-Fetch-Mode': 'navigate',
+          'Sec-Fetch-Site': 'same-origin',
+        },
+      })
+      const parts = (resp as { url?: string }).url?.split('/')
+      shortcode = parts?.[parts.length - 2] ?? shortcode
+    }
+
+    this.log.debug(`Fetching post: ${shortcode}`)
+    yield* this.api.media(shortcode)
+  }
+}
+
+export class InstagramUserExtractor extends InstagramExtractor {
+  static readonly subcategory = 'user'
+  static override pattern = re(BASE_RE, /(\/[^/?#]+)\/?(?:$|[?#])/)
+
+  override readonly subcategory = InstagramUserExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramUserExtractor | null {
+    const match = InstagramUserExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramUserExtractor({ ...opts, url, match })
+  }
+
+  override async* items(): MessageIter {
+    await this.login()
+
+    const userPath = this.groups[0] ?? '/'
+    const base = `${this.root}${userPath}/`
+    const storiesUrl = `${this.root}/stories/${userPath.slice(1)}/`
+
+    const include = this._cfg('include', ['posts']) as string[] | string
+    const categories: string[]
+      = include === 'all'
+        ? [
+            'posts',
+            'reels',
+            'tagged',
+            'stories',
+            'highlights',
+            'info',
+            'avatar',
+          ]
+        : typeof include === 'string'
+          ? include.replace(/\s+/g, '').split(',')
+          : include
+
+    const extractors: Record<
+      string,
+      { cls: ExtractorClass, url: string }
+    > = {
+      info: {
+        cls: InstagramInfoExtractor as unknown as ExtractorClass,
+        url: `${base}info/`,
+      },
+      avatar: {
+        cls: InstagramAvatarExtractor as unknown as ExtractorClass,
+        url: `${base}avatar/`,
+      },
+      stories: {
+        cls: InstagramStoriesExtractor as unknown as ExtractorClass,
+        url: storiesUrl,
+      },
+      highlights: {
+        cls: InstagramHighlightsExtractor as unknown as ExtractorClass,
+        url: `${base}highlights/`,
+      },
+      posts: {
+        cls: InstagramPostsExtractor as unknown as ExtractorClass,
+        url: `${base}posts/`,
+      },
+      reels: {
+        cls: InstagramReelsExtractor as unknown as ExtractorClass,
+        url: `${base}reels/`,
+      },
+      tagged: {
+        cls: InstagramTaggedExtractor as unknown as ExtractorClass,
+        url: `${base}tagged/`,
+      },
+    }
+
+    for (const cat of categories) {
+      const entry = extractors[cat]
+      if (entry) {
+        yield queue(entry.url, { _extractor: entry.cls })
+      }
+      else {
+        this.log.warn(`Invalid include '${cat}'`)
+      }
+    }
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {}
+}
+
+export class InstagramPostsExtractor extends InstagramExtractor {
+  static readonly subcategory = 'posts'
+  static override pattern = re(BASE_RE, /(\/[^/?#]+)\/posts/)
+
+  override readonly subcategory = InstagramPostsExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramPostsExtractor | null {
+    const match = InstagramPostsExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramPostsExtractor({ ...opts, url, match })
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    const screenName = (this.groups[0] ?? '').replace(/^\//, '')
+    const uid = await this.api.userId(screenName)
+    yield* this.api.userFeed(uid)
+  }
+}
+
+export class InstagramReelsExtractor extends InstagramExtractor {
+  static readonly subcategory = 'reels'
+  static override pattern = re(BASE_RE, /(\/[^/?#]+)\/reels/)
+
+  override readonly subcategory = InstagramReelsExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramReelsExtractor | null {
+    const match = InstagramReelsExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramReelsExtractor({ ...opts, url, match })
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    const screenName = (this.groups[0] ?? '').replace(/^\//, '')
+    const uid = await this.api.userId(screenName)
+    yield* this.api.userClips(uid)
+  }
+}
+
+export class InstagramTaggedExtractor extends InstagramExtractor {
+  static readonly subcategory = 'tagged'
+  static override pattern = re(BASE_RE, /(\/[^/?#]+)\/tagged/)
+
+  override readonly subcategory = InstagramTaggedExtractor.subcategory
+  private _taggedUserId = ''
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramTaggedExtractor | null {
+    const match = InstagramTaggedExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramTaggedExtractor({ ...opts, url, match })
+  }
+
+  override async metadata(): Promise<Record<string, unknown>> {
+    const screenName = (this.groups[0] ?? '').replace(/^\//, '')
+    let user: InstagramUser
+
+    if (screenName.startsWith('id:')) {
+      this._taggedUserId = screenName.slice(3)
+      user = await this.api.userById(screenName.slice(3))
+    }
+    else {
+      this._taggedUserId = await this.api.userId(screenName)
+      user = await this.api.userByScreenName(screenName)
+    }
+
+    return {
+      tagged_owner_id: user.id ?? user.pk,
+      tagged_username: user.username,
+      tagged_full_name: user.full_name,
+    }
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    if (!this._taggedUserId)
+      await this.metadata()
+    yield* this.api.userTagged(this._taggedUserId)
+  }
+}
+
+export class InstagramStoriesExtractor extends InstagramExtractor {
+  static readonly subcategory = 'stories'
+  static override pattern
+    = /^(?:https?:\/\/)?(?:www\.)?instagram\.com\/(?:stories\/(?:highlights\/(\d+)|([^/?#]+)(?:\/(\d+))?)|\/(aGlnaGxpZ2h0[^?#]+)(?:\?story_media_id=(\d+))?)/
+
+  override readonly subcategory = InstagramStoriesExtractor.subcategory
+  private highlightId: string | null = null
+  private mediaId: string | null = null
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+    const groups = this.groups as string[]
+    const h1 = groups[0]
+    const user = groups[1]
+    const m1 = groups[2]
+    const h2 = groups[3]
+    const m2 = groups[4]
+
+    if (user) {
+      (this as { subcategory: string }).subcategory = 'stories'
+      this.highlightId = null
+    }
+    else {
+      (this as { subcategory: string }).subcategory = 'highlights'
+      this.highlightId = h1
+        ? `highlight:${h1}`
+        : `highlight:${Buffer.from(h2 ?? '', 'base64').toString('utf-8')}`
+    }
+
+    this.mediaId = m1 ?? m2 ?? null
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramStoriesExtractor | null {
+    const match = InstagramStoriesExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramStoriesExtractor({ ...opts, url, match })
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    const reelId = this.highlightId
+      ? this.highlightId
+      : await this.api.userId((this.groups[1] ?? '').toString())
+
+    const reels = await this.api.reelsMedia([reelId])
+    if (!reels.length)
+      return
+
+    if (this.mediaId) {
+      const reel = reels[0]!
+      for (const item of reel.items ?? []) {
+        if (item.pk === this.mediaId) {
+          reel.items = [item]
+          break
+        }
+      }
+      yield reel
+      return
+    }
+
+    const split = this._cfg('split', false) as boolean
+    if (split) {
+      const reel = reels[0]!
+      for (const item of reel.items ?? []) {
+        const copy = { ...reel }
+        copy.items = [item]
+        yield copy
+      }
+    }
+    else {
+      yield* reels
+    }
+  }
+}
+
+export class InstagramHighlightsExtractor extends InstagramExtractor {
+  static readonly subcategory = 'highlights'
+  static override pattern = re(BASE_RE, /(\/[^/?#]+)\/highlights/)
+
+  override readonly subcategory = InstagramHighlightsExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramHighlightsExtractor | null {
+    const match = InstagramHighlightsExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramHighlightsExtractor({ ...opts, url, match })
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    const screenName = (this.groups[0] ?? '').replace(/^\//, '')
+    const uid = await this.api.userId(screenName)
+    yield* this.api.highlightsMedia(uid)
+  }
+}
+
+export class InstagramTagExtractor extends InstagramExtractor {
+  static readonly subcategory = 'tag'
+  static override pattern = re(BASE_RE, /\/explore\/tags\/([^/?#]+)/)
+
+  override readonly subcategory = InstagramTagExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramTagExtractor | null {
+    const match = InstagramTagExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramTagExtractor({ ...opts, url, match })
+  }
+
+  override async metadata(): Promise<Record<string, unknown>> {
+    const tag = this.groups[0] ?? ''
+    return { tag: decodeURIComponent(tag) }
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    const tag = this.groups[0] ?? ''
+    yield* this.api.tagsMedia(decodeURIComponent(tag))
+  }
+}
+
+export class InstagramInfoExtractor extends InstagramExtractor {
+  static readonly subcategory = 'info'
+  static override pattern = re(BASE_RE, /(\/[^/?#]+)\/info/)
+
+  override readonly subcategory = InstagramInfoExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramInfoExtractor | null {
+    const match = InstagramInfoExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramInfoExtractor({ ...opts, url, match })
+  }
+
+  override async* items(): MessageIter {
+    const screenName = (this.groups[0] ?? '').replace(/^\//, '')
+    let user: InstagramUser
+
+    if (screenName.startsWith('id:')) {
+      user = await this.api.userById(screenName.slice(3))
+    }
+    else {
+      user = await this.api.userByScreenName(screenName)
+    }
+
+    yield directory(user as unknown as Metadata)
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {}
+}
+
+export class InstagramAvatarExtractor extends InstagramExtractor {
+  static readonly subcategory = 'avatar'
+  static override pattern = re(BASE_RE, /(\/[^/?#]+)\/avatar/)
+
+  override readonly subcategory = InstagramAvatarExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramAvatarExtractor | null {
+    const match = InstagramAvatarExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramAvatarExtractor({ ...opts, url, match })
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    const screenName = (this.groups[0] ?? '').replace(/^\//, '')
+    let user: InstagramUser
+
+    if (screenName.startsWith('id:')) {
+      user = await this.api.userById(screenName.slice(3))
+    }
+    else {
+      user = await this.api.userByScreenName(screenName)
+    }
+
+    const avatar
+      = user.hd_profile_pic_url_info
+        ?? user.hd_profile_pic_versions?.[user.hd_profile_pic_versions.length - 1]
+        ?? { url: user.profile_pic_url ?? '', width: 0, height: 0 }
+
+    let pk = user.profile_pic_id?.split('_')[0]
+    let code: string
+
+    if (pk) {
+      code = shortcodeFromId(pk)
+    }
+    else {
+      pk = `avatar:${user.pk}`
+      code = pk
+    }
+
+    yield {
+      pk,
+      code,
+      user,
+      caption: null,
+      like_count: 0,
+      image_versions2: { candidates: [avatar] },
+    } as unknown as InstagramPost
+  }
+}
+
+export class InstagramSavedExtractor extends InstagramExtractor {
+  static readonly subcategory = 'saved'
+  static override pattern = re(
+    BASE_RE,
+    /(\/[^/?#]+)\/saved(?:\/all-posts)?\/?$/,
+  )
+
+  override readonly subcategory = InstagramSavedExtractor.subcategory
+
+  constructor(opts: InstagramExtractorOptions) {
+    super(opts)
+  }
+
+  static fromURL(
+    url: string,
+    opts: InstagramExtractorOptions,
+  ): InstagramSavedExtractor | null {
+    const match = InstagramSavedExtractor.pattern.exec(url)
+    if (!match)
+      return null
+    return new InstagramSavedExtractor({ ...opts, url, match })
+  }
+
+  override async* posts(): AsyncGenerator<InstagramPost> {
+    yield* this.api.userSaved()
+  }
+}

package/instagram/index.ts

@@ -0,0 +1,43 @@
+/**
+ * Instagram module — public API surface.
+ */
+
+// API client
+export { InstagramRestAPI } from './api'
+
+// Base extractor
+export { InstagramExtractor } from './base'
+export type { InstagramExtractorOptions } from './base'
+
+// Extractors
+export {
+  InstagramAvatarExtractor,
+  InstagramHighlightsExtractor,
+  InstagramInfoExtractor,
+  InstagramPostExtractor,
+  InstagramPostsExtractor,
+  InstagramReelsExtractor,
+  InstagramSavedExtractor,
+  InstagramStoriesExtractor,
+  InstagramTagExtractor,
+  InstagramTaggedExtractor,
+  InstagramUserExtractor,
+} from './extractors'
+
+// Parsers
+export { extractAudio, extractTaggedUsers, parsePostGraphql, parsePostRest } from './parsers'
+
+// Types
+export type {
+  Coauthor,
+  ImageCandidate,
+  InstagramCarouselItem,
+  InstagramLocation,
+  InstagramPost,
+  InstagramUser,
+  ParsedMedia,
+  ParsedPost,
+  ParserConfig,
+  TaggedUser,
+  VideoVersion,
+} from './types'