pinterest-dl 2.0.3 → 2.0.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2ba41b3cdd44c6463cfffde77fc4b9db8185f47b1d62e3a9555b9408bfd4f197
-  data.tar.gz: b032e121ffb9750ed128b0a0c6ce5d0266dea8db4fbc99ee1bd6579e8f680d1a
+  metadata.gz: d14e37c46d9bf7d3b8851c433a430da6811a040151e1103912b300985569b038
+  data.tar.gz: f19617c2661b7cb970206eabe8a6fb1471dfa9045f09ebf2e0393bfebdc83e5c
 SHA512:
-  metadata.gz: 11fe875c2a1eabda28fb7f7b4f342b33565dc141b5fd7ff06cac800595a834c9156abd9fe61544f75eed944f0553fc180183cad5ac7fdef6ca13a03023e179ae
-  data.tar.gz: dc3600f15608c7717f2010cceef198e636e423b3788ee5763ff4e49fdfd19f49cef545e5206d1a7a0981d658194a97745aecbc0f373da038062f381b2ad74ec6
+  metadata.gz: 8ac5b7a9d1525a24dee57a7857cc345d8febb84061cb82b018d507c35c03d04e2d214cc108584dcfb56208d6f05f524a74fa427faa9ffd1b25e34314ab597eae
+  data.tar.gz: 565a098363fbf28418ef5312972bf2e4dec08d8ade89cad70a59c425bc1a25f052c56eb4c335e48b15237097db9487c833384d4ede470acd685cec986ed98e82

data/CHANGELOG.md

@@ -3,7 +3,32 @@
 All notable changes to this project are documented here.
 This project follows [Semantic Versioning](https://semver.org/).
 
-## [2.0.2]
+## [2.0.5]
+
+### Fixed
+- Pin URLs on country domains (e.g. `de.pinterest.com/pin/...`) didn't download
+  while `pin.it` links did. Pin URLs are now normalised to
+  `www.pinterest.com/pin/ID/`, and the HTTP client follows redirects itself
+  (max 5) with a small cookie jar, so a redirect loop can no longer happen.
+- `Client#raw_get` returned the first redirect response instead of the final
+  one, which lost the session cookie / CSRF token.
+- Pin extraction could pick a related pin instead of the pin you asked for.
+  The pin is now matched by its id.
+- Video pins returned "No video found". The best `.mp4` is picked now (not just
+  the first format, which could be an `.m3u8`), story pins are supported, and
+  `og:video` meta tags are read in any attribute order.
+- `board` never worked: Pinterest needs the real board id, we sent an empty
+  one. The id is now looked up first, the session is created once instead of
+  once per page, links with `?invite_code=...` are parsed correctly, and
+  `pin.it` links work. Sections use `BoardSectionPinsResource`.
+
+### Added
+- `--type all|image|video|pin` for `search` and `board`
+  (`PinterestDL.search(..., type: :video)`, `PinterestDL.board(..., type: :image)`).
+  `pin` prints the pin links (saved to `pins.txt` with `-o`).
+- With `-o` and type `all`, video pins are now saved as videos, not thumbnails.
+
+## [2.0.3]
 
 ### Fixed
 - **`PinterestDL::Board` was broken.** `lib/pinterest_dl/board.rb` accidentally
@@ -11,6 +36,11 @@ This project follows [Semantic Versioning](https://semver.org/).
   search logic), so it silently collided with `PinterestDL::Search` at load
   time and `PinterestDL.board(...)` never worked correctly. Restored the
   correct `Board` class backed by Pinterest's `BoardFeedResource`.
+- `PinterestDL.board` requests were missing a session cookie and CSRF token,
+  which Pinterest's internal resource API requires — every board request
+  came back as HTTP 403. `Board` now loads the board page first to obtain
+  a session cookie + CSRF token before calling the resource API, the same
+  way `Search` already did.
 - `pin.it/...` short links were rejected by `get_image_url` / `get_video_url` /
   `get_media` even though Pinterest itself redirects them to a normal pin
   page. `PIN_URL_PATTERN` now accepts `pin.it/...` links too.

data/README.md

@@ -160,10 +160,13 @@ pinterest-dl video https://www.pinterest.com/pin/example/ -o pin.mp4
 pinterest-dl search "cyberpunk city" -n 20
 pinterest-dl search "cyberpunk city" -n 20 -o ./downloads
 pinterest-dl board https://www.pinterest.com/username/board-name/ -n 50 -o ./downloads
+pinterest-dl board https://www.pinterest.com/username/board-name/ -t video -o ./videos
+pinterest-dl search "cats" -t image -n 30 -o ./images
+pinterest-dl search "cats" -t pin -n 30
 pinterest-dl --help
 ```
 
-Flags: `-o/--output`, `-n/--limit`, `-q/--quality`, `-c/--cookies`, `-v/--verbose`.
+Flags: `-o/--output`, `-n/--limit`, `-t/--type` (`all`, `image`, `video`, `pin`), `-q/--quality`, `-c/--cookies`, `-v/--verbose`.
 
 
 ## Development

data/bin/pinterest-dl

@@ -10,12 +10,13 @@ def usage
     Usage:
       pinterest-dl image URL [-o FILE]
       pinterest-dl video URL [-o FILE]
-      pinterest-dl search QUERY [-n LIMIT] [-o DIR]
-      pinterest-dl board URL [-n LIMIT] [-o DIR]
+      pinterest-dl search QUERY [-n LIMIT] [-t TYPE] [-o DIR]
+      pinterest-dl board URL [-n LIMIT] [-t TYPE] [-o DIR]
 
     Options:
       -o, --output PATH   File (image/video) or directory (search/board) to save to
       -n, --limit N       Max number of results (search/board), default 25
+      -t, --type TYPE     What to collect in search/board: all, image, video, pin (default all)
       -q, --quality Q     Image quality: originals, 736x, 474x, 236x (default originals)
       -c, --cookies STR   Cookie string for authenticated requests
       -v, --verbose       Print debug logging
@@ -23,10 +24,11 @@ def usage
   USAGE
 end
 
-options = { limit: 25, quality: :originals }
+options = { limit: 25, quality: :originals, type: :all }
 parser = OptionParser.new do |opts|
   opts.on('-o PATH', '--output PATH') { |v| options[:output] = v }
   opts.on('-n N', '--limit N', Integer) { |v| options[:limit] = v }
+  opts.on('-t T', '--type T') { |v| options[:type] = v.to_sym }
   opts.on('-q Q', '--quality Q') { |v| options[:quality] = v.to_sym }
   opts.on('-c STR', '--cookies STR') { |v| options[:cookies] = v }
   opts.on('-v', '--verbose') { options[:verbose] = true }
@@ -61,6 +63,31 @@ def die(message)
   exit 1
 end
 
+def handle_pins(pins, options)
+  if pins.empty?
+    puts 'No matching pins found'
+    return
+  end
+
+  if options[:type] == :pin
+    links = pins.map { |p| p[:pin_url] }
+    if options[:output]
+      FileUtils.mkdir_p(options[:output])
+      File.write(File.join(options[:output], 'pins.txt'), "#{links.join("\n")}\n")
+      puts "Saved #{links.size} pin links to #{options[:output]}/pins.txt"
+    else
+      puts links
+    end
+  elsif options[:output]
+    urls = pins.filter_map { |p| p[:video_url] || p[:image_url] }
+    bar = PinterestDL::ProgressBar.new(total: urls.size)
+    result = PinterestDL.download_batch(urls, dir: options[:output]) { |done, _total, _| bar.update(done) }
+    puts "Downloaded #{result[:succeeded].size}/#{urls.size} files to #{options[:output]}"
+  else
+    puts JSON.pretty_generate(pins)
+  end
+end
+
 begin
   case command
   when 'image'
@@ -80,28 +107,12 @@ begin
       puts url
     end
   when 'search'
-    pins = PinterestDL.search(target, limit: options[:limit])
-    if options[:output]
-      urls = pins.filter_map { |p| p[:image_url] }
-      bar = PinterestDL::ProgressBar.new(total: urls.size)
-      result = PinterestDL.download_batch(urls, dir: options[:output]) { |done, total, _| bar.update(done) }
-      puts "Downloaded #{result[:succeeded].size}/#{urls.size} images to #{options[:output]}"
-    else
-      puts JSON.pretty_generate(pins)
-    end
+    handle_pins(PinterestDL.search(target, limit: options[:limit], type: options[:type]), options)
   when 'board'
-    pins = PinterestDL.board(target, limit: options[:limit])
-    if options[:output]
-      urls = pins.filter_map { |p| p[:image_url] }
-      bar = PinterestDL::ProgressBar.new(total: urls.size)
-      result = PinterestDL.download_batch(urls, dir: options[:output]) { |done, total, _| bar.update(done) }
-      puts "Downloaded #{result[:succeeded].size}/#{urls.size} images to #{options[:output]}"
-    else
-      puts JSON.pretty_generate(pins)
-    end
+    handle_pins(PinterestDL.board(target, limit: options[:limit], type: options[:type]), options)
   else
     die "unknown command '#{command}'\n\n#{usage}"
   end
-rescue PinterestDL::Error => e
+rescue PinterestDL::Error, ArgumentError => e
   die e.message
 end

data/lib/pinterest-dl.rb

@@ -6,6 +6,7 @@ require_relative 'pinterest_dl/version'
 require_relative 'pinterest_dl/errors'
 require_relative 'pinterest_dl/configuration'
 require_relative 'pinterest_dl/client'
+require_relative 'pinterest_dl/media_type'
 require_relative 'pinterest_dl/extractors/image'
 require_relative 'pinterest_dl/extractors/video'
 require_relative 'pinterest_dl/extractors/pin_data'

data/lib/pinterest_dl/api.rb

@@ -19,29 +19,29 @@ module PinterestDL
     end
 
     def get_image_url(pin_url, quality: config.quality, client: Client.new(config: config))
-      validate_pin_url!(pin_url)
-      html = client.get(pin_url)
-      url = extract_pin_image(html, quality: quality)
+      html, pin_id = load_pin(pin_url, client)
+      url = extract_pin_image(html, pin_id, quality: quality)
       raise NotFoundError, "No image found at #{pin_url}" unless url
 
       url
     end
 
     def get_video_url(pin_url, client: Client.new(config: config))
-      validate_pin_url!(pin_url)
-      html = client.get(pin_url)
-      url = extract_pin_video(html)
-      raise NotFoundError, "No video found at #{pin_url}" unless url
+      html, pin_id = load_pin(pin_url, client)
+      url = extract_pin_video(html, pin_id)
+      unless url
+        hint = extract_pin_image(html, pin_id, quality: :originals) ? ' (this pin is an image, try the image command)' : ''
+        raise NotFoundError, "No video found at #{pin_url}#{hint}"
+      end
 
       url
     end
 
     def get_media(pin_url, quality: config.quality, client: Client.new(config: config))
-      validate_pin_url!(pin_url)
-      html = client.get(pin_url)
+      html, pin_id = load_pin(pin_url, client)
       {
-        image_url: extract_pin_image(html, quality: quality),
-        video_url: extract_pin_video(html)
+        image_url: extract_pin_image(html, pin_id, quality: quality),
+        video_url: extract_pin_video(html, pin_id)
       }
     end
 
@@ -53,8 +53,10 @@ module PinterestDL
       Downloader.new(config: config).download_batch(urls, dir: dir, filename: filename, &progress)
     end
 
-    def search(query, limit: 25, pages: 5)
-      Search.new(config: config).call(query, limit: limit, pages: pages)
+    def search(query, limit: 25, pages: 5, type: :all)
+      type = MediaType.parse(type)
+      pages = [pages, 15].max unless type == :all
+      Search.new(config: config).call(query, limit: limit, pages: pages, type: type)
     end
 
     def search_json(query, limit: 25)
@@ -62,21 +64,39 @@ module PinterestDL
     end
 
     def search_images(query, limit: 25)
-      search(query, limit: limit).filter_map { |p| p[:image_url] }
+      search(query, limit: limit, type: :image).filter_map { |p| p[:image_url] }
     end
 
     def search_videos(query, limit: 25)
-      search(query, limit: limit).filter_map { |p| p[:video_url] }
+      search(query, limit: limit, type: :video).filter_map { |p| p[:video_url] }
     end
 
-    def board(board_url, limit: 50, pages: 10)
-      Board.new(config: config).call(board_url, limit: limit, pages: pages)
+    def board(board_url, limit: 50, pages: 10, type: :all)
+      type = MediaType.parse(type)
+      pages = [pages, 20].max unless type == :all
+      Board.new(config: config).call(board_url, limit: limit, pages: pages, type: type)
     end
 
     private
 
-    def extract_pin_image(html, quality:)
-      pin = Extractors::PinData.extract(html)
+    def load_pin(pin_url, client)
+      validate_pin_url!(pin_url)
+      html = client.get(normalize_pin_url(pin_url))
+      [html, pin_id_from(client.last_url) || pin_id_from(pin_url)]
+    end
+
+    def normalize_pin_url(pin_url)
+      id = pin_id_from(pin_url)
+      id ? "https://www.pinterest.com/pin/#{id}/" : pin_url
+    end
+
+    def pin_id_from(url)
+      match = url.to_s.match(%r{/pin/(?:[^/?#]*--)?(\d+)})
+      match && match[1]
+    end
+
+    def extract_pin_image(html, pin_id, quality:)
+      pin = Extractors::PinData.extract(html, pin_id)
       url = pin && pin[:image_url]
       url ||= Extractors::Image.extract(html, quality: quality)
       return nil unless url
@@ -84,8 +104,8 @@ module PinterestDL
       Extractors::Image.apply_quality(url, quality)
     end
 
-    def extract_pin_video(html)
-      pin = Extractors::PinData.extract(html)
+    def extract_pin_video(html, pin_id)
+      pin = Extractors::PinData.extract(html, pin_id)
       (pin && pin[:video_url]) || Extractors::Video.extract(html)
     end
 

data/lib/pinterest_dl/board.rb

@@ -5,31 +5,31 @@ require 'uri'
 
 module PinterestDL
   class Board
-    B_URL = 'https://www.pinterest.com/resource/BoardFeedResource/get/'
-
-    BOARD_URL_PATTERN = %r{pinterest\.[a-z.]+/(?<username>[^/]+)/(?<slug>[^/]+)/?(?<section>[^/]+)?/?\z}
+    PAGE_SIZE = 50
 
     def initialize(client: PinterestDL::Client.new, config: PinterestDL.config)
       @client = client
       @config = config
     end
 
-    def call(board_url, limit: 50, pages: 10)
-      match = BOARD_URL_PATTERN.match(board_url)
-      raise PinterestDL::InvalidURLError, "Not a Pinterest board URL: #{board_url}" unless match
+    def call(board_url, limit: 50, pages: 10, type: :all)
+      username, slug, section = parse_url(board_url)
+      start_session(username, slug)
+      board_id = find_board_id(username, slug)
+      resource, options = feed_request(board_id, section)
 
       results = []
       bookmarks = nil
       pages_fetched = 0
 
       loop do
-        page = fetch_page(match, bookmarks: bookmarks)
-        results.concat(page[:pins])
+        page = fetch_page(resource, options, bookmarks)
+        results.concat(PinterestDL::MediaType.filter(page[:pins], type))
         bookmarks = page[:bookmarks]
         pages_fetched += 1
 
         break if results.size >= limit
-        break if bookmarks.nil? || bookmarks.empty? || bookmarks == ['-end-']
+        break if bookmarks.empty? || bookmarks == ['-end-']
         break if pages_fetched >= pages
         break if page[:pins].empty?
       end
@@ -39,67 +39,110 @@ module PinterestDL
 
     private
 
-    def fetch_page(match, bookmarks:)
-      body = @client.get(build_request_url(match, bookmarks), use_cache: false, headers: request_headers)
-      json = safe_parse(body)
-      raise PinterestDL::NotFoundError, "Board not found: #{match[0]}" unless json
+    def parse_url(board_url)
+      url = board_url.to_s
+      if url.include?('pin.it/')
+        @client.raw_get(url)
+        url = @client.last_url
+      end
+
+      uri = URI(url)
+      parts = uri.path.to_s.split('/').reject(&:empty?)
+      valid = uri.host.to_s.include?('pinterest') && [2, 3].include?(parts.size) && parts.first != 'pin'
+      raise PinterestDL::InvalidURLError, "Not a Pinterest board URL: #{board_url}" unless valid
+
+      parts
+    rescue URI::InvalidURIError
+      raise PinterestDL::InvalidURLError, "Not a Pinterest board URL: #{board_url}"
+    end
+
+    def start_session(username, slug)
+      @client.raw_get("https://www.pinterest.com/#{username}/#{slug}/")
+      @page_url = @client.last_url
+      @host = URI(@page_url).host
+      @source_path = URI(@page_url).path
+    end
+
+    def find_board_id(username, slug)
+      json = call_resource('BoardResource', username: username, slug: slug, field_set_key: 'detailed')
+      data = json.dig('resource_response', 'data')
+      id = data.is_a?(Hash) && data['id']
+      raise PinterestDL::NotFoundError, "Board not found: #{username}/#{slug} (private boards need --cookies)" unless id
+
+      id
+    end
+
+    def feed_request(board_id, section)
+      unless section
+        options = {
+          board_id: board_id,
+          field_set_key: 'react_grid_pin',
+          filter_section_pins: true,
+          sort: 'default',
+          layout: 'default',
+          page_size: PAGE_SIZE,
+          redux_normalize_feed: true
+        }
+        return ['BoardFeedResource', options]
+      end
+
+      ['BoardSectionPinsResource', { section_id: find_section_id(board_id, section), page_size: PAGE_SIZE }]
+    end
 
-      data_node = json.dig('resource_response', 'data') || []
-      next_bookmarks = json.dig('resource_response', 'bookmark')
+    def find_section_id(board_id, slug)
+      json = call_resource('BoardSectionsResource', board_id: board_id)
+      data = json.dig('resource_response', 'data')
+      found = data.is_a?(Array) && data.find { |s| s['slug'] == slug }
+      raise PinterestDL::NotFoundError, "Section not found: #{slug}" unless found
 
-      { pins: Array(data_node).filter_map { |r| build_pin(r) }, bookmarks: Array(next_bookmarks) }
+      found['id']
     end
 
-    def build_request_url(match, bookmarks)
-      options = {
-        board_url: "/#{match[:username]}/#{match[:slug]}/",
-        board_id: '',
-        section_slug: match[:section]
-      }.compact
-      options[:bookmarks] = bookmarks if bookmarks && !bookmarks.empty?
+    def fetch_page(resource, options, bookmarks)
+      options = options.merge(bookmarks: bookmarks) if bookmarks && !bookmarks.empty?
+      json = call_resource(resource, options)
 
+      data = json.dig('resource_response', 'data')
+      pins = data.is_a?(Array) ? data.filter_map { |r| build_pin(r) } : []
+      { pins: pins, bookmarks: Array(json.dig('resource_response', 'bookmark')) }
+    end
+
+    def call_resource(name, options)
       data = { options: options, context: {} }.to_json
-      res_uri = URI(B_URL)
-      res_uri.query = URI.encode_www_form('source_url' => options[:board_url], 'data' => data)
-      res_uri.to_s
+      uri = URI("https://#{@host}/resource/#{name}/get/")
+      uri.query = URI.encode_www_form('source_url' => @source_path, 'data' => data)
+
+      body = @client.get(uri.to_s, use_cache: false, headers: api_headers)
+      JSON.parse(body)
+    rescue JSON::ParserError, TypeError
+      raise PinterestDL::NotFoundError, "Unexpected response from Pinterest (#{name})"
     end
 
-    def request_headers
+    def api_headers
       {
         'Accept' => 'application/json, text/javascript, */*, q=0.01',
-        'X-Requested-With' => 'XMLHttpRequest'
+        'X-Requested-With' => 'XMLHttpRequest',
+        'X-Pinterest-PWS-Handler' => 'www/[username]/[slug].js',
+        'X-CSRFToken' => @client.cookie('csrftoken', @page_url) || 'undefined',
+        'Referer' => @page_url
       }
     end
 
     def build_pin(raw)
+      return nil unless raw.is_a?(Hash)
+
       id = raw['id']
       return nil unless id
 
-      orig_image = raw.dig('images', 'orig', 'url')
-      video_url = first_video_url(raw)
-
       {
         id: id,
         pin_url: "https://www.pinterest.com/pin/#{id}/",
         title: raw['title'] || raw['grid_title'],
         description: raw['description'],
         creator: raw.dig('pinner', 'username') || raw.dig('pinner', 'full_name'),
-        image_url: orig_image,
-        video_url: video_url
+        image_url: raw.dig('images', 'orig', 'url'),
+        video_url: PinterestDL::Extractors::PinData.first_video_url(raw)
       }
     end
-
-    def first_video_url(raw)
-      video_list = raw.dig('videos', 'video_list')
-      return nil unless video_list
-
-      video_list.values.first&.dig('url')
-    end
-
-    def safe_parse(body)
-      JSON.parse(body)
-    rescue JSON::ParserError, TypeError
-      nil
-    end
   end
 end

data/lib/pinterest_dl/client.rb

@@ -7,26 +7,36 @@ require 'openssl'
 
 module PinterestDL
   class Client
+    MAX_REDIRECTS = 5
+
+    attr_reader :last_url
+
     def initialize(config: PinterestDL.config)
       @config = config
       @cache = {}
+      @cookies = Hash.new { |hash, key| hash[key] = {} }
       @last_request_at = nil
+      @last_url = nil
     end
 
     def get(url, headers: {}, use_cache: true)
       cache_key = [url, headers].hash
-      return @cache[cache_key] if use_cache && @cache.key?(cache_key)
+      if use_cache && @cache.key?(cache_key)
+        body, @last_url = @cache[cache_key]
+        return body
+      end
 
-      response = request_with_retries(url, headers: headers)
-      body = handle_response(response, url)
-      @cache[cache_key] = body if use_cache
+      body = fetch(url, headers: headers).body
+      @cache[cache_key] = [body, @last_url] if use_cache
       body
     end
 
     def raw_get(url, headers: {})
-      response = request_with_retries(url, headers: headers)
-      handle_response(response, url, raw: true)
-      response
+      fetch(url, headers: headers)
+    end
+
+    def cookie(name, url = @last_url)
+      @cookies[site_key(URI(url))][name]
     end
 
     def clear_cache
@@ -35,13 +45,32 @@ module PinterestDL
 
     private
 
+    def fetch(url, headers: {})
+      (MAX_REDIRECTS + 1).times do
+        response = request_with_retries(url, headers: headers)
+        save_cookies(response, url)
+
+        unless response.is_a?(Net::HTTPRedirection)
+          check_response(response, url)
+          @last_url = url
+          return response
+        end
+
+        location = response['location']
+        raise PinterestDL::NotFoundError, "Redirected without a location header (#{url})" unless location
+
+        url = URI.join(url, location).to_s
+      end
+      raise PinterestDL::NetworkError, "Too many redirects for #{url}"
+    end
+
     def request_with_retries(url, headers: {})
       attempts = 0
       begin
         throttle!
         uri = URI(url)
         http = build_http(uri)
-        req = Net::HTTP::Get.new(uri, default_headers.merge(headers))
+        req = Net::HTTP::Get.new(uri, default_headers(uri).merge(headers))
         http.request(req)
       rescue Timeout::Error, Errno::ECONNRESET, Errno::ECONNREFUSED,
              SocketError, OpenSSL::SSL::SSLError, EOFError => e
@@ -55,10 +84,10 @@ module PinterestDL
       end
     end
 
-    def handle_response(response, url, raw: false)
+    def check_response(response, url)
       case response
       when Net::HTTPSuccess
-        raw ? response : response.body
+        response
       when Net::HTTPTooManyRequests
         raise PinterestDL::RateLimitError.new(
           "Rate limited while requesting #{url}",
@@ -66,19 +95,6 @@ module PinterestDL
         )
       when Net::HTTPNotFound
         raise PinterestDL::NotFoundError, "Nothing found at #{url}"
-      when Net::HTTPRedirection
-        location = response['location']
-        raise PinterestDL::NotFoundError, "Redirected without a location header (#{url})" unless location
-
-        location = URI.join(url, location).to_s if location.start_with?('/')
-
-        uri = URI(location)
-      if uri.host&.match?(/^(?:www\.)?pinterest\.(?!com$)[a-z.]+$/i) || uri.host == 'pin.it'
-        uri.host = 'www.pinterest.com'
-        location = uri.to_s
-      end
-
-  get_or_raw(location, raw: raw)
       when Net::HTTPForbidden
         raise PinterestDL::RateLimitError, "Request to #{url} was blocked (HTTP 403)"
       else
@@ -86,10 +102,6 @@ module PinterestDL
       end
     end
 
-    def get_or_raw(url, raw:)
-      raw ? raw_get(url) : get(url)
-    end
-
     def build_http(uri)
       http = Net::HTTP.new(uri.host, uri.port)
       http.use_ssl = (uri.scheme == 'https')
@@ -98,12 +110,37 @@ module PinterestDL
       http
     end
 
-    def default_headers
-      headers = { 'User-Agent' => @config.user_agent }
-      headers['Cookie'] = @config.cookies if @config.authenticated?
+    def default_headers(uri)
+      headers = {
+        'User-Agent' => @config.user_agent,
+        'Accept-Language' => 'en-US,en;q=0.9'
+      }
+      cookie_line = cookie_header(uri)
+      headers['Cookie'] = cookie_line unless cookie_line.empty?
       headers
     end
 
+    def cookie_header(uri)
+      parts = []
+      parts << @config.cookies.to_s if @config.authenticated?
+      parts.concat(@cookies[site_key(uri)].map { |name, value| "#{name}=#{value}" })
+      parts.join('; ')
+    end
+
+    def save_cookies(response, url)
+      jar = @cookies[site_key(URI(url))]
+      (response.get_fields('set-cookie') || []).each do |line|
+        name, value = line.split(';').first.to_s.split('=', 2)
+        next if name.nil? || value.nil?
+
+        value.empty? ? jar.delete(name.strip) : jar[name.strip] = value
+      end
+    end
+
+    def site_key(uri)
+      uri.host.to_s.split('.').last(2).join('.')
+    end
+
     def throttle!
       return if @config.rate_limit_interval <= 0
       return unless @last_request_at

data/lib/pinterest_dl/extractors/image.rb

@@ -25,8 +25,7 @@ module PinterestDL
       end
 
       def from_open_graph(html)
-        match = html.match(/<meta[^>]*property=["']og:image["'][^>]*content=["']([^"']+)["']/)
-        match && match[1]
+        PinterestDL::Extractors::Meta.content(html, 'og:image')
       end
 
       def from_raw_pattern(html)

data/lib/pinterest_dl/extractors/pin_data.rb

@@ -6,13 +6,16 @@ module PinterestDL
   module Extractors
     module PinData
       SCRIPT_PATTERN = %r{<script[^>]*id=["']__PWS_DATA__["'][^>]*>(.*?)</script>}m.freeze
+
       module_function
-      def extract(html)
+
+      def extract(html, pin_id = nil)
         data = parse_embedded_json(html)
         return nil unless data
 
-        pin = find_pin_node(data)
+        pin = find_pin_node(data, pin_id)
         return nil unless pin
+
         {
           image_url: pin.dig('images', 'orig', 'url'),
           video_url: first_video_url(pin),
@@ -29,20 +32,24 @@ module PinterestDL
         nil
       end
 
-      def find_pin_node(data)
+      def find_pin_node(data, pin_id = nil)
+        matches = []
         stack = [data]
         until stack.empty?
           node = stack.pop
           case node
           when Hash
-            return node if pin_like?(node)
-
+            if pin_id
+              matches << node if node['id'].to_s == pin_id.to_s && has_media?(node)
+            elsif pin_like?(node)
+              return node
+            end
             stack.concat(node.values)
           when Array
             stack.concat(node)
           end
         end
-        nil
+        matches.find { |n| first_video_url(n) } || matches.first
       end
 
       def pin_like?(node)
@@ -50,11 +57,42 @@ module PinterestDL
         images.is_a?(Hash) && images.dig('orig', 'url')
       end
 
+      def has_media?(node)
+        node['images'].is_a?(Hash) || node['videos'].is_a?(Hash) || node['story_pin_data'].is_a?(Hash)
+      end
+
       def first_video_url(pin)
         video_list = pin.dig('videos', 'video_list')
+        video_list = deep_find(pin['story_pin_data'], 'video_list') unless video_list.is_a?(Hash)
+        best_video(video_list)
+      end
+
+      def best_video(video_list)
         return nil unless video_list.is_a?(Hash)
 
-        video_list.values.first&.dig('url')
+        formats = video_list.values.select { |v| v.is_a?(Hash) && v['url'] }
+        mp4s = formats.select { |v| v['url'].include?('.mp4') }
+        best = mp4s.max_by { |v| v['width'].to_i * v['height'].to_i }
+        return best['url'] if best
+
+        hls = formats.find { |v| v['url'].include?('.m3u8') }
+        hls && hls['url'].sub('/hls/', '/720p/').sub(/\.m3u8.*\z/, '.mp4')
+      end
+
+      def deep_find(node, key)
+        stack = [node]
+        until stack.empty?
+          current = stack.pop
+          case current
+          when Hash
+            return current[key] if current[key].is_a?(Hash)
+
+            stack.concat(current.values)
+          when Array
+            stack.concat(current)
+          end
+        end
+        nil
       end
 
       def safe_parse(raw)

data/lib/pinterest_dl/extractors/video.rb

@@ -3,12 +3,11 @@
 # Monji024
 module PinterestDL
   module Extractors
-    # Extracts the direct video URL for a pin from raw HTML.
     module Video
       module_function
 
       def extract(html)
-        url = from_json_ld(html) || from_open_graph(html) || from_content_url(html)
+        url = from_json_ld(html) || from_open_graph(html) || from_content_url(html) || from_raw_pattern(html)
         return nil unless url
 
         unescape(url)
@@ -19,8 +18,7 @@ module PinterestDL
       end
 
       def from_open_graph(html)
-        match = html.match(/<meta[^>]*property=["']og:video["'][^>]*content=["']([^"']+)["']/)
-        match && match[1]
+        PinterestDL::Extractors::Meta.content(html, 'og:video')
       end
 
       def from_content_url(html)
@@ -28,11 +26,31 @@ module PinterestDL
         match && match[1]
       end
 
+      def from_raw_pattern(html)
+        clean = html.gsub('\\u002F', '/').gsub('\\/', '/')
+        match = clean.match(%r{https://v\d*\.pinimg\.com/videos/[^"'\s\\<>]+?\.mp4})
+        match && match[0]
+      end
+
       def unescape(url)
         url.gsub('\\u002F', '/').gsub('\\', '')
       end
     end
 
+    module Meta
+      module_function
+
+      def content(html, key)
+        html.to_s.scan(/<meta\b[^>]*>/i).each do |tag|
+          next unless tag.match?(/(?:property|name)=["']#{Regexp.escape(key)}["']/i)
+
+          match = tag.match(/content=["']([^"']+)["']/i)
+          return match[1] if match
+        end
+        nil
+      end
+    end
+
     module JsonLd
       module_function
 

data/lib/pinterest_dl/media_type.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+# Monji024
+module PinterestDL
+  module MediaType
+    TYPES = %i[all image video pin].freeze
+
+    module_function
+
+    def parse(value)
+      type = (value || :all).to_sym
+      return type if TYPES.include?(type)
+
+      raise ArgumentError, "type must be one of: #{TYPES.join(', ')}"
+    end
+
+    def filter(pins, type)
+      case parse(type)
+      when :image then pins.select { |p| p[:image_url] && !p[:video_url] }
+      when :video then pins.select { |p| p[:video_url] }
+      else pins
+      end
+    end
+  end
+end

data/lib/pinterest_dl/search.rb

@@ -13,7 +13,7 @@ module PinterestDL
       @config = config
     end
 
-    def call(query, limit: 25, pages: 5)
+    def call(query, limit: 25, pages: 5, type: :all)
       results = []
       bookmarks = nil
       pages_fetched = 0
@@ -21,7 +21,7 @@ module PinterestDL
       loop do
         page = fetch_page(query, bookmarks: bookmarks)
         pins = page[:pins]
-        results.concat(pins)
+        results.concat(PinterestDL::MediaType.filter(pins, type))
         bookmarks = page[:bookmarks]
         pages_fetched += 1
 
@@ -50,7 +50,8 @@ module PinterestDL
       parse_search_response(json)
     end
 
-
+    # Pinterest's search resource requires a session cookie + CSRF token
+    # obtained by first loading the human-facing search page.
     def establish_session(query)
       source_path = "/search/pins/?q=#{URI.encode_www_form_component(query)}"
       home_uri = URI("https://www.pinterest.com#{source_path}")
@@ -128,10 +129,7 @@ module PinterestDL
     end
 
     def first_video_url(raw)
-      video_list = raw.dig('videos', 'video_list')
-      return nil unless video_list
-
-      video_list.values.first&.dig('url')
+      PinterestDL::Extractors::PinData.first_video_url(raw)
     end
 
     def safe_parse(body)

data/lib/pinterest_dl/version.rb

@@ -2,5 +2,5 @@
 
 # Monji024
 module PinterestDL
-  VERSION = '2.0.3'
+  VERSION = '2.0.5'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pinterest-dl
 version: !ruby/object:Gem::Version
-  version: 2.0.3
+  version: 2.0.5
 platform: ruby
 authors:
 - Monji
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-08-29 00:00:00.000000000 Z
+date: 2026-10-12 00:00:00.000000000 Z
 dependencies: []
 description: A Ruby toolkit for getting direct image and video URLs from Pinterest
   pins, searching pins, fetching board contents, and downloading media to disk — as
@@ -33,6 +33,7 @@ files:
 - lib/pinterest_dl/extractors/image.rb
 - lib/pinterest_dl/extractors/pin_data.rb
 - lib/pinterest_dl/extractors/video.rb
+- lib/pinterest_dl/media_type.rb
 - lib/pinterest_dl/progress_bar.rb
 - lib/pinterest_dl/search.rb
 - lib/pinterest_dl/version.rb
@@ -60,7 +61,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.3.15
+rubygems_version: 3.4.20
 signing_key:
 specification_version: 4
 summary: Extract direct image/video URLs, search, and download from Pinterest