pinterest-dl 2.0.1 → 2.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b4d94f0a5bbac63a983308d541b5e50a6a13c1d5bee422c356e56395c686ebcd
-  data.tar.gz: 148473c309ec313a3e0d7ddb0864f0388d22a2bf9edee39e819e54ccaab10745
+  metadata.gz: ab49a0f8fc2500835b07ec8abf935c4a643d7a9fa8f6e2de9bdb5250f10ef24c
+  data.tar.gz: 9fb506ca599f3a1625989ab178c7ed4f0c0e5ddad236f416f514666c90741bf0
 SHA512:
-  metadata.gz: 4e26a226009b81da00d41f46f9ccfcdca2f058aa7ff93210b488e1f2082c27f927103f5a0329dcea921ea34a62c49c3ad70c77d411d4a1b569b4b97fbba8c4cf
-  data.tar.gz: 66b3ddf6d10555f9e018198a3de376578b22fc504dfe386d8a49f06bdcc213519e0b273c5f123d34bcb33eb3217da51af7609712a2f1a0a3b67004be6858eb9b
+  metadata.gz: e3962e2ea256111c1c9178947a0aa026daec6fed0c1cfefb2f81d92dd77299eebc9c2634f2ecf9e74ede3edfacb327ce28381c9aeffa951f9960c81f8883dddd
+  data.tar.gz: '09407f8541a4042be42584393524b8aee0147654d46c0abdc1d0c17ad9467bbaa17406fd260099fe214db43aa946553f20a696524616e66e2817477d6cd897b8'

data/CHANGELOG.md

@@ -3,6 +3,33 @@
 All notable changes to this project are documented here.
 This project follows [Semantic Versioning](https://semver.org/).
 
+## [2.0.3]
+
+### Fixed
+- **`PinterestDL::Board` was broken.** `lib/pinterest_dl/board.rb` accidentally
+  defined `class Search` instead of `class Board` (with a full copy of the old
+  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.
+- The raw-HTML fallback regex used to find an image when no `og:image` or
+  JSON-LD tag was present could "bleed" past the actual image URL into
+  trailing inline CSS (e.g. `...img.png)}._YsBbF{border:0...`), producing a
+  garbage, unusable URL. The regex now requires a real image extension and
+  only matches valid URL characters, so it stops exactly at the image URL.
+- Image/video extraction now tries Pinterest's embedded page-state JSON
+  (`<script id="__PWS_DATA__">`) first via a new `Extractors::PinData`
+  module, which is far more reliable than scraping meta tags/regex and is
+  immune to the CSS-bleed issue above; the old scraping strategies remain as
+  a fallback if a page doesn't embed that data.
+
 ## [2.0.0]
 
 ### Added

data/lib/pinterest-dl.rb

@@ -1,12 +1,14 @@
 # frozen_string_literal: true
 
 # Monji024
+require 'json'
 require_relative 'pinterest_dl/version'
 require_relative 'pinterest_dl/errors'
 require_relative 'pinterest_dl/configuration'
 require_relative 'pinterest_dl/client'
 require_relative 'pinterest_dl/extractors/image'
 require_relative 'pinterest_dl/extractors/video'
+require_relative 'pinterest_dl/extractors/pin_data'
 require_relative 'pinterest_dl/search'
 require_relative 'pinterest_dl/board'
 require_relative 'pinterest_dl/downloader'

data/lib/pinterest_dl/api.rb

@@ -2,7 +2,7 @@
 
 # Monji024
 module PinterestDL
-  PIN_URL_PATTERN = %r{pinterest\.[a-z.]+/pin/}i.freeze
+  PIN_URL_PATTERN = %r{(pinterest\.[a-z.]+/pin/|pin\.it/)}i.freeze
 
   class << self
     def config
@@ -21,7 +21,7 @@ module PinterestDL
     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 = Extractors::Image.extract(html, quality: quality)
+      url = extract_pin_image(html, quality: quality)
       raise NotFoundError, "No image found at #{pin_url}" unless url
 
       url
@@ -30,18 +30,18 @@ module PinterestDL
     def get_video_url(pin_url, client: Client.new(config: config))
       validate_pin_url!(pin_url)
       html = client.get(pin_url)
-      url = Extractors::Video.extract(html)
+      url = extract_pin_video(html)
       raise NotFoundError, "No video found at #{pin_url}" unless url
 
       url
     end
 
-    def get_media(pin_url, quality: config.quality)
-      client = Client.new(config: config)
+    def get_media(pin_url, quality: config.quality, client: Client.new(config: config))
+      validate_pin_url!(pin_url)
       html = client.get(pin_url)
       {
-        image_url: Extractors::Image.extract(html, quality: quality),
-        video_url: Extractors::Video.extract(html)
+        image_url: extract_pin_image(html, quality: quality),
+        video_url: extract_pin_video(html)
       }
     end
 
@@ -74,6 +74,19 @@ module PinterestDL
     end
 
     private
+    def extract_pin_image(html, quality:)
+      pin = Extractors::PinData.extract(html)
+      url = pin && pin[:image_url]
+      url ||= Extractors::Image.extract(html, quality: quality)
+      return nil unless url
+
+      Extractors::Image.apply_quality(url, quality)
+    end
+
+    def extract_pin_video(html)
+      pin = Extractors::PinData.extract(html)
+      (pin && pin[:video_url]) || Extractors::Video.extract(html)
+    end
 
     def validate_pin_url!(pin_url)
       return if pin_url.to_s.match?(PIN_URL_PATTERN)

data/lib/pinterest_dl/board.rb

@@ -1,35 +1,37 @@
 # frozen_string_literal: true
 
 require 'json'
-require 'net/http'
 require 'uri'
 
 module PinterestDL
-  class Search
-    B_URL = 'https://www.pinterest.com/resource/BaseSearchResource/get/'
+  class Board
+    B_URL = 'https://www.pinterest.com/resource/BoardFeedResource/get/'
+
+    BOARD_URL_PATTERN = %r{pinterest\.[a-z.]+/(?<username>[^/]+)/(?<slug>[^/]+)/?(?<section>[^/]+)?/?\z}
 
     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(query, limit: 25, pages: 5)
       results = []
       bookmarks = nil
       pages_fetched = 0
 
       loop do
-        page = fetch_page(query, bookmarks: bookmarks)
-        pins = page[:pins]
-        results.concat(pins)
+        page = fetch_page(match, bookmarks: bookmarks)
+        results.concat(page[:pins])
         bookmarks = page[:bookmarks]
         pages_fetched += 1
 
         break if results.size >= limit
         break if bookmarks.nil? || bookmarks.empty? || bookmarks == ['-end-']
         break if pages_fetched >= pages
-        break if pins.empty?
+        break if page[:pins].empty?
       end
 
       results.first(limit)
@@ -37,40 +39,62 @@ module PinterestDL
 
     private
 
-    def fetch_page(query, bookmarks: nil)
-      source_path = "/search/pins/?q=#{URI.encode_www_form_component(query)}"
-      home_uri = URI("https://www.pinterest.com#{source_path}")
+    def fetch_page(match, bookmarks:)
+      session = establish_session(match)
+      body = @client.get(
+        build_request_url(match, bookmarks),
+        use_cache: false,
+        headers: board_headers(session)
+      )
 
-      home_response = @client.raw_get(home_uri.to_s)
-      session_cookies = (home_response.get_fields('set-cookie') || []).map { |c| c.split(';').first }
-      cookie_header = [@config.cookies, session_cookies.join('; ')].compact.reject(&:empty?).join('; ')
-      csrf = session_cookies.find { |c| c.start_with?('csrftoken=') }&.split('=', 2)&.last || 'undefined'
+      json = safe_parse(body)
+      raise PinterestDL::NotFoundError, "Board not found: #{match[0]}" unless json
 
-      options = { query: query, scope: 'pins', page_size: 25 }
-      options[:bookmarks] = bookmarks if bookmarks && !bookmarks.empty?
-      data = { options: options, context: {} }.to_json
+      data_node = json.dig('resource_response', 'data') || []
+      next_bookmarks = json.dig('resource_response', 'bookmark')
 
-      res_uri = URI(B_URL)
-      res_uri.query = URI.encode_www_form('source_url' => source_path, 'data' => data)
+      { pins: Array(data_node).filter_map { |r| build_pin(r) }, bookmarks: Array(next_bookmarks) }
+    end
 
-      body = @client.get(res_uri.to_s, use_cache: false, headers: {
-                            'Accept' => 'application/json, text/javascript, */*, q=0.01',
-                            'X-Requested-With' => 'XMLHttpRequest',
-                            'X-Pinterest-PWS-Handler' => 'www/search/[scope].js',
-                            'X-CSRFToken' => csrf,
-                            'Referer' => home_uri.to_s,
-                            'Cookie' => cookie_header
-                          })
+    def establish_session(match)
+      board_path = "/#{match[:username]}/#{match[:slug]}/"
+      board_uri = URI("https://www.pinterest.com#{board_path}")
+      home_response = @client.raw_get(board_uri.to_s)
+      cookies = (home_response.get_fields('set-cookie') || []).map { |c| c.split(';').first }
 
-      json = safe_parse(body)
-      raise PinterestDL::NotFoundError, "No search results for \"#{query}\"" unless json
+      { board_path: board_path, board_uri: board_uri, cookies: cookies, csrf: csrf_token(cookies) }
+    end
 
-      data_node = json.dig('resource_response', 'data') || {}
-      raw_results = data_node['results'] || []
-      next_bookmarks = json.dig('resource_response', 'bookmark') ||
-                        json.dig('resource', 'options', 'bookmarks')
+    def csrf_token(cookies)
+      token_cookie = cookies.find { |c| c.start_with?('csrftoken=') }
+      token_cookie ? token_cookie.split('=', 2).last : 'undefined'
+    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?
 
-      { pins: raw_results.filter_map { |r| build_pin(r) }, bookmarks: Array(next_bookmarks) }
+      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
+    end
+
+    def board_headers(session)
+      cookie_header = [@config.cookies, session[:cookies].join('; ')].compact.reject(&:empty?).join('; ')
+
+      {
+        'Accept' => 'application/json, text/javascript, */*, q=0.01',
+        'X-Requested-With' => 'XMLHttpRequest',
+        'X-Pinterest-PWS-Handler' => 'www/[username]/[slug].js',
+        'X-CSRFToken' => session[:csrf],
+        'Referer' => session[:board_uri].to_s,
+        'Cookie' => cookie_header
+      }
     end
 
     def build_pin(raw)
@@ -78,8 +102,7 @@ module PinterestDL
       return nil unless id
 
       orig_image = raw.dig('images', 'orig', 'url')
-      fallback_image = raw['images']&.values&.map { |v| v.is_a?(Hash) ? v['url'] : nil }&.compact&.last
-      video_url = raw.dig('videos', 'video_list')&.values&.first&.dig('url')
+      video_url = first_video_url(raw)
 
       {
         id: id,
@@ -87,11 +110,18 @@ module PinterestDL
         title: raw['title'] || raw['grid_title'],
         description: raw['description'],
         creator: raw.dig('pinner', 'username') || raw.dig('pinner', 'full_name'),
-        image_url: orig_image || fallback_image,
+        image_url: orig_image,
         video_url: video_url
       }
     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

data/lib/pinterest_dl/extractors/image.rb

@@ -30,7 +30,9 @@ module PinterestDL
       end
 
       def from_raw_pattern(html)
-        match = html.match(%r{https://i\.pinimg\.com/(?:originals|736x|474x|236x)/[^"'\s\\]+})
+        match = html.match(
+          %r{https://i\.pinimg\.com/(?:originals|736x|474x|236x)/[A-Za-z0-9_\-/]+\.(?:jpg|jpeg|png|gif|webp)}i
+        )
         match && match[0]
       end
 

data/lib/pinterest_dl/extractors/pin_data.rb

@@ -0,0 +1,70 @@
+# frozen_string_literal: true
+
+require 'json'
+
+module PinterestDL
+  module Extractors
+    module PinData
+      SCRIPT_PATTERN = %r{<script[^>]*id=["']__PWS_DATA__["'][^>]*>(.*?)</script>}m.freeze
+
+      module_function
+
+      def extract(html)
+        data = parse_embedded_json(html)
+        return nil unless data
+
+        pin = find_pin_node(data)
+        return nil unless pin
+
+        {
+          image_url: pin.dig('images', 'orig', 'url'),
+          video_url: first_video_url(pin),
+          title: pin['title'] || pin['grid_title'],
+          description: pin['description']
+        }
+      end
+
+      def parse_embedded_json(html)
+        html.to_s.scan(SCRIPT_PATTERN).each do |(raw)|
+          parsed = safe_parse(raw)
+          return parsed if parsed
+        end
+        nil
+      end
+
+      def find_pin_node(data)
+        stack = [data]
+        until stack.empty?
+          node = stack.pop
+          case node
+          when Hash
+            return node if pin_like?(node)
+
+            stack.concat(node.values)
+          when Array
+            stack.concat(node)
+          end
+        end
+        nil
+      end
+
+      def pin_like?(node)
+        images = node['images']
+        images.is_a?(Hash) && images.dig('orig', 'url')
+      end
+
+      def first_video_url(pin)
+        video_list = pin.dig('videos', 'video_list')
+        return nil unless video_list.is_a?(Hash)
+
+        video_list.values.first&.dig('url')
+      end
+
+      def safe_parse(raw)
+        JSON.parse(raw)
+      rescue JSON::ParserError
+        nil
+      end
+    end
+  end
+end

data/lib/pinterest_dl/extractors/video.rb

@@ -3,7 +3,6 @@
 # Monji024
 module PinterestDL
   module Extractors
-    # Extracts the direct video URL for a pin from raw HTML.
     module Video
       module_function
 
@@ -33,8 +32,6 @@ module PinterestDL
       end
     end
 
-    # Shared helper for both extractors: pulls a `contentUrl` out of an
-    # `application/ld+json` <script> tag matching the given @type.
     module JsonLd
       module_function
 

data/lib/pinterest_dl/version.rb

@@ -2,5 +2,5 @@
 
 # Monji024
 module PinterestDL
-  VERSION = '2.0.1'
+  VERSION = '2.0.4'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pinterest-dl
 version: !ruby/object:Gem::Version
-  version: 2.0.1
+  version: 2.0.4
 platform: ruby
 authors:
 - Monji
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-08-28 00:00:00.000000000 Z
+date: 2026-08-30 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
@@ -31,6 +31,7 @@ files:
 - lib/pinterest_dl/downloader.rb
 - lib/pinterest_dl/errors.rb
 - lib/pinterest_dl/extractors/image.rb
+- lib/pinterest_dl/extractors/pin_data.rb
 - lib/pinterest_dl/extractors/video.rb
 - lib/pinterest_dl/progress_bar.rb
 - lib/pinterest_dl/search.rb