context.dev 1.32.0 → 1.35.0

data/rbi/context_dev/models/web_web_scrape_md_response.rbi

@@ -15,6 +15,17 @@ module ContextDev
       sig { returns(String) }
       attr_accessor :markdown
 
+      # Metadata extracted from the scraped page HTML.
+      sig { returns(ContextDev::Models::WebWebScrapeMdResponse::Metadata) }
+      attr_reader :metadata
+
+      sig do
+        params(
+          metadata: ContextDev::Models::WebWebScrapeMdResponse::Metadata::OrHash
+        ).void
+      end
+      attr_writer :metadata
+
       # Indicates success
       sig do
         returns(
@@ -47,6 +58,8 @@ module ContextDev
       sig do
         params(
           markdown: String,
+          metadata:
+            ContextDev::Models::WebWebScrapeMdResponse::Metadata::OrHash,
           success:
             ContextDev::Models::WebWebScrapeMdResponse::Success::OrBoolean,
           url: String,
@@ -57,6 +70,8 @@ module ContextDev
       def self.new(
         # Page content converted to GitHub Flavored Markdown
         markdown:,
+        # Metadata extracted from the scraped page HTML.
+        metadata:,
         # Indicates success
         success:,
         # The URL that was scraped
@@ -71,6 +86,7 @@ module ContextDev
         override.returns(
           {
             markdown: String,
+            metadata: ContextDev::Models::WebWebScrapeMdResponse::Metadata,
             success:
               ContextDev::Models::WebWebScrapeMdResponse::Success::TaggedBoolean,
             url: String,
@@ -82,6 +98,464 @@ module ContextDev
       def to_hash
       end
 
+      class Metadata < ContextDev::Internal::Type::BaseModel
+        OrHash =
+          T.type_alias do
+            T.any(
+              ContextDev::Models::WebWebScrapeMdResponse::Metadata,
+              ContextDev::Internal::AnyHash
+            )
+          end
+
+        # Final URL scraped after redirects or scraper fallback, when known. Falls back to
+        # sourceUrl when unavailable.
+        sig { returns(String) }
+        attr_accessor :final_url
+
+        # Original URL requested by the caller.
+        sig { returns(String) }
+        attr_accessor :source_url
+
+        # Additional non-social meta tags not promoted to top-level metadata fields.
+        sig do
+          returns(
+            T.nilable(
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::AdditionalMeta::Variants
+              ]
+            )
+          )
+        end
+        attr_reader :additional_meta
+
+        sig do
+          params(
+            additional_meta:
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::AdditionalMeta::Variants
+              ]
+          ).void
+        end
+        attr_writer :additional_meta
+
+        # Resolved alternate links from link rel=alternate tags.
+        sig do
+          returns(
+            T.nilable(
+              T::Array[
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::Alternate
+              ]
+            )
+          )
+        end
+        attr_reader :alternates
+
+        sig do
+          params(
+            alternates:
+              T::Array[
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::Alternate::OrHash
+              ]
+          ).void
+        end
+        attr_writer :alternates
+
+        # Author metadata, when present.
+        sig { returns(T.nilable(String)) }
+        attr_reader :author
+
+        sig { params(author: String).void }
+        attr_writer :author
+
+        # Resolved canonical URL, when present.
+        sig { returns(T.nilable(String)) }
+        attr_reader :canonical_url
+
+        sig { params(canonical_url: String).void }
+        attr_writer :canonical_url
+
+        # Best description extracted from standard, Open Graph, or Twitter metadata.
+        sig { returns(T.nilable(String)) }
+        attr_reader :description
+
+        sig { params(description: String).void }
+        attr_writer :description
+
+        # Resolved favicon URL, when present.
+        sig { returns(T.nilable(String)) }
+        attr_reader :favicon
+
+        sig { params(favicon: String).void }
+        attr_writer :favicon
+
+        # Primary resolved preview image from Open Graph, Twitter, or image metadata.
+        sig { returns(T.nilable(String)) }
+        attr_reader :image
+
+        sig { params(image: String).void }
+        attr_writer :image
+
+        # JSON-LD structured data blocks parsed from the page.
+        sig { returns(T.nilable(T::Array[T::Hash[Symbol, T.anything]])) }
+        attr_reader :json_ld
+
+        sig { params(json_ld: T::Array[T::Hash[Symbol, T.anything]]).void }
+        attr_writer :json_ld
+
+        # Keywords extracted from the page's keywords meta tag.
+        sig { returns(T.nilable(T::Array[String])) }
+        attr_reader :keywords
+
+        sig { params(keywords: T::Array[String]).void }
+        attr_writer :keywords
+
+        # Language extracted from html lang or language meta tags.
+        sig { returns(T.nilable(String)) }
+        attr_reader :language
+
+        sig { params(language: String).void }
+        attr_writer :language
+
+        # Modified timestamp/date from page metadata, when present.
+        sig { returns(T.nilable(String)) }
+        attr_reader :modified_time
+
+        sig { params(modified_time: String).void }
+        attr_writer :modified_time
+
+        # Open Graph metadata with the og: prefix removed and keys camel-cased.
+        sig do
+          returns(
+            T.nilable(
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::OpenGraph::Variants
+              ]
+            )
+          )
+        end
+        attr_reader :open_graph
+
+        sig do
+          params(
+            open_graph:
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::OpenGraph::Variants
+              ]
+          ).void
+        end
+        attr_writer :open_graph
+
+        # Published timestamp/date from page metadata, when present.
+        sig { returns(T.nilable(String)) }
+        attr_reader :published_time
+
+        sig { params(published_time: String).void }
+        attr_writer :published_time
+
+        # Robots meta directive, when present.
+        sig { returns(T.nilable(String)) }
+        attr_reader :robots
+
+        sig { params(robots: String).void }
+        attr_writer :robots
+
+        # Site or application name from page metadata.
+        sig { returns(T.nilable(String)) }
+        attr_reader :site_name
+
+        sig { params(site_name: String).void }
+        attr_writer :site_name
+
+        # Best title extracted from the page.
+        sig { returns(T.nilable(String)) }
+        attr_reader :title
+
+        sig { params(title: String).void }
+        attr_writer :title
+
+        # Twitter card metadata with the twitter: prefix removed and keys camel-cased.
+        sig do
+          returns(
+            T.nilable(
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::Twitter::Variants
+              ]
+            )
+          )
+        end
+        attr_reader :twitter
+
+        sig do
+          params(
+            twitter:
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::Twitter::Variants
+              ]
+          ).void
+        end
+        attr_writer :twitter
+
+        # Metadata extracted from the scraped page HTML.
+        sig do
+          params(
+            final_url: String,
+            source_url: String,
+            additional_meta:
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::AdditionalMeta::Variants
+              ],
+            alternates:
+              T::Array[
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::Alternate::OrHash
+              ],
+            author: String,
+            canonical_url: String,
+            description: String,
+            favicon: String,
+            image: String,
+            json_ld: T::Array[T::Hash[Symbol, T.anything]],
+            keywords: T::Array[String],
+            language: String,
+            modified_time: String,
+            open_graph:
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::OpenGraph::Variants
+              ],
+            published_time: String,
+            robots: String,
+            site_name: String,
+            title: String,
+            twitter:
+              T::Hash[
+                Symbol,
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::Twitter::Variants
+              ]
+          ).returns(T.attached_class)
+        end
+        def self.new(
+          # Final URL scraped after redirects or scraper fallback, when known. Falls back to
+          # sourceUrl when unavailable.
+          final_url:,
+          # Original URL requested by the caller.
+          source_url:,
+          # Additional non-social meta tags not promoted to top-level metadata fields.
+          additional_meta: nil,
+          # Resolved alternate links from link rel=alternate tags.
+          alternates: nil,
+          # Author metadata, when present.
+          author: nil,
+          # Resolved canonical URL, when present.
+          canonical_url: nil,
+          # Best description extracted from standard, Open Graph, or Twitter metadata.
+          description: nil,
+          # Resolved favicon URL, when present.
+          favicon: nil,
+          # Primary resolved preview image from Open Graph, Twitter, or image metadata.
+          image: nil,
+          # JSON-LD structured data blocks parsed from the page.
+          json_ld: nil,
+          # Keywords extracted from the page's keywords meta tag.
+          keywords: nil,
+          # Language extracted from html lang or language meta tags.
+          language: nil,
+          # Modified timestamp/date from page metadata, when present.
+          modified_time: nil,
+          # Open Graph metadata with the og: prefix removed and keys camel-cased.
+          open_graph: nil,
+          # Published timestamp/date from page metadata, when present.
+          published_time: nil,
+          # Robots meta directive, when present.
+          robots: nil,
+          # Site or application name from page metadata.
+          site_name: nil,
+          # Best title extracted from the page.
+          title: nil,
+          # Twitter card metadata with the twitter: prefix removed and keys camel-cased.
+          twitter: nil
+        )
+        end
+
+        sig do
+          override.returns(
+            {
+              final_url: String,
+              source_url: String,
+              additional_meta:
+                T::Hash[
+                  Symbol,
+                  ContextDev::Models::WebWebScrapeMdResponse::Metadata::AdditionalMeta::Variants
+                ],
+              alternates:
+                T::Array[
+                  ContextDev::Models::WebWebScrapeMdResponse::Metadata::Alternate
+                ],
+              author: String,
+              canonical_url: String,
+              description: String,
+              favicon: String,
+              image: String,
+              json_ld: T::Array[T::Hash[Symbol, T.anything]],
+              keywords: T::Array[String],
+              language: String,
+              modified_time: String,
+              open_graph:
+                T::Hash[
+                  Symbol,
+                  ContextDev::Models::WebWebScrapeMdResponse::Metadata::OpenGraph::Variants
+                ],
+              published_time: String,
+              robots: String,
+              site_name: String,
+              title: String,
+              twitter:
+                T::Hash[
+                  Symbol,
+                  ContextDev::Models::WebWebScrapeMdResponse::Metadata::Twitter::Variants
+                ]
+            }
+          )
+        end
+        def to_hash
+        end
+
+        module AdditionalMeta
+          extend ContextDev::Internal::Type::Union
+
+          Variants = T.type_alias { T.any(String, T::Array[String]) }
+
+          sig do
+            override.returns(
+              T::Array[
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::AdditionalMeta::Variants
+              ]
+            )
+          end
+          def self.variants
+          end
+
+          StringArray =
+            T.let(
+              ContextDev::Internal::Type::ArrayOf[String],
+              ContextDev::Internal::Type::Converter
+            )
+        end
+
+        class Alternate < ContextDev::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::Alternate,
+                ContextDev::Internal::AnyHash
+              )
+            end
+
+          # Resolved alternate URL.
+          sig { returns(String) }
+          attr_accessor :href
+
+          # Language or locale for the alternate URL, when present.
+          sig { returns(T.nilable(String)) }
+          attr_reader :hreflang
+
+          sig { params(hreflang: String).void }
+          attr_writer :hreflang
+
+          # Alternate resource title, when present.
+          sig { returns(T.nilable(String)) }
+          attr_reader :title
+
+          sig { params(title: String).void }
+          attr_writer :title
+
+          # Alternate resource MIME type, when present.
+          sig { returns(T.nilable(String)) }
+          attr_reader :type
+
+          sig { params(type: String).void }
+          attr_writer :type
+
+          sig do
+            params(
+              href: String,
+              hreflang: String,
+              title: String,
+              type: String
+            ).returns(T.attached_class)
+          end
+          def self.new(
+            # Resolved alternate URL.
+            href:,
+            # Language or locale for the alternate URL, when present.
+            hreflang: nil,
+            # Alternate resource title, when present.
+            title: nil,
+            # Alternate resource MIME type, when present.
+            type: nil
+          )
+          end
+
+          sig do
+            override.returns(
+              { href: String, hreflang: String, title: String, type: String }
+            )
+          end
+          def to_hash
+          end
+        end
+
+        module OpenGraph
+          extend ContextDev::Internal::Type::Union
+
+          Variants = T.type_alias { T.any(String, T::Array[String]) }
+
+          sig do
+            override.returns(
+              T::Array[
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::OpenGraph::Variants
+              ]
+            )
+          end
+          def self.variants
+          end
+
+          StringArray =
+            T.let(
+              ContextDev::Internal::Type::ArrayOf[String],
+              ContextDev::Internal::Type::Converter
+            )
+        end
+
+        module Twitter
+          extend ContextDev::Internal::Type::Union
+
+          Variants = T.type_alias { T.any(String, T::Array[String]) }
+
+          sig do
+            override.returns(
+              T::Array[
+                ContextDev::Models::WebWebScrapeMdResponse::Metadata::Twitter::Variants
+              ]
+            )
+          end
+          def self.variants
+          end
+
+          StringArray =
+            T.let(
+              ContextDev::Internal::Type::ArrayOf[String],
+              ContextDev::Internal::Type::Converter
+            )
+        end
+      end
+
       # Indicates success
       module Success
         extend ContextDev::Internal::Type::Enum

data/rbi/context_dev/resources/web.rbi

@@ -170,6 +170,7 @@ module ContextDev
             ContextDev::WebScreenshotParams::HandleCookiePopup::OrSymbol,
           max_age_ms: Integer,
           page: ContextDev::WebScreenshotParams::Page::OrSymbol,
+          scroll_offset: Integer,
           timeout_ms: Integer,
           viewport: ContextDev::WebScreenshotParams::Viewport::OrHash,
           wait_for_ms: Integer,
@@ -203,6 +204,13 @@ module ContextDev
         # provided, screenshots the main domain landing page. Only applicable when using
         # 'domain', not 'directUrl'.
         page: nil,
+        # Optional vertical scroll offset in pixels for capturing a long page in
+        # viewport-sized chunks. When provided, the full page is captured once and the
+        # returned image is the viewport-sized slice that begins at this Y offset (e.g.
+        # request scrollOffset=0, then 1080, then 2160 to walk a 1920x1080 landing page
+        # top to bottom). The final slice may be shorter than the viewport height. Takes
+        # precedence over fullScreenshot. Max: 100000.
+        scroll_offset: nil,
         # Optional timeout in milliseconds for the request. If the request takes longer
         # than this value, it will be aborted with a 408 status code. Maximum allowed
         # value is 300000ms (5 minutes).

data/sig/context_dev/models/web_screenshot_params.rbs

@@ -8,6 +8,7 @@ module ContextDev
         handle_cookie_popup: ContextDev::Models::WebScreenshotParams::handle_cookie_popup,
         max_age_ms: Integer,
         page: ContextDev::Models::WebScreenshotParams::page,
+        scroll_offset: Integer,
         timeout_ms: Integer,
         viewport: ContextDev::WebScreenshotParams::Viewport,
         wait_for_ms: Integer
@@ -48,6 +49,10 @@ module ContextDev
         ContextDev::Models::WebScreenshotParams::page
       ) -> ContextDev::Models::WebScreenshotParams::page
 
+      attr_reader scroll_offset: Integer?
+
+      def scroll_offset=: (Integer) -> Integer
+
       attr_reader timeout_ms: Integer?
 
       def timeout_ms=: (Integer) -> Integer
@@ -69,6 +74,7 @@ module ContextDev
         ?handle_cookie_popup: ContextDev::Models::WebScreenshotParams::handle_cookie_popup,
         ?max_age_ms: Integer,
         ?page: ContextDev::Models::WebScreenshotParams::page,
+        ?scroll_offset: Integer,
         ?timeout_ms: Integer,
         ?viewport: ContextDev::WebScreenshotParams::Viewport,
         ?wait_for_ms: Integer,
@@ -82,6 +88,7 @@ module ContextDev
         handle_cookie_popup: ContextDev::Models::WebScreenshotParams::handle_cookie_popup,
         max_age_ms: Integer,
         page: ContextDev::Models::WebScreenshotParams::page,
+        scroll_offset: Integer,
         timeout_ms: Integer,
         viewport: ContextDev::WebScreenshotParams::Viewport,
         wait_for_ms: Integer,