html-to-markdown 3.8.0 → 3.8.2

data/lib/html_to_markdown/native.rb

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:411256ebae58a5a5ab1b77fe1f91fe158a88eecdc7709575c93a6e4ed499f324
+# alef:hash:df03cfa1ea58fc063ec7e305db2e9dcdc739bb4b063899fc4750f8584c5bd37e
 # To regenerate: alef generate
 # To verify freshness: alef verify --exit-code
 # frozen_string_literal: true
@@ -7,6 +7,7 @@
 require "json"
 require "sorbet-runtime"
 require "html_to_markdown_rb"
+
 module HtmlToMarkdown
   # Re-export public types from the native extension (curated list, excludes Update/Builder types)
   ConversionOptions = HtmlToMarkdownRs.const_get(:ConversionOptions)
@@ -29,8 +30,11 @@ module HtmlToMarkdown
   TableGrid = HtmlToMarkdownRs.const_get(:TableGrid)
   TextAnnotation = HtmlToMarkdownRs.const_get(:TextAnnotation)
   # Re-export public module functions from the native extension (curated list)
-  define_singleton_method(:convert) { |*args, **kwargs, &blk| HtmlToMarkdownRs.public_send(:convert, *args, **kwargs, &blk) }
+  define_singleton_method(:convert) { |*args, **kwargs, &blk|
+    HtmlToMarkdownRs.public_send(:convert, *args, **kwargs, &blk)
+  }
 end
+
 module HtmlToMarkdown
   # The semantic content type of a document node.
   #
@@ -48,20 +52,34 @@ module HtmlToMarkdown
     def self.from_hash(hash)
       discriminator = hash[:node_type] || hash["node_type"]
       case discriminator
-      when "heading" then NodeContentHeading.from_hash(hash)
-      when "paragraph" then NodeContentParagraph.from_hash(hash)
-      when "list" then NodeContentList.from_hash(hash)
-      when "list_item" then NodeContentListItem.from_hash(hash)
-      when "table" then NodeContentTable.from_hash(hash)
-      when "image" then NodeContentImage.from_hash(hash)
-      when "code" then NodeContentCode.from_hash(hash)
-      when "quote" then NodeContentQuote.from_hash(hash)
-      when "definition_list" then NodeContentDefinitionList.from_hash(hash)
-      when "definition_item" then NodeContentDefinitionItem.from_hash(hash)
-      when "raw_block" then NodeContentRawBlock.from_hash(hash)
-      when "metadata_block" then NodeContentMetadataBlock.from_hash(hash)
-      when "group" then NodeContentGroup.from_hash(hash)
-      else raise "Unknown discriminator: #{discriminator}"
+      when "heading"
+        NodeContentHeading.from_hash(hash)
+      when "paragraph"
+        NodeContentParagraph.from_hash(hash)
+      when "list"
+        NodeContentList.from_hash(hash)
+      when "list_item"
+        NodeContentListItem.from_hash(hash)
+      when "table"
+        NodeContentTable.from_hash(hash)
+      when "image"
+        NodeContentImage.from_hash(hash)
+      when "code"
+        NodeContentCode.from_hash(hash)
+      when "quote"
+        NodeContentQuote.from_hash(hash)
+      when "definition_list"
+        NodeContentDefinitionList.from_hash(hash)
+      when "definition_item"
+        NodeContentDefinitionItem.from_hash(hash)
+      when "raw_block"
+        NodeContentRawBlock.from_hash(hash)
+      when "metadata_block"
+        NodeContentMetadataBlock.from_hash(hash)
+      when "group"
+        NodeContentGroup.from_hash(hash)
+      else
+        raise "Unknown discriminator: #{discriminator}"
       end
     end
   end
@@ -72,10 +90,12 @@ module HtmlToMarkdown
 
     # Heading level (1-6).
     sig { returns(Integer) }
-    def level = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def level = super
     # The heading text content.
     sig { returns(String) }
-    def text = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def text = super
     sig { returns(T::Boolean) }
     def heading? = true
     sig { returns(T::Boolean) }
@@ -116,7 +136,8 @@ module HtmlToMarkdown
 
     # The paragraph text content.
     sig { returns(String) }
-    def text = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def text = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -157,7 +178,8 @@ module HtmlToMarkdown
 
     # Whether this is an ordered list.
     sig { returns(T::Boolean) }
-    def ordered = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def ordered = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -198,7 +220,8 @@ module HtmlToMarkdown
 
     # The list item text content.
     sig { returns(String) }
-    def text = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def text = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -239,7 +262,8 @@ module HtmlToMarkdown
 
     # The table grid structure.
     sig { returns(TableGrid) }
-    def grid = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def grid = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -280,13 +304,16 @@ module HtmlToMarkdown
 
     # Alt text or caption.
     sig { returns(T.nilable(String)) }
-    def description = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def description = super
     # Image source URL.
     sig { returns(T.nilable(String)) }
-    def src = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def src = super
     # Index into `ConversionResult.images` when image extraction is enabled.
     sig { returns(T.nilable(Integer)) }
-    def image_index = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def image_index = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -317,7 +344,11 @@ module HtmlToMarkdown
     # @return [self]
     sig { params(hash: T::Hash[T.untyped, T.untyped]).returns(T.attached_class) }
     def self.from_hash(hash)
-      new(description: hash[:description] || hash["description"], src: hash[:src] || hash["src"], image_index: hash[:image_index] || hash["image_index"])
+      new(
+        description: hash[:description] || hash["description"],
+        src: hash[:src] || hash["src"],
+        image_index: hash[:image_index] || hash["image_index"]
+      )
     end
   end
   ## A code block or inline code.
@@ -327,10 +358,12 @@ module HtmlToMarkdown
 
     # The code text content.
     sig { returns(String) }
-    def text = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def text = super
     # Programming language (from class="language-*" or similar).
     sig { returns(T.nilable(String)) }
-    def language = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def language = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -447,10 +480,12 @@ module HtmlToMarkdown
 
     # The term being defined.
     sig { returns(String) }
-    def term = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def term = super
     # The definition text.
     sig { returns(String) }
-    def definition = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def definition = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -491,10 +526,12 @@ module HtmlToMarkdown
 
     # The format of the raw content (e.g. "html", "css", "javascript").
     sig { returns(String) }
-    def format = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def format = super
     # The raw content.
     sig { returns(String) }
-    def content = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def content = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -535,7 +572,8 @@ module HtmlToMarkdown
 
     # Key-value metadata pairs.
     sig { returns(T::Array[MetadataEntry]) }
-    def entries = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def entries = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -576,13 +614,16 @@ module HtmlToMarkdown
 
     # Optional section label.
     sig { returns(T.nilable(String)) }
-    def label = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def label = super
     # The heading level that created this group.
     sig { returns(T.nilable(Integer)) }
-    def heading_level = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def heading_level = super
     # The heading text that created this group.
     sig { returns(T.nilable(String)) }
-    def heading_text = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def heading_text = super
     sig { returns(T::Boolean) }
     def heading? = false
     sig { returns(T::Boolean) }
@@ -613,7 +654,11 @@ module HtmlToMarkdown
     # @return [self]
     sig { params(hash: T::Hash[T.untyped, T.untyped]).returns(T.attached_class) }
     def self.from_hash(hash)
-      new(label: hash[:label] || hash["label"], heading_level: hash[:heading_level] || hash["heading_level"], heading_text: hash[:heading_text] || hash["heading_text"])
+      new(
+        label: hash[:label] || hash["label"],
+        heading_level: hash[:heading_level] || hash["heading_level"],
+        heading_text: hash[:heading_text] || hash["heading_text"]
+      )
     end
   end
 end
@@ -635,16 +680,26 @@ module HtmlToMarkdown
     def self.from_hash(hash)
       discriminator = hash[:annotation_type] || hash["annotation_type"]
       case discriminator
-      when "bold" then AnnotationKindBold.from_hash(hash)
-      when "italic" then AnnotationKindItalic.from_hash(hash)
-      when "underline" then AnnotationKindUnderline.from_hash(hash)
-      when "strikethrough" then AnnotationKindStrikethrough.from_hash(hash)
-      when "code" then AnnotationKindCode.from_hash(hash)
-      when "subscript" then AnnotationKindSubscript.from_hash(hash)
-      when "superscript" then AnnotationKindSuperscript.from_hash(hash)
-      when "highlight" then AnnotationKindHighlight.from_hash(hash)
-      when "link" then AnnotationKindLink.from_hash(hash)
-      else raise "Unknown discriminator: #{discriminator}"
+      when "bold"
+        AnnotationKindBold.from_hash(hash)
+      when "italic"
+        AnnotationKindItalic.from_hash(hash)
+      when "underline"
+        AnnotationKindUnderline.from_hash(hash)
+      when "strikethrough"
+        AnnotationKindStrikethrough.from_hash(hash)
+      when "code"
+        AnnotationKindCode.from_hash(hash)
+      when "subscript"
+        AnnotationKindSubscript.from_hash(hash)
+      when "superscript"
+        AnnotationKindSuperscript.from_hash(hash)
+      when "highlight"
+        AnnotationKindHighlight.from_hash(hash)
+      when "link"
+        AnnotationKindLink.from_hash(hash)
+      else
+        raise "Unknown discriminator: #{discriminator}"
       end
     end
   end
@@ -900,13 +955,15 @@ module HtmlToMarkdown
     # written in the source. Callers that need an absolute URL must resolve it against the
     # document base URL themselves.
     sig { returns(String) }
-    def url = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def url = super
     # The `title` attribute of the `<a>` element, if present.
     #
     # `None` when the `<a>` tag has no `title="..."` attribute. When present, the value
     # is copied verbatim — HTML entities within the title are not decoded.
     sig { returns(T.nilable(String)) }
-    def title = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def title = super
     sig { returns(T::Boolean) }
     def bold? = false
     sig { returns(T::Boolean) }

data/lib/html_to_markdown/version.rb

@@ -1,10 +1,10 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:411256ebae58a5a5ab1b77fe1f91fe158a88eecdc7709575c93a6e4ed499f324
+# alef:hash:df03cfa1ea58fc063ec7e305db2e9dcdc739bb4b063899fc4750f8584c5bd37e
 # To regenerate: alef generate
 # To verify freshness: alef verify --exit-code
 # frozen_string_literal: true
 
 module HtmlToMarkdown
   ## The version string for this package.
-  VERSION = "3.8.0"
+  VERSION = "3.8.2"
 end

data/lib/html_to_markdown.rb

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:411256ebae58a5a5ab1b77fe1f91fe158a88eecdc7709575c93a6e4ed499f324
+# alef:hash:df03cfa1ea58fc063ec7e305db2e9dcdc739bb4b063899fc4750f8584c5bd37e
 # To regenerate: alef generate
 # To verify freshness: alef verify --exit-code
 # frozen_string_literal: true
@@ -16,3 +16,12 @@ require_relative "html_to_markdown/native"
 module HtmlToMarkdown
   # Re-export all types and functions from native extension
 end
+
+# Bring top-level HtmlToMarkdown classes into the global namespace so callers
+# (and the generated e2e suite) can reference them unqualified. The native
+# extension has already been required above, so every type constant is defined
+# under HtmlToMarkdown by this point.
+HtmlToMarkdown.constants.each do |const_name|
+  value = HtmlToMarkdown.const_get(const_name)
+  ::Object.const_set(const_name, value) if value.is_a?(Module) && !::Object.const_defined?(const_name)
+end

data/sig/html_to_markdown/cli.rbs

@@ -1,24 +1,24 @@
 module HtmlToMarkdown
-  module CLI
-    # Module method (module_function creates both module and instance methods)
-    #
-    # Run the CLI with the given arguments
-    #
-    # @param argv Command-line arguments (defaults to ARGV)
-    # @param stdout Output stream for standard output
-    # @param stderr Output stream for standard error
-    # @return Exit code (0 for success, non-zero for failure)
-    def self.run: (
-      ?Array[String] argv,
-      ?stdout: IO,
-      ?stderr: IO
-    ) -> Integer
+    module CLI
+        # Module method (module_function creates both module and instance methods)
+        #
+        # Run the CLI with the given arguments
+        #
+        # @param argv Command-line arguments (defaults to ARGV)
+        # @param stdout Output stream for standard output
+        # @param stderr Output stream for standard error
+        # @return Exit code (0 for success, non-zero for failure)
+        def self.run: (
+        ?Array[String] argv,
+            ?stdout: IO,
+            ?stderr: IO
+        ) -> Integer
 
-    # Instance method version (created by module_function)
-    def run: (
-      ?Array[String] argv,
-      ?stdout: IO,
-      ?stderr: IO
-    ) -> Integer
-  end
+        # Instance method version (created by module_function)
+        def run: (
+        ?Array[String] argv,
+            ?stdout: IO,
+            ?stderr: IO
+        ) -> Integer
+    end
 end

data/sig/html_to_markdown/cli_proxy.rbs

@@ -1,48 +1,48 @@
 module HtmlToMarkdown
-  module CLIProxy
-    # Base error class
-    class Error < StandardError
-    end
+    module CLIProxy
+        # Base error class
+        class Error < StandardError
+        end
 
-    # Error when CLI binary is not found
-    class MissingBinaryError < Error
-    end
+        # Error when CLI binary is not found
+        class MissingBinaryError < Error
+        end
 
-    # Error when CLI execution fails
-    class CLIExecutionError < Error
-      attr_reader stderr: String
-      attr_reader status: Integer?
+        # Error when CLI execution fails
+        class CLIExecutionError < Error
+            attr_reader stderr: String
+            attr_reader status: Integer?
 
-      def initialize: (String message, stderr: String, status: Integer?) -> void
-    end
+        def initialize: (String message, stderr: String, status: Integer?) -> void
+        end
 
-    # Module methods (module_function creates both module and instance methods)
+        # Module methods (module_function creates both module and instance methods)
 
-    # Execute CLI with given arguments
-    def self.call: (Array[String] argv) -> String
+        # Execute CLI with given arguments
+def self.call: (Array[String] argv) -> String
 
-    # Find the CLI binary in search paths
+        # Find the CLI binary in search paths
     def self.find_cli_binary: () -> Pathname
 
-    # Get root path of the gem
+        # Get root path of the gem
     def self.root_path: () -> Pathname
 
-    # Get lib path of the gem
+        # Get lib path of the gem
     def self.lib_path: () -> Pathname
 
-    # Get search paths for CLI binary
-    def self.search_paths: (String binary_name) -> Array[Pathname]
+        # Get search paths for CLI binary
+def self.search_paths: (String binary_name) -> Array[Pathname]
 
-    # Get error message for missing binary
+        # Get error message for missing binary
     def self.missing_binary_message: () -> String
 
-    # Instance method versions (created by module_function)
+        # Instance method versions (created by module_function)
 
-    def call: (Array[String] argv) -> String
+def call: (Array[String] argv) -> String
     def find_cli_binary: () -> Pathname
     def root_path: () -> Pathname
     def lib_path: () -> Pathname
-    def search_paths: (String binary_name) -> Array[Pathname]
+def search_paths: (String binary_name) -> Array[Pathname]
     def missing_binary_message: () -> String
-  end
+    end
 end

data/sig/open3.rbs

@@ -1,12 +1,12 @@
 # Type signature for Open3 standard library
 module Open3
-  # Execute command and capture stdout, stderr, and status
-  #
-  # @param cmd Command to execute
-  # @param args Command arguments
-  # @return Array containing stdout (String), stderr (String), and status (Process::Status)
-  def self.capture3: (
-    String cmd,
-    *String args
-  ) -> [String, String, Process::Status]
+    # Execute command and capture stdout, stderr, and status
+    #
+    # @param cmd Command to execute
+    # @param args Command arguments
+    # @return Array containing stdout (String), stderr (String), and status (Process::Status)
+    def self.capture3: (
+        String cmd,
+        *String args
+) -> [String, String, Process::Status]
 end