kreuzberg 4.0.0.pre.rc.29 → 4.0.0.rc1

data/lib/kreuzberg/error_context.rb

@@ -1,136 +0,0 @@
-# frozen_string_literal: true
-
-require 'json'
-
-module Kreuzberg
-  module ErrorContext
-    class << self
-      # @return [Integer] Error code constant (ERROR_CODE_* values), or 0 on success
-      # @example Check last error
-      def last_error_code
-        Kreuzberg._last_error_code_native
-      rescue StandardError
-        0
-      end
-
-      # Get panic context information from the last error.
-      #
-      # Returns a {Errors::PanicContext} object containing detailed information about
-      # the last panic that occurred in the Rust core. Includes file path, line number,
-      # function name, error message, and timestamp.
-      #
-      # @return [Errors::PanicContext, nil] Panic context if a panic occurred, nil otherwise
-      #
-      # @example Get panic details
-      #   panic = Kreuzberg::ErrorContext.last_panic_context
-      #   if panic
-      #     puts "Panic at #{panic.file}:#{panic.line} in #{panic.function}"
-      #     puts "Message: #{panic.message}"
-      #     puts "Time: #{panic.timestamp_secs}"
-      #   end
-      def last_panic_context
-        json_str = Kreuzberg._last_panic_context_json_native
-        return nil unless json_str
-
-        Errors::PanicContext.from_json(json_str)
-      rescue StandardError
-        nil
-      end
-
-      # Get panic context as raw JSON string.
-      #
-      # Returns the panic context information as a JSON string for raw access or
-      # custom parsing. Returns nil if no panic has occurred.
-      #
-      # @return [String, nil] JSON-serialized panic context, or nil if no panic
-      #
-      # @example Get raw JSON panic context
-      #   json = Kreuzberg::ErrorContext.last_panic_context_json
-      #   if json
-      #     panic_data = JSON.parse(json)
-      #     puts panic_data
-      #   end
-      def last_panic_context_json
-        Kreuzberg._last_panic_context_json_native
-      rescue StandardError
-        nil
-      end
-
-      # Get detailed error information from the last operation.
-      #
-      # Returns comprehensive error details including message, code, type, source location,
-      # and panic information.
-      #
-      # @return [Hash] Hash with keys: :message, :error_code, :error_type, :source_file,
-      #   :source_function, :source_line, :context_info, :is_panic
-      #
-      # @example Get error details
-      #   details = Kreuzberg::ErrorContext.error_details
-      #   puts "Error: #{details[:message]}"
-      #   puts "Code: #{details[:error_code]}"
-      #   puts "Type: #{details[:error_type]}"
-      def error_details
-        Kreuzberg._get_error_details_native
-      rescue StandardError
-        {}
-      end
-
-      # Classify an error message into a Kreuzberg error code.
-      #
-      # Analyzes an error message and returns the most likely error code (0-7).
-      # Useful for converting third-party error messages into Kreuzberg categories.
-      #
-      # @param message [String] The error message to classify
-      # @return [Integer] Error code (0-7)
-      #
-      # Error code mapping:
-      # - 0: Validation
-      # - 1: Parsing
-      # - 2: OCR
-      # - 3: MissingDependency
-      # - 4: IO
-      # - 5: Plugin
-      # - 6: UnsupportedFormat
-      # - 7: Internal
-      #
-      # @example Classify an error
-      #   code = Kreuzberg::ErrorContext.classify_error("File not found")
-      #   if code == 4
-      #     puts "This is an I/O error"
-      #   end
-      def classify_error(message)
-        Kreuzberg._classify_error_native(message)
-      rescue StandardError
-        7
-      end
-
-      # Get the human-readable name of an error code.
-      #
-      # @param code [Integer] Numeric error code (0-7)
-      # @return [String] Human-readable error code name (e.g., "validation", "io")
-      #
-      # @example Get error code name
-      #   name = Kreuzberg::ErrorContext.error_code_name(0)
-      #   puts name  # => "validation"
-      def error_code_name(code)
-        Kreuzberg._error_code_name_native(code)
-      rescue StandardError
-        'unknown'
-      end
-
-      # Get the description of an error code.
-      #
-      # @param code [Integer] Numeric error code (0-7)
-      # @return [String] Description of the error code
-      #
-      # @example Get error code description
-      #   desc = Kreuzberg::ErrorContext.error_code_description(0)
-      #   puts desc  # => "Input validation error"
-      def error_code_description(code)
-        Kreuzberg._error_code_description_native(code)
-      rescue StandardError
-        'Unknown error code'
-      end
-    end
-  end
-end

data/lib/kreuzberg/types.rb

@@ -1,170 +0,0 @@
-# frozen_string_literal: true
-
-require 'sorbet-runtime'
-
-module Kreuzberg
-  # @example
-  class HtmlMetadata < T::Struct
-    extend T::Sig
-
-    const :title, T.nilable(String)
-
-    const :description, T.nilable(String)
-
-    const :author, T.nilable(String)
-
-    const :copyright, T.nilable(String)
-
-    const :keywords, T::Array[String]
-
-    const :canonical_url, T.nilable(String)
-
-    const :language, T.nilable(String)
-
-    const :text_direction, T.nilable(String)
-
-    const :mime_type, T.nilable(String)
-
-    const :charset, T.nilable(String)
-
-    const :generator, T.nilable(String)
-
-    const :viewport, T.nilable(String)
-
-    const :theme_color, T.nilable(String)
-
-    const :application_name, T.nilable(String)
-
-    const :robots, T.nilable(String)
-
-    const :open_graph, T::Hash[String, String]
-
-    const :twitter_card, T::Hash[String, String]
-
-    const :meta_tags, T::Hash[String, String]
-
-    const :headers, T::Array[HeaderMetadata]
-
-    const :links, T::Array[LinkMetadata]
-
-    const :images, T::Array[ImageMetadata]
-
-    const :structured_data, T::Array[StructuredData]
-  end
-
-  # Header/Heading metadata
-  #
-  # Represents a heading element found in the HTML document
-  #
-  # @example
-  #   header = Kreuzberg::HeaderMetadata.new(
-  #     level: 1,
-  #     text: "Main Title",
-  #     id: "main-title",
-  #     depth: 0,
-  #     html_offset: 245
-  #   )
-  #   puts "#{header.text} (H#{header.level})"
-  #
-  class HeaderMetadata < T::Struct
-    extend T::Sig
-
-    const :level, Integer
-
-    const :text, String
-
-    const :id, T.nilable(String)
-
-    const :depth, Integer
-
-    const :html_offset, Integer
-  end
-
-  # Link metadata
-  #
-  # Represents a link element found in the HTML document
-  #
-  # @example
-  #   link = Kreuzberg::LinkMetadata.new(
-  #     href: "https://example.com",
-  #     text: "Example",
-  #     title: "Example Site",
-  #     link_type: "external",
-  #     rel: ["noopener", "noreferrer"],
-  #     attributes: { "data-id" => "123" }
-  #   )
-  #   puts "#{link.text} -> #{link.href}"
-  #
-  class LinkMetadata < T::Struct
-    extend T::Sig
-
-    const :href, String
-
-    const :text, String
-
-    const :title, T.nilable(String)
-
-    const :link_type, String
-
-    const :rel, T::Array[String]
-
-    const :attributes, T::Hash[String, String]
-  end
-
-  # Image metadata
-  #
-  # Represents an image element found in the HTML document
-  #
-  # @example
-  #   image = Kreuzberg::ImageMetadata.new(
-  #     src: "images/logo.png",
-  #     alt: "Company Logo",
-  #     title: nil,
-  #     dimensions: [200, 100],
-  #     image_type: "png",
-  #     attributes: { "loading" => "lazy" }
-  #   )
-  #   if image.dimensions
-  #     width, height = image.dimensions
-  #     puts "#{width}x#{height}"
-  #   end
-  #
-  class ImageMetadata < T::Struct
-    extend T::Sig
-
-    const :src, String
-
-    const :alt, T.nilable(String)
-
-    const :title, T.nilable(String)
-
-    const :dimensions, T.nilable(T::Array[Integer])
-
-    const :image_type, String
-
-    const :attributes, T::Hash[String, String]
-  end
-
-  # Structured data metadata
-  #
-  # Represents structured data (JSON-LD, microdata, etc.) found in the HTML document
-  #
-  # @example
-  #   structured = Kreuzberg::StructuredData.new(
-  #     data_type: "json-ld",
-  #     raw_json: '{"@context":"https://schema.org","@type":"Article",...}',
-  #     schema_type: "Article"
-  #   )
-  #   data = JSON.parse(structured.raw_json)
-  #   puts data['@type']
-  #
-  class StructuredData < T::Struct
-    extend T::Sig
-
-    const :data_type, String
-
-    const :raw_json, String
-
-    const :schema_type, T.nilable(String)
-  end
-end