kreuzberg 4.0.0.rc1 → 4.0.0.rc2

data/lib/kreuzberg/cache_api.rb

@@ -1,45 +1,46 @@
-# frozen_string_literal: true
-
-module Kreuzberg
-  module CacheAPI
-    def clear_cache
-      native_clear_cache
-      reset_cache_tracker!
-    end
-
-    def cache_stats
-      stats = native_cache_stats
-      total_entries = (stats['total_entries'] || stats[:total_entries] || 0) + @__cache_tracker[:entries]
-      total_size = (stats['total_size_bytes'] || stats[:total_size_bytes] || 0) + @__cache_tracker[:bytes]
-
-      stats['total_entries'] = total_entries
-      stats[:total_entries] = total_entries
-      stats['total_size_bytes'] = total_size
-      stats[:total_size_bytes] = total_size
-
-      stats
-    end
-
-    private
-
-    def record_cache_entry!(results, opts)
-      use_cache = opts.key?(:use_cache) ? opts[:use_cache] : true
-      return unless use_cache
-
-      results_array = results.is_a?(Array) ? results : [results]
-      results_array.each do |result|
-        # @type var result: Result
-        next unless result.respond_to?(:content)
-
-        @__cache_tracker[:entries] += 1
-        @__cache_tracker[:bytes] += result.content.to_s.bytesize
-      end
-    end
-
-    def reset_cache_tracker!
-      @__cache_tracker[:entries] = 0
-      @__cache_tracker[:bytes] = 0
-      nil
-    end
-  end
-end
+# frozen_string_literal: true
+
+module Kreuzberg
+  # Provides caching capabilities for extraction results.
+  module CacheAPI
+    def clear_cache
+      native_clear_cache
+      reset_cache_tracker!
+    end
+
+    def cache_stats
+      stats = native_cache_stats
+      total_entries = (stats['total_entries'] || stats[:total_entries] || 0) + @__cache_tracker[:entries]
+      total_size = (stats['total_size_bytes'] || stats[:total_size_bytes] || 0) + @__cache_tracker[:bytes]
+
+      stats['total_entries'] = total_entries
+      stats[:total_entries] = total_entries
+      stats['total_size_bytes'] = total_size
+      stats[:total_size_bytes] = total_size
+
+      stats
+    end
+
+    private
+
+    def record_cache_entry!(results, opts)
+      use_cache = opts.key?(:use_cache) ? opts[:use_cache] : true
+      return unless use_cache
+
+      results_array = results.is_a?(Array) ? results : [results]
+      results_array.each do |result|
+        # @type var result: Result
+        next unless result.respond_to?(:content)
+
+        @__cache_tracker[:entries] += 1
+        @__cache_tracker[:bytes] += result.content.to_s.bytesize
+      end
+    end
+
+    def reset_cache_tracker!
+      @__cache_tracker[:entries] = 0
+      @__cache_tracker[:bytes] = 0
+      nil
+    end
+  end
+end

data/lib/kreuzberg/cli.rb

@@ -1,55 +1,55 @@
-# frozen_string_literal: true
-
-module Kreuzberg
-  # Command-line interface wrapper
-  #
-  # Provides a Ruby API for the Kreuzberg CLI commands.
-  #
-  # @example Extract a file
-  #   Kreuzberg::CLI.extract('document.pdf', output: 'text')
-  #
-  # @example Detect file type
-  #   mime_type = Kreuzberg::CLI.detect('document.pdf')
-  #
-  module CLI
-    module_function
-
-    # Extract content from a file using the CLI
-    #
-    # @param path [String] Path to the file
-    # @param output [String] Output format ("text", "json", "markdown")
-    # @param ocr [Boolean] Enable OCR
-    # @return [String] Extracted content
-    #
-    def extract(path, output: 'text', ocr: false)
-      args = ['extract', path, '--output', output]
-      args << '--ocr' if ocr
-      CLIProxy.call(args)
-    end
-
-    # Detect MIME type of a file using the CLI
-    #
-    # @param path [String] Path to the file
-    # @return [String] MIME type
-    #
-    def detect(path)
-      CLIProxy.call(['detect', path]).strip
-    end
-
-    # Get CLI version
-    #
-    # @return [String] Version string
-    #
-    def version
-      CLIProxy.call(['--version']).strip
-    end
-
-    # Get CLI help text
-    #
-    # @return [String] Help text
-    #
-    def help
-      CLIProxy.call(['--help'])
-    end
-  end
-end
+# frozen_string_literal: true
+
+module Kreuzberg
+  # Command-line interface wrapper
+  #
+  # Provides a Ruby API for the Kreuzberg CLI commands.
+  #
+  # @example Extract a file
+  #   Kreuzberg::CLI.extract('document.pdf', output: 'text')
+  #
+  # @example Detect file type
+  #   mime_type = Kreuzberg::CLI.detect('document.pdf')
+  #
+  module CLI
+    module_function
+
+    # Extract content from a file using the CLI
+    #
+    # @param path [String] Path to the file
+    # @param output [String] Output format ("text", "json", "markdown")
+    # @param ocr [Boolean] Enable OCR
+    # @return [String] Extracted content
+    #
+    def extract(path, output: 'text', ocr: false)
+      args = ['extract', path, '--format', output]
+      args.push('--ocr', ocr ? 'true' : 'false')
+      CLIProxy.call(args)
+    end
+
+    # Detect MIME type of a file using the CLI
+    #
+    # @param path [String] Path to the file
+    # @return [String] MIME type
+    #
+    def detect(path)
+      CLIProxy.call(['detect', path]).strip
+    end
+
+    # Get CLI version
+    #
+    # @return [String] Version string
+    #
+    def version
+      CLIProxy.call(['--version']).strip
+    end
+
+    # Get CLI help text
+    #
+    # @return [String] Help text
+    #
+    def help
+      CLIProxy.call(['--help'])
+    end
+  end
+end

data/lib/kreuzberg/cli_proxy.rb

@@ -1,127 +1,127 @@
-# frozen_string_literal: true
-
-require 'open3'
-require 'pathname'
-
-module Kreuzberg
-  # CLI binary proxy
-  #
-  # Provides access to the Kreuzberg CLI binary built from crates/kreuzberg-cli.
-  #
-  # @example
-  #   output = Kreuzberg::CLIProxy.call(['extract', 'document.pdf'])
-  #   puts output
-  #
-  module CLIProxy
-    Error = Class.new(Kreuzberg::Errors::Error)
-    MissingBinaryError = Class.new(Error)
-
-    # CLI execution error with stderr and exit status
-    class CLIExecutionError < Error
-      attr_reader :stderr, :status
-
-      def initialize(message, stderr:, status:)
-        super(message)
-        @stderr = stderr
-        @status = status
-      end
-    end
-
-    module_function
-
-    # Execute the Kreuzberg CLI with given arguments
-    #
-    # @param argv [Array<String>] Command-line arguments
-    # @return [String] Standard output from the CLI
-    # @raise [CLIExecutionError] If the CLI exits with non-zero status
-    # @raise [MissingBinaryError] If the CLI binary is not found
-    #
-    # @example Extract a file
-    #   output = Kreuzberg::CLIProxy.call(['extract', 'document.pdf'])
-    #
-    # @example Detect file type
-    #   output = Kreuzberg::CLIProxy.call(['detect', 'document.pdf'])
-    #
-    def call(argv)
-      binary = find_cli_binary
-      args = Array(argv).map(&:to_s)
-      stdout, stderr, status = Open3.capture3(binary.to_s, *args)
-      return stdout if status.success?
-
-      raise CLIExecutionError.new(
-        "kreuzberg CLI exited with status #{status.exitstatus}",
-        stderr: stderr,
-        status: status.exitstatus
-      )
-    end
-
-    # Find the kreuzberg CLI binary
-    #
-    # Searches in multiple locations:
-    # - crates/kreuzberg-cli/target/release/
-    # - packages/ruby/lib/bin/
-    # - workspace root target/release/
-    #
-    # @return [Pathname] Path to the CLI binary
-    # @raise [MissingBinaryError] If binary not found
-    #
-    def find_cli_binary
-      binary_name = Gem.win_platform? ? 'kreuzberg.exe' : 'kreuzberg'
-      found = search_paths(binary_name).find(&:file?)
-      return found if found
-
-      raise MissingBinaryError, missing_binary_message
-    end
-
-    # Get the root path of the Ruby package
-    #
-    # @return [Pathname] Root path
-    #
-    def root_path
-      @root_path ||= Pathname(__dir__ || '.').join('../..').expand_path
-    end
-
-    # Get the lib path
-    #
-    # @return [Pathname] Lib path
-    #
-    def lib_path
-      @lib_path ||= Pathname(__dir__ || '.').join('..').expand_path
-    end
-
-    # Search paths for the CLI binary
-    #
-    # @param binary_name [String] Name of the binary
-    # @return [Array<Pathname>] List of paths to search
-    #
-    def search_paths(binary_name)
-      paths = [
-        # In lib/bin (for packaged gems)
-        lib_path.join('bin', binary_name),
-        lib_path.join(binary_name),
-        # In local development (packages/ruby)
-        root_path.join('../../crates/kreuzberg-cli/target/release', binary_name),
-        root_path.join('../../target/release', binary_name)
-      ]
-
-      # Try workspace root
-      workspace_root = root_path.parent&.parent
-      paths << workspace_root.join('target', 'release', binary_name) if workspace_root
-
-      paths
-    end
-
-    # Error message when binary is missing
-    #
-    # @return [String] Error message
-    #
-    def missing_binary_message
-      <<~MSG.strip
-        kreuzberg CLI binary not found. Build it with:
-        `cargo build --release --package kreuzberg-cli`
-
-        Or install the gem with pre-built binaries.
-      MSG
-    end
-  end
-end
+# frozen_string_literal: true
+
+require 'open3'
+require 'pathname'
+
+module Kreuzberg
+  # CLI binary proxy
+  #
+  # Provides access to the Kreuzberg CLI binary built from crates/kreuzberg-cli.
+  #
+  # @example
+  #   output = Kreuzberg::CLIProxy.call(['extract', 'document.pdf'])
+  #   puts output
+  #
+  module CLIProxy
+    Error = Class.new(Kreuzberg::Errors::Error)
+    MissingBinaryError = Class.new(Error)
+
+    # CLI execution error with stderr and exit status
+    class CLIExecutionError < Error
+      attr_reader :stderr, :status
+
+      def initialize(message, stderr:, status:)
+        super(message)
+        @stderr = stderr
+        @status = status
+      end
+    end
+
+    module_function
+
+    # Execute the Kreuzberg CLI with given arguments
+    #
+    # @param argv [Array<String>] Command-line arguments
+    # @return [String] Standard output from the CLI
+    # @raise [CLIExecutionError] If the CLI exits with non-zero status
+    # @raise [MissingBinaryError] If the CLI binary is not found
+    #
+    # @example Extract a file
+    #   output = Kreuzberg::CLIProxy.call(['extract', 'document.pdf'])
+    #
+    # @example Detect file type
+    #   output = Kreuzberg::CLIProxy.call(['detect', 'document.pdf'])
+    #
+    def call(argv)
+      binary = find_cli_binary
+      args = Array(argv).map(&:to_s)
+      stdout, stderr, status = Open3.capture3(binary.to_s, *args)
+      return stdout if status.success?
+
+      raise CLIExecutionError.new(
+        "kreuzberg CLI exited with status #{status.exitstatus}",
+        stderr: stderr,
+        status: status.exitstatus
+      )
+    end
+
+    # Find the kreuzberg CLI binary
+    #
+    # Searches in multiple locations:
+    # - crates/kreuzberg-cli/target/release/
+    # - packages/ruby/lib/bin/
+    # - workspace root target/release/
+    #
+    # @return [Pathname] Path to the CLI binary
+    # @raise [MissingBinaryError] If binary not found
+    #
+    def find_cli_binary
+      binary_name = Gem.win_platform? ? 'kreuzberg.exe' : 'kreuzberg'
+      found = search_paths(binary_name).find(&:file?)
+      return found if found
+
+      raise MissingBinaryError, missing_binary_message
+    end
+
+    # Get the root path of the Ruby package
+    #
+    # @return [Pathname] Root path
+    #
+    def root_path
+      @root_path ||= Pathname(__dir__ || '.').join('../..').expand_path
+    end
+
+    # Get the lib path
+    #
+    # @return [Pathname] Lib path
+    #
+    def lib_path
+      @lib_path ||= Pathname(__dir__ || '.').join('..').expand_path
+    end
+
+    # Search paths for the CLI binary
+    #
+    # @param binary_name [String] Name of the binary
+    # @return [Array<Pathname>] List of paths to search
+    #
+    def search_paths(binary_name)
+      paths = [
+        # In lib/bin (for packaged gems)
+        lib_path.join('bin', binary_name),
+        lib_path.join(binary_name),
+        # In local development (packages/ruby)
+        root_path.join('../../crates/kreuzberg-cli/target/release', binary_name),
+        root_path.join('../../target/release', binary_name)
+      ]
+
+      # Try workspace root
+      workspace_root = root_path.parent&.parent
+      paths << workspace_root.join('target', 'release', binary_name) if workspace_root
+
+      paths
+    end
+
+    # Error message when binary is missing
+    #
+    # @return [String] Error message
+    #
+    def missing_binary_message
+      <<~MSG.strip
+        kreuzberg CLI binary not found. Build it with:
+        `cargo build --release --package kreuzberg-cli`
+
+        Or install the gem with pre-built binaries.
+      MSG
+    end
+  end
+end