kreuzberg 4.0.0.pre.rc.13 → 4.0.0.pre.rc.15

data/lib/kreuzberg/api_proxy.rb

@@ -1,142 +1,142 @@
-# frozen_string_literal: true
-
-require 'open3'
-require 'pathname'
-
-module Kreuzberg
-  # API server proxy
-  #
-  # Starts and manages the Kreuzberg API server (Litestar/Python-based or Rust-based).
-  #
-  # @example Start the server
-  #   server = Kreuzberg::APIProxy.new(port: 8000)
-  #   server.start
-  #   # Server runs in background
-  #   server.stop
-  #
-  # @example With block
-  #   Kreuzberg::APIProxy.run(port: 8000) do |server|
-  #     # Server runs while block executes
-  #     response = Net::HTTP.get(URI('http://localhost:8000/health'))
-  #   end
-  #
-  module APIProxy
-    Error = Class.new(Kreuzberg::Errors::Error)
-    MissingBinaryError = Class.new(Error)
-    ServerError = Class.new(Error)
-
-    # API server instance
-    class Server
-      attr_reader :port, :host, :pid
-
-      # Initialize server
-      #
-      # @param port [Integer] Port to run on (default: 8000)
-      # @param host [String] Host to bind to (default: "0.0.0.0")
-      #
-      def initialize(port: 8000, host: '0.0.0.0')
-        @port = port
-        @host = host
-        @pid = nil
-        @process = nil
-      end
-
-      # Start the server in the background
-      #
-      # @return [Integer] Process ID
-      # @raise [ServerError] If server fails to start
-      #
-      def start
-        binary = APIProxy.find_api_binary
-        @pid = spawn(
-          binary.to_s,
-          'api',
-          '--host', @host,
-          '--port', @port.to_s,
-          out: $stdout,
-          err: $stderr
-        )
-        Process.detach(@pid)
-        sleep 1 # Give server time to start
-        @pid
-      end
-
-      # Stop the server
-      #
-      # @return [void]
-      #
-      def stop
-        return unless @pid
-
-        Process.kill('TERM', @pid)
-        Process.wait(@pid)
-      rescue Errno::ESRCH, Errno::ECHILD
-        # Process already dead
-      ensure
-        @pid = nil
-      end
-
-      # Check if server is running
-      #
-      # @return [Boolean]
-      #
-      def running?
-        return false unless @pid
-
-        Process.kill(0, @pid)
-        true
-      rescue Errno::ESRCH, Errno::EPERM
-        false
-      end
-    end
-
-    module_function
-
-    # Run server with a block
-    #
-    # @param port [Integer] Port to run on
-    # @param host [String] Host to bind to
-    # @yield [Server] Yields server instance
-    # @return [Object] Block result
-    #
-    # @example
-    #   Kreuzberg::APIProxy.run(port: 8000) do |server|
-    #     # Make API requests
-    #   end
-    #
-    def run(port: 8000, host: '0.0.0.0')
-      server = Server.new(port: port, host: host)
-      server.start
-      yield server
-    ensure
-      server&.stop
-    end
-
-    # Find the API binary
-    #
-    # @return [Pathname] Path to binary
-    # @raise [MissingBinaryError] If not found
-    #
-    def find_api_binary
-      # API might be served by kreuzberg CLI or a separate binary
-      binary_name = Gem.win_platform? ? 'kreuzberg.exe' : 'kreuzberg'
-      found = CLIProxy.search_paths(binary_name).find(&:file?)
-      return found if found
-
-      raise MissingBinaryError, missing_binary_message
-    end
-
-    # Error message for missing binary
-    #
-    # @return [String]
-    #
-    def missing_binary_message
-      <<~MSG.strip
-        kreuzberg binary not found for API server. Build it with:
-        `cargo build --release --package kreuzberg-cli`
-
-        Or ensure kreuzberg is installed with API support.
-      MSG
-    end
-  end
-end
+# frozen_string_literal: true
+
+require 'open3'
+require 'pathname'
+
+module Kreuzberg
+  # API server proxy
+  #
+  # Starts and manages the Kreuzberg API server (Litestar/Python-based or Rust-based).
+  #
+  # @example Start the server
+  #   server = Kreuzberg::APIProxy.new(port: 8000)
+  #   server.start
+  #   # Server runs in background
+  #   server.stop
+  #
+  # @example With block
+  #   Kreuzberg::APIProxy.run(port: 8000) do |server|
+  #     # Server runs while block executes
+  #     response = Net::HTTP.get(URI('http://localhost:8000/health'))
+  #   end
+  #
+  module APIProxy
+    Error = Class.new(Kreuzberg::Errors::Error)
+    MissingBinaryError = Class.new(Error)
+    ServerError = Class.new(Error)
+
+    # API server instance
+    class Server
+      attr_reader :port, :host, :pid
+
+      # Initialize server
+      #
+      # @param port [Integer] Port to run on (default: 8000)
+      # @param host [String] Host to bind to (default: "0.0.0.0")
+      #
+      def initialize(port: 8000, host: '0.0.0.0')
+        @port = port
+        @host = host
+        @pid = nil
+        @process = nil
+      end
+
+      # Start the server in the background
+      #
+      # @return [Integer] Process ID
+      # @raise [ServerError] If server fails to start
+      #
+      def start
+        binary = APIProxy.find_api_binary
+        @pid = spawn(
+          binary.to_s,
+          'api',
+          '--host', @host,
+          '--port', @port.to_s,
+          out: $stdout,
+          err: $stderr
+        )
+        Process.detach(@pid)
+        sleep 1 # Give server time to start
+        @pid
+      end
+
+      # Stop the server
+      #
+      # @return [void]
+      #
+      def stop
+        return unless @pid
+
+        Process.kill('TERM', @pid)
+        Process.wait(@pid)
+      rescue Errno::ESRCH, Errno::ECHILD
+        # Process already dead
+      ensure
+        @pid = nil
+      end
+
+      # Check if server is running
+      #
+      # @return [Boolean]
+      #
+      def running?
+        return false unless @pid
+
+        Process.kill(0, @pid)
+        true
+      rescue Errno::ESRCH, Errno::EPERM
+        false
+      end
+    end
+
+    module_function
+
+    # Run server with a block
+    #
+    # @param port [Integer] Port to run on
+    # @param host [String] Host to bind to
+    # @yield [Server] Yields server instance
+    # @return [Object] Block result
+    #
+    # @example
+    #   Kreuzberg::APIProxy.run(port: 8000) do |server|
+    #     # Make API requests
+    #   end
+    #
+    def run(port: 8000, host: '0.0.0.0')
+      server = Server.new(port: port, host: host)
+      server.start
+      yield server
+    ensure
+      server&.stop
+    end
+
+    # Find the API binary
+    #
+    # @return [Pathname] Path to binary
+    # @raise [MissingBinaryError] If not found
+    #
+    def find_api_binary
+      # API might be served by kreuzberg CLI or a separate binary
+      binary_name = Gem.win_platform? ? 'kreuzberg.exe' : 'kreuzberg'
+      found = CLIProxy.search_paths(binary_name).find(&:file?)
+      return found if found
+
+      raise MissingBinaryError, missing_binary_message
+    end
+
+    # Error message for missing binary
+    #
+    # @return [String]
+    #
+    def missing_binary_message
+      <<~MSG.strip
+        kreuzberg binary not found for API server. Build it with:
+        `cargo build --release --package kreuzberg-cli`
+
+        Or ensure kreuzberg is installed with API support.
+      MSG
+    end
+  end
+end

data/lib/kreuzberg/cache_api.rb

@@ -1,81 +1,81 @@
-# frozen_string_literal: true
-
-module Kreuzberg
-  # Provides caching capabilities for extraction results.
-  #
-  # This module manages the cache for document extraction results. Results are cached
-  # based on document content, configuration, and MIME type, improving performance for
-  # repeated extractions of the same documents.
-  module CacheAPI
-    # Clear all cached extraction results.
-    #
-    # Removes all entries from both the native Rust cache and the local tracking state.
-    # After calling this method, all extraction results will be recomputed on subsequent
-    # requests (unless caching is disabled).
-    #
-    # @return [void] No meaningful return value
-    #
-    # @example Clear cache
-    #   Kreuzberg.clear_cache
-    #   puts "Cache cleared"
-    def clear_cache
-      native_clear_cache
-      reset_cache_tracker!
-    end
-
-    # Retrieve cache statistics.
-    #
-    # Returns information about the current state of the extraction result cache,
-    # including the number of cached entries and total memory used. Statistics include
-    # both native Rust cache metrics and local tracker metrics.
-    #
-    # @return [Hash{Symbol | String => Integer}] Cache statistics hash containing:
-    #   - :total_entries [Integer] Total number of cached extraction results
-    #   - :total_size_bytes [Integer] Total memory used by cached results in bytes
-    #
-    # @example Get cache statistics
-    #   stats = Kreuzberg.cache_stats
-    #   puts "Cached entries: #{stats[:total_entries]}"
-    #   puts "Cache size: #{stats[:total_size_bytes]} bytes"
-    #
-    # @example Check if cache is full
-    #   stats = Kreuzberg.cache_stats
-    #   if stats[:total_size_bytes] > 1_000_000_000  # 1GB
-    #     Kreuzberg.clear_cache
-    #   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.
+  #
+  # This module manages the cache for document extraction results. Results are cached
+  # based on document content, configuration, and MIME type, improving performance for
+  # repeated extractions of the same documents.
+  module CacheAPI
+    # Clear all cached extraction results.
+    #
+    # Removes all entries from both the native Rust cache and the local tracking state.
+    # After calling this method, all extraction results will be recomputed on subsequent
+    # requests (unless caching is disabled).
+    #
+    # @return [void] No meaningful return value
+    #
+    # @example Clear cache
+    #   Kreuzberg.clear_cache
+    #   puts "Cache cleared"
+    def clear_cache
+      native_clear_cache
+      reset_cache_tracker!
+    end
+
+    # Retrieve cache statistics.
+    #
+    # Returns information about the current state of the extraction result cache,
+    # including the number of cached entries and total memory used. Statistics include
+    # both native Rust cache metrics and local tracker metrics.
+    #
+    # @return [Hash{Symbol | String => Integer}] Cache statistics hash containing:
+    #   - :total_entries [Integer] Total number of cached extraction results
+    #   - :total_size_bytes [Integer] Total memory used by cached results in bytes
+    #
+    # @example Get cache statistics
+    #   stats = Kreuzberg.cache_stats
+    #   puts "Cached entries: #{stats[:total_entries]}"
+    #   puts "Cache size: #{stats[:total_size_bytes]} bytes"
+    #
+    # @example Check if cache is full
+    #   stats = Kreuzberg.cache_stats
+    #   if stats[:total_size_bytes] > 1_000_000_000  # 1GB
+    #     Kreuzberg.clear_cache
+    #   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, '--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
+# 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