kreuzberg 4.3.5-aarch64-linux

data/lib/kreuzberg/api_proxy.rb

@@ -0,0 +1,125 @@
+# frozen_string_literal: true
+
+require 'open3'
+
+module Kreuzberg
+  # @example Start the server
+  # @example With block
+  module APIProxy
+    class Error < Kreuzberg::Errors::Error; end
+    class MissingBinaryError < Error; end
+    class ServerError < Error; end
+
+    # 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
+        @pid
+      end
+
+      # Stop the server
+      #
+      # @return [void]
+      #
+      def stop
+        return unless @pid
+
+        Process.kill('TERM', @pid)
+        Process.wait(@pid)
+      rescue Errno::ESRCH, Errno::ECHILD # rubocop:disable Lint/SuppressedException
+      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
+      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

@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+module Kreuzberg
+  module CacheAPI
+    # @return [void] No meaningful return value
+    # @example Clear cache
+    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

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+module Kreuzberg
+  # @example Extract a file
+  # @example Detect file type
+  module CLI
+    module_function
+
+    # Extract content from a file using the CLI
+    #
+    # @param path_or_nil [String, nil] Path to the file (positional, for backward compatibility)
+    # @param path [String] Path to the file (keyword argument)
+    # @param output [String] Output format ("text", "json", "markdown")
+    # @param ocr [Boolean] Enable OCR
+    # @return [String] Extracted content
+    #
+    def extract(path_or_nil = nil, path: nil, output: 'text', ocr: false)
+      # Support both positional and keyword argument for path (backward compatibility)
+      actual_path = path_or_nil || path
+      raise ArgumentError, 'path is required' if actual_path.nil?
+
+      args = ['extract', actual_path, '--format', output]
+      args.push('--ocr', ocr ? 'true' : 'false')
+      CLIProxy.call(args)
+    end
+
+    # Detect MIME type of a file using the CLI
+    #
+    # @param path_or_nil [String, nil] Path to the file (positional, for backward compatibility)
+    # @param path [String] Path to the file (keyword argument)
+    # @return [String] MIME type
+    #
+    def detect(path_or_nil = nil, path: nil)
+      # Support both positional and keyword argument for path (backward compatibility)
+      actual_path = path_or_nil || path
+      raise ArgumentError, 'path is required' if actual_path.nil?
+
+      CLIProxy.call(['detect', actual_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

@@ -0,0 +1,118 @@
+# frozen_string_literal: true
+
+require 'open3'
+
+module Kreuzberg
+  # @example
+  module CLIProxy
+    class Error < Kreuzberg::Errors::Error; end
+    class MissingBinaryError < Error; end
+
+    # 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 = [
+        lib_path.join('bin', binary_name),
+        lib_path.join(binary_name),
+        root_path.join('../../crates/kreuzberg-cli/target/release', binary_name),
+        root_path.join('../../target/release', binary_name),
+        root_path.join('../../target/debug', binary_name)
+      ]
+
+      workspace_root = root_path.parent&.parent
+      paths << workspace_root.join('target', 'release', binary_name) if workspace_root
+      paths << workspace_root.join('target', 'debug', 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