dify_llm 1.8.2 → 1.9.1

data/lib/ruby_llm/stream_accumulator.rb

@@ -8,8 +8,10 @@ module RubyLLM
     def initialize
       @content = +''
       @tool_calls = {}
-      @input_tokens = 0
-      @output_tokens = 0
+      @input_tokens = nil
+      @output_tokens = nil
+      @cached_tokens = nil
+      @cache_creation_tokens = nil
       @latest_tool_call_id = nil
     end
 
@@ -35,8 +37,10 @@ module RubyLLM
         model_id: model_id,
         conversation_id: conversation_id,
         tool_calls: tool_calls_from_stream,
-        input_tokens: @input_tokens.positive? ? @input_tokens : nil,
-        output_tokens: @output_tokens.positive? ? @output_tokens : nil,
+        input_tokens: @input_tokens,
+        output_tokens: @output_tokens,
+        cached_tokens: @cached_tokens,
+        cache_creation_tokens: @cache_creation_tokens,
         raw: response
       )
     end
@@ -92,6 +96,8 @@ module RubyLLM
     def count_tokens(chunk)
       @input_tokens = chunk.input_tokens if chunk.input_tokens
       @output_tokens = chunk.output_tokens if chunk.output_tokens
+      @cached_tokens = chunk.cached_tokens if chunk.cached_tokens
+      @cache_creation_tokens = chunk.cache_creation_tokens if chunk.cache_creation_tokens
     end
   end
 end

data/lib/ruby_llm/tool.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require 'ruby_llm/schema'
+
 module RubyLLM
   # Parameter definition for Tool methods.
   class Parameter
@@ -29,6 +31,8 @@ module RubyLLM
     end
 
     class << self
+      attr_reader :params_schema_definition
+
       def description(text = nil)
         return @description unless text
 
@@ -42,6 +46,20 @@ module RubyLLM
       def parameters
         @parameters ||= {}
       end
+
+      def params(schema = nil, &block)
+        @params_schema_definition = SchemaDefinition.new(schema:, block:)
+        self
+      end
+
+      def with_params(**params)
+        @provider_params = params
+        self
+      end
+
+      def provider_params
+        @provider_params ||= {}
+      end
     end
 
     def name
@@ -63,6 +81,23 @@ module RubyLLM
       self.class.parameters
     end
 
+    def provider_params
+      self.class.provider_params
+    end
+
+    def params_schema
+      return @params_schema if defined?(@params_schema)
+
+      @params_schema = begin
+        definition = self.class.params_schema_definition
+        if definition&.present?
+          definition.json_schema
+        elsif parameters.any?
+          SchemaDefinition.from_parameters(parameters)&.json_schema
+        end
+      end
+    end
+
     def call(args)
       RubyLLM.logger.debug "Tool #{name} called with: #{args.inspect}"
       result = execute(**args.transform_keys(&:to_sym))
@@ -79,5 +114,96 @@ module RubyLLM
     def halt(message)
       Halt.new(message)
     end
+
+    # Wraps schema handling for tool parameters, supporting JSON Schema hashes,
+    # RubyLLM::Schema instances/classes, and DSL blocks.
+    class SchemaDefinition
+      def self.from_parameters(parameters)
+        return nil if parameters.nil? || parameters.empty?
+
+        properties = parameters.to_h do |name, param|
+          schema = {
+            type: map_type(param.type),
+            description: param.description
+          }.compact
+
+          schema[:items] = default_items_schema if schema[:type] == 'array'
+
+          [name.to_s, schema]
+        end
+
+        required = parameters.select { |_, param| param.required }.keys.map(&:to_s)
+
+        json_schema = {
+          type: 'object',
+          properties: properties,
+          required: required,
+          additionalProperties: false,
+          strict: true
+        }
+
+        new(schema: json_schema)
+      end
+
+      def self.map_type(type)
+        case type.to_s
+        when 'integer', 'int' then 'integer'
+        when 'number', 'float', 'double' then 'number'
+        when 'boolean' then 'boolean'
+        when 'array' then 'array'
+        when 'object' then 'object'
+        else
+          'string'
+        end
+      end
+
+      def self.default_items_schema
+        { type: 'string' }
+      end
+
+      def initialize(schema: nil, block: nil)
+        @schema = schema
+        @block = block
+      end
+
+      def present?
+        @schema || @block
+      end
+
+      def json_schema
+        @json_schema ||= RubyLLM::Utils.deep_stringify_keys(resolve_schema)
+      end
+
+      private
+
+      def resolve_schema
+        return resolve_direct_schema(@schema) if @schema
+        return build_from_block(&@block) if @block
+
+        nil
+      end
+
+      def resolve_direct_schema(schema)
+        return extract_schema(schema.to_json_schema) if schema.respond_to?(:to_json_schema)
+        return RubyLLM::Utils.deep_dup(schema) if schema.is_a?(Hash)
+        if schema.is_a?(Class) && schema.instance_methods.include?(:to_json_schema)
+          return extract_schema(schema.new.to_json_schema)
+        end
+
+        nil
+      end
+
+      def build_from_block(&)
+        schema_class = RubyLLM::Schema.create(&)
+        extract_schema(schema_class.new.to_json_schema)
+      end
+
+      def extract_schema(schema_hash)
+        return nil unless schema_hash.is_a?(Hash)
+
+        schema = schema_hash[:schema] || schema_hash['schema'] || schema_hash
+        RubyLLM::Utils.deep_dup(schema)
+      end
+    end
   end
 end

data/lib/ruby_llm/transcription.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  # Represents a transcription of audio content.
+  class Transcription
+    attr_reader :text, :model, :language, :duration, :segments, :input_tokens, :output_tokens
+
+    def initialize(text:, model:, **attributes)
+      @text = text
+      @model = model
+      @language = attributes[:language]
+      @duration = attributes[:duration]
+      @segments = attributes[:segments]
+      @input_tokens = attributes[:input_tokens]
+      @output_tokens = attributes[:output_tokens]
+    end
+
+    def self.transcribe(audio_file, **kwargs)
+      model = kwargs.delete(:model)
+      language = kwargs.delete(:language)
+      provider = kwargs.delete(:provider)
+      assume_model_exists = kwargs.delete(:assume_model_exists) { false }
+      context = kwargs.delete(:context)
+      options = kwargs
+
+      config = context&.config || RubyLLM.config
+      model ||= config.default_transcription_model
+      model, provider_instance = Models.resolve(model, provider: provider, assume_exists: assume_model_exists,
+                                                       config: config)
+      model_id = model.id
+
+      provider_instance.transcribe(audio_file, model: model_id, language:, **options)
+    end
+  end
+end

data/lib/ruby_llm/utils.rb

@@ -41,5 +41,51 @@ module RubyLLM
         end
       end
     end
+
+    def deep_dup(value)
+      case value
+      when Hash
+        value.each_with_object({}) do |(key, val), duped|
+          duped[deep_dup(key)] = deep_dup(val)
+        end
+      when Array
+        value.map { |item| deep_dup(item) }
+      else
+        begin
+          value.dup
+        rescue TypeError
+          value
+        end
+      end
+    end
+
+    def deep_stringify_keys(value)
+      case value
+      when Hash
+        value.each_with_object({}) do |(key, val), result|
+          result[key.to_s] = deep_stringify_keys(val)
+        end
+      when Array
+        value.map { |item| deep_stringify_keys(item) }
+      when Symbol
+        value.to_s
+      else
+        value
+      end
+    end
+
+    def deep_symbolize_keys(value)
+      case value
+      when Hash
+        value.each_with_object({}) do |(key, val), result|
+          symbolized_key = key.respond_to?(:to_sym) ? key.to_sym : key
+          result[symbolized_key] = deep_symbolize_keys(val)
+        end
+      when Array
+        value.map { |item| deep_symbolize_keys(item) }
+      else
+        value
+      end
+    end
   end
 end

data/lib/ruby_llm/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RubyLLM
-  VERSION = '1.8.2'
+  VERSION = '1.9.1'
 end

data/lib/ruby_llm.rb

@@ -3,10 +3,12 @@
 require 'base64'
 require 'event_stream_parser'
 require 'faraday'
+require 'faraday/multipart'
 require 'faraday/retry'
 require 'faraday/multipart'
 require 'json'
 require 'logger'
+require 'marcel'
 require 'securerandom'
 require 'zeitwerk'
 
@@ -28,6 +30,7 @@ loader.inflector.inflect(
 )
 loader.ignore("#{__dir__}/tasks")
 loader.ignore("#{__dir__}/generators")
+loader.ignore("#{__dir__}/ruby_llm/railtie.rb")
 loader.ignore("#{__dir__}/dify_llm.rb")
 loader.setup
 
@@ -58,6 +61,10 @@ module RubyLLM
       Image.paint(...)
     end
 
+    def transcribe(...)
+      Transcription.transcribe(...)
+    end
+
     def models
       Models.instance
     end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: dify_llm
 version: !ruby/object:Gem::Version
-  version: 1.8.2
+  version: 1.9.1
 platform: ruby
 authors:
 - Carmine Paolino
@@ -107,6 +107,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '1.0'
+- !ruby/object:Gem::Dependency
+  name: ruby_llm-schema
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: 0.2.1
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: 0.2.1
 - !ruby/object:Gem::Dependency
   name: zeitwerk
   requirement: !ruby/object:Gem::Requirement
@@ -169,6 +183,8 @@ files:
 - lib/generators/ruby_llm/install/templates/tool_call_model.rb.tt
 - lib/generators/ruby_llm/upgrade_to_v1_7/templates/migration.rb.tt
 - lib/generators/ruby_llm/upgrade_to_v1_7/upgrade_to_v1_7_generator.rb
+- lib/generators/ruby_llm/upgrade_to_v1_9/templates/add_v1_9_message_columns.rb.tt
+- lib/generators/ruby_llm/upgrade_to_v1_9/upgrade_to_v1_9_generator.rb
 - lib/ruby_llm.rb
 - lib/ruby_llm/active_record/acts_as.rb
 - lib/ruby_llm/active_record/acts_as_legacy.rb
@@ -203,6 +219,7 @@ files:
 - lib/ruby_llm/providers/anthropic.rb
 - lib/ruby_llm/providers/anthropic/capabilities.rb
 - lib/ruby_llm/providers/anthropic/chat.rb
+- lib/ruby_llm/providers/anthropic/content.rb
 - lib/ruby_llm/providers/anthropic/embeddings.rb
 - lib/ruby_llm/providers/anthropic/media.rb
 - lib/ruby_llm/providers/anthropic/models.rb
@@ -237,6 +254,7 @@ files:
 - lib/ruby_llm/providers/gemini/models.rb
 - lib/ruby_llm/providers/gemini/streaming.rb
 - lib/ruby_llm/providers/gemini/tools.rb
+- lib/ruby_llm/providers/gemini/transcription.rb
 - lib/ruby_llm/providers/gpustack.rb
 - lib/ruby_llm/providers/gpustack/chat.rb
 - lib/ruby_llm/providers/gpustack/media.rb
@@ -260,6 +278,7 @@ files:
 - lib/ruby_llm/providers/openai/moderation.rb
 - lib/ruby_llm/providers/openai/streaming.rb
 - lib/ruby_llm/providers/openai/tools.rb
+- lib/ruby_llm/providers/openai/transcription.rb
 - lib/ruby_llm/providers/openrouter.rb
 - lib/ruby_llm/providers/openrouter/models.rb
 - lib/ruby_llm/providers/perplexity.rb
@@ -271,11 +290,13 @@ files:
 - lib/ruby_llm/providers/vertexai/embeddings.rb
 - lib/ruby_llm/providers/vertexai/models.rb
 - lib/ruby_llm/providers/vertexai/streaming.rb
+- lib/ruby_llm/providers/vertexai/transcription.rb
 - lib/ruby_llm/railtie.rb
 - lib/ruby_llm/stream_accumulator.rb
 - lib/ruby_llm/streaming.rb
 - lib/ruby_llm/tool.rb
 - lib/ruby_llm/tool_call.rb
+- lib/ruby_llm/transcription.rb
 - lib/ruby_llm/utils.rb
 - lib/ruby_llm/version.rb
 - lib/tasks/models.rake
@@ -294,8 +315,8 @@ metadata:
   funding_uri: https://github.com/sponsors/crmne
   rubygems_mfa_required: 'true'
 post_install_message: |
-  Upgrading from RubyLLM <= 1.6.x? Check the upgrade guide for new features and migration instructions
-  --> https://rubyllm.com/upgrading-to-1-7/
+  Upgrading from RubyLLM <= 1.8.x? Check the upgrade guide for new features and migration instructions
+  --> https://rubyllm.com/upgrading/
 rdoc_options: []
 require_paths:
 - lib