ruby_llm 0.1.0.pre30 → 0.1.0.pre33

data/lib/ruby_llm/providers/openai/media.rb

@@ -0,0 +1,52 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    module OpenAI
+      # Handles formatting of media content (images, audio) for OpenAI APIs
+      module Media
+        module_function
+
+        def format_content(content) # rubocop:disable Metrics/MethodLength
+          return content unless content.is_a?(Array)
+
+          content.map do |part|
+            case part[:type]
+            when 'image'
+              format_image(part)
+            when 'input_audio'
+              format_audio(part)
+            else
+              part
+            end
+          end
+        end
+
+        def format_image(part)
+          {
+            type: 'image_url',
+            image_url: {
+              url: format_data_url(part[:source]),
+              detail: 'auto'
+            }
+          }
+        end
+
+        def format_audio(part)
+          {
+            type: 'input_audio',
+            input_audio: part[:input_audio]
+          }
+        end
+
+        def format_data_url(source)
+          if source[:type] == 'base64'
+            "data:#{source[:media_type]};base64,#{source[:data]}"
+          else
+            source[:url]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/openai/models.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    module OpenAI
+      # Models methods of the OpenAI API integration
+      module Models
+        module_function
+
+        def models_url
+          'models'
+        end
+
+        def parse_list_models_response(response) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
+          (response.body['data'] || []).map do |model|
+            ModelInfo.new(
+              id: model['id'],
+              created_at: model['created'] ? Time.at(model['created']) : nil,
+              display_name: capabilities.format_display_name(model['id']),
+              provider: slug,
+              type: capabilities.model_type(model['id']),
+              family: capabilities.model_family(model['id']),
+              metadata: {
+                object: model['object'],
+                owned_by: model['owned_by']
+              },
+              context_window: capabilities.context_window_for(model['id']),
+              max_tokens: capabilities.max_tokens_for(model['id']),
+              supports_vision: capabilities.supports_vision?(model['id']),
+              supports_functions: capabilities.supports_functions?(model['id']),
+              supports_json_mode: capabilities.supports_json_mode?(model['id']),
+              input_price_per_million: capabilities.input_price_for(model['id']),
+              output_price_per_million: capabilities.output_price_for(model['id'])
+            )
+          end
+        end
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/openai/streaming.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    module OpenAI
+      # Streaming methods of the OpenAI API integration
+      module Streaming
+        module_function
+
+        def stream_url
+          completion_url
+        end
+
+        def handle_stream(&block) # rubocop:disable Metrics/MethodLength
+          to_json_stream do |data|
+            block.call(
+              Chunk.new(
+                role: :assistant,
+                model_id: data['model'],
+                content: data.dig('choices', 0, 'delta', 'content'),
+                tool_calls: parse_tool_calls(data.dig('choices', 0, 'delta', 'tool_calls'), parse_arguments: false),
+                input_tokens: data.dig('usage', 'prompt_tokens'),
+                output_tokens: data.dig('usage', 'completion_tokens')
+              )
+            )
+          end
+        end
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/openai/tools.rb

@@ -0,0 +1,69 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    module OpenAI
+      # Tools methods of the OpenAI API integration
+      module Tools
+        module_function
+
+        def tool_for(tool) # rubocop:disable Metrics/MethodLength
+          {
+            type: 'function',
+            function: {
+              name: tool.name,
+              description: tool.description,
+              parameters: {
+                type: 'object',
+                properties: tool.parameters.transform_values { |param| param_schema(param) },
+                required: tool.parameters.select { |_, p| p.required }.keys
+              }
+            }
+          }
+        end
+
+        def param_schema(param)
+          {
+            type: param.type,
+            description: param.description
+          }.compact
+        end
+
+        def format_tool_calls(tool_calls) # rubocop:disable Metrics/MethodLength
+          return nil unless tool_calls&.any?
+
+          tool_calls.map do |_, tc|
+            {
+              id: tc.id,
+              type: 'function',
+              function: {
+                name: tc.name,
+                arguments: JSON.generate(tc.arguments)
+              }
+            }
+          end
+        end
+
+        def parse_tool_calls(tool_calls, parse_arguments: true) # rubocop:disable Metrics/MethodLength
+          return nil unless tool_calls&.any?
+
+          tool_calls.to_h do |tc|
+            [
+              tc['id'],
+              ToolCall.new(
+                id: tc['id'],
+                name: tc.dig('function', 'name'),
+                arguments: if parse_arguments
+                             JSON.parse(tc.dig('function',
+                                               'arguments'))
+                           else
+                             tc.dig('function', 'arguments')
+                           end
+              )
+            ]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/openai.rb

@@ -5,17 +5,28 @@ module RubyLLM
     # OpenAI API integration. Handles chat completion, function calling,
     # and OpenAI's unique streaming format. Supports GPT-4, GPT-3.5,
     # and other OpenAI models.
-    class OpenAI # rubocop:disable Metrics/ClassLength
-      include Provider
-
-      def parse_error(response)
-        return if response.body.empty?
-
-        body = try_parse_json(response.body)
-        body.is_a?(Hash) ? body.dig('error', 'message') : body
-      end
-
-      private
+    module OpenAI
+      extend Provider
+      extend OpenAI::Chat
+      extend OpenAI::Embeddings
+      extend OpenAI::Models
+      extend OpenAI::Streaming
+      extend OpenAI::Tools
+      extend OpenAI::Images
+      extend OpenAI::Media
+
+      def self.extended(base)
+        base.extend(Provider)
+        base.extend(OpenAI::Chat)
+        base.extend(OpenAI::Embeddings)
+        base.extend(OpenAI::Models)
+        base.extend(OpenAI::Streaming)
+        base.extend(OpenAI::Tools)
+        base.extend(OpenAI::Images)
+        base.extend(OpenAI::Media)
+      end
+
+      module_function
 
       def api_base
         'https://api.openai.com/v1'
@@ -26,195 +37,6 @@ module RubyLLM
           'Authorization' => "Bearer #{RubyLLM.config.openai_api_key}"
         }
       end
-
-      def completion_url
-        'chat/completions'
-      end
-
-      def stream_url
-        completion_url
-      end
-
-      def models_url
-        'models'
-      end
-
-      def embedding_url
-        'embeddings'
-      end
-
-      def build_payload(messages, tools:, temperature:, model:, stream: false) # rubocop:disable Metrics/MethodLength
-        {
-          model: model,
-          messages: format_messages(messages),
-          temperature: temperature,
-          stream: stream
-        }.tap do |payload|
-          if tools.any?
-            payload[:tools] = tools.map { |_, tool| tool_for(tool) }
-            payload[:tool_choice] = 'auto'
-          end
-          payload[:stream_options] = { include_usage: true } if stream
-        end
-      end
-
-      def format_messages(messages)
-        messages.map do |msg|
-          {
-            role: format_role(msg.role),
-            content: msg.content,
-            tool_calls: format_tool_calls(msg.tool_calls),
-            tool_call_id: msg.tool_call_id
-          }.compact
-        end
-      end
-
-      def format_role(role)
-        case role
-        when :system
-          'developer'
-        else
-          role.to_s
-        end
-      end
-
-      def build_embedding_payload(text, model:)
-        {
-          model: model,
-          input: text
-        }
-      end
-
-      def parse_embedding_response(response)
-        embeddings = response.body['data'].map { |d| d['embedding'] }
-        embeddings.size == 1 ? embeddings.first : embeddings
-      end
-
-      def format_tool_calls(tool_calls) # rubocop:disable Metrics/MethodLength
-        return nil unless tool_calls&.any?
-
-        tool_calls.map do |_, tc|
-          {
-            id: tc.id,
-            type: 'function',
-            function: {
-              name: tc.name,
-              arguments: JSON.generate(tc.arguments)
-            }
-          }
-        end
-      end
-
-      def tool_for(tool) # rubocop:disable Metrics/MethodLength
-        {
-          type: 'function',
-          function: {
-            name: tool.name,
-            description: tool.description,
-            parameters: {
-              type: 'object',
-              properties: tool.parameters.transform_values { |param| param_schema(param) },
-              required: tool.parameters.select { |_, p| p.required }.keys
-            }
-          }
-        }
-      end
-
-      def param_schema(param)
-        {
-          type: param.type,
-          description: param.description
-        }.compact
-      end
-
-      def parse_completion_response(response) # rubocop:disable Metrics/MethodLength
-        data = response.body
-        return if data.empty?
-
-        message_data = data.dig('choices', 0, 'message')
-        return unless message_data
-
-        Message.new(
-          role: :assistant,
-          content: message_data['content'],
-          tool_calls: parse_tool_calls(message_data['tool_calls']),
-          input_tokens: data['usage']['prompt_tokens'],
-          output_tokens: data['usage']['completion_tokens'],
-          model_id: data['model']
-        )
-      end
-
-      def parse_tool_calls(tool_calls, parse_arguments: true) # rubocop:disable Metrics/MethodLength
-        return nil unless tool_calls&.any?
-
-        tool_calls.to_h do |tc|
-          [
-            tc['id'],
-            ToolCall.new(
-              id: tc['id'],
-              name: tc.dig('function', 'name'),
-              arguments: parse_arguments ? JSON.parse(tc.dig('function', 'arguments')) : tc.dig('function', 'arguments')
-            )
-          ]
-        end
-      end
-
-      def parse_models_response(response) # rubocop:disable Metrics/MethodLength
-        (response.body['data'] || []).map do |model|
-          model_info = begin
-            Models.find(model['id'])
-          rescue StandardError
-            nil
-          end
-          next unless model_info
-
-          model_info.tap do |info|
-            info.metadata.merge!(
-              object: model['object'],
-              owned_by: model['owned_by']
-            )
-          end
-        end.compact
-      end
-
-      def handle_stream(&block) # rubocop:disable Metrics/MethodLength
-        to_json_stream do |data|
-          block.call(
-            Chunk.new(
-              role: :assistant,
-              model_id: data['model'],
-              content: data.dig('choices', 0, 'delta', 'content'),
-              tool_calls: parse_tool_calls(data.dig('choices', 0, 'delta', 'tool_calls'), parse_arguments: false),
-              input_tokens: data.dig('usage', 'prompt_tokens'),
-              output_tokens: data.dig('usage', 'completion_tokens')
-            )
-          )
-        end
-      end
-
-      def parse_list_models_response(response) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
-        (response.body['data'] || []).map do |model|
-          ModelInfo.new(
-            id: model['id'],
-            created_at: model['created'] ? Time.at(model['created']) : nil,
-            display_name: capabilities.format_display_name(model['id']),
-            provider: slug,
-            type: capabilities.model_type(model['id']),
-            family: capabilities.model_family(model['id']),
-            metadata: {
-              object: model['object'],
-              owned_by: model['owned_by']
-            },
-            context_window: capabilities.context_window_for(model['id']),
-            max_tokens: capabilities.max_tokens_for(model['id']),
-            supports_vision: capabilities.supports_vision?(model['id']),
-            supports_functions: capabilities.supports_functions?(model['id']),
-            supports_json_mode: capabilities.supports_json_mode?(model['id']),
-            input_price_per_million: capabilities.input_price_for(model['id']),
-            output_price_per_million: capabilities.output_price_for(model['id'])
-          )
-        end
-      end
     end
   end
 end

data/lib/ruby_llm/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RubyLLM
-  VERSION = '0.1.0.pre30'
+  VERSION = '0.1.0.pre33'
 end

data/lib/ruby_llm.rb

@@ -1,11 +1,13 @@
 # frozen_string_literal: true
 
-require 'zeitwerk'
+require 'base64'
+require 'event_stream_parser'
 require 'faraday'
+require 'faraday/retry'
 require 'json'
 require 'logger'
-require 'event_stream_parser'
 require 'securerandom'
+require 'zeitwerk'
 
 loader = Zeitwerk::Loader.for_gem
 loader.inflector.inflect(
@@ -32,6 +34,10 @@ module RubyLLM
       Embedding.embed(...)
     end
 
+    def paint(...)
+      Image.paint(...)
+    end
+
     def models
       Models
     end

data/ruby_llm.gemspec

@@ -8,11 +8,11 @@ Gem::Specification.new do |spec|
   spec.authors       = ['Carmine Paolino']
   spec.email         = ['carmine@paolino.me']
 
-  spec.summary       = 'Clean Ruby interface to modern AI language models'
-  spec.description   = 'A delightful Ruby way to work with AI language models. Provides a unified interface to OpenAI' \
-                       ', Anthropic, Google, and DeepSeek models with automatic token counting, proper streaming' \
-                       ' support, and a focus on developer happiness. No wrapping your head around multiple APIs' \
-                       ' - just clean Ruby code that works.'
+  spec.summary       = 'Beautiful Ruby interface to modern AI'
+  spec.description   = 'A delightful Ruby way to work with AI. Chat in text, analyze and generate images, understand' \
+                       ' audio, and use tools through a unified interface to OpenAI, Anthropic, Google, and DeepSeek.' \
+                       ' Built for developer happiness with automatic token counting, proper streaming, and Rails' \
+                       ' integration. No wrapping your head around multiple APIs - just clean Ruby code that works.'
   spec.homepage      = 'https://github.com/crmne/ruby_llm'
   spec.license       = 'MIT'
   spec.required_ruby_version = Gem::Requirement.new('>= 3.1.0')
@@ -36,6 +36,7 @@ Gem::Specification.new do |spec|
   spec.add_dependency 'event_stream_parser', '>= 0.3.0', '< 2.0.0'
   spec.add_dependency 'faraday', '>= 2.0'
   spec.add_dependency 'faraday-multipart', '>= 1.0'
+  spec.add_dependency 'faraday-retry', '>= 2.0'
   spec.add_dependency 'zeitwerk', '>= 2.6'
 
   # Rails integration dependencies
@@ -56,6 +57,7 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency 'rubocop', '>= 1.0'
   spec.add_development_dependency 'rubocop-rake', '>= 0.6'
   spec.add_development_dependency 'simplecov', '>= 0.21'
+  spec.add_development_dependency 'sqlite3'
   spec.add_development_dependency 'webmock', '~> 3.18'
   spec.add_development_dependency 'yard', '>= 0.9'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ruby_llm
 version: !ruby/object:Gem::Version
-  version: 0.1.0.pre30
+  version: 0.1.0.pre33
 platform: ruby
 authors:
 - Carmine Paolino
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2025-02-17 00:00:00.000000000 Z
+date: 2025-02-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: event_stream_parser
@@ -58,6 +58,20 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '1.0'
+- !ruby/object:Gem::Dependency
+  name: faraday-retry
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '2.0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '2.0'
 - !ruby/object:Gem::Dependency
   name: zeitwerk
   requirement: !ruby/object:Gem::Requirement
@@ -294,6 +308,20 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0.21'
+- !ruby/object:Gem::Dependency
+  name: sqlite3
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
 - !ruby/object:Gem::Dependency
   name: webmock
   requirement: !ruby/object:Gem::Requirement
@@ -322,21 +350,22 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0.9'
-description: A delightful Ruby way to work with AI language models. Provides a unified
-  interface to OpenAI, Anthropic, Google, and DeepSeek models with automatic token
-  counting, proper streaming support, and a focus on developer happiness. No wrapping
-  your head around multiple APIs - just clean Ruby code that works.
+description: A delightful Ruby way to work with AI. Chat in text, analyze and generate
+  images, understand audio, and use tools through a unified interface to OpenAI, Anthropic,
+  Google, and DeepSeek. Built for developer happiness with automatic token counting,
+  proper streaming, and Rails integration. No wrapping your head around multiple APIs
+  - just clean Ruby code that works.
 email:
 - carmine@paolino.me
 executables: []
 extensions: []
 extra_rdoc_files: []
 files:
-- ".github/workflows/gem-push.yml"
-- ".github/workflows/test.yml"
+- ".github/workflows/cicd.yml"
 - ".gitignore"
 - ".overcommit.yml"
 - ".rspec"
+- ".rspec_status"
 - ".rubocop.yml"
 - Gemfile
 - LICENSE
@@ -349,21 +378,36 @@ files:
 - lib/ruby_llm/chat.rb
 - lib/ruby_llm/chunk.rb
 - lib/ruby_llm/configuration.rb
+- lib/ruby_llm/content.rb
 - lib/ruby_llm/embedding.rb
 - lib/ruby_llm/error.rb
+- lib/ruby_llm/image.rb
 - lib/ruby_llm/message.rb
-- lib/ruby_llm/model_capabilities/anthropic.rb
-- lib/ruby_llm/model_capabilities/deepseek.rb
-- lib/ruby_llm/model_capabilities/gemini.rb
-- lib/ruby_llm/model_capabilities/openai.rb
 - lib/ruby_llm/model_info.rb
 - lib/ruby_llm/models.json
 - lib/ruby_llm/models.rb
 - lib/ruby_llm/provider.rb
 - 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/embeddings.rb
+- lib/ruby_llm/providers/anthropic/models.rb
+- lib/ruby_llm/providers/anthropic/streaming.rb
+- lib/ruby_llm/providers/anthropic/tools.rb
 - lib/ruby_llm/providers/deepseek.rb
+- lib/ruby_llm/providers/deepseek/capabilites.rb
 - lib/ruby_llm/providers/gemini.rb
+- lib/ruby_llm/providers/gemini/capabilities.rb
+- lib/ruby_llm/providers/gemini/models.rb
 - lib/ruby_llm/providers/openai.rb
+- lib/ruby_llm/providers/openai/capabilities.rb
+- lib/ruby_llm/providers/openai/chat.rb
+- lib/ruby_llm/providers/openai/embeddings.rb
+- lib/ruby_llm/providers/openai/images.rb
+- lib/ruby_llm/providers/openai/media.rb
+- lib/ruby_llm/providers/openai/models.rb
+- lib/ruby_llm/providers/openai/streaming.rb
+- lib/ruby_llm/providers/openai/tools.rb
 - lib/ruby_llm/railtie.rb
 - lib/ruby_llm/stream_accumulator.rb
 - lib/ruby_llm/tool.rb
@@ -398,5 +442,5 @@ requirements: []
 rubygems_version: 3.5.22
 signing_key:
 specification_version: 4
-summary: Clean Ruby interface to modern AI language models
+summary: Beautiful Ruby interface to modern AI
 test_files: []

data/.github/workflows/test.yml

@@ -1,35 +0,0 @@
-name: Test
-
-
-
-on:
-  push:
-    branches: [ "main" ]
-  pull_request:
-    branches: [ "main" ]
-  workflow_call:
-
-jobs:
-  test:
-    runs-on: ubuntu-latest
-    strategy:
-      matrix:
-        ruby-version: ['3.1', '3.2', '3.3']
-
-    steps:
-    - uses: actions/checkout@v4
-
-    - name: Set up Ruby
-      uses: ruby/setup-ruby@v1
-      with:
-        ruby-version: ${{ matrix.ruby-version }}
-        bundler-cache: true
-
-    - name: Install dependencies
-      run: bundle install
-
-    - name: Check code format
-      run: bundle exec rubocop
-
-    # - name: Run tests
-    #   run: bundle exec rspec