ruby_llm 0.1.0.pre39 → 0.1.0.pre41

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 0c887462ad2c35c5e831991b8b6cd78095c3f286861fd4913afddd5887833f3b
-  data.tar.gz: 382a2dd05532510ed5ac8ae5deb7a81febb831e3bbd8a11376eba0ad266451c3
+  metadata.gz: e6a58fa5d1286364224b46aced4e7150aa935988ca5e6a5f5a12d760da76c214
+  data.tar.gz: 7d24ed72b786fea8b030191e796facd27c6ac6d45efe499ea2a25d3a5d01fab1
 SHA512:
-  metadata.gz: b8c372a54f0bb94ecf6b22ead70f34a324974f4f61ad4610ba15ffee64ed8ec226fac24a74b73667bc07ea73a14fb5ee5d813759ab60f86c54c7d00cfc3905aa
-  data.tar.gz: 868fadbdaf583aa911a98cce36b6b61208a2065d92f32c31f1a2ea0908a37aa91f2047d2ea7aa77911093a7fcbb4ae10ace87f7ddafa2d39856e67a54d5bf846
+  metadata.gz: 293c57036acd3a7b0b5fdeda7866d0bf1599f52138aeac058bb248bc53d4b687caff047ab84aa57b1b50ae034e8cfb25729bd1e027af3febd1c50ce640bebf44
+  data.tar.gz: 762c75a2d0e7dc02829c2b7d7171c6fc57932656c72890ca0d1656fefc7d9d5cfa5989726ee40fa3b5d63cac81b1a2dc66c0439cb487272e70288718d993c21e

data/.rspec_status

@@ -7,6 +7,8 @@ example_id                                         | status | run_time        |
 ./spec/ruby_llm/chat_content_spec.rb[1:1:3]        | passed | 2.54 seconds    |
 ./spec/ruby_llm/chat_content_spec.rb[1:2:1]        | passed | 2.77 seconds    |
 ./spec/ruby_llm/chat_content_spec.rb[1:2:2]        | passed | 2.1 seconds     |
+./spec/ruby_llm/chat_pdf_spec.rb[1:1:1]            | passed | 7.75 seconds    |
+./spec/ruby_llm/chat_pdf_spec.rb[1:1:2]            | passed | 13.88 seconds   |
 ./spec/ruby_llm/chat_spec.rb[1:1:1:1]              | passed | 1.02 seconds    |
 ./spec/ruby_llm/chat_spec.rb[1:1:1:2]              | passed | 3.95 seconds    |
 ./spec/ruby_llm/chat_spec.rb[1:1:2:1]              | passed | 0.4854 seconds  |

data/README.md

@@ -16,7 +16,6 @@ A delightful Ruby way to work with AI. Chat in text, analyze and generate images
   <a href="https://badge.fury.io/rb/ruby_llm"><img src="https://badge.fury.io/rb/ruby_llm.svg" alt="Gem Version" /></a>
   <a href="https://github.com/testdouble/standard"><img src="https://img.shields.io/badge/code_style-standard-brightgreen.svg" alt="Ruby Style Guide" /></a>
   <a href="https://rubygems.org/gems/ruby_llm"><img alt="Gem Total Downloads" src="https://img.shields.io/gem/dt/ruby_llm"></a>
-  <a href="https://github.com/crmne/ruby_llm/actions/workflows/cicd.yml"><img src="https://github.com/crmne/ruby_llm/actions/workflows/cicd.yml/badge.svg" alt="CI" /></a>
   <a href="https://codecov.io/gh/crmne/ruby_llm"><img src="https://codecov.io/gh/crmne/ruby_llm/branch/main/graph/badge.svg" alt="codecov" /></a>
 </p>
 
@@ -25,6 +24,7 @@ A delightful Ruby way to work with AI. Chat in text, analyze and generate images
 ## Features
 
 - 💬 **Beautiful Chat Interface** - Converse with AI models as easily as `RubyLLM.chat.ask "teach me Ruby"`
+- 📄 **PDF Analysis** - Analyze PDF documents directly with Claude models using `chat.ask "What's in this?", with: { pdf: "document.pdf" }`
 - 🎵 **Audio Analysis** - Get audio transcription and understanding with `chat.ask "what's said here?", with: { audio: "clip.wav" }`
 - 👁️ **Vision Understanding** - Let AIs analyze images with a simple `chat.ask "what's in this?", with: { image: "photo.jpg" }`
 - 🌊 **Streaming** - Real-time responses with proper Ruby streaming with `chat.ask "hello" do |chunk| puts chunk.content end`
@@ -116,6 +116,13 @@ chat.ask "What's being said in this recording?", with: { audio: "meeting.wav" }
 # Combine multiple pieces of content
 chat.ask "Compare these diagrams", with: { image: ["diagram1.png", "diagram2.png"] }
 
+# Ask about PDFs (currently supported with Claude models)
+chat = RubyLLM.chat(model: 'claude-3-7-sonnet-20250219')
+chat.ask "Summarize this research paper", with: { pdf: "research.pdf" }
+
+# Multiple PDFs work too
+chat.ask "Compare these contracts", with: { pdf: ["contract1.pdf", "contract2.pdf"] }
+
 # Check token usage
 last_message = chat.messages.last
 puts "Conversation used #{last_message.input_tokens} input tokens and #{last_message.output_tokens} output tokens"

data/lib/ruby_llm/content.rb

@@ -5,7 +5,7 @@ module RubyLLM
   # Stores data in a standard internal format, letting providers
   # handle their own formatting needs.
   class Content
-    def initialize(text = nil, attachments = {})
+    def initialize(text = nil, attachments = {}) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
       @parts = []
       @parts << { type: 'text', text: text } unless text.nil? || text.empty?
 
@@ -16,6 +16,10 @@ module RubyLLM
       Array(attachments[:audio]).each do |source|
         @parts << attach_audio(source)
       end
+
+      Array(attachments[:pdf]).each do |source|
+        @parts << attach_pdf(source)
+      end
     end
 
     def to_a
@@ -64,6 +68,33 @@ module RubyLLM
       }
     end
 
+    def attach_pdf(source) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
+      source = File.expand_path(source) unless source.start_with?('http')
+
+      pdf_data = {
+        type: 'pdf',
+        source: source
+      }
+
+      # For local files, validate they exist
+      unless source.start_with?('http')
+        raise Error, "PDF file not found: #{source}" unless File.exist?(source)
+
+        # Simple check for PDF file type (could be more robust)
+        unless source.downcase.end_with?('.pdf') || File.read(source, 5) == '%PDF-'
+          RubyLLM.logger.warn "File may not be a valid PDF: #{source}"
+        end
+
+        # Preload file content for providers that need it
+        pdf_data[:content] = File.read(source)
+      end
+
+      pdf_data
+    rescue StandardError => e
+      RubyLLM.logger.error "Error attaching PDF #{source}: #{e.message}"
+      raise Error, "Failed to attach PDF: #{e.message}"
+    end
+
     def encode_file(source)
       if source.start_with?('http')
         response = Faraday.get(source)

data/lib/ruby_llm/provider.rb

@@ -143,11 +143,20 @@ module RubyLLM
       maybe_json
     end
 
-    def parse_error(response)
+    def parse_error(response) # rubocop:disable Metrics/MethodLength
       return if response.body.empty?
 
       body = try_parse_json(response.body)
-      body.is_a?(Hash) ? body.dig('error', 'message') : body
+      case body
+      when Hash
+        body.dig('error', 'message')
+      when Array
+        body.map do |part|
+          part.dig('error', 'message')
+        end.join('. ')
+      else
+        body
+      end
     end
 
     class << self

data/lib/ruby_llm/providers/anthropic/chat.rb

@@ -62,7 +62,7 @@ module RubyLLM
         def format_basic_message(msg)
           {
             role: convert_role(msg.role),
-            content: msg.content
+            content: Media.format_content(msg.content)
           }
         end
 

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

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    module Anthropic
+      # Handles formatting of media content (images, PDFs, audio) for Anthropic
+      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 'pdf'
+              format_pdf(part)
+            when 'text'
+              format_text_block(part[:text])
+            else
+              part
+            end
+          end
+        end
+
+        def format_image(part)
+          # Handle image formatting for Anthropic
+          # This is just a placeholder - implement based on Anthropic's requirements
+          part
+        end
+
+        def format_pdf(part) # rubocop:disable Metrics/MethodLength
+          source = part[:source]
+
+          if source.start_with?('http')
+            # For URLs
+            {
+              type: 'document',
+              source: { url: source }
+            }
+          else
+            # For local files
+            data = Base64.strict_encode64(part[:content])
+
+            {
+              type: 'document',
+              source: {
+                type: 'base64',
+                media_type: 'application/pdf',
+                data: data
+              }
+            }
+          end
+        end
+
+        def format_text_block(text)
+          {
+            type: 'text',
+            text: text
+          }
+        end
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/anthropic.rb

@@ -8,6 +8,7 @@ module RubyLLM
       extend Provider
       extend Anthropic::Chat
       extend Anthropic::Embeddings
+      extend Anthropic::Media
       extend Anthropic::Models
       extend Anthropic::Streaming
       extend Anthropic::Tools

data/lib/ruby_llm/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RubyLLM
-  VERSION = '0.1.0.pre39'
+  VERSION = '0.1.0.pre41'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ruby_llm
 version: !ruby/object:Gem::Version
-  version: 0.1.0.pre39
+  version: 0.1.0.pre41
 platform: ruby
 authors:
 - Carmine Paolino
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2025-02-25 00:00:00.000000000 Z
+date: 2025-02-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: event_stream_parser
@@ -129,6 +129,7 @@ files:
 - 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/media.rb
 - lib/ruby_llm/providers/anthropic/models.rb
 - lib/ruby_llm/providers/anthropic/streaming.rb
 - lib/ruby_llm/providers/anthropic/tools.rb