ragdoll-rails 0.1.11 → 0.1.12

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1672f9baacb258a7c8842d8f3632011b5e4f064b0fbab5b58ac803cbac12d196
-  data.tar.gz: 4ae58b4aa072fd99e1a48593812b6dcdde99b5877b9e9cdcce04e3614ff09d76
+  metadata.gz: 490d43fbee26a4ebe582b4ce30e6d09050bb59b2e3edfb79ab7d12c964061e6f
+  data.tar.gz: d6f36e0b81eb7c9c1f0eb60657e3c372fa83dd3119555e776fcaf6469ac8c622
 SHA512:
-  metadata.gz: 9233cdc22a07420f550c992c6ba7b0ec53eb36d72f6b2aafa7ca8771802fdffe9a067364842ad40e318321260d7478853663ec39978c2926fc796152996b447d
-  data.tar.gz: fcff07ebf80eba9f1e6c4cbbd48e995e2b29593b775185e15b179589cf21de935ef524c88fb823f985bdf04cf00a3dbfa806b5042d85c8bcae0b3a7d2fd93ce7
+  metadata.gz: ddb0ea79bdc41ffddb7aed78acdb27434f0e9325608ac7e4c854421731a0dea3bf16d100a998a6258f32ca5a8afaf837112873051c087f963b1a2ea34b28a114
+  data.tar.gz: db740b880b21b5e93ed7fd357c42382dc5bdb71d636e63ae3291d355fe030caf953fccd220c8af6841e78779fd2cc893e711510dc55157d62a3aa3e54c269893

data/README.md

@@ -31,9 +31,10 @@ See Also:
 
 ## ✨ Features
 
-- 🔍 **Semantic Search** - Vector similarity search with flexible embedding models and pgvector
+- 🔄 **Unified Text-Based RAG** - All media types converted to searchable text before vectorization
+- 🔍 **Cross-Modal Search** - Find images by descriptions, audio by transcripts, documents by content
 - 🤖 **Multi-Provider Support** - OpenAI, Anthropic, Google, Azure, Ollama, HuggingFace via ruby_llm
-- 📄 **Multi-format Support** - PDF, DOCX, text, HTML, JSON, XML, CSV document parsing
+- 📄 **Multi-format Support** - PDF, DOCX, text, HTML, JSON, XML, CSV, images (AI descriptions), audio (transcripts)
 - 🧠 **Context Enhancement** - Automatically enhance AI prompts with relevant context
 - ⚡ **Background Processing** - Asynchronous document processing with Sidekiq
 - 🎛️ **Simple API** - Clean, intuitive interface for Rails integration

data/app/controllers/ragdoll/search_controller.rb

@@ -99,8 +99,11 @@ module Ragdoll
               }
               
               # Add document type filter if specified
+              # NOTE: Document type filtering is deprecated in unified text-based architecture
+              # All media types are now converted to text for unified cross-modal search
               if @filters[:document_type].present?
                 search_params[:document_type] = @filters[:document_type]
+                ::Rails.logger.warn "⚠️ Document type filtering is deprecated in unified text-based RAG architecture"
               end
               
               # Add status filter if specified
@@ -207,8 +210,11 @@ module Ragdoll
             }
             
             # Add document type filter if specified
+            # NOTE: Document type filtering is deprecated in unified text-based architecture
+            # All media types are now converted to text for unified cross-modal search
             if @filters[:document_type].present?
               fulltext_params[:document_type] = @filters[:document_type]
+              ::Rails.logger.warn "⚠️ Document type filtering is deprecated in unified text-based RAG architecture"
             end
             
             # Add status filter if specified

data/app/views/ragdoll/documents/index.html.erb

@@ -21,7 +21,10 @@
             <%= form.text_field :search, placeholder: "Search documents...", value: params[:search], class: "form-control" %>
           </div>
           <div class="col-md-3">
-            <%= form.select :document_type, options_for_select([["All Types", ""]] + @document_types.map { |type| [type.titleize, type] }, params[:document_type]), {}, { class: "form-select" } %>
+            <%= form.select :document_type, options_for_select([["All Types", ""]] + @document_types.map { |type| [type.titleize, type] }, params[:document_type]), {}, { class: "form-select", title: "Document Type filtering is deprecated in unified text-based architecture" } %>
+            <small class="form-text text-warning">
+              <i class="fas fa-exclamation-triangle"></i> Note: All media types are now converted to text for unified search
+            </small>
           </div>
           <div class="col-md-3">
             <%= form.select :status, options_for_select([["All Statuses", ""]] + @statuses.map { |status| [status.titleize, status] }, params[:status]), {}, { class: "form-select" } %>

data/app/views/ragdoll/search/index.html.erb

@@ -38,10 +38,13 @@
             <div class="col-md-6">
               <div class="mb-3">
                 <label class="form-label">Document Type</label>
-                <%= form.select :document_type, 
-                    options_for_select([["All Types", ""]] + ::Ragdoll::Document.distinct.pluck(:document_type).compact.map { |type| [type.titleize, type] }, @filters[:document_type]), 
-                    {}, 
-                    { class: "form-select" } %>
+                <%= form.select :document_type,
+                    options_for_select([["All Types", ""]] + ::Ragdoll::Document.distinct.pluck(:document_type).compact.map { |type| [type.titleize, type] }, @filters[:document_type]),
+                    {},
+                    { class: "form-select", title: "Document Type filtering is deprecated in unified text-based architecture" } %>
+                <small class="form-text text-warning">
+                  <i class="fas fa-exclamation-triangle"></i> Note: All media types are now converted to text for unified cross-modal search
+                </small>
               </div>
             </div>
             

data/lib/generators/ragdoll/init/templates/ragdoll_config.rb

@@ -1,7 +1,8 @@
 # frozen_string_literal: true
 
-# Ragdoll RAG (Retrieval-Augmented Generation) Configuration
+# Ragdoll Unified Text-Based RAG Configuration
 # This initializer configures the Ragdoll Rails engine for your application.
+# All media types (images, audio, documents) are converted to searchable text.
 
 Ragdoll.configure do |config|
   # LLM Provider Configuration
@@ -39,9 +40,13 @@ Ragdoll.configure do |config|
     }
   }
 
-  # Embedding Model Configuration
+  # Unified Embedding Model Configuration
+  # Single model for all content types (converted to text)
   # Examples: 'text-embedding-3-small', 'text-embedding-3-large', 'text-embedding-ada-002'
-  config.embedding_model = 'text-embedding-3-small'
+  config.embedding_model = 'text-embedding-3-large'
+
+  # Enable unified text-based architecture
+  config.use_unified_content = true
   
   # Default model for chat/completion
   config.default_model = 'gpt-4o-mini'
@@ -50,6 +55,13 @@ Ragdoll.configure do |config|
   config.chunk_size = 1000
   config.chunk_overlap = 200
 
+  # Text Conversion Settings (for unified architecture)
+  config.text_conversion = {
+    image_detail_level: :comprehensive,  # :minimal, :standard, :comprehensive, :analytical
+    audio_transcription_provider: :openai,  # :openai, :azure, :google, :whisper_local
+    enable_fallback_descriptions: true
+  }
+
   # Search Configuration
   config.search_similarity_threshold = 0.7
   config.max_search_results = 10
@@ -92,5 +104,6 @@ Ragdoll::Rails.configure do |config|
   config.max_file_size = 10.megabytes
   
   # Allowed file types for document upload
-  config.allowed_file_types = %w[pdf docx txt md html htm json xml csv]
+  # All types are converted to text: images -> descriptions, audio -> transcripts
+  config.allowed_file_types = %w[pdf docx txt md html htm json xml csv jpg jpeg png gif mp3 wav m4a]
 end

data/lib/ragdoll/rails/configuration.rb

@@ -12,7 +12,8 @@ module Ragdoll
         @job_adapter = :sidekiq
         @queue_name = :ragdoll
         @max_file_size = 10 * 1024 * 1024 # 10MB
-        @allowed_file_types = %w[pdf docx txt md html htm json xml csv]
+        # Unified text-based architecture supports all media types converted to text
+        @allowed_file_types = %w[pdf docx txt md html htm json xml csv jpg jpeg png gif mp3 wav m4a]
       end
 
       def configure_core

data/lib/ragdoll/rails/version.rb

@@ -4,6 +4,6 @@
 
 module Ragdoll
   module Rails
-    VERSION = "0.1.11"
+    VERSION = "0.1.12"
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ragdoll-rails
 version: !ruby/object:Gem::Version
-  version: 0.1.11
+  version: 0.1.12
 platform: ruby
 authors:
 - Dewayne VanHoozer
@@ -206,7 +206,8 @@ dependencies:
       - !ruby/object:Gem::Version
         version: '0'
 description: Rails engine providing ActiveRecord integration, background jobs, and
-  UI components for the Ragdoll RAG (Retrieval-Augmented Generation) system
+  UI components for Ragdoll's unified text-based RAG system. Converts all media types
+  to searchable text for powerful cross-modal search capabilities.
 email:
 - dvanhoozer@gmail.com
 executables: []
@@ -305,7 +306,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.7.1
+rubygems_version: 3.7.2
 specification_version: 4
-summary: Rails engine for Ragdoll RAG system
+summary: Rails engine for Ragdoll unified text-based RAG system
 test_files: []