Support for text files attachments.

Author: crmne

lib/ruby_llm/attachments.rb

@@ -53,6 +53,10 @@ def encoded
         Base64.strict_encode64(content)
       end
 
+      def text?
+        content.is_a?(String) || RubyLLM::MimeTypes.text?(mime_type)
+      end
+
       def mime_type
         RubyLLM::MimeTypes.detect_from_path(@filename)
       end

lib/ruby_llm/attachments/text.rb

@@ -0,0 +1,9 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Attachments
+    # Represents a Text attachment
+    class Text < Base
+    end
+  end
+end

lib/ruby_llm/content.rb

@@ -30,6 +30,11 @@ def add_pdf(source)
       self
     end
 
+    def add_text(source)
+      @attachments << Attachments::Text.new(source)
+      self
+    end
+
     def format
       if @text && @attachments.empty?
         @text
@@ -57,6 +62,7 @@ def process_attachments_hash(attachments)
       Array(attachments[:image]).each { |source| add_image(source) }
       Array(attachments[:audio]).each { |source| add_audio(source) }
       Array(attachments[:pdf]).each { |source| add_pdf(source) }
+      Array(attachments[:text]).each { |source| add_text(source) }
     end
 
     def process_attachments_array_or_string(attachments)
@@ -66,8 +72,10 @@ def process_attachments_array_or_string(attachments)
           add_image file
         elsif RubyLLM::MimeTypes.audio?(mime_type)
           add_audio file
+        elsif RubyLLM::MimeTypes.pdf?(mime_type)
+          add_pdf file
         else
-          add_pdf file # Default to PDF for unknown types for now
+          add_text file
         end
       end
     end

lib/ruby_llm/error.rb

@@ -24,6 +24,7 @@ class ConfigurationError < StandardError; end
   class InvalidRoleError < StandardError; end
   class ModelNotFoundError < StandardError; end
   class UnsupportedFunctionsError < StandardError; end
+  class UnsupportedAttachmentError < StandardError; end
 
   # Error classes for different HTTP status codes
   class BadRequestError < Error; end

lib/ruby_llm/mime_types.rb

@@ -21,26 +21,36 @@ def mime_type(ext, fallback = 'application/octet-stream')
     # @param identifier [String] File extension, path, or MIME type
     # @return [Boolean] true if content is an image
     def image?(identifier)
-      type = identifier.include?('/') ? identifier : mime_type(".#{identifier.to_s.downcase.delete('.')}")
+      type = mime_type_from_identifier(identifier)
       type.start_with?('image/')
     end
 
     # Detect if content is audio based on extension or MIME type
     # @param identifier [String] File extension, path, or MIME type
     # @return [Boolean] true if content is audio
     def audio?(identifier)
-      type = identifier.include?('/') ? identifier : mime_type(".#{identifier.to_s.downcase.delete('.')}")
+      type = mime_type_from_identifier(identifier)
       type.start_with?('audio/')
     end
 
     # Detect if content is a PDF based on extension or MIME type
     # @param identifier [String] File extension, path, or MIME type
     # @return [Boolean] true if content is a PDF
     def pdf?(identifier)
-      type = identifier.include?('/') ? identifier : mime_type(".#{identifier.to_s.downcase.delete('.')}")
+      type = mime_type_from_identifier(identifier)
       type == 'application/pdf'
     end
 
+    # Detect if content is text based on extension or MIME type
+    # @param identifier [String] File extension, path, or MIME type
+    # @return [Boolean] true if content is text
+    def text?(identifier)
+      type = mime_type_from_identifier(identifier)
+      type.start_with?('text/') ||
+        TEXT_SUFFIXES.any? { |suffix| type.end_with?(suffix) } ||
+        NON_TEXT_PREFIX_TEXT_MIME_TYPES.include?(type)
+    end
+
     # Extract extension from filename or path
     # @param path [String] File path or name
     # @return [String] Extension with leading dot
@@ -62,6 +72,11 @@ def extension_for_mime_type(mime)
       MIME_TYPES.invert[mime]
     end
 
+    # Helper method to get MIME type from identifier
+    def mime_type_from_identifier(identifier)
+      identifier.include?('/') ? identifier : mime_type(".#{identifier.to_s.downcase.delete('.')}")
+    end
+
     # List of most common mime-types, selected various sources
     # according to their usefulness in a webserving scope for Ruby
     # users.
@@ -347,6 +362,7 @@ def extension_for_mime_type(mime)
       '.mc1' => 'application/vnd.medcalcdata',
       '.mcd' => 'application/vnd.mcd',
       '.mdb' => 'application/x-msaccess',
+      '.md' => 'text/markdown',
       '.mdi' => 'image/vnd.ms-modi',
       '.mdoc' => 'text/troff',
       '.me' => 'text/troff',
@@ -709,5 +725,38 @@ def extension_for_mime_type(mime)
       '.zip' => 'application/zip',
       '.zmm' => 'application/vnd.handheld-entertainment+xml'
     }.freeze
+
+    NON_TEXT_PREFIX_TEXT_MIME_TYPES = [
+      'application/json', # Base type, even if specific ones end with +json
+      'application/xml',  # Base type, even if specific ones end with +xml
+      'application/javascript',
+      'application/ecmascript',
+      'application/rtf',
+      'application/sql',
+      'application/x-sh',
+      'application/x-csh',
+      'application/x-httpd-php',
+      'application/sdp',
+      'application/sparql-query',
+      'application/graphql',
+      'application/yang', # Data modeling language, often serialized as XML/JSON but the type itself is distinct
+      'application/mbox', # Mailbox format
+      'application/x-tex',
+      'application/x-latex',
+      'application/x-perl',
+      'application/x-python',
+      'application/x-tcl',
+      'application/pgp-signature', # Often ASCII armored
+      'application/pgp-keys',      # Often ASCII armored
+      'application/vnd.coffeescript',
+      'application/vnd.dart',
+      'application/vnd.oai.openapi', # Base for OpenAPI, often with +json or +yaml suffix
+      'application/vnd.zul',         # ZK User Interface Language (can be XML-like)
+      'application/x-yaml',          # Common non-standard for YAML
+      'application/yaml',            # Standard for YAML
+      'application/toml'             # TOML configuration files
+    ].freeze
+
+    TEXT_SUFFIXES = ['+json', '+xml', '+html', '+yaml', '+csv', '+plain', '+javascript', '+svg'].freeze
   end
 end

lib/ruby_llm/providers/anthropic/media.rb

@@ -19,6 +19,10 @@ def format_content(content)
               parts << format_image(attachment)
             when Attachments::PDF
               parts << format_pdf(attachment)
+            when Attachments::Text
+              parts << format_text_file(attachment)
+            else
+              raise UnsupportedAttachmentError, attachment.class
             end
           end
 
@@ -73,6 +77,13 @@ def format_pdf(pdf)
             }
           end
         end
+
+        def format_text_file(text_file)
+          {
+            type: 'text',
+            text: Utils.format_text_file_for_llm(text_file)
+          }
+        end
       end
     end
   end

lib/ruby_llm/providers/bedrock/media.rb

@@ -4,6 +4,7 @@ module RubyLLM
   module Providers
     module Bedrock
       # Media handling methods for the Bedrock API integration
+      # NOTE: Bedrock does not support url attachments
       module Media
         extend Anthropic::Media
 
@@ -21,8 +22,10 @@ def format_content(content)
               parts << format_image(attachment)
             when Attachments::PDF
               parts << format_pdf(attachment)
+            when Attachments::Text
+              parts << Anthropic::Media.format_text_file(attachment)
             else
-              raise "Unsupported attachment type: #{attachment.class}"
+              raise UnsupportedAttachmentError, attachment.class
             end
           end
 

lib/ruby_llm/providers/gemini/media.rb

@@ -14,43 +14,29 @@ def format_content(content)
           parts << format_text(content.text) if content.text
 
           content.attachments.each do |attachment|
-            case attachment
-            when Attachments::Image
-              parts << format_image(attachment)
-            when Attachments::PDF
-              parts << format_pdf(attachment)
-            when Attachments::Audio
-              parts << format_audio(attachment)
-            end
+            parts << case attachment
+                     when Attachments::Text
+                       format_text_file(attachment)
+                     else
+                       format_attachment(attachment)
+                     end
           end
 
           parts
         end
 
-        def format_image(image)
+        def format_attachment(attachment)
           {
             inline_data: {
-              mime_type: image.mime_type,
-              data: image.encoded
+              mime_type: attachment.mime_type,
+              data: attachment.encoded
             }
           }
         end
 
-        def format_pdf(pdf)
+        def format_text_file(text_file)
           {
-            inline_data: {
-              mime_type: pdf.mime_type,
-              data: pdf.encoded
-            }
-          }
-        end
-
-        def format_audio(audio)
-          {
-            inline_data: {
-              mime_type: audio.mime_type,
-              data: audio.encoded
-            }
+            text: Utils.format_text_file_for_llm(text_file)
           }
         end
 

lib/ruby_llm/providers/openai/media.rb

@@ -21,6 +21,10 @@ def format_content(content)
               parts << format_pdf(attachment)
             when Attachments::Audio
               parts << format_audio(attachment)
+            when Attachments::Text
+              parts << format_text_file(attachment)
+            else
+              raise UnsupportedAttachmentError, attachment.class
             end
           end
 
@@ -47,6 +51,13 @@ def format_pdf(pdf)
           }
         end
 
+        def format_text_file(text_file)
+          {
+            type: 'text',
+            text: Utils.format_text_file_for_llm(text_file)
+          }
+        end
+
         def format_audio(audio)
           {
             type: 'input_audio',

lib/ruby_llm/utils.rb

@@ -18,5 +18,9 @@ def deep_symbolize_keys(value)
         value
       end
     end
+
+    def format_text_file_for_llm(text_file)
+      "<file name='#{File.basename(text_file.source)}'>#{text_file.content}</file>"
+    end
   end
 end