ox-ai-workers 1.1.0.1 → 1.1.1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e3e0b6d4d785c54691cb19806854a74664a48aa512e0e11eb39865cb07df3636
-  data.tar.gz: 80bd4adbe5694349bb5710fbc36f72195c181d6781701d49d834a68d48c864b0
+  metadata.gz: 5c834fb5bda0eea743e7a05d9f9f895f17f0ba566a19933e5e8fab2d55239e2a
+  data.tar.gz: cee14714d3c1fffb86eae51f23f036b7bc9bfa8cac57e87c51fd314406ed4eca
 SHA512:
-  metadata.gz: 7f017fab582104cced95b80e802a1f06ca5ecbb9bbd6533328ecca9b0d93468c7c873ff28871e9c31ebdff2d51f23ac433eb8b7d27ff14a4cc139f7f4e1808be
-  data.tar.gz: 1799968565739b120e1ca012d23cbaf5d16ff2744374c50cc43e5acb7cdd21df781662fc0bb49ad7c34c705e1dc207180abf1e66f6a3c557ec3f1311a939dbd2
+  metadata.gz: 0231e2b17097087d373e0108993425be918e189cd2ad8acd587ea33ae5492965162f2cf11ef02ef241b3358db29843c6ff26a5d5c90afeb11bb708adc3ebef51
+  data.tar.gz: 31f417a2415d3424e61ace88d4f3876e93843d06d8dcf6101e1f2f1b936729eca15f5c699c764ffdbb848c644746ec4fde4892ed19201daf492fb7d684966175

data/CHANGELOG.md

@@ -1,3 +1,7 @@
+## [1.1.1] - 2025-05-26
+
+- Removed task appending in `Iterator`
+
 ## [1.1.0] - 2025-05-21
 
 - Added `add_base64` and `add_url` for Models

data/lib/oxaiworkers/iterator.rb

@@ -131,7 +131,7 @@ module OxAiWorkers
         @worker.append(role: :user, content: tool.context) if tool.respond_to?(:context) && tool.context.present?
       end
       @worker.append(messages: @messages)
-      @tasks.each { |task| @worker.append(role: :user, content: "<task>\n#{task}\n</task>") }
+      # @tasks.each { |task| @worker.append(role: :user, content: "<task>\n#{task}\n</task>") }
       @worker.tools = [function_schemas]
       return unless @tools.present?
 
@@ -264,15 +264,15 @@ module OxAiWorkers
       add_raw_context({ role:, content: text })
     end
 
-    def add_file(pdf:, filename:, text:, role: :user)
-      content = @worker.model.add_base64(binary: pdf, filename:, text:, mime_type: 'application/pdf')
+    def add_file(pdf:, text:, role: :user)
+      content = @worker.model.add_base64(binary: pdf, text:, mime_type: 'application/pdf')
       add_raw_context({ role:, content: })
     end
 
     def add_image(text:, url: nil, binary: nil, role: :user, detail: 'auto', mime_type: 'image/png')
       content = []
       if binary.present?
-        content = @worker.model.add_base64(binary:, filename:, text:, mime_type:, detail:)
+        content = @worker.model.add_base64(binary:, text:, mime_type:, detail:)
       elsif url.present?
         content = @worker.model.add_url(url:, text:, detail:, mime_type:)
       end

data/lib/oxaiworkers/models/anthropic_max.rb

@@ -70,7 +70,7 @@ module OxAiWorkers
         ]
       end
 
-      def add_base64(binary:, filename:, text:, mime_type:)
+      def add_base64(binary:, text:, mime_type:)
         content = []
         content << { type: 'text', text: } if text.present?
         content << {

data/lib/oxaiworkers/models/gemini_pro.rb

@@ -0,0 +1,151 @@
+# frozen_string_literal: true
+
+module OxAiWorkers
+  module Models
+    class GeminiPro < LLMBase
+      def initialize(uri_base: nil, api_key: nil, model: nil, max_tokens: nil, temperature: nil, frequency_penalty: nil)
+        @model = model || 'gemini-pro'
+        @uri_base = uri_base # || 'https://api.anthropic.com/v1/'
+        @api_key = api_key || OxAiWorkers.configuration.access_token_gemini
+        super(uri_base: @uri_base, api_key: @api_key, model: @model, max_tokens:, temperature:, frequency_penalty:)
+      end
+
+      def client
+        Gemini.new(
+          credentials: {
+            service: 'generative-language-api',
+            api_key: ENV['GOOGLE_API_KEY']
+          },
+          options: { model: 'gemini-pro', server_sent_events: true }
+        )
+      end
+
+      def request(parameters)
+        client.messages(parameters:)
+      end
+
+      def build_parameters(messages:, tools: [], filtered_functions: [], tool_choice: nil)
+        parameters = {
+          model: @model,
+          system: messages.select { |m| m[:role] == :system }.map { |m| m[:content] }.join("\n\n"),
+          messages: messages.reject { |m| m[:role] == :system },
+          temperature: @temperature,
+          max_tokens: @max_tokens
+        }
+        if tools.present?
+          functions = tools.map(&:to_anthropic_format).flatten
+
+          @names = functions.map { |f| f[:name] }
+
+          parameters[:tools] = functions.reject { |f| filtered_functions.include?(f[:name]) }
+
+          parameters[:tool_choice] =
+            tool_choice.nil? ? { type: 'any' } : { type: 'tool', name: tool_choice }
+        end
+        parameters
+      end
+
+      def tool_call(name:, args:, call_id:, out:)
+        [
+          {
+            role: :assistant,
+            content: [
+              {
+                type: 'tool_use',
+                id: "call_#{call_id}",
+                name:,
+                input: args
+              }
+            ]
+          },
+          {
+            role: :user,
+            content: [
+              {
+                type: 'tool_result',
+                tool_use_id: "call_#{call_id}",
+                content: out.present? ? out : "Tool call #{name} successful."
+              }
+            ]
+          }
+        ]
+      end
+
+      def add_base64(binary:, filename:, text:, mime_type:)
+        content = []
+        content << { type: 'text', text: } if text.present?
+        content << {
+                      type: mime_type.include?('image') ? 'image' : 'document',
+                      source: {
+                        type: 'base64',
+                        media_type: mime_type,
+                        data: Base64.strict_encode64(binary)
+                      }
+                    }
+        content
+      end
+
+      def add_url(url:, text:, mime_type:)
+        content = []
+        content << { type: 'text', text: } if text.present?
+        content << {
+                      type: mime_type.include?('image') ? 'image' : 'document',
+                      source: {
+                        type: 'url',
+                        url: url
+                      }
+                    }
+        content
+      end
+
+      def parse_response(response, &)
+        choices = response['content']
+        @is_truncated = (response['stop_reason'] == 'max_tokens')
+        return if choices.nil? || choices.empty?
+
+        choices.each do |choice|
+          result, tool_calls = parse_one_choice(choice)
+          yield(result, @is_truncated, tool_calls)
+        end
+      end
+
+      def parse_one_choice(choice)
+        return unless choice # Skip if there's no choice
+
+        # Initialize result variables
+        @result = nil
+        @tool_calls = []
+
+        # Process content item
+        if choice['type'] == 'tool_use'
+          # Handle tool use
+          begin
+            # Attempt to parse arguments, handle potential JSON errors
+            args = JSON.parse(choice['input'].to_json, symbolize_names: true)
+          rescue JSON::ParserError => e
+            OxAiWorkers.logger.error("Failed to parse tool call arguments: #{e.message}", for: self.class)
+            OxAiWorkers.logger.debug("Raw arguments: #{choice['input']}", for: self.class)
+          end
+
+          fname = @names.find { |n| n.end_with?(choice['name']) }
+          if fname != choice['name']
+            OxAiWorkers.logger.error("Tool call name #{choice['name']} not found. Using #{fname} instead.",
+                                     for: self.class)
+          end
+
+          tool_call = {
+            class: fname.split('__').first,
+            name: fname.split('__').last,
+            args:
+          }
+          @tool_calls << tool_call
+        elsif choice['type'] == 'text'
+          # Handle text content
+          @result = choice['text']
+        end
+
+        [@result, @tool_calls]
+      end
+    end
+  end
+end

data/lib/oxaiworkers/models/llm_base.rb

@@ -66,7 +66,7 @@ module OxAiWorkers
         ]
       end
 
-      def add_base64(binary:, filename:, text:, mime_type:, detail: 'auto')
+      def add_base64(binary:, text:, mime_type:, detail: 'auto')
         content = []
         content << { type: 'text', text: } if text.present?
         content << if mime_type.include?('image')
@@ -80,7 +80,7 @@ module OxAiWorkers
                      {
                        type: 'file',
                        file: {
-                         filename:,
+                         filename: "file.#{mime_type.split('/').last}",
                          file_data: "data:#{mime_type};base64,#{Base64.strict_encode64(binary)}"
                        }
                      }

data/lib/oxaiworkers/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module OxAiWorkers
-  VERSION = '1.1.0.1'
+  VERSION = '1.1.1.1'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ox-ai-workers
 version: !ruby/object:Gem::Version
-  version: 1.1.0.1
+  version: 1.1.1.1
 platform: ruby
 authors:
 - Denis Smolev
@@ -159,6 +159,7 @@ files:
 - lib/oxaiworkers/load_i18n.rb
 - lib/oxaiworkers/models/anthropic_max.rb
 - lib/oxaiworkers/models/deepseek_max.rb
+- lib/oxaiworkers/models/gemini_pro.rb
 - lib/oxaiworkers/models/images_base.rb
 - lib/oxaiworkers/models/llm_base.rb
 - lib/oxaiworkers/models/openai_dalle3.rb