language-operator 0.1.57 → 0.1.59

data/lib/language_operator/learning/pattern_detector.rb

@@ -0,0 +1,260 @@
+# frozen_string_literal: true
+
+require 'logger'
+
+module LanguageOperator
+  module Learning
+    # Detects patterns in task execution traces and generates symbolic code
+    #
+    # The PatternDetector analyzes execution patterns from TraceAnalyzer and
+    # converts deterministic neural task behavior into symbolic Ruby DSL v1 code.
+    # This enables the learning system to automatically optimize neural tasks
+    # into faster, cheaper symbolic implementations.
+    #
+    # @example Basic usage
+    #   analyzer = TraceAnalyzer.new(endpoint: ENV['OTEL_QUERY_ENDPOINT'])
+    #   validator = Agent::Safety::ASTValidator.new
+    #   detector = PatternDetector.new(trace_analyzer: analyzer, validator: validator)
+    #
+    #   analysis = analyzer.analyze_patterns(task_name: 'fetch_user_data')
+    #   result = detector.detect_pattern(analysis_result: analysis)
+    #
+    #   if result[:success]
+    #     puts "Generated code:"
+    #     puts result[:generated_code]
+    #   end
+    class PatternDetector
+      # Default minimum consistency threshold for learning
+      DEFAULT_CONSISTENCY_THRESHOLD = 0.85
+
+      # Default minimum executions required before learning
+      DEFAULT_MIN_EXECUTIONS = 10
+
+      # Minimum pattern confidence for code generation
+      MIN_PATTERN_CONFIDENCE = 0.75
+
+      # Initialize pattern detector
+      #
+      # @param trace_analyzer [TraceAnalyzer] Analyzer for querying execution traces
+      # @param validator [Agent::Safety::ASTValidator] Validator for generated code
+      # @param logger [Logger, nil] Logger instance (creates default if nil)
+      def initialize(trace_analyzer:, validator:, logger: nil)
+        @trace_analyzer = trace_analyzer
+        @validator = validator
+        @logger = logger || ::Logger.new($stdout, level: ::Logger::WARN)
+      end
+
+      # Detect patterns and generate symbolic code
+      #
+      # Main entry point for pattern detection. Takes analysis results from
+      # TraceAnalyzer and generates validated symbolic code if the pattern
+      # meets consistency and execution count thresholds.
+      #
+      # @param analysis_result [Hash] Result from TraceAnalyzer#analyze_patterns
+      # @return [Hash] Detection result with generated code and metadata
+      def detect_pattern(analysis_result:)
+        # Validate that we can generate code from this analysis
+        return early_rejection_result(analysis_result) unless can_generate_code?(analysis_result)
+
+        # Generate symbolic code from the pattern
+        code = generate_symbolic_code(
+          pattern: analysis_result[:common_pattern],
+          task_name: analysis_result[:task_name]
+        )
+
+        # Validate the generated code
+        validation_result = validate_generated_code(code: code)
+
+        # Build result
+        {
+          success: validation_result[:valid],
+          task_name: analysis_result[:task_name],
+          generated_code: code,
+          validation_violations: validation_result[:violations],
+          consistency_score: analysis_result[:consistency_score],
+          execution_count: analysis_result[:execution_count],
+          pattern: analysis_result[:common_pattern],
+          ready_to_deploy: validation_result[:valid] && analysis_result[:consistency_score] >= 0.90,
+          generated_at: Time.now.iso8601
+        }
+      end
+
+      # Generate symbolic Ruby code from tool call pattern
+      #
+      # Converts a deterministic tool call sequence into a valid Ruby DSL v1
+      # task definition with chained execute_task calls.
+      #
+      # @param pattern [String] Tool sequence like "db_fetch → cache_get → api"
+      # @param task_name [String] Name of the task being learned
+      # @return [String] Complete Ruby DSL v1 agent definition
+      def generate_symbolic_code(pattern:, task_name:)
+        sequence = extract_tool_sequence(pattern)
+
+        # Generate the task code body with chained execute_task calls
+        task_body = generate_task_code(sequence: sequence)
+
+        # Wrap in complete agent definition
+        generate_agent_wrapper(
+          task_name: task_name,
+          task_body: task_body
+        )
+      end
+
+      # Validate generated code with ASTValidator
+      #
+      # @param code [String] Ruby code to validate
+      # @return [Hash] Validation result with violations
+      def validate_generated_code(code:)
+        violations = @validator.validate(code)
+
+        {
+          valid: violations.empty?,
+          violations: violations,
+          safe_methods_used: true
+        }
+      rescue StandardError => e
+        @logger.error("Failed to validate generated code: #{e.message}")
+        {
+          valid: false,
+          violations: [{ type: :validation_error, message: e.message }],
+          safe_methods_used: false
+        }
+      end
+
+      private
+
+      # Check if pattern analysis meets criteria for code generation
+      #
+      # @param analysis_result [Hash] Analysis result from TraceAnalyzer
+      # @return [Boolean] True if code can be generated
+      def can_generate_code?(analysis_result)
+        return false unless analysis_result.is_a?(Hash)
+
+        # Check if ready for learning flag is set
+        return false unless analysis_result[:ready_for_learning]
+
+        # Check execution count
+        return false if analysis_result[:execution_count].to_i < DEFAULT_MIN_EXECUTIONS
+
+        # Check consistency score
+        return false if analysis_result[:consistency_score].to_f < DEFAULT_CONSISTENCY_THRESHOLD
+
+        # Check pattern exists
+        return false if analysis_result[:common_pattern].nil? || analysis_result[:common_pattern].empty?
+
+        true
+      end
+
+      # Build early rejection result when criteria not met
+      #
+      # @param analysis_result [Hash, nil] Analysis result if available
+      # @return [Hash] Rejection result with reason
+      def early_rejection_result(analysis_result)
+        if analysis_result.nil? || !analysis_result.is_a?(Hash)
+          return {
+            success: false,
+            reason: 'Invalid analysis result'
+          }
+        end
+
+        reasons = []
+        if analysis_result[:execution_count].to_i < DEFAULT_MIN_EXECUTIONS
+          reasons << "Insufficient executions (#{analysis_result[:execution_count]}/#{DEFAULT_MIN_EXECUTIONS})"
+        end
+        if analysis_result[:consistency_score].to_f < DEFAULT_CONSISTENCY_THRESHOLD
+          reasons << "Low consistency (#{analysis_result[:consistency_score]}/#{DEFAULT_CONSISTENCY_THRESHOLD})"
+        end
+        reasons << 'No common pattern found' if analysis_result[:common_pattern].nil? || analysis_result[:common_pattern].empty?
+
+        {
+          success: false,
+          task_name: analysis_result[:task_name],
+          execution_count: analysis_result[:execution_count],
+          consistency_score: analysis_result[:consistency_score],
+          ready_for_learning: false,
+          reason: reasons.join('; ')
+        }
+      end
+
+      # Extract tool sequence from pattern string
+      #
+      # @param pattern [String] Pattern like "db_fetch → cache_get → api"
+      # @return [Array<Symbol>] Sequence like [:db_fetch, :cache_get, :api]
+      def extract_tool_sequence(pattern)
+        pattern.split('→').map(&:strip).map(&:to_sym)
+      end
+
+      # Generate task code body with chained execute_task calls
+      #
+      # Creates Ruby code that executes tools in sequence, passing outputs
+      # from each tool to the next one as inputs.
+      #
+      # @param sequence [Array<Symbol>] Tool sequence
+      # @return [String] Ruby code for task body
+      def generate_task_code(sequence:)
+        return '      { result: {} }' if sequence.empty?
+
+        lines = []
+
+        # First call: use original inputs
+        first_tool = sequence[0]
+        lines << "step1_result = execute_task(:#{first_tool}, inputs: inputs)"
+
+        # Middle calls: chain outputs from previous step
+        if sequence.size > 1
+          sequence[1..-2].each_with_index do |tool, index|
+            step_num = index + 2
+            prev_step = "step#{step_num - 1}_result"
+            lines << "step#{step_num}_result = execute_task(:#{tool}, inputs: #{prev_step})"
+          end
+
+          # Final call
+          final_tool = sequence[-1]
+          last_step = "step#{sequence.size - 1}_result"
+          lines << "final_result = execute_task(:#{final_tool}, inputs: #{last_step})"
+        else
+          lines << 'final_result = step1_result'
+        end
+
+        # Return statement
+        lines << '{ result: final_result }'
+
+        # Indent and join
+        lines.map { |line| "      #{line}" }.join("\n")
+      end
+
+      # Generate complete agent wrapper with task definition
+      #
+      # @param task_name [String] Name of the task
+      # @param task_body [String] Generated task code body
+      # @return [String] Complete Ruby DSL v1 agent definition
+      def generate_agent_wrapper(task_name:, task_body:)
+        # Convert task name to kebab-case for agent name
+        agent_name = task_name.to_s.gsub('_', '-')
+
+        <<~RUBY
+          # frozen_string_literal: true
+
+          require 'language_operator'
+
+          LanguageOperator::Dsl.define_agents do
+            agent "#{agent_name}-symbolic" do
+              description "Symbolic implementation of #{task_name} (learned from execution patterns)"
+
+              task :core_pattern,
+                inputs: { data: 'hash' },
+                outputs: { result: 'hash' }
+              do |inputs|
+          #{task_body}
+              end
+
+              main do |inputs|
+                execute_task(:core_pattern, inputs: inputs)
+              end
+            end
+          end
+        RUBY
+      end
+    end
+  end
+end

data/lib/language_operator/learning/task_synthesizer.rb

@@ -0,0 +1,261 @@
+# frozen_string_literal: true
+
+require 'json'
+
+module LanguageOperator
+  module Learning
+    # Synthesizes deterministic code for neural tasks using LLM analysis
+    #
+    # TaskSynthesizer uses an LLM to analyze task definitions and execution traces,
+    # then generates optimized Ruby code if the task can be made deterministic.
+    # This approach can handle inconsistent traces better than pure pattern matching
+    # because the LLM can understand intent and unify variations.
+    #
+    # @example Basic usage
+    #   synthesizer = TaskSynthesizer.new(
+    #     llm_client: my_llm_client,
+    #     validator: ASTValidator.new
+    #   )
+    #
+    #   result = synthesizer.synthesize(
+    #     task_definition: task_def,
+    #     traces: execution_traces,
+    #     available_tools: tool_list
+    #   )
+    #
+    #   if result[:is_deterministic]
+    #     puts result[:code]
+    #   end
+    class TaskSynthesizer
+      # Template file name
+      TEMPLATE_FILE = 'task_synthesis.tmpl'
+
+      # Initialize synthesizer
+      #
+      # @param llm_client [Object] Client for LLM API calls (must respond to #chat)
+      # @param validator [Agent::Safety::ASTValidator] Code validator
+      # @param logger [Logger, nil] Logger instance
+      def initialize(llm_client:, validator:, logger: nil)
+        @llm_client = llm_client
+        @validator = validator
+        @logger = logger || ::Logger.new($stdout, level: ::Logger::WARN)
+      end
+
+      # Synthesize deterministic code for a task
+      #
+      # @param task_definition [Dsl::TaskDefinition] Task to optimize
+      # @param traces [Array<Hash>] Execution traces from TraceAnalyzer
+      # @param available_tools [Array<String>] List of available tool names
+      # @param consistency_score [Float] Current consistency score
+      # @param common_pattern [String, nil] Most common tool pattern
+      # @return [Hash] Synthesis result with :is_deterministic, :code, :explanation, :confidence
+      def synthesize(task_definition:, traces:, available_tools: [], consistency_score: 0.0, common_pattern: nil)
+        # Build prompt from template
+        prompt = build_prompt(
+          task_definition: task_definition,
+          traces: traces,
+          available_tools: available_tools,
+          consistency_score: consistency_score,
+          common_pattern: common_pattern
+        )
+
+        @logger.debug("Task synthesis prompt:\n#{prompt}")
+
+        # Call LLM
+        response = call_llm(prompt)
+
+        # Parse JSON response
+        result = parse_response(response)
+
+        # Validate generated code if present
+        if result[:is_deterministic] && result[:code]
+          validation = validate_code(result[:code])
+          unless validation[:valid]
+            @logger.warn("Generated code failed validation: #{validation[:errors].join(', ')}")
+            result[:validation_errors] = validation[:errors]
+            result[:is_deterministic] = false
+            result[:explanation] = "Generated code failed safety validation: #{validation[:errors].first}"
+          end
+        end
+
+        result
+      rescue StandardError => e
+        @logger.error("Task synthesis failed: #{e.message}")
+        @logger.error(e.backtrace&.first(10)&.join("\n"))
+        {
+          is_deterministic: false,
+          confidence: 0.0,
+          explanation: "Synthesis error: #{e.message}",
+          code: nil
+        }
+      end
+
+      private
+
+      # Build synthesis prompt from template
+      #
+      # @return [String] Rendered prompt
+      def build_prompt(task_definition:, traces:, available_tools:, consistency_score:, common_pattern:)
+        template = load_template
+
+        # Pre-render traces since our template engine doesn't support loops
+        traces_text = format_traces(traces)
+        inputs_text = format_schema(task_definition.inputs)
+        outputs_text = format_schema(task_definition.outputs)
+        tools_text = available_tools.map { |t| "- #{t}" }.join("\n")
+
+        # Count unique patterns
+        unique_patterns = traces.map { |t| (t[:tool_calls] || []).map { |tc| tc[:tool_name] }.join(' → ') }.uniq.size
+
+        # Build data hash for template substitution
+        data = {
+          'TaskName' => task_definition.name.to_s,
+          'Instructions' => task_definition.instructions || '(none)',
+          'Inputs' => inputs_text,
+          'Outputs' => outputs_text,
+          'TaskCode' => format_task_code(task_definition),
+          'TraceCount' => traces.size.to_s,
+          'Traces' => traces_text,
+          'CommonPattern' => common_pattern || '(none detected)',
+          'ConsistencyScore' => (consistency_score * 100).round(1).to_s,
+          'UniquePatternCount' => unique_patterns.to_s,
+          'ToolsList' => tools_text.empty? ? '(none available)' : tools_text
+        }
+
+        render_template(template, data)
+      end
+
+      # Load template file
+      #
+      # @return [String] Template content
+      def load_template
+        template_path = File.join(__dir__, '..', 'templates', TEMPLATE_FILE)
+        File.read(template_path)
+      end
+
+      # Render Go-style template with data
+      #
+      # @param template [String] Template content
+      # @param data [Hash] Data to substitute
+      # @return [String] Rendered content
+      def render_template(template, data)
+        result = template.dup
+
+        # Remove range blocks (we pre-render these)
+        result.gsub!(/{{range.*?}}.*?{{end}}/m, '')
+
+        # Remove if blocks for empty values
+        result.gsub!(/{{if not \.Inputs}}.*?{{end}}/m, '')
+        result.gsub!(/{{if \.InputSummary}}.*?{{end}}/m, '')
+
+        # Replace simple variables {{.Variable}}
+        data.each do |key, value|
+          result.gsub!("{{.#{key}}}", value.to_s)
+        end
+
+        result
+      end
+
+      # Format traces for template
+      #
+      # @param traces [Array<Hash>] Execution traces
+      # @return [String] Formatted traces text
+      def format_traces(traces)
+        return '(no traces available)' if traces.empty?
+
+        traces.first(10).each_with_index.map do |trace, idx|
+          tool_sequence = trace[:tool_calls]&.map { |tc| tc[:tool_name] }&.join(' → ') || '(no tools)'
+          duration = trace[:duration_ms]&.round(1) || 'unknown'
+          inputs_summary = trace[:inputs]&.keys&.join(', ') || 'none'
+
+          <<~TRACE
+            ### Execution #{idx + 1}
+            - **Tool Sequence:** #{tool_sequence}
+            - **Duration:** #{duration}ms
+            - **Inputs:** #{inputs_summary}
+          TRACE
+        end.join("\n")
+      end
+
+      # Format schema hash as text
+      #
+      # @param schema [Hash] Input/output schema
+      # @return [String] Formatted text
+      def format_schema(schema)
+        return '(none)' if schema.nil? || schema.empty?
+
+        schema.map { |k, v| "- #{k}: #{v}" }.join("\n")
+      end
+
+      # Format task definition as Ruby code
+      #
+      # @param task_def [Dsl::TaskDefinition] Task definition
+      # @return [String] Ruby code representation
+      def format_task_code(task_def)
+        inputs_str = (task_def.inputs || {}).map { |k, v| "#{k}: '#{v}'" }.join(', ')
+        outputs_str = (task_def.outputs || {}).map { |k, v| "#{k}: '#{v}'" }.join(', ')
+
+        <<~RUBY
+          task :#{task_def.name},
+            instructions: "#{task_def.instructions}",
+            inputs: { #{inputs_str} },
+            outputs: { #{outputs_str} }
+        RUBY
+      end
+
+      # Call LLM with prompt
+      #
+      # @param prompt [String] Synthesis prompt
+      # @return [String] LLM response
+      def call_llm(prompt)
+        @llm_client.chat(prompt)
+      end
+
+      # Parse JSON response from LLM
+      #
+      # @param response [String] LLM response text
+      # @return [Hash] Parsed result
+      def parse_response(response)
+        # Extract JSON from response (may be wrapped in markdown code block)
+        json_match = response.match(/```json\s*(.*?)\s*```/m) ||
+                     response.match(/\{.*\}/m)
+
+        json_str = json_match ? json_match[1] || json_match[0] : response
+
+        parsed = JSON.parse(json_str, symbolize_names: true)
+
+        {
+          is_deterministic: parsed[:is_deterministic] == true,
+          confidence: parsed[:confidence].to_f,
+          explanation: parsed[:explanation] || 'No explanation provided',
+          code: parsed[:code]
+        }
+      rescue JSON::ParserError => e
+        @logger.warn("Failed to parse LLM response as JSON: #{e.message}")
+        {
+          is_deterministic: false,
+          confidence: 0.0,
+          explanation: "Failed to parse synthesis response: #{e.message}",
+          code: nil
+        }
+      end
+
+      # Validate generated code
+      #
+      # @param code [String] Ruby code to validate
+      # @return [Hash] Validation result with :valid and :errors
+      def validate_code(code)
+        errors = @validator.validate(code)
+        {
+          valid: errors.empty?,
+          errors: errors
+        }
+      rescue StandardError => e
+        {
+          valid: false,
+          errors: ["Validation error: #{e.message}"]
+        }
+      end
+    end
+  end
+end