docit 0.5.0 → 0.7.0

data/lib/docit/ai/system_insight_generator.rb

@@ -1,127 +0,0 @@
-# frozen_string_literal: true
-
-require "json"
-
-module Docit
-  module Ai
-    class SystemInsightGenerator
-      # mode: :nodes  -> explain an arbitrary selection (diagram "AI Explain")
-      #       :section -> explain one resource and how its endpoints work together
-      def initialize(graph:, selected_node_ids: [], mode: :nodes)
-        @graph = graph
-        @selected_node_ids = selected_node_ids
-        @mode = mode
-      end
-
-      def generate
-        config = Configuration.load
-        Client.for(config).generate(prompt)
-      end
-
-      private
-
-      attr_reader :graph, :selected_node_ids, :mode
-
-      def prompt
-        mode == :section ? section_prompt : nodes_prompt
-      end
-
-      def nodes_prompt
-        <<~PROMPT
-          You are a senior engineer explaining a Rails system architecture to a developer. Keep your explanation extremely concise, professional, and clear. Avoid fluff, unnecessary details, or general tutorial information. Focus only on the provided components.
-
-          FORMAT YOUR RESPONSE EXACTLY LIKE THIS:
-
-          ## Overview
-          A 1-2 sentence plain-English summary of what this component/action does.
-
-          ## Data Flow
-          Show a simple, linear flow diagram using text and arrows (→). Keep it to 1 line if possible.
-          Example:
-            Client → GET /api/v1/users → UsersController#index → queries User model
-
-          ## Connections & Interactions
-          List only the direct, relevant relationships from the graph (max 3 bullets):
-          - **Component** (type): Action details/purpose.
-
-          Do not invent or assume anything outside of the provided graph. Keep the total response under 150 words.
-
-          ---
-
-          Selected node ids:
-          #{selected_node_ids.join("\n")}
-
-          Graph JSON:
-          #{JSON.pretty_generate(compact_graph)}
-        PROMPT
-      end
-
-      def section_prompt
-        <<~PROMPT
-          You are a senior engineer writing the introduction to a section of API documentation. The section covers ONE resource and all of its endpoints. Explain, for a developer new to this codebase, what the resource is for and how its endpoints work together as a workflow.
-
-          Use the documented summaries where available. Where an endpoint has no documentation, infer cautiously from its HTTP method and path, and do not fabricate behavior.
-
-          FORMAT YOUR RESPONSE EXACTLY LIKE THIS:
-
-          ## What this section does
-          1-2 sentences on the resource and its overall purpose.
-
-          ## How the endpoints work together
-          A short narrative (2-4 sentences) describing the typical flow across these endpoints — e.g. how a client lists, creates, then updates this resource. Reference endpoints by their HTTP method and path.
-
-          ## Notes
-          Up to 2 bullets on related models/services or anything a consumer must know. Omit this section if there is nothing concrete to say.
-
-          Do not invent endpoints, fields, or behavior not present in the graph. Keep the total response under 180 words.
-
-          ---
-
-          Endpoint node ids in this section:
-          #{selected_node_ids.join("\n")}
-
-          Graph JSON:
-          #{JSON.pretty_generate(compact_graph)}
-        PROMPT
-      end
-
-      def compact_graph
-        nodes = selected_nodes
-        node_ids = nodes.map { |node| node[:id] }
-
-        # Include edges where at least one end is in the selection
-        related_edges = graph[:edges].select do |edge|
-          node_ids.include?(edge[:source]) || node_ids.include?(edge[:target])
-        end
-
-        # Also include neighbor nodes (one hop away) for context
-        neighbor_ids = Set.new(node_ids)
-        related_edges.each do |edge|
-          neighbor_ids.add(edge[:source])
-          neighbor_ids.add(edge[:target])
-        end
-
-        neighbor_nodes = graph[:nodes].select { |node| neighbor_ids.include?(node[:id]) }
-
-        {
-          selected_nodes: nodes.map { |node| compact_hash(node, %i[id type label status file metadata]) },
-          context_nodes: (neighbor_nodes - nodes).map { |node| compact_hash(node, %i[id type label status]) },
-          edges: related_edges.map { |edge| compact_hash(edge, %i[source target type confidence evidence]) },
-          stats: graph[:stats]
-        }
-      end
-
-      def selected_nodes
-        return graph[:nodes] if selected_node_ids.empty?
-
-        graph[:nodes].select { |node| selected_node_ids.include?(node[:id]) }
-      end
-
-      def compact_hash(hash, keys)
-        keys.each_with_object({}) do |key, result|
-          result[key] = hash[key] if hash.key?(key)
-        end
-      end
-    end
-  end
-end