endpoints

Author: romanrizzi

app/controllers/discourse_ai/admin/ai_embeddings_controller.rb

@@ -0,0 +1,116 @@
+# frozen_string_literal: true
+
+module DiscourseAi
+  module Admin
+    class AiEmbeddingsController < ::Admin::AdminController
+      requires_plugin ::DiscourseAi::PLUGIN_NAME
+
+      def index
+        embedding_defs = EmbeddingDefinition.all.order(:display_name)
+
+        render json: {
+                 ai_embeddings:
+                   ActiveModel::ArraySerializer.new(
+                     embedding_defs,
+                     each_serializer: AiEmbeddingDefinitionSerializer,
+                     root: false,
+                   ).as_json,
+                 meta: {
+                   provider_params: EmbeddingDefinition.provider_params,
+                   providers: EmbeddingDefinition.provider_names,
+                   tokenizers:
+                     EmbeddingDefinition.tokenizer_names.map { |tn|
+                       { id: tn, name: tn.split("::").last }
+                     },
+                 },
+               }
+      end
+
+      def new
+      end
+
+      def edit
+        embedding_def = EmbeddingDefinition.find(params[:id])
+        render json: AiEmbeddingDefinitionSerializer.new(embedding_def)
+      end
+
+      def create
+        embedding_def = EmbeddingDefinition.new(ai_embeddings_params)
+
+        if embedding_def.save
+          render json: AiEmbeddingDefinitionSerializer.new(embedding_def), status: :created
+        else
+          render_json_error embedding_def
+        end
+      end
+
+      def update
+        embedding_def = EmbeddingDefinition.find(params[:id])
+
+        if embedding_def.update(ai_embeddings_params)
+          render json: AiEmbeddingDefinitionSerializer.new(embedding_def)
+        else
+          render_json_error embedding_def
+        end
+      end
+
+      def destroy
+        embedding_def = EmbeddingDefinition.find(params[:id])
+
+        if embedding_def.id == SiteSetting.ai_embeddings_selected_model.to_i
+          return render_json_error(I18n.t("discourse_ai.embeddings.delete_failed"), status: 409)
+        end
+
+        if embedding_def.destroy
+          head :no_content
+        else
+          render_json_error embedding_def
+        end
+      end
+
+      def test
+        RateLimiter.new(
+          current_user,
+          "ai_embeddings_test_#{current_user.id}",
+          3,
+          1.minute,
+        ).performed!
+
+        embedding_def = EmbeddingDefinition.new(ai_embeddings_params)
+        DiscourseAi::Embeddings::Vector.new(embedding_def).vector_from("this is a test")
+
+        render json: { success: true }
+      rescue Net::HTTPBadResponse => e
+        render json: { success: false, error: e.message }
+      end
+
+      private
+
+      def ai_embeddings_params
+        permitted =
+          params.require(:ai_embedding).permit(
+            :display_name,
+            :dimensions,
+            :max_sequence_length,
+            :pg_function,
+            :provider,
+            :url,
+            :api_key,
+            :tokenizer_class,
+          )
+
+        extra_field_names = EmbeddingDefinition.provider_params.dig(permitted[:provider]&.to_sym)
+        if extra_field_names.present?
+          received_prov_params =
+            params.dig(:ai_embedding, :provider_params)&.slice(*extra_field_names.keys)
+
+          if received_prov_params.present?
+            permitted[:provider_params] = received_prov_params.permit!
+          end
+        end
+
+        permitted
+      end
+    end
+  end
+end

app/models/embedding_definition.rb

@@ -23,14 +23,15 @@ def tokenizer_names
       ].map(&:name)
     end
 
-    def self.provider_params
+    def provider_params
       { discourse: { model_name: :text }, open_ai: { model_name: :text } }
     end
   end
 
   validates :provider, presence: true, inclusion: provider_names
   validates :display_name, presence: true, length: { maximum: 100 }
   validates :tokenizer_class, presence: true, inclusion: tokenizer_names
+  validates_presence_of :url, :api_key, :dimensions, :max_sequence_length, :pg_function
 
   def tokenizer
     tokenizer_class.constantize
@@ -60,7 +61,7 @@ def lookup_custom_param(key)
   def endpoint_url
     return url if !url.starts_with?("srv://")
 
-    service = DiscourseAi::Utils::DnsSrv.lookup(url)
+    service = DiscourseAi::Utils::DnsSrv.lookup(url.sub("srv://", ""))
     "https://#{service.target}:#{service.port}"
   end
 

app/serializers/ai_embedding_definition_serializer.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+class AiEmbeddingDefinitionSerializer < ApplicationSerializer
+  root "ai_embedding"
+
+  attributes :display_name,
+             :dimensions,
+             :max_sequence_length,
+             :pg_function,
+             :provider,
+             :url,
+             :api_key,
+             :tokenizer_class,
+             :provider_params
+end

config/locales/server.en.yml

@@ -434,6 +434,7 @@ en:
       cannot_edit_builtin: "You can't edit a built-in model."
 
     embeddings:
+      delete_failed: "This model is currently in use. Update the `ai embeddings selected model` first."
       configuration:
         disable_embeddings: "You have to disable 'ai embeddings enabled' first."
         choose_model: "Set 'ai embeddings selected model' first."

config/routes.rb

@@ -96,6 +96,13 @@
               controller: "discourse_ai/admin/ai_llm_quotas",
               path: "quotas",
               only: %i[index create update destroy]
+
+    resources :ai_embeddings,
+              only: %i[index new create edit update destroy],
+              path: "ai-embeddings",
+              controller: "discourse_ai/admin/ai_embeddings" do
+      collection { get :test }
+    end
   end
 end
 

lib/inference/cloudflare_workers_ai.rb

@@ -9,14 +9,6 @@ def initialize(endpoint, api_token, referer = Discourse.base_url)
         @referer = referer
       end
 
-      def self.instance(model)
-        new(
-          SiteSetting.ai_cloudflare_workers_account_id,
-          SiteSetting.ai_cloudflare_workers_api_token,
-          model,
-        )
-      end
-
       attr_reader :endpoint, :api_token, :referer
 
       def perform!(content)
@@ -40,7 +32,7 @@ def perform!(content)
           Rails.logger.warn(
             "Cloudflare Workers AI Embeddings failed with status: #{response.status} body: #{response.body}",
           )
-          raise Net::HTTPBadResponse
+          raise Net::HTTPBadResponse.new(response.body.to_s)
         end
       end
     end

lib/inference/discourse_classifier.rb

@@ -10,25 +10,6 @@ def initialize(endpoint, api_key, model, referer = Discourse.base_url)
         @referer = referer
       end
 
-      def self.instance(model)
-        endpoint =
-          if SiteSetting.ai_embeddings_discourse_service_api_endpoint_srv.present?
-            service =
-              DiscourseAi::Utils::DnsSrv.lookup(
-                SiteSetting.ai_embeddings_discourse_service_api_endpoint_srv,
-              )
-            "https://#{service.target}:#{service.port}"
-          else
-            SiteSetting.ai_embeddings_discourse_service_api_endpoint
-          end
-
-        new(
-          "#{endpoint}/api/v1/classify",
-          SiteSetting.ai_embeddings_discourse_service_api_key,
-          model,
-        )
-      end
-
       attr_reader :endpoint, :api_key, :model, :referer
 
       def perform!(content)
@@ -38,7 +19,9 @@ def perform!(content)
         conn = Faraday.new { |f| f.adapter FinalDestination::FaradayAdapter }
         response = conn.post(endpoint, { model: model, content: content }.to_json, headers)
 
-        raise Net::HTTPBadResponse if ![200, 415].include?(response.status)
+        if ![200, 415].include?(response.status)
+          raise raise Net::HTTPBadResponse.new(response.body.to_s)
+        end
 
         JSON.parse(response.body, symbolize_names: true)
       end

lib/inference/gemini_embeddings.rb

@@ -3,13 +3,6 @@
 module ::DiscourseAi
   module Inference
     class GeminiEmbeddings
-      def self.instance
-        new(
-          "https://generativelanguage.googleapis.com/v1beta/models/embedding-001:embedContent",
-          SiteSetting.ai_gemini_api_key,
-        )
-      end
-
       def initialize(embedding_url, api_key, referer = Discourse.base_url)
         @api_key = api_key
         @embedding_url = embedding_url
@@ -35,7 +28,7 @@ def perform!(content)
           Rails.logger.warn(
             "Google Gemini Embeddings failed with status: #{response.status} body: #{response.body}",
           )
-          raise Net::HTTPBadResponse
+          raise Net::HTTPBadResponse.new(response.body.to_s)
         end
       end
     end

lib/inference/hugging_face_text_embeddings.rb

@@ -12,19 +12,6 @@ def initialize(endpoint, key, referer = Discourse.base_url)
       attr_reader :endpoint, :key, :referer
 
       class << self
-        def instance
-          endpoint =
-            if SiteSetting.ai_hugging_face_tei_endpoint_srv.present?
-              service =
-                DiscourseAi::Utils::DnsSrv.lookup(SiteSetting.ai_hugging_face_tei_endpoint_srv)
-              "https://#{service.target}:#{service.port}"
-            else
-              SiteSetting.ai_hugging_face_tei_endpoint
-            end
-
-          new(endpoint, SiteSetting.ai_hugging_face_tei_api_key)
-        end
-
         def configured?
           SiteSetting.ai_hugging_face_tei_endpoint.present? ||
             SiteSetting.ai_hugging_face_tei_endpoint_srv.present?
@@ -100,7 +87,7 @@ def perform!(content)
         conn = Faraday.new { |f| f.adapter FinalDestination::FaradayAdapter }
         response = conn.post(endpoint, body, headers)
 
-        raise Net::HTTPBadResponse if ![200].include?(response.status)
+        raise Net::HTTPBadResponse.new(response.body.to_s) if ![200].include?(response.status)
 
         JSON.parse(response.body, symbolize_names: true).first
       end

lib/inference/open_ai_embeddings.rb

@@ -12,10 +12,6 @@ def initialize(endpoint, api_key, model, dimensions)
 
       attr_reader :endpoint, :api_key, :model, :dimensions
 
-      def self.instance(model:, dimensions: nil)
-        new(SiteSetting.ai_openai_embeddings_url, SiteSetting.ai_openai_api_key, model, dimensions)
-      end
-
       def perform!(content)
         headers = { "Content-Type" => "application/json" }
 
@@ -29,7 +25,7 @@ def perform!(content)
         payload[:dimensions] = dimensions if dimensions.present?
 
         conn = Faraday.new { |f| f.adapter FinalDestination::FaradayAdapter }
-        response = conn.post(SiteSetting.ai_openai_embeddings_url, payload.to_json, headers)
+        response = conn.post(endpoint, payload.to_json, headers)
 
         case response.status
         when 200
@@ -40,7 +36,7 @@ def perform!(content)
           Rails.logger.warn(
             "OpenAI Embeddings failed with status: #{response.status} body: #{response.body}",
           )
-          raise Net::HTTPBadResponse
+          raise Net::HTTPBadResponse.new(response.body.to_s)
         end
       end
     end