From 2ade066468edd418fd21a815091275204d275543 Mon Sep 17 00:00:00 2001
From: Rodribm10 <rodrigobm10@me.com>
Date: Fri, 1 May 2026 16:08:18 -0300
Subject: [PATCH] feat(captain): EmbeddingService aceita provider de embedding
 dedicado
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Permite trocar provider de embedding sem afetar o provider de chat. Útil
quando OpenAI key tradicional está fora (ban, billing, etc) mas você
quer usar outro provider OpenAI-compatible só pra embeddings — exemplo
clássico: Gemini OpenAI-compatible em
https://generativelanguage.googleapis.com/v1beta/openai com modelo
gemini-embedding-001 + dimensions=1536 (pra bater com schema pgvector).

Env vars novas (com fallback pro legacy_openai_settings se não setadas):

  CAPTAIN_EMBEDDING_API_KEY     — API key dedicada pra embeddings
  CAPTAIN_EMBEDDING_ENDPOINT    — base URL sem /v1 (default herda OpenAI)
  CAPTAIN_EMBEDDING_DIMENSIONS  — força redução do vector (ex: 1536)

Quando CAPTAIN_EMBEDDING_API_KEY está vazia, comportamento é idêntico ao
de antes (legacy_openai_settings). Backward-compatible.

Também aceita as variáveis via InstallationConfig (UI) ou ENV — ENV tem
precedência (padrão Chatwoot).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 .../services/captain/llm/embedding_service.rb | 42 ++++++++++++++++---
 1 file changed, 36 insertions(+), 6 deletions(-)

diff --git a/enterprise/app/services/captain/llm/embedding_service.rb b/enterprise/app/services/captain/llm/embedding_service.rb
index 7cbb3c03c..420ddb027 100644
--- a/enterprise/app/services/captain/llm/embedding_service.rb
+++ b/enterprise/app/services/captain/llm/embedding_service.rb
@@ -26,17 +26,47 @@ class Captain::Llm::EmbeddingService
 
   private
 
-  # Embeddings sempre vão direto pra OpenAI tradicional — o endpoint Codex
-  # via ChatGPT OAuth não expõe /embeddings.
+  # Embeddings vão pra OpenAI tradicional por default (o endpoint Codex
+  # via ChatGPT OAuth não expõe /embeddings). Override opcional via env vars
+  # dedicadas — útil pra trocar provider de embedding (ex: Gemini
+  # OpenAI-compatible) sem alterar o provider de chat:
+  #
+  #   CAPTAIN_EMBEDDING_API_KEY     — sobrescreve API key
+  #   CAPTAIN_EMBEDDING_ENDPOINT    — sobrescreve base URL (sem /v1 no final)
+  #   CAPTAIN_EMBEDDING_DIMENSIONS  — força reduction (ex: 1536 pra Gemini
+  #                                    bater com schema pgvector(1536))
   def embed_with_legacy_openai(content, model)
-    legacy = Captain::Llm::ProviderConfig.legacy_openai_settings
-    api_base = legacy[:api_base].present? ? "#{legacy[:api_base]}/v1" : nil
+    settings = embedding_settings
+    api_base = settings[:api_base].present? ? "#{settings[:api_base]}/v1" : nil
+    embed_options = embed_extra_options
 
-    Llm::Config.with_api_key(legacy[:api_key], api_base: api_base) do |ctx|
-      ctx.embed(content, model: model).vectors
+    Llm::Config.with_api_key(settings[:api_key], api_base: api_base) do |ctx|
+      ctx.embed(content, model: model, **embed_options).vectors
     end
   end
 
+  def embedding_settings
+    custom_key = installation_config_value('CAPTAIN_EMBEDDING_API_KEY')
+    return Captain::Llm::ProviderConfig.legacy_openai_settings if custom_key.blank?
+
+    {
+      api_key: custom_key,
+      api_base: installation_config_value('CAPTAIN_EMBEDDING_ENDPOINT')&.chomp('/')
+    }
+  end
+
+  def embed_extra_options
+    dims = installation_config_value('CAPTAIN_EMBEDDING_DIMENSIONS')
+    return {} if dims.blank?
+
+    { dimensions: dims.to_i }
+  end
+
+  def installation_config_value(name)
+    ENV.fetch(name, nil) ||
+      InstallationConfig.find_by(name: name)&.value
+  end
+
   def instrumentation_params(content, model)
     {
       span_name: 'llm.captain.embedding',