Update LLM configuration to support FastGPT integration. Modify requirements to include fastgpt-python-sdk, enhance greeting messages, and adjust LLM service creation to handle app_id. Implement welcome text fetching for FastGPT and improve context handling in the pipeline based on LLM provider. Update related configurations and properties for better integration.

engine/services.py

@@ -54,12 +54,13 @@ def create_llm_service(
     session_variables: dict | None = None,
     greeting_prompt: str | None = None,
 ):
-    if config.provider == "fastgpt":
+    if config.is_fastgpt:
         variables = {**config.variables, **(session_variables or {})}
         return FastGPTLLMService(
             api_key=config.api_key,
             base_url=config.base_url or "http://localhost:3000",
             chat_id=chat_id or config.chat_id,
+            app_id=config.app_id,
             send_system_prompt=config.send_system_prompt,
             greeting_prompt=greeting_prompt,
             timeout=config.timeout_sec,
@@ -70,7 +71,11 @@ def create_llm_service(
             ),
         )
 
-    _require_provider(config.provider, "openai", "llm")
+    if not config.is_openai:
+        supported = ", ".join(sorted(("openai", "fastgpt", "llm")))
+        raise ValueError(
+            f"Unsupported llm provider {config.provider!r}; expected one of: {supported}"
+        )
     return OpenAILLMService(
         api_key=config.api_key or None,
         base_url=config.base_url,