Use system_prompt in solver configuration (breaking change)

ovos_solver_openai_persona/__init__.py

@@ -1,59 +1,29 @@
-from typing import Optional
-
+import warnings
 from ovos_solver_openai_persona.engines import OpenAIChatCompletionsSolver
 
-
 class OpenAIPersonaSolver(OpenAIChatCompletionsSolver):
-    """default "Persona" engine"""
-
-    def __init__(self, config=None):
-        # defaults to gpt-3.5-turbo
-        super().__init__(config=config)
-        self.default_persona = config.get("persona") or "helpful, creative, clever, and very friendly."
-
-    def get_chat_history(self, persona=None):
-        persona = persona or self.default_persona
-        initial_prompt = f"You are a helpful assistant. " \
-                         f"You give short and factual answers. " \
-                         f"You are {persona}"
-        return super().get_chat_history(initial_prompt)
-
-    # officially exported Solver methods
-    def get_spoken_answer(self, query: str,
-                          lang: Optional[str] = None,
-                          units: Optional[str] = None) -> Optional[str]:
-        """
-        Obtain the spoken answer for a given query.
-
-        Args:
-            query (str): The query text.
-            lang (Optional[str]): Optional language code. Defaults to None.
-            units (Optional[str]): Optional units for the query. Defaults to None.
-
-        Returns:
-            str: The spoken answer as a text response.
-        """
-        answer = super().get_spoken_answer(query, lang, units)
-        if not answer or not answer.strip("?") or not answer.strip("_"):
-            return None
-        return answer
-
+    def __init__(self, *args, **kwargs):
+           warnings.warn(
+              "use OpenAIChatCompletionsSolver instead",
+              DeprecationWarning,
+              stacklevel=2,
+           )
+           super().__init__(*args, **kwargs)
 
 # for ovos-persona
 LLAMA_DEMO = {
   "name": "Remote LLama",
   "solvers": [
-    "ovos-solver-openai-plugin"
+    "ovos-solver-openai-persona-plugin"
   ],
-  "ovos-solver-openai-plugin": {
+  "ovos-solver-openai-persona-plugin": {
     "api_url": "https://llama.smartgic.io/v1",
     "key": "sk-xxxx"
   }
 }
 
-
 if __name__ == "__main__":
-    bot = OpenAIPersonaSolver(LLAMA_DEMO["ovos-solver-openai-plugin"])
+    bot = OpenAIChatCompletionsSolver(LLAMA_DEMO["ovos-solver-openai-persona-plugin"])
     #for utt in bot.stream_utterances("describe quantum mechanics in simple terms"):
     #    print(utt)
         #  Quantum mechanics is a branch of physics that studies the behavior of atoms and particles at the smallest scales.

ovos_solver_openai_persona/dialog_transformers.py

@@ -12,7 +12,7 @@ def __init__(self, name="ovos-dialog-transformer-openai-plugin", priority=10, co
             "key": self.config.get("key"),
             'api_url': self.config.get('api_url', 'https://api.openai.com/v1'),
             "enable_memory": False,
-            "initial_prompt": "your task is to rewrite text as if it was spoken by a different character"
+            "system_prompt": self.config.get("system_prompt") or "Your task is to rewrite text as if it was spoken by a different character"
         })
 
     def transform(self, dialog: str, context: dict = None) -> Tuple[str, dict]:

ovos_solver_openai_persona/engines.py

@@ -24,7 +24,7 @@ def __init__(self, config=None,
                          enable_tx=enable_tx, enable_cache=enable_cache,
                          internal_lang=internal_lang)
         self.api_url = f"{self.config.get('api_url', 'https://api.openai.com/v1')}/completions"
-        self.engine = self.config.get("model", "text-davinci-002")  # "ada" cheaper and faster, "davinci" better
+        self.engine = self.config.get("model", "gpt-4o-mini")
         self.key = self.config.get("key")
         if not self.key:
             LOG.error("key not set in config")
@@ -107,7 +107,12 @@ def __init__(self, config=None,
         self.memory = config.get("enable_memory", True)
         self.max_utts = config.get("memory_size", 3)
         self.qa_pairs = []  # tuple of q+a
-        self.initial_prompt = config.get("initial_prompt", "You are a helpful assistant.")
+        if "initial_prompt" in config:
+            LOG.warning("'initial_prompt' config option is deprecated, use 'system_prompt' instead")
+        self.system_prompt = config.get("system_prompt") or config.get("initial_prompt")
+        if not self.system_prompt:
+            self.system_prompt =  "You are a helpful assistant."
+            LOG.error(f"system prompt not set in config! defaulting to '{self.system_prompt}'")
 
     # OpenAI API integration
     def _do_api_request(self, messages):
@@ -179,19 +184,19 @@ def _do_streaming_api_request(self, messages):
                     continue
                 yield chunk["choices"][0]["delta"]["content"]
 
-    def get_chat_history(self, initial_prompt=None):
+    def get_chat_history(self, system_prompt=None):
         qa = self.qa_pairs[-1 * self.max_utts:]
-        initial_prompt = initial_prompt or self.initial_prompt or "You are a helpful assistant."
+        system_prompt = system_prompt or self.system_prompt or "You are a helpful assistant."
         messages = [
-            {"role": "system", "content": initial_prompt},
+            {"role": "system", "content": system_prompt},
         ]
         for q, a in qa:
             messages.append({"role": "user", "content": q})
             messages.append({"role": "assistant", "content": a})
         return messages
 
-    def get_messages(self, utt, initial_prompt=None) -> MessageList:
-        messages = self.get_chat_history(initial_prompt)
+    def get_messages(self, utt, system_prompt=None) -> MessageList:
+        messages = self.get_chat_history(system_prompt)
         messages.append({"role": "user", "content": utt})
         return messages
 
@@ -209,6 +214,8 @@ def continue_chat(self, messages: MessageList,
         Returns:
             Optional[str]: The generated response or None if no response could be generated.
         """
+        if not messages or messages[0]["role"] != "system":
+            messages = [{"role": "system", "content": self.system_prompt }] + messages
         response = self._do_api_request(messages)
         answer = post_process_sentence(response)
         if not answer or not answer.strip("?") or not answer.strip("_"):
@@ -218,7 +225,7 @@ def continue_chat(self, messages: MessageList,
             self.qa_pairs.append((query, answer))
         return answer
 
-    def stream_chat_utterances(self, messages: List[Dict[str, str]],
+    def stream_chat_utterances(self, messages: MessageList,
                                lang: Optional[str] = None,
                                units: Optional[str] = None) -> Iterable[str]:
         """
@@ -232,6 +239,8 @@ def stream_chat_utterances(self, messages: List[Dict[str, str]],
         Returns:
             Iterable[str]: An iterable of utterances.
         """
+        if not messages or messages[0]["role"] != "system":
+            messages = [{"role": "system", "content": self.system_prompt }] + messages
         answer = ""
         query = messages[-1]["content"]
         if self.memory:

setup.py

@@ -48,7 +48,7 @@ def get_version():
 
 
 PERSONA_ENTRY_POINT = 'Remote Llama=ovos_solver_openai_persona:LLAMA_DEMO'
-PLUGIN_ENTRY_POINT = 'ovos-solver-openai-plugin=ovos_solver_openai_persona:OpenAIPersonaSolver'
+PLUGIN_ENTRY_POINT = 'ovos-solver-openai-plugin=ovos_solver_openai_persona.engines:OpenAICompletionsSolver'
 DIALOG_PLUGIN_ENTRY_POINT = 'ovos-dialog-transformer-openai-plugin=ovos_solver_openai_persona.dialog_transformers:OpenAIDialogTransformer'
 SUMMARIZER_ENTRY_POINT = 'ovos-summarizer-openai-plugin=ovos_solver_openai_persona.summarizer:OpenAISummarizer'