feat: test knowledge profile

Author: sjy3

.gitignore

@@ -199,4 +199,5 @@ cython_debug/
 **./output
 
 *.mp3
-*.pcm
+*.pcm
+.trae

veadk/knowledgebase/knowledgebase.py

@@ -303,6 +303,6 @@ async def generate_profiles(self, files: list[str], profile_path: str = ""):
         profile_names = [profile.name for profile in profiles]
 
         write_string_to_file(
-            profile_path + "/profile_list.json",
+            profile_path + "/all_profile_list.json",
             json.dumps(profile_names, indent=4, ensure_ascii=False),
         )

veadk/tools/builtin_tools/load_kb_queries.py

@@ -51,8 +51,19 @@ def load_kb_queries(profile_names: list[str], tool_context: ToolContext) -> list
 
     index = tool_context._invocation_context.agent.knowledgebase.index
 
+    all_profile_names = []
+
+    with open(
+            f"./profiles/knowledgebase/profiles_{index}/all_profile_list.json",
+            "r",
+    ) as f:
+        all_profile_names = json.load(f)
+
     recommanded_queries = []
     for profile_name in profile_names:
+        if profile_name not in all_profile_names:
+            logger.info(f"{profile_name} Invalid")
+            continue
         profile_path = Path(
             f"./profiles/knowledgebase/profiles_{index}/profile_{profile_name}.json"
         )

veadk/tools/builtin_tools/load_knowledgebase.py

@@ -27,6 +27,7 @@
 from veadk.knowledgebase.entry import KnowledgebaseEntry
 from veadk.tools.builtin_tools.load_kb_queries import load_profile
 from veadk.utils.logger import get_logger
+from veadk import Agent
 
 logger = get_logger(__name__)
 
@@ -47,6 +48,46 @@ def __init__(self, knowledgebase: KnowledgeBase):
             self.custom_metadata = {}
         self.custom_metadata["backend"] = knowledgebase.backend
 
+    #     if self.knowledgebase.enable_profile:
+    #         from pathlib import Path
+    #
+    #         profile_names = []
+    #         profile_descriptions = []
+    #
+    #         with open(
+    #                 f"./profiles/knowledgebase/profiles_{self.knowledgebase.index}/all_profile_list.json",
+    #                 "r",
+    #         ) as f:
+    #             profile_names = json.load(f)
+    #
+    #         for profile_name in profile_names:
+    #             profile_descriptions.append(
+    #
+    #                 load_profile(
+    #                     Path(
+    #                         f"./profiles/knowledgebase/profiles_{self.knowledgebase.index}/profile_{profile_name}.json"
+    #                     ),
+    #                 )["description"]
+    #             )
+    #
+    #         profiles_text = "\n".join(
+    #             f"- profile_name: **{name}**\n  profile_description: {profile_descriptions[idx]}"
+    #             for idx, name in enumerate(profile_names)
+    #         )
+    #
+    #         self.select_agent = Agent(
+    #             name="DocRetrieveAgent",
+    #             description="A helpful assistant for retrieving documents relevant to user query.",
+    #             instruction=f"""
+    # The knowledgebase is divided into the following profiles:
+    #
+    # {profiles_text}
+    #
+    # You should choose some profile_name which are most relevant to the user question.
+    # """,
+    #             model_name="doubao-seed-1-6-251015",
+    #         )
+
     @override
     def _get_declaration(self) -> types.FunctionDeclaration | None:
         return types.FunctionDeclaration(
@@ -65,10 +106,10 @@ def _get_declaration(self) -> types.FunctionDeclaration | None:
 
     @override
     async def process_llm_request(
-        self,
-        *,
-        tool_context: ToolContext,
-        llm_request: LlmRequest,
+            self,
+            *,
+            tool_context: ToolContext,
+            llm_request: LlmRequest,
     ) -> None:
         await super().process_llm_request(
             tool_context=tool_context, llm_request=llm_request
@@ -82,13 +123,14 @@ async def process_llm_request(
             profile_descriptions = []
 
             with open(
-                f"./profiles/knowledgebase/profiles_{index}/profile_list.json",
-                "r",
+                    f"./profiles/knowledgebase/profiles_{index}/all_profile_list.json",
+                    "r",
             ) as f:
                 profile_names = json.load(f)
 
             for profile_name in profile_names:
                 profile_descriptions.append(
+
                     load_profile(
                         Path(
                             f"./profiles/knowledgebase/profiles_{index}/profile_{profile_name}.json"
@@ -97,30 +139,30 @@ async def process_llm_request(
                 )
 
             profiles_text = "\n".join(
-                f"- profile_name: {name}\n  profile_description: {profile_descriptions[idx]}"
+                f"- profile_name: **{name}**\n  profile_description: {profile_descriptions[idx]}"
                 for idx, name in enumerate(profile_names)
             )
 
         # Tell the model about the knowledgebase.
         llm_request.append_instructions(
             [
                 f"""
-You have a knowledgebase (knowledegebase name is `{self.knowledgebase.name}`, knowledgebase description is `{self.knowledgebase.description}`). You can use it to answer questions. If any questions need
-you to look up the knowledgebase, you should call load_knowledgebase function with a query.
-"""
+    You have a knowledgebase (knowledegebase name is `{self.knowledgebase.name}`, knowledgebase description is `{self.knowledgebase.description}`). You can use it to answer questions. If any questions need
+    you to look up the knowledgebase, you should call load_knowledgebase function with a query.
+    """
             ],
         )
 
         if self.knowledgebase.enable_profile:
             llm_request.append_instructions(
                 [
                     f"""
-The knowledgebase is divided into the following profiles: 
-
-{profiles_text}
-
-You should choose some profiles which are relevant to the user question. Before load the knowledgebase, you must call `load_kb_queries` to load the recommanded queries of the knowledgebase profiles. You should generate final knowledgebase queries based on the user question and recommanded queries.
-"""
+    The knowledgebase is divided into the following profiles: 
+    
+    {profiles_text}
+    
+    You should choose some profile_name which are most relevant to the user question. Before load the knowledgebase, you must call `load_kb_queries` to load the recommanded queries of the knowledgebase profiles. You should generate final knowledgebase queries based on the user question and recommanded queries.
+    """
                 ]
             )
 
@@ -140,17 +182,17 @@ async def process_llm_request(
                 llm_request.append_instructions(
                     [
                         f"""
-Please generate the knowledgebase queries based on the user profile (description) at the same time. For example, for a query `quick sort algorithm`, you should generate `quick sort algorithm for python` if the user is a python developer, or `quick sort algorithm friendly introduction` if the user is a beginner.
-
-The user profile is : 
-
-{user_profile}
-"""
+    Please generate the knowledgebase queries based on the user profile (description) at the same time. For example, for a query `quick sort algorithm`, you should generate `quick sort algorithm for python` if the user is a python developer, or `quick sort algorithm friendly introduction` if the user is a beginner.
+    
+    The user profile is : 
+    
+    {user_profile}
+    """
                     ]
                 )
 
     async def load_knowledgebase(
-        self, query: str, tool_context: ToolContext
+            self, query: str, tool_context: ToolContext
     ) -> LoadKnowledgebaseResponse:
         """Loads the knowledgebase for the user.