Method __init__

src/hyperagent/tools/nav_tools.py:471–514

Semantic code search tool allows you to search for code using natural language. It's useful when the query is a sentance, semantic and vague. If exact search such as code search failed after multiple tries, try this Args: path (_type_): relative path to the repo lang

(self, path, language: str="python", db_path: Optional[str] = None, build: bool = False)

class SemanticCodeSearchTool(Tool):
    def __init__(self, path, language: str="python", db_path: Optional[str] = None, build: bool = False):
        """Semantic code search tool allows you to search for code using natural language. It's useful when the query is a sentance, semantic and vague. If exact search such as code search failed after multiple tries, try this

        Args:
            path (_type_): relative path to the repo
            language (str, optional): we have 4 options: python, rust, csharp, java. Defaults to "python".
        """
        if db_path == None:
            # randomize db_path
            db_path = os.path.join(SEMANTIC_CODE_SEARCH_DB_PATH, str(uuid.uuid4()))
        if not os.path.exists(db_path):
            os.makedirs(db_path)
        if build:
            extension = identify_extension(language)
            splitter = RecursiveCharacterTextSplitter.from_language(
                language=language, chunk_size=400, chunk_overlap=100
            )

            non_utf8_files = find_non_utf8_files(path)

            loader = GenericLoader.from_filesystem(
                path,
                suffixes=[extension],
                exclude=non_utf8_files,
                parser=LanguageParser(language=language, parser_threshold=500),
                show_progress=True,
            )
            
            documents = loader.load()
            texts = splitter.split_documents(documents)
            db = Chroma.from_documents(texts, CohereEmbeddings(model="embed-english-light-v3.0", cohere_api_key=os.getenv("COHERE_API_KEY")), persist_directory=db_path)
            db.persist()
        else:
            db = Chroma(persist_directory=db_path, embedding_function=CohereEmbeddings(model="embed-english-light-v3.0", cohere_api_key=os.getenv("COHERE_API_KEY")))
        
        def semantic_code_search(query):
            retrieved_docs = db.similarity_search(query, k=3)
            return [doc.page_content for doc in retrieved_docs]
        
        super().__init__(
            name="Semantic Code Search",
            func=semantic_code_search,
            description="useful for when the query is a sentance, semantic and vague. If exact search such as code search failed after multiple tries, try this",
        )

class FindFileArgs(BaseModel):
    file_name: str = Field(..., description="The name of the file you want to find")