agno 0.1.2__py3-none-any.whl → 2.3.13__py3-none-any.whl

agno/utils/functions.py

@@ -1,8 +1,10 @@
 import json
-from typing import Any, Dict, Optional
+from typing import Any, Callable, Dict, Optional, TypeVar
 
 from agno.tools.function import Function, FunctionCall
-from agno.utils.log import logger
+from agno.utils.log import log_debug, log_error
+
+T = TypeVar("T")
 
 
 def get_function_call(
@@ -11,7 +13,6 @@ def get_function_call(
     call_id: Optional[str] = None,
     functions: Optional[Dict[str, Function]] = None,
 ) -> Optional[FunctionCall]:
-    logger.debug(f"Getting function {name}")
     if functions is None:
         return None
 
@@ -19,7 +20,7 @@ def get_function_call(
     if name in functions:
         function_to_call = functions[name]
     if function_to_call is None:
-        logger.error(f"Function {name} not found")
+        log_error(f"Function {name} not found")
         return None
 
     function_call = FunctionCall(function=function_to_call)
@@ -27,16 +28,14 @@ def get_function_call(
         function_call.call_id = call_id
     if arguments is not None and arguments != "":
         try:
-            if function_to_call.sanitize_arguments:
-                if "None" in arguments:
-                    arguments = arguments.replace("None", "null")
-                if "True" in arguments:
-                    arguments = arguments.replace("True", "true")
-                if "False" in arguments:
-                    arguments = arguments.replace("False", "false")
-            _arguments = json.loads(arguments)
+            try:
+                _arguments = json.loads(arguments)
+            except Exception:
+                import ast
+
+                _arguments = ast.literal_eval(arguments)
         except Exception as e:
-            logger.error(f"Unable to decode function arguments:\n{arguments}\nError: {e}")
+            log_error(f"Unable to decode function arguments:\n{arguments}\nError: {e}")
             function_call.error = (
                 f"Error while decoding function arguments: {e}\n\n"
                 f"Please make sure we can json.loads() the arguments and retry."
@@ -44,7 +43,7 @@ def get_function_call(
             return function_call
 
         if not isinstance(_arguments, dict):
-            logger.error(f"Function arguments are not a valid JSON object: {arguments}")
+            log_error(f"Function arguments are not a valid JSON object: {arguments}")
             function_call.error = "Function arguments are not a valid JSON object.\n\n Please fix and retry."
             return function_call
 
@@ -66,31 +65,102 @@ def get_function_call(
 
             function_call.arguments = clean_arguments
         except Exception as e:
-            logger.error(f"Unable to parsing function arguments:\n{arguments}\nError: {e}")
+            log_error(f"Unable to parsing function arguments:\n{arguments}\nError: {e}")
             function_call.error = f"Error while parsing function arguments: {e}\n\n Please fix and retry."
             return function_call
     return function_call
 
 
-# def run_function(func, *args, **kwargs):
-#     if asyncio.iscoroutinefunction(func):
-#         logger.debug("Running asynchronous function")
-#         try:
-#             loop = asyncio.get_running_loop()
-#         except RuntimeError as e:  # No running event loop
-#             logger.debug(f"Could not get running event loop: {e}")
-#             logger.debug("Running with a new event loop")
-#             loop = asyncio.new_event_loop()
-#             asyncio.set_event_loop(loop)
-#             result = loop.run_until_complete(func(*args, **kwargs))
-#             loop.close()
-#             logger.debug("Done running with a new event loop")
-#             return result
-#         else:  # There is a running event loop
-#             logger.debug("Running in existing event loop")
-#             result = loop.run_until_complete(func(*args, **kwargs))
-#             logger.debug("Done running in existing event loop")
-#             return result
-#     else:  # The function is a synchronous function
-#         logger.debug("Running synchronous function")
-#         return func(*args, **kwargs)
+def cache_result(enable_cache: bool = True, cache_dir: Optional[str] = None, cache_ttl: int = 3600):
+    """
+    Decorator factory that creates a file-based caching decorator for function results.
+
+    Args:
+        enable_cache (bool): Enable caching of function results.
+        cache_dir (Optional[str]): Directory to store cache files. Defaults to system temp dir.
+        cache_ttl (int): Time-to-live for cached results in seconds.
+
+    Returns:
+        A decorator function that caches results on the filesystem.
+    """
+    import functools
+    import hashlib
+    import json
+    import os
+    import tempfile
+    import time
+
+    def decorator(func: Callable[..., T]) -> Callable[..., T]:
+        @functools.wraps(func)
+        def wrapper(*args, **kwargs):
+            # First argument might be 'self' but we don't need to handle it specially
+            instance = args[0] if args else None
+
+            # Skip caching if cache_results is False (only for class methods)
+            if instance and hasattr(instance, "cache_results") and not instance.cache_results:
+                return func(*args, **kwargs)
+
+            if not enable_cache:
+                return func(*args, **kwargs)
+
+            # Get cache directory
+            instance_cache_dir = (
+                getattr(instance, "cache_dir", cache_dir) if hasattr(instance, "cache_dir") else cache_dir
+            )
+            base_cache_dir = instance_cache_dir or os.path.join(tempfile.gettempdir(), "agno_cache")
+
+            # Create cache directory if it doesn't exist
+            func_cache_dir = os.path.join(base_cache_dir, func.__module__, func.__qualname__)
+            os.makedirs(func_cache_dir, exist_ok=True)
+
+            # Create a cache key using all arguments
+            # Convert args and kwargs to strings and join them
+            args_str = str(args)
+            kwargs_str = str(sorted(kwargs.items()))
+
+            # Create a hash for potentially large input
+            key_str = f"{func.__module__}.{func.__qualname__}:{args_str}:{kwargs_str}"
+            cache_key = hashlib.md5(key_str.encode()).hexdigest()
+
+            # Define cache file path
+            cache_file = os.path.join(func_cache_dir, f"{cache_key}.json")
+
+            # Check for cached result
+            if os.path.exists(cache_file):
+                try:
+                    with open(cache_file, "r") as f:
+                        cache_data = json.load(f)
+
+                    timestamp = cache_data.get("timestamp", 0)
+                    result = cache_data.get("result")
+
+                    # Use instance ttl if available, otherwise use decorator ttl
+                    effective_ttl = (
+                        getattr(instance, "cache_ttl", cache_ttl) if hasattr(instance, "cache_ttl") else cache_ttl
+                    )
+
+                    if time.time() - timestamp <= effective_ttl:
+                        log_debug(f"Cache hit for: {func.__name__}")
+                        return result
+
+                    # Remove expired entry
+                    os.remove(cache_file)
+                except Exception as e:
+                    log_error(f"Error reading cache: {e}")
+                    # Continue with function execution if cache read fails
+
+            # Execute the function and cache the result
+            result = func(*args, **kwargs)
+
+            try:
+                with open(cache_file, "w") as f:
+                    json.dump({"timestamp": time.time(), "result": result}, f)
+            except Exception as e:
+                log_error(f"Error writing cache: {e}")
+                # Continue even if cache write fails
+
+            return result
+
+        return wrapper
+
+    return decorator

agno/utils/gemini.py

@@ -0,0 +1,426 @@
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Type, Union
+
+from pydantic import BaseModel
+
+from agno.media import Image
+from agno.utils.log import log_error, log_warning
+
+try:
+    from google.genai.types import (
+        FunctionDeclaration,
+        Schema,
+        Tool,
+    )
+    from google.genai.types import (
+        Type as GeminiType,
+    )
+except ImportError:
+    raise ImportError("`google-genai` not installed. Please install it using `pip install google-genai`")
+
+
+def prepare_response_schema(pydantic_model: Type[BaseModel]) -> Union[Type[BaseModel], Schema]:
+    """
+    Prepare a Pydantic model for use as Gemini response schema.
+
+    Returns the model directly if Gemini can handle it natively,
+    otherwise converts to Gemini's Schema format.
+
+    Args:
+        pydantic_model: A Pydantic model class
+
+    Returns:
+        Either the original Pydantic model or a converted Schema object
+    """
+    schema_dict = pydantic_model.model_json_schema()
+
+    # Convert to Gemini Schema if the model has problematic patterns
+    if needs_conversion(schema_dict):
+        try:
+            converted = convert_schema(schema_dict)
+        except Exception as e:
+            log_warning(f"Failed to convert schema for {pydantic_model}: {e}")
+            converted = None
+
+        if converted is None:
+            # If conversion fails, let Gemini handle it directly
+            return pydantic_model
+        return converted
+
+    # Gemini can handle this model directly
+    return pydantic_model
+
+
+def needs_conversion(schema_dict: Dict[str, Any]) -> bool:
+    """
+    Check if a schema needs conversion for Gemini.
+
+    Returns True if the schema has:
+    - Self-references or circular references
+    - Dict fields (additionalProperties) that Gemini doesn't handle well
+    - Empty object definitions that Gemini rejects
+    """
+    # Check for dict fields (additionalProperties) anywhere in the schema
+    if has_additional_properties(schema_dict):
+        return True
+
+    # Check if schema has $defs with circular references
+    if "$defs" in schema_dict:
+        defs = schema_dict["$defs"]
+        for def_name, def_schema in defs.items():
+            ref_path = f"#/$defs/{def_name}"
+            if has_self_reference(def_schema, ref_path):
+                return True
+
+    return False
+
+
+def has_additional_properties(schema: Any) -> bool:
+    """Check if schema has additionalProperties (Dict fields)"""
+    if isinstance(schema, dict):
+        # Direct check
+        if "additionalProperties" in schema:
+            return True
+
+        # Check properties recursively
+        if "properties" in schema:
+            for prop_schema in schema["properties"].values():
+                if has_additional_properties(prop_schema):
+                    return True
+
+        # Check array items
+        if "items" in schema:
+            if has_additional_properties(schema["items"]):
+                return True
+
+    return False
+
+
+def has_self_reference(schema: Dict, target_ref: str) -> bool:
+    """Check if a schema references itself (directly or indirectly)"""
+    if isinstance(schema, dict):
+        # Direct self-reference
+        if schema.get("$ref") == target_ref:
+            return True
+
+        # Check properties
+        if "properties" in schema:
+            for prop_schema in schema["properties"].values():
+                if has_self_reference(prop_schema, target_ref):
+                    return True
+
+        # Check array items
+        if "items" in schema:
+            if has_self_reference(schema["items"], target_ref):
+                return True
+
+        # Check anyOf/oneOf/allOf
+        for key in ["anyOf", "oneOf", "allOf"]:
+            if key in schema:
+                for sub_schema in schema[key]:
+                    if has_self_reference(sub_schema, target_ref):
+                        return True
+
+    return False
+
+
+def format_image_for_message(image: Image) -> Optional[Dict[str, Any]]:
+    # Case 1: Image is a URL
+    # Download the image from the URL and add it as base64 encoded data
+    if image.url is not None:
+        content_bytes = image.get_content_bytes()  # type: ignore
+        if content_bytes is not None:
+            try:
+                import base64
+
+                image_data = {
+                    "mime_type": "image/jpeg",
+                    "data": base64.b64encode(content_bytes).decode("utf-8"),
+                }
+                return image_data
+            except Exception as e:
+                log_warning(f"Failed to download image from {image}: {e}")
+                return None
+        else:
+            log_warning(f"Unsupported image format: {image}")
+            return None
+
+    # Case 2: Image is a local path
+    elif image.filepath is not None:
+        try:
+            image_path = Path(image.filepath)
+            if image_path.exists() and image_path.is_file():
+                with open(image_path, "rb") as f:
+                    content_bytes = f.read()
+            else:
+                log_error(f"Image file {image_path} does not exist.")
+                raise
+            return {
+                "mime_type": "image/jpeg",
+                "data": content_bytes,
+            }
+        except Exception as e:
+            log_warning(f"Failed to load image from {image.filepath}: {e}")
+            return None
+
+    # Case 3: Image is a bytes object
+    # Add it as base64 encoded data
+    elif image.content is not None and isinstance(image.content, bytes):
+        import base64
+
+        image_data = {"mime_type": "image/jpeg", "data": base64.b64encode(image.content).decode("utf-8")}
+        return image_data
+    else:
+        log_warning(f"Unknown image type: {type(image)}")
+        return None
+
+
+def convert_schema(
+    schema_dict: Dict[str, Any], root_schema: Optional[Dict[str, Any]] = None, visited_refs: Optional[set] = None
+) -> Optional[Schema]:
+    """
+    Recursively convert a JSON-like schema dictionary to a types.Schema object.
+
+    Parameters:
+        schema_dict (dict): The JSON schema dictionary with keys like "type", "description",
+                            "properties", and "required".
+        root_schema (dict, optional): The root schema containing $defs for resolving $ref
+        visited_refs (set, optional): Set of visited $ref paths to detect circular references
+
+    Returns:
+        types.Schema: The converted schema.
+    """
+
+    # If this is the initial call, set root_schema to self and initialize visited_refs
+    if root_schema is None:
+        root_schema = schema_dict
+    if visited_refs is None:
+        visited_refs = set()
+
+    # Handle $ref references with cycle detection
+    if "$ref" in schema_dict:
+        ref_path = schema_dict["$ref"]
+
+        # Check for circular reference
+        if ref_path in visited_refs:
+            # Return a basic object schema to break the cycle
+            return Schema(
+                type=GeminiType.OBJECT,
+                description=f"Circular reference to {ref_path}",
+            )
+
+        if ref_path.startswith("#/$defs/"):
+            def_name = ref_path.split("/")[-1]
+            if "$defs" in root_schema and def_name in root_schema["$defs"]:
+                # Add to visited set before recursing
+                new_visited = visited_refs.copy()
+                new_visited.add(ref_path)
+
+                referenced_schema = root_schema["$defs"][def_name]
+                return convert_schema(referenced_schema, root_schema, new_visited)
+        # If we can't resolve the reference, return None
+        return None
+
+    schema_type = schema_dict.get("type", "")
+    if schema_type is None or schema_type == "null":
+        return None
+    description = schema_dict.get("description", None)
+    title = schema_dict.get("title", None)
+    default = schema_dict.get("default", None)
+
+    # Handle enum types
+    if "enum" in schema_dict:
+        enum_values = schema_dict["enum"]
+        return Schema(type=GeminiType.STRING, enum=enum_values, description=description, default=default, title=title)
+
+    if schema_type == "object":
+        # Handle regular objects with properties
+        if "properties" in schema_dict:
+            properties = {}
+            for key, prop_def in schema_dict["properties"].items():
+                # Process nullable types
+                prop_type = prop_def.get("type", "")
+                is_nullable = False
+                if isinstance(prop_type, list) and "null" in prop_type:
+                    prop_def["type"] = prop_type[0]
+                    is_nullable = True
+
+                # Process property schema (pass root_schema and visited_refs for $ref resolution)
+                converted_schema = convert_schema(prop_def, root_schema, visited_refs)
+                if converted_schema is not None:
+                    if is_nullable:
+                        converted_schema.nullable = True
+                    properties[key] = converted_schema
+                else:
+                    properties[key] = Schema(
+                        title=prop_def.get("title", None), description=prop_def.get("description", None)
+                    )
+
+            required = schema_dict.get("required", [])
+
+            if properties:
+                return Schema(
+                    type=GeminiType.OBJECT,
+                    properties=properties,
+                    required=required,
+                    description=description,
+                    default=default,
+                    title=title,
+                )
+            else:
+                return Schema(type=GeminiType.OBJECT, description=description, default=default, title=title)
+
+        # Handle Dict types (objects with additionalProperties but no properties)
+        elif "additionalProperties" in schema_dict:
+            additional_props = schema_dict["additionalProperties"]
+
+            # If additionalProperties is a schema object (Dict[str, T] case)
+            if isinstance(additional_props, dict) and "type" in additional_props:
+                # For Gemini, we need to represent Dict[str, T] as an object with at least one property
+                # to avoid the "properties should be non-empty" error.
+                # We'll create a generic property that represents the dictionary structure
+
+                # Handle both single types and union types (arrays) from Zod schemas
+                type_value = additional_props.get("type", "string")
+                if isinstance(type_value, list):
+                    value_type = type_value[0].upper() if type_value else "STRING"
+                    union_types = ", ".join(type_value)
+                    type_description_suffix = f" (supports union types: {union_types})"
+                else:
+                    # Single type
+                    value_type = type_value.upper()
+                    type_description_suffix = ""
+
+                # Create a placeholder property to satisfy Gemini's requirements
+                # This is a workaround since Gemini doesn't support additionalProperties directly
+                placeholder_properties = {
+                    "example_key": Schema(
+                        type=value_type,
+                        description=f"Example key-value pair. This object can contain any number of keys with {value_type.lower()} values{type_description_suffix}.",
+                    )
+                }
+                if value_type == "ARRAY":
+                    placeholder_properties["example_key"].items = {}  # type: ignore
+
+                return Schema(
+                    type=GeminiType.OBJECT,
+                    properties=placeholder_properties,
+                    description=description
+                    or f"Dictionary with {value_type.lower()} values{type_description_suffix}. Can contain any number of key-value pairs.",
+                    default=default,
+                )
+            else:
+                # additionalProperties is false or true
+                return Schema(type=GeminiType.OBJECT, description=description, default=default, title=title)
+
+        # Handle empty objects
+        else:
+            return Schema(type=GeminiType.OBJECT, description=description, default=default, title=title)
+
+    elif schema_type == "array" and "items" in schema_dict:
+        if not schema_dict["items"]:  # Handle empty {}
+            items = Schema(type=GeminiType.STRING)
+        else:
+            converted_items = convert_schema(schema_dict["items"], root_schema, visited_refs)
+            items = converted_items if converted_items is not None else Schema(type=GeminiType.STRING)
+        min_items = schema_dict.get("minItems")
+        max_items = schema_dict.get("maxItems")
+        return Schema(
+            type=GeminiType.ARRAY,
+            description=description,
+            items=items,
+            min_items=min_items,
+            max_items=max_items,
+            title=title,
+        )
+
+    elif schema_type == "string":
+        schema_kwargs = {
+            "type": GeminiType.STRING,
+            "description": description,
+            "default": default,
+            "title": title,
+        }
+        if "format" in schema_dict:
+            schema_kwargs["format"] = schema_dict["format"]
+        return Schema(**schema_kwargs)
+
+    elif schema_type in ("integer", "number"):
+        schema_kwargs = {
+            "type": schema_type.upper(),
+            "description": description,
+            "default": default,
+            "title": title,
+        }
+        if "maximum" in schema_dict:
+            schema_kwargs["maximum"] = schema_dict["maximum"]
+        if "minimum" in schema_dict:
+            schema_kwargs["minimum"] = schema_dict["minimum"]
+        return Schema(**schema_kwargs)
+
+    elif schema_type == "" and "anyOf" in schema_dict:
+        any_of = []
+        for sub_schema in schema_dict["anyOf"]:
+            sub_schema_converted = convert_schema(sub_schema, root_schema, visited_refs)
+            any_of.append(sub_schema_converted)
+
+        is_nullable = False
+        filtered_any_of = []
+
+        for schema in any_of:
+            if schema is None:
+                is_nullable = True
+            else:
+                filtered_any_of.append(schema)
+
+        any_of = filtered_any_of  # type: ignore
+        if len(any_of) == 1 and any_of[0] is not None:
+            any_of[0].nullable = is_nullable
+            return any_of[0]
+        else:
+            return Schema(
+                any_of=any_of,
+                description=description,
+                default=default,
+                title=title,
+            )
+    else:
+        if isinstance(schema_type, list):
+            non_null_types = [t for t in schema_type if t != "null"]
+            if non_null_types:
+                schema_type = non_null_types[0]
+            else:
+                schema_type = ""
+        # Only convert to uppercase if schema_type is not empty
+        if schema_type:
+            schema_type = schema_type.upper()
+            return Schema(type=schema_type, description=description, default=default, title=title)
+        else:
+            # If we get here with an empty type and no other handlers matched,
+            # something is wrong with the schema
+            return None
+
+
+def format_function_definitions(tools_list: List[Dict[str, Any]]) -> Optional[Tool]:
+    function_declarations = []
+
+    for tool in tools_list:
+        if tool.get("type") == "function":
+            func_info = tool.get("function", {})
+            name = func_info.get("name")
+            description = func_info.get("description", "")
+            parameters_dict = func_info.get("parameters", {})
+
+            parameters_schema = convert_schema(parameters_dict)
+            # Create a FunctionDeclaration instance
+            function_decl = FunctionDeclaration(
+                name=name,
+                description=description,
+                parameters=parameters_schema,
+            )
+
+            function_declarations.append(function_decl)
+    if function_declarations:
+        return Tool(function_declarations=function_declarations)
+    else:
+        return None