veadk-python 0.2.4__py3-none-any.whl → 0.2.5__py3-none-any.whl

veadk/agent.py

@@ -28,9 +28,10 @@ from typing_extensions import Any
 
 from veadk.config import getenv
 from veadk.consts import (
-    DEFALUT_MODEL_AGENT_PROVIDER,
+    DEFAULT_MODEL_AGENT_PROVIDER,
     DEFAULT_MODEL_AGENT_API_BASE,
     DEFAULT_MODEL_AGENT_NAME,
+    DEFAULT_MODEL_EXTRA_HEADERS,
 )
 from veadk.evaluation import EvalSetRecorder
 from veadk.knowledgebase import KnowledgeBase
@@ -64,7 +65,7 @@ class Agent(LlmAgent):
     model_name: str = getenv("MODEL_AGENT_NAME", DEFAULT_MODEL_AGENT_NAME)
     """The name of the model for agent running."""
 
-    model_provider: str = getenv("MODEL_AGENT_PROVIDER", DEFALUT_MODEL_AGENT_PROVIDER)
+    model_provider: str = getenv("MODEL_AGENT_PROVIDER", DEFAULT_MODEL_AGENT_PROVIDER)
     """The provider of the model for agent running."""
 
     model_api_base: str = getenv("MODEL_AGENT_API_BASE", DEFAULT_MODEL_AGENT_API_BASE)
@@ -73,6 +74,9 @@ class Agent(LlmAgent):
     model_api_key: str = Field(default_factory=lambda: getenv("MODEL_AGENT_API_KEY"))
     """The api key of the model for agent running."""
 
+    model_extra_config: dict = Field(default_factory=dict)
+    """The extra config to include in the model requests."""
+
     tools: list[ToolUnion] = []
     """The tools provided to agent."""
 
@@ -96,11 +100,27 @@ class Agent(LlmAgent):
 
     def model_post_init(self, __context: Any) -> None:
         super().model_post_init(None)  # for sub_agents init
-        self.model = LiteLlm(
-            model=f"{self.model_provider}/{self.model_name}",
-            api_key=self.model_api_key,
-            api_base=self.model_api_base,
-        )
+
+        # add model request source (veadk) in extra headers
+        if self.model_extra_config and "extra_headers" in self.model_extra_config:
+            self.model_extra_config["extra_headers"] |= DEFAULT_MODEL_EXTRA_HEADERS
+        else:
+            self.model_extra_config["extra_headers"] = DEFAULT_MODEL_EXTRA_HEADERS
+
+        if not self.model:
+            self.model = LiteLlm(
+                model=f"{self.model_provider}/{self.model_name}",
+                api_key=self.model_api_key,
+                api_base=self.model_api_base,
+                **self.model_extra_config,
+            )
+            logger.debug(
+                f"LiteLLM client created with config: {self.model_extra_config}"
+            )
+        else:
+            logger.warning(
+                "You are trying to use your own LiteLLM client, some default request headers may be missing."
+            )
 
         if self.knowledgebase:
             from veadk.tools import load_knowledgebase_tool
@@ -117,7 +137,7 @@ class Agent(LlmAgent):
 
         logger.info(f"{self.__class__.__name__} `{self.name}` init done.")
         logger.debug(
-            f"Agent: {self.model_dump(include={'name', 'model_name', 'model_api_base', 'tools', 'serve_url'})}"
+            f"Agent: {self.model_dump(include={'name', 'model_name', 'model_api_base', 'tools'})}"
         )
 
     async def _run(

veadk/cli/cli_deploy.py

@@ -29,7 +29,9 @@ TEMP_PATH = "/tmp"
     default=None,
     help="Volcengine secret key",
 )
-@click.option("--vefaas-app-name", help="Expected Volcengine FaaS application name")
+@click.option(
+    "--vefaas-app-name", required=True, help="Expected Volcengine FaaS application name"
+)
 @click.option(
     "--veapig-instance-name", default="", help="Expected Volcengine APIG instance name"
 )

veadk/cloud/cloud_app.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 
 import json
+import time
 from typing import Any
 from uuid import uuid4
 
@@ -60,9 +61,11 @@ class CloudApp:
         if not vefaas_endpoint:
             self.vefaas_endpoint = self._get_vefaas_endpoint()
 
-        if not self.vefaas_endpoint.startswith(
-            "http"
-        ) and not self.vefaas_endpoint.startswith("https"):
+        if (
+            self.vefaas_endpoint
+            and not self.vefaas_endpoint.startswith("http")
+            and not self.vefaas_endpoint.startswith("https")
+        ):
             raise ValueError(
                 f"Invalid endpoint: {vefaas_endpoint}. The endpoint must start with `http` or `https`."
             )
@@ -92,12 +95,13 @@ class CloudApp:
             raise ValueError(
                 f"VeFaaS CloudAPP with application_id `{self.vefaas_application_id}` or application_name `{self.vefaas_application_name}` not found."
             )
-        cloud_resource = json.loads(app["CloudResource"])
 
         try:
+            cloud_resource = json.loads(app["CloudResource"])
             vefaas_endpoint = cloud_resource["framework"]["url"]["system_url"]
         except Exception as e:
-            raise ValueError(f"VeFaaS cloudAPP could not get endpoint. Error: {e}")
+            logger.warning(f"VeFaaS cloudAPP could not get endpoint. Error: {e}")
+            vefaas_endpoint = ""
         return vefaas_endpoint
 
     def _get_vefaas_application_id_by_name(self) -> str:
@@ -167,7 +171,18 @@ class CloudApp:
 
             vefaas_client = VeFaaS(access_key=volcengine_ak, secret_key=volcengine_sk)
             vefaas_client.delete(self.vefaas_application_id)
-            print(f"Cloud app {self.vefaas_application_id} is deleting...")
+            print(
+                f"Cloud app {self.vefaas_application_id} delete request has been sent to VeFaaS"
+            )
+            while True:
+                try:
+                    id = self._get_vefaas_application_id_by_name()
+                    if not id:
+                        break
+                    time.sleep(3)
+                except Exception as _:
+                    break
+            print("Delete application done.")
 
     async def message_send(
         self, message: str, session_id: str, user_id: str, timeout: float = 600.0

veadk/consts.py

@@ -12,6 +12,19 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from veadk.version import VERSION
+
 DEFAULT_MODEL_AGENT_NAME = "doubao-seed-1-6-250615"
-DEFALUT_MODEL_AGENT_PROVIDER = "openai"
+DEFAULT_MODEL_AGENT_PROVIDER = "openai"
 DEFAULT_MODEL_AGENT_API_BASE = "https://ark.cn-beijing.volces.com/api/v3/"
+DEFAULT_MODEL_EXTRA_HEADERS = {"veadk-source": "veadk", "veadk-version": VERSION}
+
+DEFAULT_APMPLUS_OTEL_EXPORTER_ENDPOINT = "http://apmplus-cn-beijing.volces.com:4317"
+DEFAULT_APMPLUS_OTEL_EXPORTER_SERVICE_NAME = "veadk_tracing"
+
+DEFAULT_COZELOOP_OTEL_EXPORTER_ENDPOINT = (
+    "https://api.coze.cn/v1/loop/opentelemetry/v1/traces"
+)
+
+DEFAULT_TLS_OTEL_EXPORTER_ENDPOINT = "https://tls-cn-beijing.volces.com:4318/v1/traces"
+DEFAULT_TLS_OTEL_EXPORTER_REGION = "cn-beijing"

veadk/database/viking/viking_database.py

@@ -387,9 +387,9 @@ class VikingDatabase(BaseModel, BaseDatabase):
             logger.error(f"Error in list_collections: {result['message']}")
             raise ValueError(f"Error in list_collections: {result['message']}")
 
-        collections = result["data"]["collection_list"]
-        if not collections:
-            raise ValueError(f"No collections found in project {self.config.project}.")
+        collections = result["data"].get("collection_list", [])
+        if len(collections) == 0:
+            return False
 
         collection_list = set()
 

veadk/integrations/ve_faas/template/{{cookiecutter.local_dir_name}}/clean.py

@@ -0,0 +1,23 @@
+# Copyright (c) 2025 Beijing Volcano Engine Technology Co., Ltd. and/or its affiliates.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from veadk.cloud.cloud_app import CloudApp
+
+def main() -> None:
+    cloud_app = CloudApp(vefaas_application_name="{{cookiecutter.vefaas_application_name}}")
+    cloud_app.delete_self()
+
+
+if __name__ == "__main__":
+    main()

veadk/integrations/ve_faas/template/{{cookiecutter.local_dir_name}}/src/app.py

@@ -26,6 +26,7 @@ from fastmcp import FastMCP
 from starlette.routing import Route
 
 from google.adk.a2a.utils.agent_card_builder import AgentCardBuilder
+from a2a.types import AgentProvider
 
 from veadk.a2a.ve_a2a_server import init_app
 from veadk.runner import Runner
@@ -46,7 +47,9 @@ app_name = agent_run_config.app_name
 agent = agent_run_config.agent
 short_term_memory = agent_run_config.short_term_memory
 
-agent_card_builder = AgentCardBuilder(agent=agent)
+VEFAAS_REGION = os.getenv("APP_REGION", "cn-beijing")
+VEFAAS_FUNC_ID = os.getenv("_FAAS_FUNC_ID", "")
+agent_card_builder = AgentCardBuilder(agent=agent, provider=AgentProvider(organization="Volcengine Agent Development Kit (VeADK)", url=f"https://console.volcengine.com/vefaas/region:vefaas+{VEFAAS_REGION}/function/detail/{VEFAAS_FUNC_ID}"))
 
 
 def load_tracer() -> None:

veadk/integrations/ve_faas/template/{{cookiecutter.local_dir_name}}/src/run.sh

@@ -34,7 +34,17 @@ while [[ $# -gt 0 ]]; do
 done
 
 # in case of deployment deps not installed in user's requirements.txt
-python3 -m pip install uvicorn[standard] fastapi
+if pip list | grep -q "^fastapi \|^uvicorn "; then
+    echo "fastapi and uvicorn already installed"
+else
+    python3 -m pip install uvicorn[standard] fastapi
+fi
+
+# Check if MODEL_AGENT_API_KEY is set
+if [ -z "$MODEL_AGENT_API_KEY" ]; then
+    echo "MODEL_AGENT_API_KEY is not set. Please set it in your environment variables."
+    exit 1
+fi
 
 USE_ADK_WEB=${USE_ADK_WEB:-False}
 

veadk/integrations/ve_tos/ve_tos.py

@@ -0,0 +1,176 @@
+# Copyright (c) 2025 Beijing Volcano Engine Technology Co., Ltd. and/or its affiliates.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+from veadk.config import getenv
+from veadk.utils.logger import get_logger
+import tos
+import asyncio
+from typing import Union
+from pydantic import BaseModel, Field
+from typing import Any
+from urllib.parse import urlparse
+from datetime import datetime
+
+logger = get_logger(__name__)
+
+
+class TOSConfig(BaseModel):
+    region: str = Field(
+        default_factory=lambda: getenv("DATABASE_TOS_REGION"),
+        description="TOS region",
+    )
+    ak: str = Field(
+        default_factory=lambda: getenv("VOLCENGINE_ACCESS_KEY"),
+        description="Volcengine access key",
+    )
+    sk: str = Field(
+        default_factory=lambda: getenv("VOLCENGINE_SECRET_KEY"),
+        description="Volcengine secret key",
+    )
+    bucket_name: str = Field(
+        default_factory=lambda: getenv("DATABASE_TOS_BUCKET"),
+        description="TOS bucket name",
+    )
+
+
+class VeTOS(BaseModel):
+    config: TOSConfig = Field(default_factory=TOSConfig)
+
+    def model_post_init(self, __context: Any) -> None:
+        try:
+            self._client = tos.TosClientV2(
+                self.config.ak,
+                self.config.sk,
+                endpoint=f"tos-{self.config.region}.volces.com",
+                region=self.config.region,
+            )
+            logger.info("Connected to TOS successfully.")
+        except Exception as e:
+            logger.error(f"Client initialization failed:{e}")
+            return None
+
+    def create_bucket(self) -> bool:
+        """If the bucket does not exist, create it"""
+        try:
+            self._client.head_bucket(self.config.bucket_name)
+            logger.info(f"Bucket {self.config.bucket_name} already exists")
+            return True
+        except tos.exceptions.TosServerError as e:
+            if e.status_code == 404:
+                self._client.create_bucket(
+                    bucket=self.config.bucket_name,
+                    storage_class=tos.StorageClassType.Storage_Class_Standard,
+                    acl=tos.ACLType.ACL_Private,
+                )
+                logger.info(f"Bucket {self.config.bucket_name} created successfully")
+                return True
+        except Exception as e:
+            logger.error(f"Bucket creation failed: {str(e)}")
+            return False
+
+    def build_tos_url(
+        self, user_id: str, app_name: str, session_id: str, data_path: str
+    ) -> tuple[str, str]:
+        """generate TOS object key"""
+        parsed_url = urlparse(data_path)
+
+        if parsed_url.scheme and parsed_url.scheme in ("http", "https", "ftp", "ftps"):
+            file_name = os.path.basename(parsed_url.path)
+        else:
+            file_name = os.path.basename(data_path)
+
+        timestamp: str = datetime.now().strftime("%Y%m%d%H%M%S%f")[:-3]
+        object_key: str = f"{app_name}-{user_id}-{session_id}/{timestamp}-{file_name}"
+        tos_url: str = f"https://{self.config.bucket_name}.tos-{self.config.region}.volces.com/{object_key}"
+
+        return object_key, tos_url
+
+    def upload(
+        self,
+        object_key: str,
+        data: Union[str, bytes],
+    ):
+        if isinstance(data, str):
+            data_type = "file"
+        elif isinstance(data, bytes):
+            data_type = "bytes"
+        else:
+            error_msg = f"Upload failed: data type error. Only str (file path) and bytes are supported, got {type(data)}"
+            logger.error(error_msg)
+            raise ValueError(error_msg)
+        if data_type == "file":
+            return asyncio.to_thread(self._do_upload_file, object_key, data)
+        elif data_type == "bytes":
+            return asyncio.to_thread(self._do_upload_bytes, object_key, data)
+
+    def _do_upload_bytes(self, object_key: str, bytes: bytes) -> bool:
+        try:
+            if not self._client:
+                return False
+            if not self.create_bucket():
+                return False
+            self._client.put_object(
+                bucket=self.config.bucket_name, key=object_key, content=bytes
+            )
+            logger.debug(f"Upload success, object_key: {object_key}")
+            self._close()
+            return True
+        except Exception as e:
+            logger.error(f"Upload failed: {e}")
+            self._close()
+            return False
+
+    def _do_upload_file(self, object_key: str, file_path: str) -> bool:
+        try:
+            if not self._client:
+                return False
+            if not self.create_bucket():
+                return False
+
+            self._client.put_object_from_file(
+                bucket=self.config.bucket_name, key=object_key, file_path=file_path
+            )
+            self._close()
+            logger.debug(f"Upload success, object_key: {object_key}")
+            return True
+        except Exception as e:
+            logger.error(f"Upload failed: {e}")
+            self._close()
+            return False
+
+    def download(self, object_key: str, save_path: str) -> bool:
+        """download image from TOS"""
+        try:
+            object_stream = self._client.get_object(self.config.bucket_name, object_key)
+
+            save_dir = os.path.dirname(save_path)
+            if save_dir and not os.path.exists(save_dir):
+                os.makedirs(save_dir, exist_ok=True)
+
+            with open(save_path, "wb") as f:
+                for chunk in object_stream:
+                    f.write(chunk)
+
+            logger.debug(f"Image download success, saved to: {save_path}")
+            return True
+
+        except Exception as e:
+            logger.error(f"Image download failed: {str(e)}")
+
+            return False
+
+    def _close(self):
+        if self._client:
+            self._client.close()

veadk/runner.py

@@ -11,9 +11,11 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+import asyncio
 from typing import Union
 
 from google.adk.agents import RunConfig
+from google.adk.agents.invocation_context import LlmCallsLimitExceededError
 from google.adk.agents.run_config import StreamingMode
 from google.adk.plugins.base_plugin import BasePlugin
 from google.adk.runners import Runner as ADKRunner
@@ -30,6 +32,7 @@ from veadk.memory.short_term_memory import ShortTermMemory
 from veadk.types import MediaMessage
 from veadk.utils.logger import get_logger
 from veadk.utils.misc import read_png_to_bytes
+from veadk.integrations.ve_tos.ve_tos import VeTOS
 
 logger = get_logger(__name__)
 
@@ -49,20 +52,25 @@ class Runner:
     def __init__(
         self,
         agent: VeAgent,
-        short_term_memory: ShortTermMemory,
+        short_term_memory: ShortTermMemory | None = None,
         plugins: list[BasePlugin] | None = None,
         app_name: str = "veadk_default_app",
         user_id: str = "veadk_default_user",
     ):
-        # basic settings
         self.app_name = app_name
         self.user_id = user_id
 
-        # agent settings
         self.agent = agent
 
-        self.short_term_memory = short_term_memory
-        self.session_service = short_term_memory.session_service
+        if not short_term_memory:
+            logger.info(
+                "No short term memory provided, using a in-memory memory by default."
+            )
+            self.short_term_memory = ShortTermMemory()
+        else:
+            self.short_term_memory = short_term_memory
+
+        self.session_service = self.short_term_memory.session_service
 
         # prevent VeRemoteAgent has no long-term memory attr
         if isinstance(self.agent, Agent):
@@ -78,13 +86,25 @@ class Runner:
             plugins=plugins,
         )
 
-    def _convert_messages(self, messages) -> list:
+    def _convert_messages(self, messages, session_id) -> list:
         if isinstance(messages, str):
             messages = [types.Content(role="user", parts=[types.Part(text=messages)])]
         elif isinstance(messages, MediaMessage):
             assert messages.media.endswith(".png"), (
                 "The MediaMessage only supports PNG format file for now."
             )
+            data = read_png_to_bytes(messages.media)
+
+            ve_tos = VeTOS()
+            object_key, tos_url = ve_tos.build_tos_url(
+                self.user_id, self.app_name, session_id, messages.media
+            )
+            try:
+                asyncio.create_task(ve_tos.upload(object_key, data))
+            except Exception as e:
+                logger.error(f"Upload to TOS failed: {e}")
+                tos_url = None
+
             messages = [
                 types.Content(
                     role="user",
@@ -92,8 +112,8 @@ class Runner:
                         types.Part(text=messages.text),
                         types.Part(
                             inline_data=Blob(
-                                display_name=messages.media,
-                                data=read_png_to_bytes(messages.media),
+                                display_name=tos_url,
+                                data=data,
                                 mime_type="image/png",
                             )
                         ),
@@ -103,7 +123,7 @@ class Runner:
         elif isinstance(messages, list):
             converted_messages = []
             for message in messages:
-                converted_messages.extend(self._convert_messages(message))
+                converted_messages.extend(self._convert_messages(message, session_id))
             messages = converted_messages
         else:
             raise ValueError(f"Unknown message type: {type(messages)}")
@@ -114,35 +134,44 @@ class Runner:
         self,
         session_id: str,
         message: types.Content,
+        run_config: RunConfig | None = None,
         stream: bool = False,
     ):
         stream_mode = StreamingMode.SSE if stream else StreamingMode.NONE
 
-        async def event_generator():
-            async for event in self.runner.run_async(
-                user_id=self.user_id,
-                session_id=session_id,
-                new_message=message,
-                run_config=RunConfig(streaming_mode=stream_mode),
-            ):
-                if event.get_function_calls():
-                    for function_call in event.get_function_calls():
-                        logger.debug(f"Function call: {function_call}")
-                elif (
-                    event.content is not None
-                    and event.content.parts
-                    and event.content.parts[0].text is not None
-                    and len(event.content.parts[0].text.strip()) > 0
-                ):
-                    yield event.content.parts[0].text
+        if run_config is not None:
+            stream_mode = run_config.streaming_mode
+        else:
+            run_config = RunConfig(streaming_mode=stream_mode)
+        try:
 
-        final_output = ""
-        async for chunk in event_generator():
+            async def event_generator():
+                async for event in self.runner.run_async(
+                    user_id=self.user_id,
+                    session_id=session_id,
+                    new_message=message,
+                    run_config=run_config,
+                ):
+                    if event.get_function_calls():
+                        for function_call in event.get_function_calls():
+                            logger.debug(f"Function call: {function_call}")
+                    elif (
+                        event.content is not None
+                        and event.content.parts
+                        and event.content.parts[0].text is not None
+                        and len(event.content.parts[0].text.strip()) > 0
+                    ):
+                        yield event.content.parts[0].text
+
+            final_output = ""
+            async for chunk in event_generator():
+                if stream:
+                    print(chunk, end="", flush=True)
+                final_output += chunk
             if stream:
-                print(chunk, end="", flush=True)
-            final_output += chunk
-        if stream:
-            print()  # end with a new line
+                print()  # end with a new line
+        except LlmCallsLimitExceededError as e:
+            logger.warning(f"Max number of llm calls limit exceeded: {e}")
 
         return final_output
 
@@ -151,9 +180,10 @@ class Runner:
         messages: RunnerMessage,
         session_id: str,
         stream: bool = False,
+        run_config: RunConfig | None = None,
         save_tracing_data: bool = False,
     ):
-        converted_messages: list = self._convert_messages(messages)
+        converted_messages: list = self._convert_messages(messages, session_id)
 
         await self.short_term_memory.create_session(
             app_name=self.app_name, user_id=self.user_id, session_id=session_id
@@ -163,7 +193,9 @@ class Runner:
 
         final_output = ""
         for converted_message in converted_messages:
-            final_output = await self._run(session_id, converted_message, stream)
+            final_output = await self._run(
+                session_id, converted_message, run_config, stream
+            )
 
         # try to save tracing file
         if save_tracing_data:
@@ -193,6 +225,47 @@ class Runner:
             logger.warning(f"Get tracer id failed as {e}")
             return "<unknown_trace_id>"
 
+    async def run_with_raw_message(
+        self,
+        message: types.Content,
+        session_id: str,
+        run_config: RunConfig | None = None,
+    ):
+        run_config = RunConfig() if not run_config else run_config
+
+        await self.short_term_memory.create_session(
+            app_name=self.app_name, user_id=self.user_id, session_id=session_id
+        )
+
+        try:
+
+            async def event_generator():
+                async for event in self.runner.run_async(
+                    user_id=self.user_id,
+                    session_id=session_id,
+                    new_message=message,
+                    run_config=run_config,
+                ):
+                    if event.get_function_calls():
+                        for function_call in event.get_function_calls():
+                            logger.debug(f"Function call: {function_call}")
+                    elif (
+                        event.content is not None
+                        and event.content.parts
+                        and event.content.parts[0].text is not None
+                        and len(event.content.parts[0].text.strip()) > 0
+                    ):
+                        yield event.content.parts[0].text
+
+            final_output = ""
+
+            async for chunk in event_generator():
+                final_output += chunk
+        except LlmCallsLimitExceededError as e:
+            logger.warning(f"Max number of llm calls limit exceeded: {e}")
+
+        return final_output
+
     def _print_trace_id(self) -> None:
         if not isinstance(self.agent, Agent):
             logger.warning(