change pipeline plugin tool (#248)

* change pipeline plugin tool

* Standardized input and output

* Standardized input and output

* Standardized input and output

Author: mushenL

modelscope_agent/tools/pipeline_tool.py

@@ -1,40 +1,30 @@
-from modelscope.pipelines import pipeline
-from .tool import Tool
+from modelscope_agent.tools.base import BaseTool, register_tool
+import json
+import requests
+import os
 
-
-class ModelscopePipelineTool(Tool):
-
-    default_model: str = ''
-    task: str = ''
-    model_revision = None
+@register_tool('pipeline')
+class ModelscopePipelineTool(BaseTool):
+    API_URL = ""
+    API_KEY = ""
 
     def __init__(self, cfg):
-
+        """
+        初始化一个ModelscopePipelineTool类
+        Initialize a ModelscopePipelineTool class.
+        参数：
+        cfg (Dict[str, object]): 配置字典，包含了初始化对象所需要的参数
+        """
         super().__init__(cfg)
-        self.model = self.cfg.get('model', None) or self.default_model
-        self.model_revision = self.cfg.get('model_revision',
-                                           None) or self.model_revision
-
-        self.pipeline_params = self.cfg.get('pipeline_params', {})
-        self.pipeline = None
-        self.is_initialized = False
-
-    def setup(self):
-
-        # only initialize when this tool is really called to save memory
-        if not self.is_initialized:
-            self.pipeline = pipeline(
-                task=self.task,
-                model=self.model,
-                model_revision=self.model_revision,
-                **self.pipeline_params)
-        self.is_initialized = True
-
-    def _local_call(self, *args, **kwargs):
-
-        self.setup()
+        self.API_URL = self.cfg.get(self.name, {}).get('url',None) or self.API_URL
+        self.API_KEY = os.getenv('MODELSCOPE_API_KEY', None) or self.API_KEY
+
+    
+    def call(self, params: str, **kwargs) -> str:
+        params = self._verify_args(params)
+        data = json.dumps(params)
+        headers = {"Authorization": f"Bearer {self.API_KEY}"}
+        response = requests.request("POST", self.API_URL, headers=headers,data=data)
+        result = json.loads(response.content.decode("utf-8"))
+        return result
 
-        parsed_args, parsed_kwargs = self._local_parse_input(*args, **kwargs)
-        origin_result = self.pipeline(*parsed_args, **parsed_kwargs)
-        final_result = self._parse_output(origin_result, remote=False)
-        return final_result

modelscope_agent/tools/plugin_tool.py

@@ -1,9 +1,16 @@
 from copy import deepcopy
+from modelscope_agent.tools.base import BaseTool, register_tool
 
-from .tool import Tool
-
-
-class LangchainTool(Tool):
+@register_tool('plugin')
+class LangchainTool(BaseTool):
+    description = '通过调用langchain插件来支持对语言模型的输入输出格式进行处理，输入文本字符，输出经过格式处理的结果'
+    name = 'plugin'
+    parameters: list = [{
+        'name': 'commands',
+        'description': '需要进行格式处理的文本字符列表',
+        'required': True,
+        'type': "string"
+    }]
 
     def __init__(self, langchain_tool):
         from langchain.tools import BaseTool
@@ -23,8 +30,11 @@ def parse_langchain_schema(self):
             tool_arg = deepcopy(arg)
             tool_arg['name'] = name
             tool_arg['required'] = True
+            tool_arg['type'] = arg['anyOf'][0].get("type","string")
             tool_arg.pop('title')
             self.parameters.append(tool_arg)
 
-    def _local_call(self, *args, **kwargs):
-        return {'result': self.langchain_tool.run(kwargs)}
+    def call(self, params: str, **kwargs):
+        params = self._verify_args(params)
+        res = self.langchain_tool.run(params)
+        return res

modelscope_agent/tools/text_to_speech_tool.py

@@ -1,8 +1,5 @@
-from modelscope_agent.output_wrapper import AudioWrapper
-
-from modelscope.utils.constant import Tasks
 from .pipeline_tool import ModelscopePipelineTool
-
+from modelscope_agent.output_wrapper import AudioWrapper
 
 class TexttoSpeechTool(ModelscopePipelineTool):
     default_model = 'damo/speech_sambert-hifigan_tts_zh-cn_16k'
@@ -11,35 +8,16 @@ class TexttoSpeechTool(ModelscopePipelineTool):
     parameters: list = [{
         'name': 'input',
         'description': '要转成语音的文本',
-        'required': True
+        'required': True,
+        'type': 'string'
     }, {
         'name': 'gender',
         'description': '用户身份',
-        'required': True
+        'required': True,
+        'type': 'string'
     }]
-    task = Tasks.text_to_speech
-
-    def _local_parse_input(self, *args, **kwargs):
-        if 'gender' not in kwargs:
-            kwargs['gender'] = 'man'
-        voice = 'zhizhe_emo' if kwargs['gender'] == 'man' else 'zhiyan_emo'
-        kwargs['voice'] = voice
-        if 'text' in kwargs and 'input' not in kwargs:
-            kwargs['input'] = kwargs['text']
-            kwargs.pop('text')
-        kwargs.pop('gender')
-        return args, kwargs
-
-    def _remote_parse_input(self, *args, **kwargs):
-        if 'gender' not in kwargs:
-            kwargs['gender'] = 'man'
-        voice = 'zhizhe_emo' if kwargs['gender'] == 'man' or kwargs[
-            'gender'] == 'male' else 'zhiyan_emo'
-        kwargs['parameters'] = {'voice': voice}
-        kwargs.pop('gender')
-        return kwargs
-
-    def _parse_output(self, origin_result, remote=True):
 
-        audio = origin_result['output_wav']
-        return {'result': AudioWrapper(audio)}
+    def call(self, params: str, **kwargs) -> str:
+        result = super().call(params, **kwargs)
+        audio = result['Data']['output_wav']
+        return AudioWrapper(audio)

tests/tools/test_langchain_tool.py

@@ -14,5 +14,9 @@ def test_is_langchain_tool():
 def test_run_langchin_tool():
     # test run langchain tool
     shell_tool = LangchainTool(ShellTool())
-    res = shell_tool(commands=["echo 'Hello World!'"])
-    assert res['result'] == 'Hello World!\n'
+    input = """{'commands': ["echo 'Hello World!'"]}"""
+    res = shell_tool.call(input)
+    print(res)
+    assert res == 'Hello World!\n'
+
+test_run_langchin_tool()

tests/tools/test_pipeline_tool.py

@@ -0,0 +1,18 @@
+from modelscope_agent.tools.pipeline_tool import ModelscopePipelineTool
+from modelscope.utils.config import Config
+import os
+
+cfg = Config.from_file('config/cfg_tool_template.json')
+# 请用自己的SDK令牌替换{YOUR_MODELSCOPE_SDK_TOKEN}（包括大括号）
+os.environ['MODELSCOPE_API_KEY'] = f"{YOUR_MODELSCOPE_SDK_TOKEN}"
+
+def test_modelscope_speech_generation():
+    from modelscope_agent.tools.text_to_speech_tool import TexttoSpeechTool
+    kwargs = """{'input': '北京今天天气怎样?', 'gender': 'man'}"""
+    txt2speech = TexttoSpeechTool(cfg)
+    res = txt2speech.call(kwargs)
+    print(res)
+
+
+test_modelscope_speech_generation()
+