feat: +unit test

metagpt/learn/text_to_image.py

@@ -6,6 +6,7 @@
 @File    : text_to_image.py
 @Desc    : Text-to-Image skill, which provides text-to-image functionality.
 """
+import base64
 
 from metagpt.config import CONFIG
 from metagpt.const import BASE64_FORMAT
@@ -25,11 +26,12 @@ async def text_to_image(text, size_type: str = "512x512", openai_api_key="", mod
     """
     image_declaration = "data:image/png;base64,"
     if CONFIG.METAGPT_TEXT_TO_IMAGE_MODEL_URL or model_url:
-        base64_data = await oas3_metagpt_text_to_image(text, size_type, model_url)
+        binary_data = await oas3_metagpt_text_to_image(text, size_type, model_url)
     elif CONFIG.OPENAI_API_KEY or openai_api_key:
-        base64_data = await oas3_openai_text_to_image(text, size_type)
+        binary_data = await oas3_openai_text_to_image(text, size_type)
     else:
         raise ValueError("Missing necessary parameters.")
+    base64_data = base64.b64encode(binary_data).decode("utf-8")
 
     s3 = S3()
     url = await s3.cache(data=base64_data, file_ext=".png", format=BASE64_FORMAT) if s3.is_valid else ""

metagpt/learn/text_to_speech.py

@@ -6,7 +6,6 @@
 @File    : text_to_speech.py
 @Desc    : Text-to-Speech skill, which provides text-to-speech functionality
 """
-import openai
 
 from metagpt.config import CONFIG
 from metagpt.const import BASE64_FORMAT
@@ -66,7 +65,6 @@ async def text_to_speech(
             return f"[{text}]({url})"
         return audio_declaration + base64_data if base64_data else base64_data
 
-    raise openai.InvalidRequestError(
-        message="AZURE_TTS_SUBSCRIPTION_KEY, AZURE_TTS_REGION, IFLYTEK_APP_ID, IFLYTEK_API_KEY, IFLYTEK_API_SECRET error",
-        param={},
+    raise ValueError(
+        "AZURE_TTS_SUBSCRIPTION_KEY, AZURE_TTS_REGION, IFLYTEK_APP_ID, IFLYTEK_API_KEY, IFLYTEK_API_SECRET error"
     )

metagpt/tools/azure_tts.py

@@ -96,9 +96,10 @@ async def oas3_azsure_tts(text, lang="", voice="", style="", role="", subscripti
         async with aiofiles.open(filename, mode="rb") as reader:
             data = await reader.read()
             base64_string = base64.b64encode(data).decode("utf-8")
-        filename.unlink()
     except Exception as e:
         logger.error(f"text:{text}, error:{e}")
         return ""
+    finally:
+        filename.unlink(missing_ok=True)
 
     return base64_string

metagpt/tools/hello.py

@@ -7,7 +7,7 @@
 @Desc    : Implement the OpenAPI Specification 3.0 demo and use the following command to test the HTTP service:
 
         curl -X 'POST' \
-        'http://localhost:8080/openapi/greeting/dave' \
+        'http://localhost:8082/openapi/greeting/dave' \
         -H 'accept: text/plain' \
         -H 'Content-Type: application/json' \
         -d '{}'
@@ -26,4 +26,4 @@ if __name__ == "__main__":
     specification_dir = Path(__file__).parent.parent.parent / ".well-known"
     app = connexion.AsyncApp(__name__, specification_dir=str(specification_dir))
     app.add_api("openapi.yaml", arguments={"title": "Hello World Example"})
-    app.run(port=8080)
+    app.run(port=8082)

metagpt/tools/iflytek_tts.py

@@ -6,7 +6,6 @@
 @File    : iflytek_tts.py
 @Desc    : iFLYTEK TTS OAS3 api, which provides text-to-speech functionality
 """
-import asyncio
 import base64
 import hashlib
 import hmac
@@ -74,12 +73,13 @@ class IFlyTekTTS(object):
             await websocket.send(req)
 
             # receive frames
-            async with aiofiles.open(str(output_file), "w") as writer:
+            async with aiofiles.open(str(output_file), "wb") as writer:
                 while True:
                     v = await websocket.recv()
                     rsp = IFlyTekTTSResponse(**json.loads(v))
                     if rsp.data:
-                        await writer.write(rsp.data.audio)
+                        binary_data = base64.b64decode(rsp.data.audio)
+                        await writer.write(binary_data)
                         if rsp.data.status != IFlyTekTTSStatus.STATUS_LAST_FRAME.value:
                             continue
                     break
@@ -140,23 +140,13 @@ async def oas3_iflytek_tts(text: str, voice: str = "", app_id: str = "", api_key
     try:
         tts = IFlyTekTTS(app_id=app_id, api_key=api_key, api_secret=api_secret)
         await tts.synthesize_speech(text=text, output_file=str(filename), voice=voice)
-        async with aiofiles.open(str(filename), mode="r") as reader:
-            base64_string = await reader.read()
+        async with aiofiles.open(str(filename), mode="rb") as reader:
+            data = await reader.read()
+            base64_string = base64.b64encode(data).decode("utf-8")
     except Exception as e:
         logger.error(f"text:{text}, error:{e}")
         base64_string = ""
     finally:
-        filename.unlink()
+        filename.unlink(missing_ok=True)
 
     return base64_string
-
-
-if __name__ == "__main__":
-    asyncio.get_event_loop().run_until_complete(
-        oas3_iflytek_tts(
-            text="你好，hello",
-            app_id="f7acef62",
-            api_key="fda72e3aa286042a492525816a5efa08",
-            api_secret="ZDk3NjdiMDBkODJlOWQ1NjRjMGI2NDY4",
-        )
-    )

metagpt/tools/metagpt_oas3_api_svc.py

@@ -6,39 +6,21 @@
 @File    : metagpt_oas3_api_svc.py
 @Desc    : MetaGPT OpenAPI Specification 3.0 REST API service
 """
-import asyncio
-import sys
+
 from pathlib import Path
 
 import connexion
 
-sys.path.append(str(Path(__file__).resolve().parent.parent.parent))  # fix-bug: No module named 'metagpt'
-
 
 def oas_http_svc():
     """Start the OAS 3.0 OpenAPI HTTP service"""
-    app = connexion.AioHttpApp(__name__, specification_dir="../../.well-known/")
+    print("http://localhost:8080/oas3/ui/")
+    specification_dir = Path(__file__).parent.parent.parent / ".well-known"
+    app = connexion.AsyncApp(__name__, specification_dir=str(specification_dir))
     app.add_api("metagpt_oas3_api.yaml")
     app.add_api("openapi.yaml")
     app.run(port=8080)
 
 
-async def async_main():
-    """Start the OAS 3.0 OpenAPI HTTP service in the background."""
-    loop = asyncio.get_event_loop()
-    loop.run_in_executor(None, oas_http_svc)
-
-    # TODO: replace following codes:
-    while True:
-        await asyncio.sleep(1)
-        print("sleep")
-
-
-def main():
-    print("http://localhost:8080/oas3/ui/")
-    oas_http_svc()
-
-
 if __name__ == "__main__":
-    # asyncio.run(async_main())
-    main()
+    oas_http_svc()

metagpt/tools/metagpt_text_to_image.py

@@ -6,7 +6,6 @@
 @File    : metagpt_text_to_image.py
 @Desc    : MetaGPT Text-to-Image OAS3 api, which provides text-to-image functionality.
 """
-import asyncio
 import base64
 from typing import Dict, List
 
@@ -14,7 +13,7 @@ import aiohttp
 import requests
 from pydantic import BaseModel
 
-from metagpt.config import CONFIG, Config
+from metagpt.config import CONFIG
 from metagpt.logs import logger
 
 
@@ -75,11 +74,12 @@ class MetaGPTText2Image:
                 async with session.post(self.model_url, headers=headers, json=data) as response:
                     result = ImageResult(**await response.json())
             if len(result.images) == 0:
-                return ""
-            return result.images[0]
+                return 0
+            data = base64.b64decode(result.images[0])
+            return data
         except requests.exceptions.RequestException as e:
             logger.error(f"An error occurred:{e}")
-        return ""
+        return 0
 
 
 # Export
@@ -96,15 +96,3 @@ async def oas3_metagpt_text_to_image(text, size_type: str = "512x512", model_url
     if not model_url:
         model_url = CONFIG.METAGPT_TEXT_TO_IMAGE_MODEL_URL
     return await MetaGPTText2Image(model_url).text_2_image(text, size_type=size_type)
-
-
-if __name__ == "__main__":
-    Config()
-    loop = asyncio.new_event_loop()
-    task = loop.create_task(oas3_metagpt_text_to_image("Panda emoji"))
-    v = loop.run_until_complete(task)
-    print(v)
-    data = base64.b64decode(v)
-    with open("tmp.png", mode="wb") as writer:
-        writer.write(data)
-    print(v)

metagpt/tools/moderation.py

@@ -5,7 +5,6 @@
 @Author  : zhanglei
 @File    : moderation.py
 """
-import asyncio
 from typing import Union
 
 from metagpt.llm import LLM
@@ -15,6 +14,38 @@ class Moderation:
     def __init__(self):
         self.llm = LLM()
 
+    def handle_moderation_results(self, results):
+        resp = []
+        for item in results:
+            categories = item.categories.dict()
+            true_categories = [category for category, item_flagged in categories.items() if item_flagged]
+            resp.append({"flagged": item.flagged, "true_categories": true_categories})
+        return resp
+
+    def moderation_with_categories(self, content: Union[str, list[str]]):
+        resp = []
+        if content:
+            moderation_results = self.llm.moderation(content=content)
+            resp = self.handle_moderation_results(moderation_results.results)
+        return resp
+
+    async def amoderation_with_categories(self, content: Union[str, list[str]]):
+        resp = []
+        if content:
+            moderation_results = await self.llm.amoderation(content=content)
+            resp = self.handle_moderation_results(moderation_results.results)
+        return resp
+
+    def moderation(self, content: Union[str, list[str]]):
+        resp = []
+        if content:
+            moderation_results = self.llm.moderation(content=content)
+            results = moderation_results.results
+            for item in results:
+                resp.append(item.flagged)
+
+        return resp
+
     async def amoderation(self, content: Union[str, list[str]]):
         resp = []
         if content:
@@ -24,15 +55,3 @@ class Moderation:
                 resp.append(item.flagged)
 
         return resp
-
-
-async def main():
-    moderation = Moderation()
-    rsp = await moderation.amoderation(
-        content=["I will kill you", "The weather is really nice today", "I want to hit you"]
-    )
-    print(rsp)
-
-
-if __name__ == "__main__":
-    asyncio.run(main())

metagpt/tools/openai_text_to_embedding.py

@@ -7,14 +7,13 @@
 @Desc    : OpenAI Text-to-Embedding OAS3 api, which provides text-to-embedding functionality.
             For more details, checkout: `https://platform.openai.com/docs/api-reference/embeddings/object`
 """
-import asyncio
 from typing import List
 
 import aiohttp
 import requests
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
-from metagpt.config import CONFIG, Config
+from metagpt.config import CONFIG
 from metagpt.logs import logger
 
 
@@ -29,15 +28,18 @@ class Embedding(BaseModel):
 
 
 class Usage(BaseModel):
-    prompt_tokens: int
-    total_tokens: int
+    prompt_tokens: int = 0
+    total_tokens: int = 0
 
 
 class ResultEmbedding(BaseModel):
-    object: str
-    data: List[Embedding]
-    model: str
-    usage: Usage
+    class Config:
+        alias = {"object_": "object"}
+
+    object_: str = ""
+    data: List[Embedding] = []
+    model: str = ""
+    usage: Usage = Field(default_factory=Usage)
 
 
 class OpenAIText2Embedding:
@@ -45,7 +47,7 @@ class OpenAIText2Embedding:
         """
         :param openai_api_key: OpenAI API key, For more details, checkout: `https://platform.openai.com/account/api-keys`
         """
-        self.openai_api_key = openai_api_key if openai_api_key else CONFIG.OPENAI_API_KEY
+        self.openai_api_key = openai_api_key or CONFIG.OPENAI_API_KEY
 
     async def text_2_embedding(self, text, model="text-embedding-ada-002"):
         """Text to embedding
@@ -55,15 +57,18 @@ class OpenAIText2Embedding:
         :return: A json object of :class:`ResultEmbedding` class if successful, otherwise `{}`.
         """
 
+        proxies = {"proxy": CONFIG.openai_proxy} if CONFIG.openai_proxy else {}
         headers = {"Content-Type": "application/json", "Authorization": f"Bearer {self.openai_api_key}"}
         data = {"input": text, "model": model}
+        url = "https://api.openai.com/v1/embeddings"
         try:
             async with aiohttp.ClientSession() as session:
-                async with session.post("https://api.openai.com/v1/embeddings", headers=headers, json=data) as response:
-                    return await response.json()
+                async with session.post(url, headers=headers, json=data, **proxies) as response:
+                    data = await response.json()
+                    return ResultEmbedding(**data)
         except requests.exceptions.RequestException as e:
             logger.error(f"An error occurred:{e}")
-        return {}
+        return ResultEmbedding()
 
 
 # Export
@@ -80,11 +85,3 @@ async def oas3_openai_text_to_embedding(text, model="text-embedding-ada-002", op
     if not openai_api_key:
         openai_api_key = CONFIG.OPENAI_API_KEY
     return await OpenAIText2Embedding(openai_api_key).text_2_embedding(text, model=model)
-
-
-if __name__ == "__main__":
-    Config()
-    loop = asyncio.new_event_loop()
-    task = loop.create_task(oas3_openai_text_to_embedding("Panda emoji"))
-    v = loop.run_until_complete(task)
-    print(v)

metagpt/tools/openai_text_to_image.py

@@ -6,13 +6,10 @@
 @File    : openai_text_to_image.py
 @Desc    : OpenAI Text-to-Image OAS3 api, which provides text-to-image functionality.
 """
-import asyncio
-import base64
 
 import aiohttp
 import requests
 
-from metagpt.config import Config
 from metagpt.llm import LLM
 from metagpt.logs import logger
 
@@ -23,7 +20,6 @@ class OpenAIText2Image:
         :param openai_api_key: OpenAI API key, For more details, checkout: `https://platform.openai.com/account/api-keys`
         """
         self._llm = LLM()
-        self._client = self._llm.async_client
 
     def __del__(self):
         if self._llm:
@@ -37,7 +33,7 @@ class OpenAIText2Image:
         :return: The image data is returned in Base64 encoding.
         """
         try:
-            result = await self._client.images.generate(prompt=text, n=1, size=size_type)
+            result = await self._llm.async_client.images.generate(prompt=text, n=1, size=size_type)
         except Exception as e:
             logger.error(f"An error occurred:{e}")
             return ""
@@ -57,12 +53,11 @@ class OpenAIText2Image:
                 async with session.get(url) as response:
                     response.raise_for_status()  # 如果是 4xx 或 5xx 响应，会引发异常
                     image_data = await response.read()
-            base64_image = base64.b64encode(image_data).decode("utf-8")
-            return base64_image
+            return image_data
 
         except requests.exceptions.RequestException as e:
             logger.error(f"An error occurred:{e}")
-            return ""
+            return 0
 
 
 # Export
@@ -76,11 +71,3 @@ async def oas3_openai_text_to_image(text, size_type: str = "1024x1024"):
     if not text:
         return ""
     return await OpenAIText2Image().text_2_image(text, size_type=size_type)
-
-
-if __name__ == "__main__":
-    Config()
-    loop = asyncio.new_event_loop()
-    task = loop.create_task(oas3_openai_text_to_image("Panda emoji"))
-    v = loop.run_until_complete(task)
-    print(v)

metagpt/tools/web_browser_engine_selenium.py

@@ -9,7 +9,7 @@ import asyncio
 import importlib
 from concurrent import futures
 from copy import deepcopy
-from typing import Dict, Literal
+from typing import Literal
 
 from selenium.webdriver.common.by import By
 from selenium.webdriver.support import expected_conditions as EC
@@ -33,7 +33,6 @@ class SeleniumWrapper:
 
     def __init__(
         self,
-        options: Dict,
         browser_type: Literal["chrome", "firefox", "edge", "ie"] | None = None,
         launch_kwargs: dict | None = None,
         *,