diff --git a/metagpt/actions/design_api.py b/metagpt/actions/design_api.py index 3e978f823..5f973bb60 100644 --- a/metagpt/actions/design_api.py +++ b/metagpt/actions/design_api.py @@ -110,7 +110,7 @@ class WriteDesign(Action): if not data_api_design: return pathname = self.git_repo.workdir / DATA_API_DESIGN_FILE_REPO / Path(design_doc.filename).with_suffix("") - await WriteDesign._save_mermaid_file(data_api_design, pathname) + await self._save_mermaid_file(data_api_design, pathname) logger.info(f"Save class view to {str(pathname)}") async def _save_seq_flow(self, design_doc): @@ -119,13 +119,12 @@ class WriteDesign(Action): if not seq_flow: return pathname = self.git_repo.workdir / Path(SEQ_FLOW_FILE_REPO) / Path(design_doc.filename).with_suffix("") - await WriteDesign._save_mermaid_file(seq_flow, pathname) + await self._save_mermaid_file(seq_flow, pathname) logger.info(f"Saving sequence flow to {str(pathname)}") async def _save_pdf(self, design_doc): await self.file_repo.save_as(doc=design_doc, with_suffix=".md", relative_path=SYSTEM_DESIGN_PDF_FILE_REPO) - @staticmethod - async def _save_mermaid_file(data: str, pathname: Path): + async def _save_mermaid_file(self, data: str, pathname: Path): pathname.parent.mkdir(parents=True, exist_ok=True) - await mermaid_to_file(data, pathname) + await mermaid_to_file(self.config.mermaid_engine, data, pathname) diff --git a/metagpt/actions/research.py b/metagpt/actions/research.py index a635714ef..0af49a1cf 100644 --- a/metagpt/actions/research.py +++ b/metagpt/actions/research.py @@ -8,7 +8,6 @@ from typing import Callable, Optional, Union from pydantic import Field, parse_obj_as from metagpt.actions import Action -from metagpt.config import CONFIG from metagpt.config2 import config from metagpt.logs import logger from metagpt.tools.search_engine import SearchEngine @@ -216,9 +215,7 @@ class WebBrowseAndSummarize(Action): for u, content in zip([url, *urls], contents): content = content.inner_text chunk_summaries = [] - for prompt in generate_prompt_chunk( - content, prompt_template, self.llm.model, system_text, CONFIG.max_tokens_rsp - ): + for prompt in generate_prompt_chunk(content, prompt_template, self.llm.model, system_text, 4096): logger.debug(prompt) summary = await self._aask(prompt, [system_text]) if summary == "Not relevant.": diff --git a/metagpt/config2.py b/metagpt/config2.py index 2a9611627..6345c1b8c 100644 --- a/metagpt/config2.py +++ b/metagpt/config2.py @@ -74,6 +74,12 @@ class Config(CLIParams, YamlModel): mmdc: str = "mmdc" puppeteer_config: str = "" pyppeteer_executable_path: str = "" + IFLYTEK_APP_ID: str = "" + IFLYTEK_APP_SECRET: str = "" + IFLYTEK_APP_KEY: str = "" + AZURE_TTS_SUBSCRIPTION_KEY: str = "" + AZURE_TTS_REGION: str = "" + mermaid_engine: str = "nodejs" @classmethod def default(cls): diff --git a/metagpt/learn/text_to_speech.py b/metagpt/learn/text_to_speech.py index f12e52b8e..8ffafbd0e 100644 --- a/metagpt/learn/text_to_speech.py +++ b/metagpt/learn/text_to_speech.py @@ -7,7 +7,6 @@ @Desc : Text-to-Speech skill, which provides text-to-speech functionality """ -from metagpt.config import CONFIG from metagpt.config2 import config from metagpt.const import BASE64_FORMAT from metagpt.tools.azure_tts import oas3_azsure_tts @@ -45,7 +44,7 @@ async def text_to_speech( """ - if (CONFIG.AZURE_TTS_SUBSCRIPTION_KEY and CONFIG.AZURE_TTS_REGION) or (subscription_key and region): + if subscription_key and region: audio_declaration = "data:audio/wav;base64," base64_data = await oas3_azsure_tts(text, lang, voice, style, role, subscription_key, region) s3 = S3(config.s3) @@ -53,14 +52,12 @@ async def text_to_speech( if url: return f"[{text}]({url})" return audio_declaration + base64_data if base64_data else base64_data - if (CONFIG.IFLYTEK_APP_ID and CONFIG.IFLYTEK_API_KEY and CONFIG.IFLYTEK_API_SECRET) or ( - iflytek_app_id and iflytek_api_key and iflytek_api_secret - ): + if iflytek_app_id and iflytek_api_key and iflytek_api_secret: audio_declaration = "data:audio/mp3;base64," base64_data = await oas3_iflytek_tts( text=text, app_id=iflytek_app_id, api_key=iflytek_api_key, api_secret=iflytek_api_secret ) - s3 = S3() + s3 = S3(config.s3) url = await s3.cache(data=base64_data, file_ext=".mp3", format=BASE64_FORMAT) if url: return f"[{text}]({url})" diff --git a/metagpt/tools/azure_tts.py b/metagpt/tools/azure_tts.py index f4f8aa0a2..2e0e2267c 100644 --- a/metagpt/tools/azure_tts.py +++ b/metagpt/tools/azure_tts.py @@ -13,7 +13,6 @@ from uuid import uuid4 import aiofiles from azure.cognitiveservices.speech import AudioConfig, SpeechConfig, SpeechSynthesizer -from metagpt.config import CONFIG from metagpt.logs import logger @@ -25,8 +24,8 @@ class AzureTTS: :param subscription_key: key is used to access your Azure AI service API, see: `https://portal.azure.com/` > `Resource Management` > `Keys and Endpoint` :param region: This is the location (or region) of your resource. You may need to use this field when making calls to this API. """ - self.subscription_key = subscription_key if subscription_key else CONFIG.AZURE_TTS_SUBSCRIPTION_KEY - self.region = region if region else CONFIG.AZURE_TTS_REGION + self.subscription_key = subscription_key + self.region = region # 参数参考:https://learn.microsoft.com/zh-cn/azure/cognitive-services/speech-service/language-support?tabs=tts#voice-styles-and-roles async def synthesize_speech(self, lang, voice, text, output_file): @@ -83,10 +82,6 @@ async def oas3_azsure_tts(text, lang="", voice="", style="", role="", subscripti role = "Girl" if not style: style = "affectionate" - if not subscription_key: - subscription_key = CONFIG.AZURE_TTS_SUBSCRIPTION_KEY - if not region: - region = CONFIG.AZURE_TTS_REGION xml_value = AzureTTS.role_style_text(role=role, style=style, text=text) tts = AzureTTS(subscription_key=subscription_key, region=region) diff --git a/metagpt/tools/iflytek_tts.py b/metagpt/tools/iflytek_tts.py index ad2395362..6ce48826b 100644 --- a/metagpt/tools/iflytek_tts.py +++ b/metagpt/tools/iflytek_tts.py @@ -23,7 +23,6 @@ import aiofiles import websockets as websockets from pydantic import BaseModel -from metagpt.config import CONFIG from metagpt.logs import logger @@ -56,9 +55,9 @@ class IFlyTekTTS(object): :param api_key: WebAPI argument, see: `https://console.xfyun.cn/services/tts` :param api_secret: WebAPI argument, see: `https://console.xfyun.cn/services/tts` """ - self.app_id = app_id or CONFIG.IFLYTEK_APP_ID - self.api_key = api_key or CONFIG.IFLYTEK_API_KEY - self.api_secret = api_secret or CONFIG.API_SECRET + self.app_id = app_id + self.api_key = api_key + self.api_secret = api_secret async def synthesize_speech(self, text, output_file: str, voice=DEFAULT_IFLYTEK_VOICE): url = self._create_url() @@ -127,14 +126,6 @@ async def oas3_iflytek_tts(text: str, voice: str = "", app_id: str = "", api_key :return: Returns the Base64-encoded .mp3 file data if successful, otherwise an empty string. """ - if not app_id: - app_id = CONFIG.IFLYTEK_APP_ID - if not api_key: - api_key = CONFIG.IFLYTEK_API_KEY - if not api_secret: - api_secret = CONFIG.IFLYTEK_API_SECRET - if not voice: - voice = CONFIG.IFLYTEK_VOICE or DEFAULT_IFLYTEK_VOICE filename = Path(__file__).parent / (uuid.uuid4().hex + ".mp3") try: diff --git a/metagpt/tools/search_engine.py b/metagpt/tools/search_engine.py index 64388a11f..fd237d537 100644 --- a/metagpt/tools/search_engine.py +++ b/metagpt/tools/search_engine.py @@ -10,7 +10,6 @@ from typing import Callable, Coroutine, Literal, Optional, Union, overload from semantic_kernel.skill_definition import sk_function -from metagpt.config import CONFIG from metagpt.tools import SearchEngineType @@ -46,7 +45,6 @@ class SearchEngine: engine: Optional[SearchEngineType] = None, run_func: Callable[[str, int, bool], Coroutine[None, None, Union[str, list[str]]]] = None, ): - engine = engine or CONFIG.search_engine if engine == SearchEngineType.SERPAPI_GOOGLE: module = "metagpt.tools.search_engine_serpapi" run_func = importlib.import_module(module).SerpAPIWrapper().run diff --git a/metagpt/utils/mermaid.py b/metagpt/utils/mermaid.py index 893d05be0..3f6a2ef12 100644 --- a/metagpt/utils/mermaid.py +++ b/metagpt/utils/mermaid.py @@ -17,7 +17,7 @@ from metagpt.logs import logger from metagpt.utils.common import check_cmd_exists -async def mermaid_to_file(mermaid_code, output_file_without_suffix, width=2048, height=2048) -> int: +async def mermaid_to_file(engine, mermaid_code, output_file_without_suffix, width=2048, height=2048) -> int: """suffix: png/svg/pdf :param mermaid_code: mermaid code @@ -35,7 +35,6 @@ async def mermaid_to_file(mermaid_code, output_file_without_suffix, width=2048, await f.write(mermaid_code) # tmp.write_text(mermaid_code, encoding="utf-8") - engine = config.mermaid["default"].engine if engine == "nodejs": if check_cmd_exists(config.mmdc) != 0: logger.warning( diff --git a/tests/metagpt/learn/test_text_to_speech.py b/tests/metagpt/learn/test_text_to_speech.py index aca08b9a2..41611171c 100644 --- a/tests/metagpt/learn/test_text_to_speech.py +++ b/tests/metagpt/learn/test_text_to_speech.py @@ -9,34 +9,43 @@ import pytest -from metagpt.config import CONFIG +from metagpt.config2 import config from metagpt.learn.text_to_speech import text_to_speech @pytest.mark.asyncio async def test_text_to_speech(): # Prerequisites - assert CONFIG.IFLYTEK_APP_ID - assert CONFIG.IFLYTEK_API_KEY - assert CONFIG.IFLYTEK_API_SECRET - assert CONFIG.AZURE_TTS_SUBSCRIPTION_KEY and CONFIG.AZURE_TTS_SUBSCRIPTION_KEY != "YOUR_API_KEY" - assert CONFIG.AZURE_TTS_REGION + assert config.IFLYTEK_APP_ID + assert config.IFLYTEK_API_KEY + assert config.IFLYTEK_API_SECRET + assert config.AZURE_TTS_SUBSCRIPTION_KEY and config.AZURE_TTS_SUBSCRIPTION_KEY != "YOUR_API_KEY" + assert config.AZURE_TTS_REGION + i = config.copy() # test azure - data = await text_to_speech("panda emoji") + data = await text_to_speech( + "panda emoji", + subscription_key=i.AZURE_TTS_SUBSCRIPTION_KEY, + region=i.AZURE_TTS_REGION, + iflytek_api_key=i.IFLYTEK_API_KEY, + iflytek_api_secret=i.IFLYTEK_API_SECRET, + iflytek_app_id=i.IFLYTEK_APP_ID, + ) assert "base64" in data or "http" in data # test iflytek ## Mock session env - old_options = CONFIG.options.copy() - new_options = old_options.copy() - new_options["AZURE_TTS_SUBSCRIPTION_KEY"] = "" - CONFIG.set_context(new_options) - try: - data = await text_to_speech("panda emoji") - assert "base64" in data or "http" in data - finally: - CONFIG.set_context(old_options) + i.AZURE_TTS_SUBSCRIPTION_KEY = "" + data = await text_to_speech( + "panda emoji", + subscription_key=i.AZURE_TTS_SUBSCRIPTION_KEY, + region=i.AZURE_TTS_REGION, + iflytek_api_key=i.IFLYTEK_API_KEY, + iflytek_api_secret=i.IFLYTEK_API_SECRET, + iflytek_app_id=i.IFLYTEK_APP_ID, + ) + assert "base64" in data or "http" in data if __name__ == "__main__": diff --git a/tests/metagpt/tools/test_azure_tts.py b/tests/metagpt/tools/test_azure_tts.py index a33925a5c..e856d3b27 100644 --- a/tests/metagpt/tools/test_azure_tts.py +++ b/tests/metagpt/tools/test_azure_tts.py @@ -11,7 +11,6 @@ import pytest from azure.cognitiveservices.speech import ResultReason -from metagpt.config import CONFIG from metagpt.config2 import config from metagpt.tools.azure_tts import AzureTTS @@ -19,8 +18,8 @@ from metagpt.tools.azure_tts import AzureTTS @pytest.mark.asyncio async def test_azure_tts(): # Prerequisites - assert CONFIG.AZURE_TTS_SUBSCRIPTION_KEY and CONFIG.AZURE_TTS_SUBSCRIPTION_KEY != "YOUR_API_KEY" - assert CONFIG.AZURE_TTS_REGION + assert config.AZURE_TTS_SUBSCRIPTION_KEY and config.AZURE_TTS_SUBSCRIPTION_KEY != "YOUR_API_KEY" + assert config.AZURE_TTS_REGION azure_tts = AzureTTS(subscription_key="", region="") text = """ diff --git a/tests/metagpt/tools/test_iflytek_tts.py b/tests/metagpt/tools/test_iflytek_tts.py index 58d8a83ce..18af0a723 100644 --- a/tests/metagpt/tools/test_iflytek_tts.py +++ b/tests/metagpt/tools/test_iflytek_tts.py @@ -7,22 +7,22 @@ """ import pytest -from metagpt.config import CONFIG +from metagpt.config2 import config from metagpt.tools.iflytek_tts import oas3_iflytek_tts @pytest.mark.asyncio async def test_tts(): # Prerequisites - assert CONFIG.IFLYTEK_APP_ID - assert CONFIG.IFLYTEK_API_KEY - assert CONFIG.IFLYTEK_API_SECRET + assert config.IFLYTEK_APP_ID + assert config.IFLYTEK_API_KEY + assert config.IFLYTEK_API_SECRET result = await oas3_iflytek_tts( text="你好,hello", - app_id=CONFIG.IFLYTEK_APP_ID, - api_key=CONFIG.IFLYTEK_API_KEY, - api_secret=CONFIG.IFLYTEK_API_SECRET, + app_id=config.IFLYTEK_APP_ID, + api_key=config.IFLYTEK_API_KEY, + api_secret=config.IFLYTEK_API_SECRET, ) assert result diff --git a/tests/metagpt/tools/test_web_browser_engine_playwright.py b/tests/metagpt/tools/test_web_browser_engine_playwright.py index 0f2679531..32019bad9 100644 --- a/tests/metagpt/tools/test_web_browser_engine_playwright.py +++ b/tests/metagpt/tools/test_web_browser_engine_playwright.py @@ -4,7 +4,7 @@ import pytest -from metagpt.config import CONFIG +from metagpt.config2 import config from metagpt.tools import web_browser_engine_playwright from metagpt.utils.parse_html import WebPage @@ -20,11 +20,11 @@ from metagpt.utils.parse_html import WebPage ids=["chromium-normal", "firefox-normal", "webkit-normal"], ) async def test_scrape_web_page(browser_type, use_proxy, kwagrs, url, urls, proxy, capfd): - global_proxy = CONFIG.global_proxy + global_proxy = config.proxy try: if use_proxy: server, proxy = await proxy - CONFIG.global_proxy = proxy + config.proxy = proxy browser = web_browser_engine_playwright.PlaywrightWrapper(browser_type=browser_type, **kwagrs) result = await browser.run(url) assert isinstance(result, WebPage) @@ -39,7 +39,7 @@ async def test_scrape_web_page(browser_type, use_proxy, kwagrs, url, urls, proxy server.close() assert "Proxy:" in capfd.readouterr().out finally: - CONFIG.global_proxy = global_proxy + config.proxy = global_proxy if __name__ == "__main__": diff --git a/tests/metagpt/tools/test_web_browser_engine_selenium.py b/tests/metagpt/tools/test_web_browser_engine_selenium.py index 8fe365352..bd5abcb9b 100644 --- a/tests/metagpt/tools/test_web_browser_engine_selenium.py +++ b/tests/metagpt/tools/test_web_browser_engine_selenium.py @@ -4,7 +4,7 @@ import pytest -from metagpt.config import CONFIG +from metagpt.config2 import config from metagpt.tools import web_browser_engine_selenium from metagpt.utils.parse_html import WebPage @@ -23,11 +23,11 @@ async def test_scrape_web_page(browser_type, use_proxy, url, urls, proxy, capfd) # Prerequisites # firefox, chrome, Microsoft Edge - global_proxy = CONFIG.global_proxy + global_proxy = config.proxy try: if use_proxy: server, proxy = await proxy - CONFIG.global_proxy = proxy + config.proxy = proxy browser = web_browser_engine_selenium.SeleniumWrapper(browser_type=browser_type) result = await browser.run(url) assert isinstance(result, WebPage) @@ -42,7 +42,7 @@ async def test_scrape_web_page(browser_type, use_proxy, url, urls, proxy, capfd) server.close() assert "Proxy:" in capfd.readouterr().out finally: - CONFIG.global_proxy = global_proxy + config.proxy = global_proxy if __name__ == "__main__": diff --git a/tests/metagpt/utils/test_mermaid.py b/tests/metagpt/utils/test_mermaid.py index 6345e9c51..367223332 100644 --- a/tests/metagpt/utils/test_mermaid.py +++ b/tests/metagpt/utils/test_mermaid.py @@ -8,7 +8,6 @@ import pytest -from metagpt.config import CONFIG from metagpt.context import CONTEXT from metagpt.utils.common import check_cmd_exists from metagpt.utils.mermaid import MMC1, mermaid_to_file @@ -22,9 +21,8 @@ async def test_mermaid(engine): # playwright prerequisites: playwright install --with-deps chromium assert check_cmd_exists("npm") == 0 - CONFIG.mermaid_engine = engine - save_to = CONTEXT.git_repo.workdir / f"{CONFIG.mermaid_engine}/1" - await mermaid_to_file(MMC1, save_to) + save_to = CONTEXT.git_repo.workdir / f"{engine}/1" + await mermaid_to_file(engine, MMC1, save_to) # ink does not support pdf if engine == "ink":