允许自定义系统提示

This commit is contained in:
xunbu
2025-05-28 10:16:42 +08:00
parent 0d29b636da
commit 3c9b94150c
7 changed files with 137 additions and 104 deletions

139
.idea/workspace.xml generated
View File

@@ -6,7 +6,12 @@
<component name="ChangeListManager"> <component name="ChangeListManager">
<list default="true" id="6b18b44a-df57-4212-a857-9e291ebe5dd2" name="更改" comment=""> <list default="true" id="6b18b44a-df57-4212-a857-9e291ebe5dd2" name="更改" comment="">
<change beforePath="$PROJECT_DIR$/.idea/workspace.xml" beforeDir="false" afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" /> <change beforePath="$PROJECT_DIR$/.idea/workspace.xml" beforeDir="false" afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" />
<change beforePath="$PROJECT_DIR$/docutranslate/converter/converter_mineru.py" beforeDir="false" afterPath="$PROJECT_DIR$/docutranslate/converter/converter_mineru.py" afterDir="false" /> <change beforePath="$PROJECT_DIR$/docutranslate/__init__.py" beforeDir="false" afterPath="$PROJECT_DIR$/docutranslate/__init__.py" afterDir="false" />
<change beforePath="$PROJECT_DIR$/docutranslate/agents/agent.py" beforeDir="false" afterPath="$PROJECT_DIR$/docutranslate/agents/agent.py" afterDir="false" />
<change beforePath="$PROJECT_DIR$/docutranslate/agents/markdown_agent.py" beforeDir="false" afterPath="$PROJECT_DIR$/docutranslate/agents/markdown_agent.py" afterDir="false" />
<change beforePath="$PROJECT_DIR$/docutranslate/app.py" beforeDir="false" afterPath="$PROJECT_DIR$/docutranslate/app.py" afterDir="false" />
<change beforePath="$PROJECT_DIR$/docutranslate/static/index.html" beforeDir="false" afterPath="$PROJECT_DIR$/docutranslate/static/index.html" afterDir="false" />
<change beforePath="$PROJECT_DIR$/docutranslate/translater.py" beforeDir="false" afterPath="$PROJECT_DIR$/docutranslate/translater.py" afterDir="false" />
</list> </list>
<option name="SHOW_DIALOG" value="false" /> <option name="SHOW_DIALOG" value="false" />
<option name="HIGHLIGHT_CONFLICTS" value="true" /> <option name="HIGHLIGHT_CONFLICTS" value="true" />
@@ -37,63 +42,63 @@
<option name="hideEmptyMiddlePackages" value="true" /> <option name="hideEmptyMiddlePackages" value="true" />
<option name="showLibraryContents" value="true" /> <option name="showLibraryContents" value="true" />
</component> </component>
<component name="PropertiesComponent">{ <component name="PropertiesComponent"><![CDATA[{
&quot;keyToString&quot;: { "keyToString": {
&quot;DefaultHtmlFileTemplate&quot;: &quot;HTML File&quot;, "DefaultHtmlFileTemplate": "HTML File",
&quot;JavaScript 调试.output.html (1).executor&quot;: &quot;Run&quot;, "JavaScript 调试.output.html (1).executor": "Run",
&quot;JavaScript 调试.output.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.output.html.executor": "Run",
&quot;JavaScript 调试.regex.md_中文.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.regex.md_中文.html.executor": "Run",
&quot;JavaScript 调试.regex_中文.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.regex_中文.html.executor": "Run",
&quot;JavaScript 调试.test.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.test.html.executor": "Run",
&quot;JavaScript 调试.test2.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.test2.html.executor": "Run",
&quot;JavaScript 调试.test2_英文.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.test2_英文.html.executor": "Run",
&quot;JavaScript 调试.test4-1_中文.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.test4-1_中文.html.executor": "Run",
&quot;JavaScript 调试.互联网认证授权机制.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.互联网认证授权机制.html.executor": "Run",
&quot;JavaScript 调试.互联网认证授权机制_英文.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.互联网认证授权机制_英文.html.executor": "Run",
&quot;JavaScript 调试.毕业论文_英文.html.executor&quot;: &quot;Run&quot;, "JavaScript 调试.毕业论文_英文.html.executor": "Run",
&quot;ModuleVcsDetector.initialDetectionPerformed&quot;: &quot;true&quot;, "ModuleVcsDetector.initialDetectionPerformed": "true",
&quot;Python 测试.Python 测试 (markdown_mask.py 内).executor&quot;: &quot;Run&quot;, "Python 测试.Python 测试 (markdown_mask.py 内).executor": "Run",
&quot;Python 测试.markdown_mask.Test.test_basic_link_masking 的 Python 测试.executor&quot;: &quot;Run&quot;, "Python 测试.markdown_mask.Test.test_basic_link_masking 的 Python 测试.executor": "Run",
&quot;Python 测试.pytest (test_html.py 内).executor&quot;: &quot;Run&quot;, "Python 测试.pytest (test_html.py 内).executor": "Run",
&quot;Python.1test.executor&quot;: &quot;Run&quot;, "Python.1test.executor": "Run",
&quot;Python.2test2 (1).executor&quot;: &quot;Run&quot;, "Python.2test2 (1).executor": "Run",
&quot;Python.3testhtml.executor&quot;: &quot;Run&quot;, "Python.3testhtml.executor": "Run",
&quot;Python.PDFtranslater (1).executor&quot;: &quot;Run&quot;, "Python.PDFtranslater (1).executor": "Run",
&quot;Python.PDFtranslater (2).executor&quot;: &quot;Run&quot;, "Python.PDFtranslater (2).executor": "Run",
&quot;Python.agent.executor&quot;: &quot;Debug&quot;, "Python.agent.executor": "Debug",
&quot;Python.agent_utils.executor&quot;: &quot;Run&quot;, "Python.agent_utils.executor": "Run",
&quot;Python.app (1).executor&quot;: &quot;Run&quot;, "Python.app (1).executor": "Run",
&quot;Python.app.executor&quot;: &quot;Run&quot;, "Python.app.executor": "Run",
&quot;Python.app2.executor&quot;: &quot;Run&quot;, "Python.app2.executor": "Run",
&quot;Python.app_test (1).executor&quot;: &quot;Run&quot;, "Python.app_test (1).executor": "Run",
&quot;Python.convert.executor&quot;: &quot;Run&quot;, "Python.convert.executor": "Run",
&quot;Python.converter_docling.executor&quot;: &quot;Run&quot;, "Python.converter_docling.executor": "Run",
&quot;Python.converter_mineru.executor&quot;: &quot;Run&quot;, "Python.converter_mineru.executor": "Run",
&quot;Python.markdown_splitter.executor&quot;: &quot;Debug&quot;, "Python.markdown_splitter.executor": "Debug",
&quot;Python.markdown_utils.executor&quot;: &quot;Run&quot;, "Python.markdown_utils.executor": "Run",
&quot;Python.test.executor&quot;: &quot;Run&quot;, "Python.test.executor": "Run",
&quot;Python.test1.executor&quot;: &quot;Run&quot;, "Python.test1.executor": "Run",
&quot;Python.test2.executor&quot;: &quot;Run&quot;, "Python.test2.executor": "Run",
&quot;Python.test3.executor&quot;: &quot;Run&quot;, "Python.test3.executor": "Run",
&quot;Python.test4.executor&quot;: &quot;Run&quot;, "Python.test4.executor": "Run",
&quot;Python.testhtml.executor&quot;: &quot;Run&quot;, "Python.testhtml.executor": "Run",
&quot;Python.translater.executor&quot;: &quot;Run&quot;, "Python.translater.executor": "Run",
&quot;Python.切分测试.executor&quot;: &quot;Run&quot;, "Python.切分测试.executor": "Run",
&quot;RunOnceActivity.ShowReadmeOnStart&quot;: &quot;true&quot;, "RunOnceActivity.ShowReadmeOnStart": "true",
&quot;RunOnceActivity.TerminalTabsStorage.copyFrom.TerminalArrangementManager&quot;: &quot;true&quot;, "RunOnceActivity.TerminalTabsStorage.copyFrom.TerminalArrangementManager": "true",
&quot;RunOnceActivity.git.unshallow&quot;: &quot;true&quot;, "RunOnceActivity.git.unshallow": "true",
&quot;git-widget-placeholder&quot;: &quot;main&quot;, "git-widget-placeholder": "dev",
&quot;last_opened_file_path&quot;: &quot;C:/Users/jxgm/Desktop/translate/docutranslate&quot;, "last_opened_file_path": "C:/Users/jxgm/Desktop/translate/docutranslate",
&quot;list.type.of.created.stylesheet&quot;: &quot;CSS&quot;, "list.type.of.created.stylesheet": "CSS",
&quot;node.js.detected.package.eslint&quot;: &quot;true&quot;, "node.js.detected.package.eslint": "true",
&quot;node.js.detected.package.tslint&quot;: &quot;true&quot;, "node.js.detected.package.tslint": "true",
&quot;node.js.selected.package.eslint&quot;: &quot;(autodetect)&quot;, "node.js.selected.package.eslint": "(autodetect)",
&quot;node.js.selected.package.tslint&quot;: &quot;(autodetect)&quot;, "node.js.selected.package.tslint": "(autodetect)",
&quot;nodejs_package_manager_path&quot;: &quot;npm&quot;, "nodejs_package_manager_path": "npm",
&quot;settings.editor.selected.configurable&quot;: &quot;preferences.pluginManager&quot;, "settings.editor.selected.configurable": "preferences.pluginManager",
&quot;vue.rearranger.settings.migration&quot;: &quot;true&quot; "vue.rearranger.settings.migration": "true"
} }
}</component> }]]></component>
<component name="RecentsManager"> <component name="RecentsManager">
<key name="CopyFile.RECENT_KEYS"> <key name="CopyFile.RECENT_KEYS">
<recent name="C:\Users\jxgm\Desktop\translate\docutranslate" /> <recent name="C:\Users\jxgm\Desktop\translate\docutranslate" />
@@ -491,8 +496,8 @@
<recent_temporary> <recent_temporary>
<list> <list>
<item itemvalue="Python.app_test (1)" /> <item itemvalue="Python.app_test (1)" />
<item itemvalue="JavaScript 调试.test2.html" />
<item itemvalue="Python.3testhtml" /> <item itemvalue="Python.3testhtml" />
<item itemvalue="JavaScript 调试.test2.html" />
<item itemvalue="Python.translater" /> <item itemvalue="Python.translater" />
<item itemvalue="Python.2test2 (1)" /> <item itemvalue="Python.2test2 (1)" />
</list> </list>
@@ -592,7 +597,8 @@
<workItem from="1748186982208" duration="3003000" /> <workItem from="1748186982208" duration="3003000" />
<workItem from="1748220705934" duration="6440000" /> <workItem from="1748220705934" duration="6440000" />
<workItem from="1748236163786" duration="323000" /> <workItem from="1748236163786" duration="323000" />
<workItem from="1748353819402" duration="220000" /> <workItem from="1748353819402" duration="2973000" />
<workItem from="1748360287873" duration="6961000" />
</task> </task>
<servers /> <servers />
</component> </component>
@@ -610,26 +616,15 @@
</map> </map>
</option> </option>
</component> </component>
<component name="XDebuggerManager">
<breakpoint-manager>
<breakpoints>
<line-breakpoint enabled="true" suspend="THREAD" type="python-line">
<url>file://$PROJECT_DIR$/docutranslate/translater.py</url>
<line>241</line>
<option name="timeStamp" value="2" />
</line-breakpoint>
</breakpoints>
</breakpoint-manager>
</component>
<component name="com.intellij.coverage.CoverageDataManagerImpl"> <component name="com.intellij.coverage.CoverageDataManagerImpl">
<SUITE FILE_PATH="coverage/filetranslate$app_test__1_.coverage" NAME="app_test (1) 覆盖结果" MODIFIED="1748353944846" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/tests" /> <SUITE FILE_PATH="coverage/filetranslate$app_test__1_.coverage" NAME="app_test (1) 覆盖结果" MODIFIED="1748398439791" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/tests" />
<SUITE FILE_PATH="coverage/filetranslate$test.coverage" NAME="test 覆盖结果" MODIFIED="1747472297913" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/tests" /> <SUITE FILE_PATH="coverage/filetranslate$test.coverage" NAME="test 覆盖结果" MODIFIED="1747472297913" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/tests" />
<SUITE FILE_PATH="coverage/filetranslate$convert.coverage" NAME="convert 覆盖结果" MODIFIED="1746963490689" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/docutranslate/utils" /> <SUITE FILE_PATH="coverage/filetranslate$convert.coverage" NAME="convert 覆盖结果" MODIFIED="1746963490689" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/docutranslate/utils" />
<SUITE FILE_PATH="coverage/PDFtranslate$PDFtranslater__1_.coverage" NAME="PDFtranslater (1) 覆盖结果" MODIFIED="1746633258205" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/pdftranslate_packages" /> <SUITE FILE_PATH="coverage/PDFtranslate$PDFtranslater__1_.coverage" NAME="PDFtranslater (1) 覆盖结果" MODIFIED="1746633258205" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/pdftranslate_packages" />
<SUITE FILE_PATH="coverage/filetranslate$pytest__test_html_py__.coverage" NAME="pytest (test_html.py 内) 覆盖结果" MODIFIED="1747554037236" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/tests" /> <SUITE FILE_PATH="coverage/filetranslate$pytest__test_html_py__.coverage" NAME="pytest (test_html.py 内) 覆盖结果" MODIFIED="1747554037236" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/tests" />
<SUITE FILE_PATH="coverage/PDFtranslate$agent_utils.coverage" NAME="agent_utils 覆盖结果" MODIFIED="1746617703678" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/pdftranslate_packages/utils" /> <SUITE FILE_PATH="coverage/PDFtranslate$agent_utils.coverage" NAME="agent_utils 覆盖结果" MODIFIED="1746617703678" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/pdftranslate_packages/utils" />
<SUITE FILE_PATH="coverage/filetranslate$app2.coverage" NAME="app2 覆盖结果" MODIFIED="1747108180309" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/docutranslate" /> <SUITE FILE_PATH="coverage/filetranslate$app2.coverage" NAME="app2 覆盖结果" MODIFIED="1747108180309" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/docutranslate" />
<SUITE FILE_PATH="coverage/filetranslate$3testhtml.coverage" NAME="3testhtml 覆盖结果" MODIFIED="1748261775567" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/tests" /> <SUITE FILE_PATH="coverage/filetranslate$3testhtml.coverage" NAME="3testhtml 覆盖结果" MODIFIED="1748361746296" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/tests" />
<SUITE FILE_PATH="coverage/filetranslate$app.coverage" NAME="app 覆盖结果" MODIFIED="1747448464521" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/docutranslate" /> <SUITE FILE_PATH="coverage/filetranslate$app.coverage" NAME="app 覆盖结果" MODIFIED="1747448464521" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/docutranslate" />
<SUITE FILE_PATH="coverage/PDFtranslate$markdown_splitter.coverage" NAME="markdown_splitter 覆盖结果" MODIFIED="1746599883603" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/pdftranslate_packages/utils" /> <SUITE FILE_PATH="coverage/PDFtranslate$markdown_splitter.coverage" NAME="markdown_splitter 覆盖结果" MODIFIED="1746599883603" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/pdftranslate_packages/utils" />
<SUITE FILE_PATH="coverage/filetranslate$converter_mineru.coverage" NAME="converter_mineru 覆盖结果" MODIFIED="1747726229881" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/docutranslate/converter" /> <SUITE FILE_PATH="coverage/filetranslate$converter_mineru.coverage" NAME="converter_mineru 覆盖结果" MODIFIED="1747726229881" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="false" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$/docutranslate/converter" />

View File

@@ -1,4 +1,4 @@
__version__="0.2.28" __version__="0.2.29"

View File

@@ -47,7 +47,6 @@ class Agent:
self.client_async = httpx.AsyncClient() self.client_async = httpx.AsyncClient()
self.max_concurrent = max_concurrent self.max_concurrent = max_concurrent
self.timeout = timeout self.timeout = timeout
def _prepare_request_data(self, prompt: str, system_prompt: str, temperature=None, top_p=0.9): def _prepare_request_data(self, prompt: str, system_prompt: str, temperature=None, top_p=0.9):
if temperature is None: if temperature is None:
temperature = self.temperature temperature = self.temperature
@@ -97,6 +96,7 @@ class Agent:
system_prompt: str | None = None, system_prompt: str | None = None,
max_concurrent: int | None = None # 新增参数默认并发数为5 max_concurrent: int | None = None # 新增参数默认并发数为5
) -> list[str]: ) -> list[str]:
print(f"system_prompt:{self.system_prompt}")
max_concurrent = self.max_concurrent if max_concurrent is None else max_concurrent max_concurrent = self.max_concurrent if max_concurrent is None else max_concurrent
total = len(prompts) total = len(prompts)
count = 0 count = 0

View File

@@ -2,8 +2,9 @@ from typing import Unpack
from .agent import (Agent, AgentArgs) from .agent import (Agent, AgentArgs)
class MDRefineAgent(Agent): class MDRefineAgent(Agent):
def __init__(self,**kwargs:Unpack[AgentArgs]): def __init__(self, custom_prompt=None, **kwargs: Unpack[AgentArgs]):
super().__init__(**kwargs) super().__init__(**kwargs)
self.system_prompt = r""" self.system_prompt = r"""
# 角色 # 角色
@@ -35,12 +36,14 @@ code、latex和HTML保持结构
{c_0,c_1,c^2}是一个集合 {c_0,c_1,c^2}是一个集合
输出: 输出:
一道题目<ph-12asd2>:$c_0+1=2$$c_0$等于几 一道题目<ph-12asd2>:$c_0+1=2$$c_0$等于几
{$c_0$,$c_1$,$c^2$}是一个集合 {$c_0$,$c_1$,$c^2$}是一个集合"""
\no_think""" if custom_prompt:
self.system_prompt += "\n#其余要求(可能为背景或指令)\n" + custom_prompt
self.system_prompt+=r'\no_think'
class MDTranslateAgent(Agent): class MDTranslateAgent(Agent):
def __init__(self,to_lang="中文",**kwargs:Unpack[AgentArgs]): def __init__(self, custom_prompt=None, to_lang="中文", **kwargs: Unpack[AgentArgs]):
super().__init__(**kwargs) super().__init__(**kwargs)
self.system_prompt = f""" self.system_prompt = f"""
# 角色 # 角色
@@ -79,5 +82,7 @@ The equation is E=mc 2. This is famous.
volume 99, pages 173186, 1999. volume 99, pages 173186, 1999.
输出:【文献引用保持源语言】 输出:【文献引用保持源语言】
[2] M. Castro, B. Liskov, et al. Practical byzantine fault tolerance. In OSDI, [2] M. Castro, B. Liskov, et al. Practical byzantine fault tolerance. In OSDI,
volume 99, pages 173186, 1999. volume 99, pages 173186, 1999."""
\\no_think""" if custom_prompt:
self.system_prompt += "\n#其余要求(可能为背景或指令)\n" + custom_prompt
self.system_prompt+=r'\no_think'

View File

@@ -127,6 +127,7 @@ async def _perform_translation(params: Dict[str, Any], file_contents: bytes, ori
to_lang=params['to_lang'], to_lang=params['to_lang'],
formula=params['formula_ocr'], formula=params['formula_ocr'],
code=params['code_ocr'], code=params['code_ocr'],
custom_prompt_translate=params['custom_prompt_translate'],
refine=params['refine_markdown'], refine=params['refine_markdown'],
save=False save=False
) )
@@ -201,7 +202,6 @@ async def main_page(request: Request):
@app.post("/translate") @app.post("/translate")
async def handle_translate( async def handle_translate(
request: Request, # Added request for potential future use, not strictly needed now
base_url: str = Form(...), base_url: str = Form(...),
apikey: str = Form(...), apikey: str = Form(...),
model_id: str = Form(...), model_id: str = Form(...),
@@ -209,8 +209,9 @@ async def handle_translate(
formula_ocr: bool = Form(False), formula_ocr: bool = Form(False),
code_ocr: bool = Form(False), code_ocr: bool = Form(False),
refine_markdown: bool = Form(False), refine_markdown: bool = Form(False),
convert_engin: str = Form(...), # New parameter convert_engin: str = Form(...),
mineru_token: Optional[str] = Form(None), # New parameter mineru_token: Optional[str] = Form(None),
custom_prompt_translate: Optional[str] = Form(None),
file: UploadFile = File(...) file: UploadFile = File(...)
): ):
global current_state, log_queue, log_history global current_state, log_queue, log_history
@@ -276,8 +277,9 @@ async def handle_translate(
"base_url": base_url, "apikey": apikey, "model_id": model_id, "base_url": base_url, "apikey": apikey, "model_id": model_id,
"to_lang": to_lang, "formula_ocr": formula_ocr, "to_lang": to_lang, "formula_ocr": formula_ocr,
"code_ocr": code_ocr, "refine_markdown": refine_markdown, "code_ocr": code_ocr, "refine_markdown": refine_markdown,
"convert_engin": convert_engin, # Pass to task "convert_engin": convert_engin,
"mineru_token": mineru_token, # Pass to task "mineru_token": mineru_token,
"custom_prompt_translate":custom_prompt_translate,
} }
loop = asyncio.get_running_loop() loop = asyncio.get_running_loop()

View File

@@ -22,6 +22,14 @@
padding-bottom: 0.5rem; padding-bottom: 0.5rem;
} }
.prompt-area {
resize: vertical;
width: 100%;
min-height: 200px;
font-size: 0.9em;
box-sizing: border-box;
display: block;
}
.log-area { .log-area {
background-color: #f5f5f5; background-color: #f5f5f5;
@@ -398,6 +406,14 @@
<input type="text" id="model_id" name="model_id" placeholder="模型id" required> <input type="text" id="model_id" name="model_id" placeholder="模型id" required>
</div> </div>
</details> </details>
<details>
<summary>系统提示(可选)</summary>
<div class="form-group">
<label for="custom_prompt_translate"></label>
<textarea class="prompt-area" type="text" id="custom_prompt_translate"
name="custom_prompt_translate" placeholder="翻译提示"></textarea>
</div>
</details>
<button type="submit" id="submitButton" class="primary">开始翻译</button> <button type="submit" id="submitButton" class="primary">开始翻译</button>
</form> </form>
<div id="resultArea"> <div id="resultArea">

View File

@@ -76,6 +76,12 @@ class FileTranslater:
} }
return result return result
def default_refine_agent(self, custom_prompt=None) -> MDRefineAgent:
return MDRefineAgent(custom_prompt=custom_prompt, **self._default_agent_params())
def default_translate_agent(self, custom_prompt=None, to_lang="中文") -> MDTranslateAgent:
return MDTranslateAgent(custom_prompt=custom_prompt, to_lang=to_lang, **self._default_agent_params())
def _convert2markdown(self, document: Document, formula: bool, code: bool, artifact: Path = None) -> str: def _convert2markdown(self, document: Document, formula: bool, code: bool, artifact: Path = None) -> str:
translater_logger.info(f"正在使用{self.convert_engin}转换文件为markdown") translater_logger.info(f"正在使用{self.convert_engin}转换文件为markdown")
if self.convert_engin == "docling": if self.convert_engin == "docling":
@@ -202,48 +208,49 @@ class FileTranslater:
self.save_as_markdown(filename=f"{file_path.stem}.md") self.save_as_markdown(filename=f"{file_path.stem}.md")
return self return self
def refine_markdown_by_agent(self, refine_agent: Agent | None = None) -> str: def refine_markdown_by_agent(self, refine_agent: Agent | None = None, custom_prompt=None) -> str:
translater_logger.info("正在修正markdown") translater_logger.info("正在修正markdown")
self._mask_uris_in_markdown() self._mask_uris_in_markdown()
chuncks = self._split_markdown_into_chunks() chuncks = self._split_markdown_into_chunks()
if refine_agent is None: if refine_agent is None:
refine_agent = MDRefineAgent(**self._default_agent_params()) refine_agent = self.default_refine_agent(custom_prompt)
result: list[str] = refine_agent.send_prompts(chuncks) result: list[str] = refine_agent.send_prompts(chuncks)
self.markdown = join_markdown_texts(result) self.markdown = join_markdown_texts(result)
self._unmask_uris_in_markdown() self._unmask_uris_in_markdown()
translater_logger.info("markdown已修正") translater_logger.info("markdown已修正")
return self.markdown return self.markdown
def translate_markdown_by_agent(self, translate_agent: Agent | None = None, to_lang="中文"): def translate_markdown_by_agent(self, translate_agent: Agent | None = None, to_lang="中文", custom_prompt=None):
translater_logger.info("正在翻译markdown") translater_logger.info("正在翻译markdown")
self._mask_uris_in_markdown() self._mask_uris_in_markdown()
chuncks = self._split_markdown_into_chunks() chuncks = self._split_markdown_into_chunks()
if translate_agent is None: if translate_agent is None:
translate_agent = MDTranslateAgent(to_lang=to_lang, **self._default_agent_params()) translate_agent = self.default_translate_agent(custom_prompt=custom_prompt, to_lang=to_lang)
result: list[str] = translate_agent.send_prompts(chuncks) result: list[str] = translate_agent.send_prompts(chuncks)
self.markdown = join_markdown_texts(result) self.markdown = join_markdown_texts(result)
self._unmask_uris_in_markdown() self._unmask_uris_in_markdown()
translater_logger.info("翻译完成") translater_logger.info("翻译完成")
return self.markdown return self.markdown
async def refine_markdown_by_agent_async(self, refine_agent: Agent | None = None) -> str: async def refine_markdown_by_agent_async(self, refine_agent: Agent | None = None, custom_prompt=None) -> str:
translater_logger.info("正在修正markdown") translater_logger.info("正在修正markdown")
self._mask_uris_in_markdown() self._mask_uris_in_markdown()
chuncks = self._split_markdown_into_chunks() chuncks = self._split_markdown_into_chunks()
if refine_agent is None: if refine_agent is None:
refine_agent = MDRefineAgent(**self._default_agent_params()) refine_agent = self.default_refine_agent(custom_prompt=custom_prompt)
result: list[str] = await refine_agent.send_prompts_async(chuncks) result: list[str] = await refine_agent.send_prompts_async(chuncks)
self.markdown = join_markdown_texts(result) self.markdown = join_markdown_texts(result)
self._unmask_uris_in_markdown() self._unmask_uris_in_markdown()
translater_logger.info("markdown已修正") translater_logger.info("markdown已修正")
return self.markdown return self.markdown
async def translate_markdown_by_agent_async(self, translate_agent: Agent | None = None, to_lang="中文"): async def translate_markdown_by_agent_async(self, translate_agent: Agent | None = None, to_lang="中文",
custom_prompt=None):
translater_logger.info("正在翻译markdown") translater_logger.info("正在翻译markdown")
self._mask_uris_in_markdown() self._mask_uris_in_markdown()
chuncks = self._split_markdown_into_chunks() chuncks = self._split_markdown_into_chunks()
if translate_agent is None: if translate_agent is None:
translate_agent = MDTranslateAgent(to_lang=to_lang, **self._default_agent_params()) translate_agent = self.default_translate_agent(to_lang=to_lang, custom_prompt=custom_prompt)
result: list[str] = await translate_agent.send_prompts_async(chuncks) result: list[str] = await translate_agent.send_prompts_async(chuncks)
self.markdown = join_markdown_texts(result) self.markdown = join_markdown_texts(result)
self._unmask_uris_in_markdown() self._unmask_uris_in_markdown()
@@ -318,7 +325,9 @@ class FileTranslater:
def translate_file(self, file_path: Path | str | None = None, to_lang="中文", output_dir="./output", def translate_file(self, file_path: Path | str | None = None, to_lang="中文", output_dir="./output",
formula=True, formula=True,
code=True, output_format: Literal["markdown", "html"] = "markdown", refine=False, code=True, output_format: Literal["markdown", "html"] = "markdown", refine=False,
refine_agent: Agent | None = None, translate_agent: Agent | None = None, save=True): custom_prompt_translate=None, refine_agent: Agent | None = None,
translate_agent: Agent | None = None,
save=True):
if file_path is None: if file_path is None:
assert self.file_path is not None, "未输入文件路径" assert self.file_path is not None, "未输入文件路径"
file_path = self.file_path file_path = self.file_path
@@ -327,7 +336,7 @@ class FileTranslater:
self.read_file(file_path, formula=formula, code=code) self.read_file(file_path, formula=formula, code=code)
if refine: if refine:
self.refine_markdown_by_agent(refine_agent) self.refine_markdown_by_agent(refine_agent)
self.translate_markdown_by_agent(translate_agent, to_lang=to_lang) self.translate_markdown_by_agent(translate_agent, to_lang=to_lang, custom_prompt=custom_prompt_translate)
if save: if save:
if output_format == "markdown": if output_format == "markdown":
filename = f"{file_path.stem}_{to_lang}.md" filename = f"{file_path.stem}_{to_lang}.md"
@@ -339,7 +348,8 @@ class FileTranslater:
async def translate_file_async(self, file_path: Path | str | None = None, to_lang="中文", output_dir="./output", async def translate_file_async(self, file_path: Path | str | None = None, to_lang="中文", output_dir="./output",
formula=True, formula=True,
code=True, output_format: Literal["markdown", "html"] = "markdown", refine=False, code=True, output_format: Literal["markdown", "html"] = "markdown",
custom_prompt_translate=None, refine=False,
refine_agent: Agent | None = None, translate_agent: Agent | None = None, save=True): refine_agent: Agent | None = None, translate_agent: Agent | None = None, save=True):
if file_path is None: if file_path is None:
assert self.file_path is not None, "未输入文件路径" assert self.file_path is not None, "未输入文件路径"
@@ -354,7 +364,8 @@ class FileTranslater:
) )
if refine: if refine:
await self.refine_markdown_by_agent_async(refine_agent) await self.refine_markdown_by_agent_async(refine_agent)
await self.translate_markdown_by_agent_async(translate_agent, to_lang=to_lang) await self.translate_markdown_by_agent_async(translate_agent, to_lang=to_lang,
custom_prompt=custom_prompt_translate)
if save: if save:
if output_format == "markdown": if output_format == "markdown":
filename = f"{file_path.stem}_{to_lang}.md" filename = f"{file_path.stem}_{to_lang}.md"
@@ -366,12 +377,14 @@ class FileTranslater:
def translate_bytes(self, name: str, file: bytes, to_lang="中文", output_dir="./output", def translate_bytes(self, name: str, file: bytes, to_lang="中文", output_dir="./output",
formula=True, formula=True,
code=True, output_format: Literal["markdown", "html"] = "markdown", refine=False, code=True, output_format: Literal["markdown", "html"] = "markdown",
custom_prompt_translate=None,
refine=False,
refine_agent: Agent | None = None, translate_agent: Agent | None = None, save=True): refine_agent: Agent | None = None, translate_agent: Agent | None = None, save=True):
self.read_bytes(name=name, file=file, formula=formula, code=code) self.read_bytes(name=name, file=file, formula=formula, code=code)
if refine: if refine:
self.refine_markdown_by_agent(refine_agent) self.refine_markdown_by_agent(refine_agent)
self.translate_markdown_by_agent(translate_agent, to_lang=to_lang) self.translate_markdown_by_agent(translate_agent, to_lang=to_lang, custom_prompt=custom_prompt_translate)
if save: if save:
if output_format == "markdown": if output_format == "markdown":
filename = f"{name}_{to_lang}.md" filename = f"{name}_{to_lang}.md"
@@ -383,13 +396,15 @@ class FileTranslater:
async def translate_bytes_async(self, name: str, file: bytes, to_lang="中文", output_dir="./output", async def translate_bytes_async(self, name: str, file: bytes, to_lang="中文", output_dir="./output",
formula=True, formula=True,
code=True, output_format: Literal["markdown", "html"] = "markdown", refine=False, code=True, output_format: Literal["markdown", "html"] = "markdown",
custom_prompt_translate=None, refine=False,
refine_agent: Agent | None = None, translate_agent: Agent | None = None, save=True): refine_agent: Agent | None = None, translate_agent: Agent | None = None, save=True):
await self.read_bytes_async(name=name, file=file, formula=formula, code=code) await self.read_bytes_async(name=name, file=file, formula=formula, code=code)
if refine: if refine:
await self.refine_markdown_by_agent_async(refine_agent) await self.refine_markdown_by_agent_async(refine_agent)
await self.translate_markdown_by_agent_async(translate_agent, to_lang=to_lang) await self.translate_markdown_by_agent_async(translate_agent, to_lang=to_lang,
custom_prompt=custom_prompt_translate)
if save: if save:
if output_format == "markdown": if output_format == "markdown":
filename = f"{name}_{to_lang}.md" filename = f"{name}_{to_lang}.md"