Leon/docutranslate
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

优化提示词

Browse Source
This commit is contained in:
xunbu
2025-08-04 19:19:09 +08:00
parent 01b5d7c0cb
commit 24f9338057
2 changed files with 9 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
docutranslate/agents/segments_agent.py
View File

@@ -1,5 +1,6 @@
import json import json
from dataclasses import dataclass from dataclasses import dataclass
from json import JSONDecodeError
from docutranslate.agents import AgentConfig, Agent from docutranslate.agents import AgentConfig, Agent
from docutranslate.utils.json_utils import flat_json_split from docutranslate.utils.json_utils import flat_json_split
@@ -27,7 +28,7 @@ class SegmentsTranslateAgent(Agent):
翻译后的片段应该与源格式尽量相同 翻译后的片段应该与源格式尽量相同
如果待翻译片段已经是目标语言，则保持原样 如果待翻译片段已经是目标语言，则保持原样
# 输出 # 输出
翻译后的片段序列，以json格式表示。其中键是片段编号，值是翻译后的片段 翻译后的片段序列，以json文本表示（文本而非代码块）。其中键是片段编号，值是翻译后的片段
# 示例 # 示例
## 输入 ## 输入
{r'{"0":"hello","1":"apple","2":true,"3":"false"}'} {r'{"0":"hello","1":"apple","2":true,"3":"false"}'}
@@ -56,6 +57,10 @@ class SegmentsTranslateAgent(Agent):
translated_chunks = await super().send_prompts_async(prompts=prompts) translated_chunks = await super().send_prompts_async(prompts=prompts)
indexed_translated = indexed_originals.copy() indexed_translated = indexed_originals.copy()
for chunk_str in translated_chunks: for chunk_str in translated_chunks:
try:
translated_part = json.loads(chunk_str) translated_part = json.loads(chunk_str)
indexed_translated.update(translated_part) indexed_translated.update(translated_part)
except JSONDecodeError as e:
self.logger.info(f"json解析错误，解析文本:{chunk_str}，错误:{e.__repr__()}")
return list(indexed_translated.values()) return list(indexed_translated.values())

3
docutranslate/exporter/xlsx/xlsx2html_exporter.py
View File

@@ -52,11 +52,10 @@ class Xlsx2HTMLExporter(XlsxExporter):
html_template = resource_path("template/xlsx.html").read_text(encoding="utf-8") html_template = resource_path("template/xlsx.html").read_text(encoding="utf-8")
pico = f'<style>{resource_path("static/pico.css").read_text(encoding="utf-8")}</style>' if not cdn else r'<link rel="stylesheet" href="https://s4.zstatic.net/ajax/libs/picocss/2.1.1/pico.min.css" integrity="sha512-+4kjFgVD0n6H3xt19Ox84B56MoS7srFn60tgdWFuO4hemtjhySKyW4LnftYZn46k3THUEiTTsbVjrHai+0MOFw==" crossorigin="anonymous" referrerpolicy="no-referrer" />' pico = f'<style>{resource_path("static/pico.css").read_text(encoding="utf-8")}</style>' if not self.cdn else r'<link rel="stylesheet" href="https://s4.zstatic.net/ajax/libs/picocss/2.1.1/pico.min.css" integrity="sha512-+4kjFgVD0n6H3xt19Ox84B56MoS7srFn60tgdWFuO4hemtjhySKyW4LnftYZn46k3THUEiTTsbVjrHai+0MOFw==" crossorigin="anonymous" referrerpolicy="no-referrer" />'
render = jinja2.Template(html_template).render( render = jinja2.Template(html_template).render(
title=document.stem, title=document.stem,
pico=pico, pico=pico,
body=table, body=table,
) )
print("\n通过openpyxl手动生成了 output_manual.html 文件！")
return Document.from_bytes(content=render.encode("utf-8"), suffix=".html", stem=document.stem) return Document.from_bytes(content=render.encode("utf-8"), suffix=".html", stem=document.stem)