翻译无效

[wx-11]

问题描述

使用测试站点/以及自建服务 谷歌/bing/openai密钥 没效果 而且gui没有预览

测试文档

033-Teacher efficacy_ its meaning and measure.pdf

[wx-11]

感觉有点奇怪 翻译文档不是需要ocr吗 这些翻译服务都具备ocr了吗
不太懂技术 没看源码 所以顺便问一句原理 有点好奇
不过oai的自带ocr不知道为啥也不行

[Byaidu]

翻译文档不需要ocr

[wx-11]

日志 感觉没什么异常

pdf2zh-1  | * Running on local URL:  http://0.0.0.0:7860
pdf2zh-1  | 
pdf2zh-1  | To create a public link, set `share=True` in `launch()`.
pdf2zh-1  | Files before translation: ['033-Teacher efficacy_ its meaning and measure.pdf']
pdf2zh-1  | {'files': ['pdf2zh_files/033-Teacher efficacy_ its meaning and measure.pdf'], 'pages': None, 'lang_in': 'en', 'lang_out': 'zh', 'service': 'google', 'output': PosixPath('pdf2zh_files'), 'thread': 4, 'callback': <function translate_file.<locals>.progress_bar at 0x7fefb0385d00>}
100%|██████████| 48/48 [00:30<00:00,  1.60it/s]
pdf2zh-1  | Files after translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']
pdf2zh-1  | Files before translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']
pdf2zh-1  | {'files': ['pdf2zh_files/033-Teacher efficacy_ its meaning and measure.pdf'], 'pages': None, 'lang_in': 'en', 'lang_out': 'zh', 'service': 'openai', 'output': PosixPath('pdf2zh_files'), 'thread': 4, 'callback': <function translate_file.<locals>.progress_bar at 0x7fefb0687b00>}
100%|██████████| 48/48 [00:36<00:00,  1.33it/s]
pdf2zh-1  | Files after translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']
pdf2zh-1  | Files before translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']
pdf2zh-1  | {'files': ['pdf2zh_files/033-Teacher efficacy_ its meaning and measure.pdf'], 'pages': None, 'lang_in': 'en', 'lang_out': 'zh', 'service': 'google', 'output': PosixPath('pdf2zh_files'), 'thread': 4, 'callback': <function translate_file.<locals>.progress_bar at 0x7fefb04d47c0>}
100%|██████████| 48/48 [00:30<00:00,  1.58it/s]
pdf2zh-1  | Files after translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']
pdf2zh-1  | Files before translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']
pdf2zh-1  | {'files': ['pdf2zh_files/033-Teacher efficacy_ its meaning and measure.pdf'], 'pages': None, 'lang_in': 'en', 'lang_out': 'zh', 'service': 'bing', 'output': PosixPath('pdf2zh_files'), 'thread': 4, 'callback': <function translate_file.<locals>.progress_bar at 0x7fefb0687ec0>}
100%|██████████| 48/48 [00:35<00:00,  1.35it/s]
pdf2zh-1  | Files after translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']
pdf2zh-1  | Files before translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']
pdf2zh-1  | {'files': ['pdf2zh_files/033-Teacher efficacy_ its meaning and measure.pdf'], 'pages': None, 'lang_in': 'en', 'lang_out': 'zh', 'service': 'bing', 'output': PosixPath('pdf2zh_files'), 'thread': 4, 'callback': <function translate_file.<locals>.progress_bar at 0x7fefb02cefc0>}
100%|██████████| 48/48 [00:32<00:00,  1.47it/s]
pdf2zh-1  | Files after translation: ['033-Teacher efficacy_ its meaning and measure.pdf', '033-Teacher efficacy_ its meaning and measure-dual.pdf', '033-Teacher efficacy_ its meaning and measure-mono.pdf']

[wx-11]

翻译文档不需要ocr

哦哦 好的谢谢 之前见过PDF文字不能复制只能ocr 还以为是内嵌图片的类型文字

那么没用到ocr 是不是说明图片类型的文档无法翻译? 后续有考虑加一个ocr的可选菜单(只在支持ocr的翻译引擎里使用比如oai)吗

[hellofinch]

#19
OCR WIP

[wx-11]

好的谢谢 请问你们可以复现我的PDF问题吗

#19 OCR WIP

[hellofinch]

可以，后边的那几页都是扫描件，暂时确实处理不了。

[wx-11]

啊 我是一个字都翻译不出来呀 难道是我部署有问题? 但是日志也没有报错 请问要怎么抓log

[hellofinch]

准确描述应该是我这里也没有出结果，但是可以看出来，后边论文正文都是扫描件。

[wx-11]

前面不是扫描也没反应 而且预览不出来 在示例站点用谷歌也不行

[Byaidu]

一眼扫描件，别试了

[hellofinch]

第一页不是扫描件，确实没出来很奇怪。

[hellofinch]

后边那些都是扫描件，第一页上有什么东西做了特别判断吗？

[wx-11]

找到问题了 抓包一下 根本没有请求api接口 也就是第一页不是扫描也没有被当做文本翻译

[Byaidu]

是这样，因为第一段根本没有正文……

[wx-11]

请问为什么关闭issue 前两页的文字没有翻译 这个是已知问题吗

因为第一段根本没有正文

有地方配置变量关闭只翻译正文的选项吗 我的理解是全文能翻译应该就翻译 , 为什么不处理非正文部分呢

[Byaidu]

没有，公式部分和装饰部分显然是不应该处理的，所以为什么不处理非正文部分呢

[wx-11]

好的 明白了 我并没有恶意 只是当做找bug反馈角度提问 没有责怪的意思 别误会 对于开源项目我是很感激的 只是想知道原因 那不打扰了