How to parse Simple Chinese PDF

[wanghang-github]

How to parse Simple Chinese PDF

[little-polka-dot]

简体中文pdf解析有什么问题吗😊

[wanghang-github]

是这样，我想解析一些中文论文，文章存在一些页眉页脚和图片部分的内容

在解析出来之后出现一些比较乱的英文字母，

我又应该如何根据字体识别出来各个章节呢。

哈哈哈，同胞啊

[little-polka-dot]

识别各个章节要么通过坐标计算的方式，要么通过视觉模型识别区域，如果你是想过滤掉页眉页脚的话，只能先对整页进行视觉版面识别。pdf本身并没有这些信息。

[little-polka-dot]

我的工作刚好是做这个的

[little-polka-dot]

至于图片下的乱码，可以获取图片坐标与seqno，获取文本的坐标和seqno，然后判断文本是不是被图片覆盖了，如果覆盖了则是不可见文本。

[little-polka-dot]

get_texttrace获取文本，get_bboxlog获取图片坐标

[little-polka-dot]

计算是否是不可见文本有以下规则
Invisibility may happen for a number of reasons:
1) text covered by something else (image, vector graphics)
2) text has same color as background (like black on black)
3) text is under control of rendering mode 3 = 3 Tr.
This is very often used by OCR: recognized text is written to the page with this rendering mode -
at the same spot where the text in the image was found, thus making it selectable / searchable.
但是目前我发现了一个疑似bug，就是get_texttrace返回的坐标不对，作者还没有回复

[JorjMcKie]

I am not sure what to make out of this post.
After using a Chinese-English translator, it appears however, that we have a typical "Discussions" thread here.
May I also ask for sticking with English as the lingua franca for IT?

[little-polka-dot]

😀😀As we all know, the vast majority of people in mainland China are not very good at English.

[wanghang-github]

Yes,I'm not good at English.Can I get yout wechat,you could add me Wh19956486056

[little-polka-dot]

Me? 17721441649