允许翻译自订的选取范围

[charles7668]

功能描述

目前的功能为自动判别文字区块 , 不好微调, 也许可以加个手动选取范围的模式

例如以 #323 的pdf为例, 可以进行如下选择以避免将行号给包含进去

同时此作法也更容易支持图片的翻译, 由于范围可以自订, 在处理图片原有文字时会比较容易

[aseaday]

There are two steps to fix that problem:

• PDFMathTranslate highly depends on the Layout Analysis from pdfminer, There may be something wrong to correctly recognize the serial number and block text even we give a fitler box. The serial number can be grouped into a textline in some cases.
• A better GUI may be a custome web app.