More info for words within detected tables -- Feature Request

[ghost]

Table.extract() returns a matrix of strings corresponding to the table cells. It would be useful to get the individual word bounding boxes (through an option) like what Page.extract_words() returns. Is this already available?

Table.extract() returns the following:
    [
        ['r1 c1', 'r1 c2', 'r1 c3'],
        ['r2 c1', 'r2 c2', 'r2 c3'],
    ]

Table.extract(word_boxes=True) could return those 12 words in the exact same format as Page.extract_words() does.

It's possible to get the page words and map them to cells. But the page words are sometimes merged ignoring column boundaries as reported here.

[jsvine]

I agree! That could be a very nice feature. I will consider adding it in the future. In the meantime, PRs are welcome on this.

[andlike]

@greddyatpt 解决了没有

[ghost]

Sorry, I deal with Chinese PDFs but don't know Chinese. Google says 解决了没有 means "Solved". Are you saying the feature is implemented?

[zhushiyuan-star]

抱歉，我处理中文PDF但不懂中文。Google说解决了没有的意思是“已解决”。您是说该功能已实现吗？

Have you implemented this function? Make the text extracted from the table contain the following information: text、fontname 、size、adv 、upright 、height 、width 、x0、x1、 y0 、y1、top 、bottom、doctop 、object_type

[wind-chh]

Sorry, I deal with Chinese PDFs but don't know Chinese. Google says 解决了没有 means "Solved". Are you saying the feature is implemented?

“解决了没有” = "解决了没有?" = “Has this feature been implemented?”

[wind-chh]

抱歉，我处理中文PDF但不懂中文。Google说解决了没有的意思是“已解决”。您是说该功能已实现吗？

Have you implemented this function? Make the text extracted from the table contain the following information: text、fontname 、size、adv 、upright 、height 、width 、x0、x1、 y0 、y1、top 、bottom、doctop 、object_type

these informations (text, fontname, size, ...) associated with chars of page. Page.extract_table() returns only text information, you can make a slight modify to return chars which contains all these informations.