Unwanted vertical lines appear

[yzt2024]

Hello everyone,
I'm trying to export the table from the following pdf, but there are vertical lines that doesn't exist in the original text when the table is divided. I tried to modify snap_tolerance but did not get the expected division result. My parameters are as follows:

table_setting={
"vertical_strategy": "lines",
"horizontal_strategy": "lines",
"snap_tolerance": 3,
"snap_tolerance": 3,
"join_tolerance": 3,
"edge_min_length": 3,
"min_words_vertical": 20,
"min_words_horizontal": 3,
"text_tolerance": 3,
"intersection_tolerance": 1,
}

new.pdf

if i use a snap_tolerance=4,then the division of the left boundary will be problematic.

So what am I supposed to do? I see from the discussion that boundaries can be determined manually by use the "explicit_vertical_lines", I haven't tried yet. I want to know if this parameter needs to specify all vertical lines, because I actually only need to modify the leftmost boundary, can I modify only one line and the other vertical lines are still automatically generated? What should I do? Thank you very much!

[jsvine]

I would suggest examining the vertical lines / rects (page.lines, page.rects) to try to determine what properties are unique to the objects you want to exclude, and then using filtered = page.filter(fn) to create a filtered version of the page before extracting. There are a few other discussions that show examples of this, including this recent one: #1219 (reply in thread)

[yzt2024]

Thank you very much! I'm checking all the rects in the picture,I found that the first cell of all tables contains three rects:

then i use the following code to delete the first rects and extend the width of the second rect.

Now the division of the first row is correct, but the first column of the other rows still has the problem, I'll try to fix them