text_x_tolerance ignored by extract_table?

[jsfenfen]

I noticed that the text_tolerance and text_x_tolerance were not behaving the way I thought they should in the context of extract tables.

I was working on an example here--(how does one link to the middle of a notebook on github?) it's essentially at the bottom.

Extracting text from the relatively small font requires a text tolerance of 1 or 2 to realize word separation, but the same text tolerance appears not to work when in a table extraction context? Possibly I'm not understanding something important.

[jsvine]

Thanks for flagging, @jsfenfen. That's a good catch and, indeed, a bug. Table.extract(...) accepts x_tolerance and y_tolerance as parameters, but Page.extract_tables(...) does not pass those parameters to that method.

In the meantime, until I've pushed a fix for the bug, you can replace this in your notebook:

result = pg.crop(under_headers_region).extract_tables(config)

... with this:

result = [ table.extract(x_tolerance = 1)
  for table in pg.crop(under_headers_region).find_tables(config) ]

Does that do it for you?

[jsfenfen]

Hey @jsvine that totally works for my purposes, thanks a ton!

[jsvine]

(Fixed in b498df2)