Incorrect row number when extract tables

[tujinshu]

Describe the bug

A clear and concise description of what the bug is.
Incorrect row number in green cycle

actual data （page 45）

Code to reproduce the problem

Paste it here, or attach a Python file.

import pdfplumber
pdf = pdfplumber.open("./wps.pdf")
p0 = pdf.pages[44]
im = p0.to_image()
im.debug_tablefinder()

PDF file

origin pdf
wps.pdf

Expected behavior

What did you expect the result should have been?
only one row should be extracted

Actual behavior

What actually happened, instead?
split to 5 rows

Screenshots

If applicable, add screenshots to help explain your problem.

Environment

• pdfplumber version: [0.9.0]
• Python version: [3.8.0]
• OS: [Linux]

Additional context

Add any other context/notes about the problem here.

[jsvine]

Hi @tujinshu, and thanks for your interest in this library. The situation you're encountering is fairly common, and typically is caused by "invisible" graphics (rectangles, lines, etc.) on the page. In this particular example, it's a series of rectangles:

im.reset().draw_rects(p0.rects)

In this particular situation, these rectangles can be identified at least a couple of ways:

• Their non_stroking_color is 1 instead of 0
• Their height and width are both greater than 1. (The other rects are very narrow, making them look more like lines.)

To demonstrate:

im.reset().draw_rects([ r for r in p0.rects if (r["width"] > 1) and (r["height"] > 1)])

To ignore these invisible rectangles when extracting the tables, you'll want to use the page.filter(...) method. For examples of how this works, see the solutions to these discussions:

• Issue with a table cell containing multiple lines of text #895
• unexpected/non-exist vertical/horizontal lines generated in tables which cause extract table result wrong #311