Text coordinate extraction error

[Number18-tong]

Description of the bug

Thanks for your great work, PDF parsing has become simpler and more convenient.
I have been use PyMuPDF for a while, and I find the problem that text coordinate extraction error in some pdfs.

Really hoping to find a way to solute this problem. Again, thanks for your great work!

How to reproduce the bug

The code:

import pymupdf
def getblock_lines_dict(fitz_dict):
linelist = []
## 获取每页的 每行文本
for block in fitz_dict["blocks"]:
if block['type'] == 0: ## block type为0时是文本
paranum = block['number']
if 'lines' in block: # 如果文本块中有内容
for line in block['lines']: ## 认为line是一行文本
for span in line['spans']:
if span['text'].strip():
linelist.append([paranum, span['bbox'], span['text']])
return linelist

if name == "main":
doc = pymupdf.open("test.pdf") # open a document
for page in doc: # iterate the document pages
dict = page.get_text("dict")
linelist = getblock_lines_dict(dict)
print(linelist)

I draw a picture for the results, basically the position coordinates of all the numbers are wrong.

There is the test pdf
number_bbox_error.pdf

PyMuPDF version

1.25.1

Operating system

Linux

Python version

3.10

[JorjMcKie]

This is a duplicate of issue #4180:
Font object definitions in the PDF specify (wrong) positive values for "descender". Since recent versions, MuPDF prefers using PDF-provided values over those provided in the embedded font binary (where they are correct in this case).