comparison tests/test_rtl.py @ 3:2c135c81b16c

MERGE: upstream PyMuPDF 1.26.4 with MuPDF 1.26.7
author Franz Glasner <fzglas.hg@dom66.de>
date Mon, 15 Sep 2025 11:44:09 +0200
parents 1d09e1dec1d9
children
comparison
equal deleted inserted replaced
0:6015a75abc2d 3:2c135c81b16c
1 import pymupdf
2
3 import os
4
5
6 def test_rtl():
7 path = os.path.normpath(f'{__file__}/../../tests/resources/test-E+A.pdf')
8 doc = pymupdf.open(path)
9 page = doc[0]
10 # set of all RTL characters
11 rtl_chars = set([chr(i) for i in range(0x590, 0x901)])
12
13 for w in page.get_text("words"):
14 # every word string must either ONLY contain RTL chars
15 cond1 = rtl_chars.issuperset(w[4])
16 # ... or NONE.
17 cond2 = rtl_chars.intersection(w[4]) == set()
18 assert cond1 or cond2