Mercurial > hgrepos > Python2 > PyMuPDF
annotate tests/test_4716.py @ 39:a6bc019ac0b2 upstream
ADD: PyMuPDF v1.26.5: the original sdist.
| author | Franz Glasner <fzglas.hg@dom66.de> |
|---|---|
| date | Sat, 11 Oct 2025 11:19:58 +0200 |
| parents | |
| children |
| rev | line source |
|---|---|
|
39
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1 import pymupdf |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2 import os |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
3 |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
4 def test_4716(): |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
5 """Confirm that ZERO WIDTH JOINER will never start a word.""" |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
6 script_dir = os.path.dirname(__file__) |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
7 filename = os.path.join(script_dir, "resources", "test_4716.pdf") |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
8 doc = pymupdf.open(filename) |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
9 expected = set(["+25.00", "Любимый", "-10.00"]) |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
10 word_text = set() |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
11 for page in doc: |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
12 words = page.get_text("words") |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
13 for w in words: |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
14 word_text.add(w[4]) |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
15 assert word_text == expected |
