wehub-resource-sync
917eedffcf
Main / Python 3.11 - Docs (push) Has been cancelled
Main / Python 3.11 - Build (push) Has been cancelled
Main / Python 3.11 - Lint (push) Has been cancelled
Main / Python 3.11 - Style (push) Has been cancelled
Main / Python 3.11 - Test (push) Has been cancelled
Main / GPU CI (push) Has been cancelled
Main / Release (push) Has been cancelled
Main / Build and Push Docker Images (push) Has been cancelled
18 lines
553 B
Python
18 lines
553 B
Python
import os
|
|
import unittest
|
|
|
|
from pypdf import PdfReader
|
|
|
|
from olmocr.filter import PdfFilter
|
|
|
|
|
|
class PdfFilterTest(unittest.TestCase):
|
|
def testFormLaterPages(self):
|
|
self.filter = PdfFilter(apply_form_check=True)
|
|
|
|
self.assertTrue(self.filter.filter_out_pdf(os.path.join(os.path.dirname(__file__), "gnarly_pdfs", "form_on_later_pages.pdf")))
|
|
|
|
self.filter = PdfFilter(apply_form_check=False)
|
|
|
|
self.assertFalse(self.filter.filter_out_pdf(os.path.join(os.path.dirname(__file__), "gnarly_pdfs", "form_on_later_pages.pdf")))
|