def test_get_pdf_htmls_content_with_encrypted(mocker, tmp_path):
mocker.patch.object(demisto, 'args', return_value={'userPassword': '******'})
from ReadPDFFileV2 import get_pdf_htmls_content
from ReadPDFFileV2 import get_images_paths_in_path
html_text = get_pdf_htmls_content(f'{CWD}/encrypted.pdf', tmp_path)
expected = 'If you are end user who wishes to use XSL Formatter yourself, you may purchase ' \
'from our Reseller or direct from Antenna<br/>House.<br/>'
assert len(get_images_paths_in_path(tmp_path)) != 0, 'Failed to get images from html'
assert expected in html_text
def test_get_pdf_htmls_content_without_encrypted(tmp_path):
from ReadPDFFileV2 import get_pdf_htmls_content
from ReadPDFFileV2 import get_images_paths_in_path
try:
get_pdf_htmls_content(f'{CWD}/encrypted.pdf', tmp_path)
raise Exception("Incorrect password exception should've been thrown")
except ShellException as e:
assert 'Command Line Error: Incorrect password\nShell error code: 1' == str(e)
html_text = get_pdf_htmls_content(f'{CWD}/hyperlinks.pdf', tmp_path)
assert 'http://www.antennahouse.com/purchase.htm' in html_text
assert len(get_images_paths_in_path(tmp_path)) != 0, 'Failed to get images from html'
def test_get_images_paths_in_path():
from ReadPDFFileV2 import get_images_paths_in_path
img_file_paths = get_images_paths_in_path('test_data')
assert 'test_data/test1.png' in img_file_paths
assert 'test_data/scanned.pdf' not in img_file_paths
