def test_get_pdf_text_without_encrypted(tmp_path):
from ReadPDFFileV2 import get_pdf_text
try:
get_pdf_text(f'{CWD}/encrypted.pdf', f'{tmp_path}/encrypted.txt')
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)
text = get_pdf_text(f'{CWD}/text-only.pdf', f'{tmp_path}/text-only.txt')
expected = "עברית"
assert expected in text
assert text.startswith('This is a pdf document with a text line within it.')
text = get_pdf_text(f'{CWD}/text-with-images.pdf', f'{tmp_path}/text-with-images.txt')
expected = 'Create an ETD Using Adobe Acrobat'
assert text.startswith(expected)
text = get_pdf_text(f'{CWD}/scanned.pdf', f'{tmp_path}/scanned.txt')
expected = '\x0c'
assert expected == text
def test_get_pdf_text_without_encrypted(tmp_path):
from ReadPDFFileV2 import get_pdf_text
# assert error raised
try:
get_pdf_text(f'{CWD}/encrypted.pdf', f'{tmp_path}/encrypted.txt')
raise Exception("Incorrect password exception should've been thrown")
except ShellException as e:
assert 'Incorrect password' in str(e)
assert 'error code: 1' in str(e)
# assert not warnings are raised
text = get_pdf_text(f'{CWD}/warning_trigger.pdf',
f'{tmp_path}/warning_trigger.txt')
assert 'Riu Plaza Berlin' in text
# assert extract file correctly
text = get_pdf_text(f'{CWD}/text-only.pdf', f'{tmp_path}/text-only.txt')
expected = "עברית"
assert expected in text
assert text.startswith(
'This is a pdf document with a text line within it.')
text = get_pdf_text(f'{CWD}/text-with-images.pdf',
f'{tmp_path}/text-with-images.txt')
expected = 'Create an ETD Using Adobe Acrobat'
assert text.startswith(expected)
text = get_pdf_text(f'{CWD}/scanned.pdf', f'{tmp_path}/scanned.txt')
expected = '\x0c'
assert expected == text
def test_get_pdf_text_with_encrypted(mocker, tmp_path):
mocker.patch.object(demisto, 'args', return_value={'userPassword': '******'})
from ReadPDFFileV2 import get_pdf_text
text = get_pdf_text(f'{CWD}/encrypted.pdf', f'{tmp_path}/encrypted.txt')
expected = "XSL FO Sample Copyright © 2002-2005 Antenna House, Inc. All rights reserved.\n\n" \
"Links in PDF\nPDF link is classified into two parts, link to the specified position in the PDF " \
"document, and link to the external document.\n" \
"The internal-destination property of fo:basic-link indicates to link to the position in the same" \
" document. The externaldestination property indicates to link to external document. " \
"Below shows the example.\n\nExample of a link to internal destination\nRefer to Purchasing " \
"Assistance to get more information.\nExample of a link to external destination\nRefer to Purchasing " \
"Assistance to get more information."
assert text.startswith(expected)
def test_get_pdf_text_with_encrypted(tmp_path):
from ReadPDFFileV2 import get_pdf_text, decrypt_pdf_file
file_path = f'{CWD}/encrypted.pdf'
dec_file_path = f'{CWD}/decrypted.pdf'
decrypt_pdf_file(file_path, '1234', dec_file_path)
text = get_pdf_text(dec_file_path, f'{tmp_path}/encrypted.txt')
expected = "XSL FO Sample Copyright © 2002-2005 Antenna House, Inc. All rights reserved.\n\n" \
"Links in PDF\nPDF link is classified into two parts, link to the specified position in the PDF " \
"document, and link to the external document.\n" \
"The internal-destination property of fo:basic-link indicates to link to the position in the same" \
" document. The externaldestination property indicates to link to external document. " \
"Below shows the example.\n\nExample of a link to internal destination\nRefer to Purchasing " \
"Assistance to get more information.\nExample of a link to external destination\nRefer to Purchasing " \
"Assistance to get more information."
if os.path.exists(dec_file_path):
os.remove(dec_file_path)
assert text.startswith(expected)
