def test_pdfimages_output(self):
zip_file = "tests/test_data/30-01-2019-RA-eLife-45644.zip"
zip_lib.unzip_zip(zip_file, self.temp_dir)
# run only if pdfimages is available
if pdf_utils.pdfimages_exists():
pdf_file = os.path.join(
self.temp_dir, "30-01-2019-RA-eLife-45644/Appendix 1figure 10.pdf"
)
result = pdf_utils.pdfimages_output(pdf_file)
self.assertIsNotNone(result.stdout)
def test_pdf_page_count_wand_runtime_error(self, mock_image_allocate):
mock_image_allocate.side_effect = wand.exceptions.WandRuntimeError()
zip_lib.unzip_zip(
"tests/test_data/30-01-2019-RA-eLife-45644.zip", self.temp_dir
)
pdf_path = "tests/tmp/30-01-2019-RA-eLife-45644/Appendix 1figure 10.pdf"
with self.assertRaises(wand.exceptions.WandRuntimeError):
self.assertIsNone(parse.pdf_page_count(pdf_path))
expected = (
"ERROR elifecleaner:parse:pdf_page_count: "
"WandRuntimeError in pdf_page_count(), imagemagick may not be installed\n"
)
log_file_lines = read_log_file_lines(self.log_file)
self.assertEqual(log_file_lines[0], expected)
def test_file_list(self):
zip_file = "tests/test_data/30-01-2019-RA-eLife-45644.zip"
asset_file_name_map = zip_lib.unzip_zip(zip_file, self.temp_dir)
xml_asset = parse.article_xml_asset(asset_file_name_map)
root = parse.parse_article_xml(xml_asset[1])
expected = read_fixture("file_list_45644.py")
files = parse.file_list(root)
self.assertEqual(files, expected)
def test_article_xml_name(self):
zip_file = "tests/test_data/30-01-2019-RA-eLife-45644.zip"
asset_file_name_map = zip_lib.unzip_zip(zip_file, self.temp_dir)
expected = (
"30-01-2019-RA-eLife-45644/30-01-2019-RA-eLife-45644.xml",
"tests/tmp/30-01-2019-RA-eLife-45644/30-01-2019-RA-eLife-45644.xml",
)
xml_asset = parse.article_xml_asset(asset_file_name_map)
self.assertEqual(xml_asset, expected)
def check_ejp_zip(zip_file, tmp_dir):
"check contents of ejp zip file"
asset_file_name_map = zip_lib.unzip_zip(zip_file, tmp_dir)
xml_asset = article_xml_asset(asset_file_name_map)
root = parse_article_xml(xml_asset[1])
files = file_list(root)
# use the zip file name as the identifier for log messages
identifer = zip_file.split(os.sep)[-1]
return check_files(files, asset_file_name_map, identifer)
def transform_ejp_zip(zip_file, tmp_dir, output_dir):
"transform ejp zip file and write a new zip file output"
zip_file_name = zip_file.split(os.sep)[-1]
# profile the zip contents
asset_file_name_map = zip_lib.unzip_zip(zip_file, tmp_dir)
# start logging
LOGGER.info("%s starting to transform", zip_file_name)
new_asset_file_name_map = transform_ejp_files(
asset_file_name_map, output_dir, zip_file_name
)
# write new zip file
new_zip_file_path = rezip(new_asset_file_name_map, output_dir, zip_file_name)
return new_zip_file_path
def test_unzip_zip(self):
zip_file = "tests/test_data/30-01-2019-RA-eLife-45644.zip"
asset_file_name_map = zip_lib.unzip_zip(zip_file, self.temp_dir)
expected = read_fixture("asset_file_name_map_45644.py")
self.assertEqual(asset_file_name_map, expected)