def transform_xml(xml_asset_path, identifier):
"modify the XML"
# remove history tags from XML for certain article types
root = parse.parse_article_xml(xml_asset_path)
soup = parser.parse_document(xml_asset_path)
root = transform_xml_history_tags(root, soup, identifier)
write_xml_file(root, xml_asset_path, identifier)
def test_parse_article_xml_entities(self):
xml_file_path = os.path.join(self.temp_dir, "test.xml")
with open(xml_file_path, "w") as open_file:
open_file.write("<article>—<>&"β</article>")
expected = b'<article>—<>&"β</article>'
root = parse.parse_article_xml(xml_file_path)
self.assertIsNotNone(root)
self.assertEqual(ElementTree.tostring(root), 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 transform_code_files(asset_file_name_map, output_dir, identifier):
"zip code files if they are not already a zip file"
# parse XML file
xml_asset = parse.article_xml_asset(asset_file_name_map)
xml_asset_path = xml_asset[1]
root = parse.parse_article_xml(xml_asset_path)
file_transformations = code_file_transformations(
root, asset_file_name_map, output_dir, identifier
)
code_file_zip(file_transformations, output_dir, identifier)
# create a new asset map
new_asset_file_name_map = transform_asset_file_name_map(
asset_file_name_map, file_transformations
)
xml_rewrite_file_tags(xml_asset_path, file_transformations, identifier)
return new_asset_file_name_map
def glencoe_xml(xml_file_path, video_data, pretty=True, indent=""):
"generate XML to be submitted to Glencoe"
# build an Article object from the XML
article, error_count = article_from_xml(xml_file_path)
# collect journal data from the XML elementtree
root = parse.parse_article_xml(xml_file_path)
journal_ids = parse.xml_journal_id_values(root)
filtered_journal_ids = {
key: value
for key, value in journal_ids.items() if key in JOURNAL_ID_TYPES
}
journal_title = parse.xml_journal_title(root)
publisher_name = parse.xml_publisher_name(root)
journal_data = {
"journal_ids": filtered_journal_ids,
"journal_title": journal_title,
"publisher_name": publisher_name,
}
return generate_xml(article, journal_data, video_data, pretty, indent)
def xml_rewrite_file_tags(xml_asset_path, file_transformations, identifier):
root = parse.parse_article_xml(xml_asset_path)
# rewrite the XML tags
LOGGER.info("%s rewriting xml tags", identifier)
root = transform_xml_file_tags(root, file_transformations)
write_xml_file(root, xml_asset_path, identifier)
def test_parse_article_xml_failure(self):
xml_file_path = os.path.join(self.temp_dir, "test.xml")
with open(xml_file_path, "w") as open_file:
open_file.write("malformed xml")
with self.assertRaises(ElementTree.ParseError):
parse.parse_article_xml(xml_file_path)
def test_parse_article_xml(self):
xml_file_path = os.path.join(self.temp_dir, "test.xml")
with open(xml_file_path, "w") as open_file:
open_file.write("<article/>")
root = parse.parse_article_xml(xml_file_path)
self.assertIsNotNone(root)