def test_iterparse_xpath(self):
with open(self.path, 'rb') as f:
elements = list(drill.iterparse(f, xpath='catalog/magazine'))
self.assertEqual(len(elements), 1)
# Make sure children of matched elements are still parsed correctly.
self.assertEqual([e.tagname for e in elements[0]], ['author', 'title', 'price', 'book'])
with open(self.path, 'rb') as f:
elements = list(drill.iterparse(f, xpath='catalog/*/title'))
self.assertEqual(len(elements), 3)
self.assertEqual([e.data for e in elements], ['Test Book', u('Él Libro'), 'Test Magazine'])
def test_custom_handler_class(self):
doc = drill.parse(self.path, handler_class=CustomHandler)
self.assertEqual(doc[0].__class__, CustomElement)
for e in drill.iterparse(open(self.path, 'rb'),
handler_class=CustomHandler):
self.assertEqual(e.__class__, CustomElement)
e.clear()
def test_iterparse(self):
parsed_tags = []
for e in drill.iterparse(open(self.path, 'rb')):
parsed_tags.append(e.tagname)
e.clear()
self.assertEqual(
parsed_tags,
[
'author',
'isbn',
'title',
'book', # The first book, children first
'author',
'isbn',
'title',
'extra_element',
'book', # The second book
'author',
'title',
'price', # Magazine elements
'title',
'isbn',
'book', # Book inside the magazine
'magazine', # The magazine
'catalog' # Finally, the root catalog
])
# The last element should be the finished root element, and it should be empty (since we cleared as we parsed).
self.assertEqual(e.tagname, 'catalog')
self.assertEqual(len(e), 0)
def test_iterparse(self):
parsed_tags = []
for e in drill.iterparse(open(self.path, 'rb')):
parsed_tags.append(e.tagname)
e.clear()
self.assertEqual(parsed_tags, [
'author', 'isbn', 'title', 'book', # The first book, children first
'author', 'isbn', 'title', 'extra_element', 'book', # The second book
'author', 'title', 'price', # Magazine elements
'title', 'isbn', 'book', # Book inside the magazine
'magazine', # The magazine
'catalog' # Finally, the root catalog
])
# The last element should be the finished root element, and it should be empty (since we cleared as we parsed).
self.assertEqual(e.tagname, 'catalog')
self.assertEqual(len(e), 0)
def test_custom_handler_class(self):
doc = drill.parse(self.path, handler_class=CustomHandler)
self.assertEqual(doc[0].__class__, CustomElement)
for e in drill.iterparse(open(self.path, 'rb'), handler_class=CustomHandler):
self.assertEqual(e.__class__, CustomElement)
e.clear()