def test_get_all_graphic_images_from_xml(self):
graphic_01 = '<graphic xlink:href="1234-5678-rctb-45-05-0110-e01.jpg"/>'
graphic_02 = '<inline-graphic xlink:href="1234-5678-rctb-45-05-0110-e02.gif"/>'
xml_file = BASE_XML.format(graphic_01, graphic_02).encode("utf-8")
xml_file_path = os.path.join(self.work_dir, self.xml_filename)
with open(xml_file_path, "wb") as fp:
fp.write(xml_file)
xml_web_optimiser = utils.XMLWebOptimiser(
self.xml_filename,
self.image_filenames,
self.mocked_read_file,
self.work_dir,
)
result = xml_web_optimiser._get_all_graphic_images_from_xml(
self.image_filenames)
expected = {
"1234-5678-rctb-45-05-0110-e01.jpg",
"1234-5678-rctb-45-05-0110-e02.gif",
"1234-5678-rctb-45-05-0110-gf03.tiff",
"1234-5678-rctb-45-05-0110-gf03.png",
"1234-5678-rctb-45-05-0110-gf03.thumbnail.jpg",
"1234-5678-rctb-45-05-0110-e04.tif",
}
for graphic_filename, expected_filename in zip(result, expected):
self.assertEqual(graphic_filename, expected_filename)
def setUp(self):
graphic_01 = '<graphic xlink:href="1234-5678-rctb-45-05-0110-e01"/>'
graphic_02 = '<inline-graphic xlink:href="1234-5678-rctb-45-05-0110-e02"/>'
self.xml_file = BASE_XML.format(graphic_01, graphic_02).encode("utf-8")
self.xml_filename = "1234-5678-rctb-45-05-0110.xml"
self.work_dir = tempfile.mkdtemp()
self.image_filenames = [
"1234-5678-rctb-45-05-0110-e01.tif",
"1234-5678-rctb-45-05-0110-e02.tiff",
"1234-5678-rctb-45-05-0110-gf03.tiff",
"1234-5678-rctb-45-05-0110-gf03.png",
"1234-5678-rctb-45-05-0110-gf03.thumbnail.jpg",
"1234-5678-rctb-45-05-0110-e04.tif",
]
image_format_seq = ["TIFF", "TIFF", "TIFF", "PNG", "JPEG", "TIFF"]
for filename, format in zip(self.image_filenames, image_format_seq):
image_file_path = os.path.join(self.work_dir, filename)
create_image_file(image_file_path, format)
xml_file_path = os.path.join(self.work_dir, self.xml_filename)
with open(xml_file_path, "wb") as fp:
fp.write(self.xml_file)
self.xml_web_optimiser = utils.XMLWebOptimiser(
self.xml_filename,
self.image_filenames,
self.mocked_read_file,
self.work_dir,
)
def test_get_all_images_to_thumbnail_does_not_return_images_with_thumbnail(
self):
graphic_01 = ""
graphic_02 = '<inline-graphic xlink:href="1234-5678-rctb-45-05-0110-e02.gif"/>'
xml_file = BASE_XML.format(graphic_01, graphic_02).encode("utf-8")
xml_file_path = os.path.join(self.work_dir, self.xml_filename)
with open(xml_file_path, "wb") as fp:
fp.write(xml_file)
xml_web_optimiser = utils.XMLWebOptimiser(
self.xml_filename,
self.image_filenames,
self.mocked_read_file,
self.work_dir,
)
images = xml_web_optimiser._get_all_images_to_thumbnail()
result = [image for image in images]
self.assertEqual(len(result), 0)
def test_add_alternative_to_alternatives_tag_add_image_tags_to_existing_alternatives(
self, ):
xml_file = BASE_XML.format("", "").encode("utf-8")
xml_file_path = os.path.join(self.work_dir, self.xml_filename)
with open(xml_file_path, "wb") as fp:
fp.write(xml_file)
xml_web_optimiser = utils.XMLWebOptimiser(
self.xml_filename,
self.image_filenames,
self.mocked_read_file,
self.work_dir,
)
XML_TAG = """<article xmlns:xlink="http://www.w3.org/1999/xlink">
<p>Bla, bla, bla
<alternatives>
<inline-graphic xlink:href="1234-5678-rctb-45-05-0110-gf03.tiff"/>
<inline-graphic xlink:href="1234-5678-rctb-45-05-0110-gf03.png" specific-use="scielo-web"/>
</alternatives>
</p>
</article>"""
xml_tag = etree.fromstring(
XML_TAG, etree.XMLParser(remove_blank_text=True, no_network=True))
xml_web_optimiser._xml_file = xml_tag
image_element = xml_tag.find(".//inline-graphic")
alternative_attr_values = (
(
"{http://www.w3.org/1999/xlink}href",
"1234-5678-rctb-45-05-0110-gf03.gif",
),
("specific-use", "scielo-web"),
)
xml_web_optimiser._add_alternative_to_alternatives_tag(
image_element, alternative_attr_values)
self.assertIsNone(xml_tag.find("p/inline-graphic"))
alternatives_tags = xml_tag.findall("p/alternatives")
self.assertIsNotNone(alternatives_tags)
self.assertEqual(len(alternatives_tags), 1)
expected_hrefs = [
"1234-5678-rctb-45-05-0110-gf03.tiff",
"1234-5678-rctb-45-05-0110-gf03.png",
"1234-5678-rctb-45-05-0110-gf03.gif",
]
for image_element in alternatives_tags[0].getchildren():
self.assertEqual(image_element.tag, "inline-graphic")
def test_get_optimised_xml_no_reader_file(self):
graphic_01 = '<graphic xlink:href="1234-5678-rctb-45-05-0110-e01.tif"/>'
graphic_02 = '<inline-graphic xlink:href="1234-5678-rctb-45-05-0110-e02.tiff"/>'
xml_file = BASE_XML.format(graphic_01, graphic_02).encode("utf-8")
xml_file_path = os.path.join(self.work_dir, self.xml_filename)
with open(xml_file_path, "wb") as fp:
fp.write(xml_file)
with self.assertRaises(exceptions.XMLWebOptimiserError) as exc_info:
xml_web_optimiser = utils.XMLWebOptimiser(
self.xml_filename,
self.image_filenames,
None,
self.work_dir,
)
self.assertEqual(
str(exc_info.exception),
"Error instantiating XMLWebOptimiser: read_file cannot be None",
)
def test_get_all_images_to_optimise_does_not_return_optimised_images(self):
graphic_01 = '<graphic xlink:href="1234-5678-rctb-45-05-0110-e01.jpg"/>'
graphic_02 = '<inline-graphic xlink:href="1234-5678-rctb-45-05-0110-e02.gif"/>'
xml_file = BASE_XML.format(graphic_01, graphic_02).encode("utf-8")
xml_file_path = os.path.join(self.work_dir, self.xml_filename)
with open(xml_file_path, "wb") as fp:
fp.write(xml_file)
xml_web_optimiser = utils.XMLWebOptimiser(
self.xml_filename,
self.image_filenames,
self.mocked_read_file,
self.work_dir,
)
images = xml_web_optimiser._get_all_images_to_optimise()
expected = ["1234-5678-rctb-45-05-0110-e04.tif"]
result = [image for image in images]
self.assertEqual(len(result), len(expected))
image_filename, __ = result[0]
self.assertEqual(expected[0], image_filename)