def build_digest(file_name,
temp_dir="tmp",
digest_config=None,
image_file_name=None):
"build a digest object from a DOCX input file"
digest = None
docx_file_name, zip_image_file_name = handle_zip(file_name, temp_dir)
LOGGER.info("build_digest file '%s' has docx_file_name: '%s'", file_name,
docx_file_name)
LOGGER.info(
"build_digest file '%s' has zip_image_file_name: '%s'",
file_name,
zip_image_file_name,
)
if not image_file_name:
image_file_name = zip_image_file_name
content = parse_content(docx_file_name)
if content:
digest = Digest()
digest.author = build_author(content)
digest.title = build_title(content)
digest.summary = build_summary(content)
digest.keywords = build_keywords(content)
digest.manuscript_number = build_manuscript_number(content)
digest.doi = build_doi(content, digest_config)
digest.text = build_text(content)
digest.image = build_image(content, image_file_name)
return digest
def test_docx_file_name(self, test_data):
"docx output file name tests for various input"
# build the Digest object
digest = Digest()
digest.author = test_data.get("author")
digest.doi = test_data.get("doi")
# set the config, if using in the test
digest_config = None
if test_data.get("use_config"):
digest_config = parse_raw_config(
raw_config(test_data.get("config_section")))
# generate the file_name
file_name = output.docx_file_name(digest, digest_config)
# test assertion
self.assertEqual(
file_name,
test_data.get("expected_file_name"),
u"failed in scenario '{scenario}', got file_name {file_name}".
format(scenario=test_data.get("scenario"), file_name=file_name),
)
# test for creating the file on disk
full_file_name = os.path.join("tmp", file_name)
output_file_name = output.digest_docx(digest, full_file_name)
self.assertEqual(
os.path.join("tmp", test_data.get("expected_file_name")),
output_file_name,
u"failed creating file in scenario '{scenario}', got file_name {file_name}"
.format(scenario=test_data.get("scenario"),
file_name=output_file_name),
)
def test_digest_figure_image_url(self):
"test figure image url formatting"
image = build_image(file_value="test.jpg")
digest = Digest()
digest.doi = "10.7554/eLife.99999"
expected = u"https://iiif.elifesciences.org/digests/99999%2Ftest.jpg/full/full/0/default.jpg"
self.assertEqual(
medium_post.digest_figure_image_url(self.digest_config, image,
digest),
expected,
)
def test_digest_json_published_value(self):
"test json output for a digest with a published value"
digest = Digest()
digest.published = "2018-10-29"
expected = OrderedDict([
("id", "None"),
("title", None),
("impactStatement", None),
("published", "2018-10-29"),
("content", []),
])
self.assertEqual(json_output.digest_json(digest, None), expected)
def test_digest_docx(self):
"test digest_docx directly for coverage of setting bold tags"
output_dir = "tmp"
output_file_name = "bold_tag_test.docx"
text = ["<b>Test</b>"]
expected_content = "DIGEST\n<b>Test</b>\n"
digest = Digest()
digest.text = text
full_file_name = os.path.join(output_dir, output_file_name)
docx_file = output.digest_docx(digest, full_file_name)
output_content = parse_content(docx_file)
self.assertEqual(output_content, expected_content)
def test_digest_json_empty(self):
"test json output for an empty digest where there is no text or image file"
digest = Digest()
# reset some lists to None for testing
digest.text = None
digest.keywords = None
digest.subjects = None
expected = OrderedDict([
("id", "None"),
("title", None),
("impactStatement", None),
("content", []),
])
self.assertEqual(json_output.digest_json(digest, None), expected)
def test_digest_jats(self):
"simple test to convert digest text to JATS XML content"
digest = Digest()
digest.text = [
"First <b>paragraph</b>.",
"Second <i>paragraph</i>.",
"",
" ",
None,
]
expected_content = ("<p>First <bold>paragraph</bold>.</p><p>Second " +
"<italic>paragraph</italic>.</p>")
jats_content = jats.digest_jats(digest)
self.assertEqual(jats_content, expected_content)
def test_digest_figure_content(self):
"test figure caption formatting"
image = build_image(caption=u"Caption. Anonymous (CC BY\xa04.0)",
file_value="test.jpg")
digest = Digest()
digest.doi = "10.7554/eLife.99999"
expected = (
u"<figure>" +
u'<img src="https://iiif.elifesciences.org/digests/99999%2Ftest.jpg/full/full/0/default.jpg" />'
+
u"<figcaption>Caption. Anonymous (CC BY\xa04.0)</figcaption></figure>"
)
self.assertEqual(
medium_post.digest_figure_content(self.digest_config, image,
digest),
expected,
)