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_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)