def test_change_filename_extension(app):
"""Test change filename extension."""
with pytest.raises(Exception) as e:
change_filename_extension('test', 'txt')
assert str(e.value) == 'test is not a valid filename'
assert change_filename_extension('test.pdf', 'txt') == 'test-pdf.txt'
def create_thumbnail(self, file):
"""Create a thumbnail for record.
This is done by getting the file with order 1 or the first file
instead.
:param file: File from which thumbnail is created.
"""
try:
# Create thumbnail
image_blob = create_thumbnail_from_file(file.file.uri,
file.mimetype)
thumbnail_key = change_filename_extension(file['key'], 'jpg')
# Store thumbnail in record's files
self.files[thumbnail_key] = BytesIO(image_blob)
self.files[thumbnail_key]['type'] = 'thumbnail'
except Exception as exception:
current_app.logger.warning(
'Error during thumbnail generation of {file} of record '
'{record}: {error}'.format(file=file['key'],
error=exception,
record=self.get(
'identifiedBy', self['pid'])))
def create_fulltext_file(self, file):
"""Create fulltext file corresponding to give file object.
:param file: File object.
"""
# If extract fulltext is disabled or file is not a PDF
if not current_app.config.get(
'SONAR_DOCUMENTS_EXTRACT_FULLTEXT_ON_IMPORT'
) or file.mimetype != 'application/pdf':
return
# Try to extract full text from file data, and generate a warning if
# it's not possible. For several cases, file is locked against fulltext
# copy.
try:
with file.file.storage().open() as pdf_file:
fulltext = extract_text_from_content(pdf_file.read())
key = change_filename_extension(file.key, 'txt')
self.files[key] = BytesIO(fulltext.encode())
self.files[key]['type'] = 'fulltext'
except Exception as exception:
current_app.logger.warning(
'Error during fulltext extraction of {file} of record '
'{record}: {error}'.format(file=file.key,
error=exception,
record=self['identifiedBy']))
def add_file(self, data, key, **kwargs):
"""Create file and add it to record.
:param data: Binary data of file
:param str key: File key
kwargs may contain some additional data such as: file label, file type,
order and url.
"""
if not current_app.config.get('SONAR_DOCUMENTS_IMPORT_FILES'):
return
# If file with the same key exists and checksum is the same as the
# registered file, we don't do anything
checksum = compute_md5_checksum(BytesIO(data))
if key in self.files and checksum == self.files[key].file.checksum:
return
# Create the file
self.files[key] = BytesIO(data)
self.files[key]['label'] = kwargs.get('label', key)
self.files[key]['type'] = kwargs.get('type', 'file')
self.files[key]['order'] = kwargs.get('order', 1)
# Embargo
if kwargs.get('restricted'):
self.files[key]['restricted'] = kwargs['restricted']
if kwargs.get('embargo_date'):
self.files[key]['embargo_date'] = kwargs['embargo_date']
# Store external file URL
if kwargs.get('url'):
self.files[key]['external_url'] = kwargs['url']
# Create thumbnail
if current_app.config.get('SONAR_DOCUMENTS_GENERATE_THUMBNAIL'):
self.create_thumbnail(self.files[key])
# Try to extract full text from file data, and generate a warning if
# it's not possible. For several cases, file is locked against fulltext
# copy.
if current_app.config.get(
'SONAR_DOCUMENTS_EXTRACT_FULLTEXT_ON_IMPORT'
) and self.files[key].mimetype == 'application/pdf':
try:
fulltext = extract_text_from_content(data)
key = change_filename_extension(key, 'txt')
self.files[key] = BytesIO(fulltext.encode())
self.files[key]['type'] = 'fulltext'
except Exception as exception:
current_app.logger.warning(
'Error during fulltext extraction of {file} of record '
'{record}: {error}'.format(file=key,
error=exception,
record=self['identifiedBy']))
def thumbnail(file, files):
"""Get thumbnail from file.
:param file: Dict of file from which thumbnail will be returned.
:param files: Liste of files of the record.
"""
key = change_filename_extension(file['key'], 'jpg')
matches = [file for file in files if file['key'] == key]
if not matches:
return None
return matches[0]
def get_thumbnail(file, record):
"""Get thumbnail from file.
If file is restricted, a restricted image is returned. If no thumbnail
found, a default image is returned.
:param file: Dict of file from which thumbnail will be returned.
:param record: Record object.
:returns: URL to thumbnail file.
"""
if file['restriction']['restricted']:
return 'static/images/restricted.png'
key = change_filename_extension(file['key'], 'jpg')
matches = [file for file in record['_files'] if file['key'] == key]
if not matches:
return 'static/images/no-image.png'
return '/documents/{pid}/files/{key}'.format(pid=record['pid'], key=key)