def test_filepath_no_date(self):
importer = mock.MagicMock(wraps=self.importer)
importer.date.return_value = None
with mock.patch('beangulp.archive.utils.getmdate',
return_value=datetime.datetime.fromtimestamp(
0, datetime.timezone.utc)):
filepath = archive.filepath(importer, path.abspath('test.pdf'))
self.assertEqual(filepath, 'Assets/Tests/1970-01-01.test.pdf')
def test_filepath_date_in_name(self):
importer = mock.MagicMock(wraps=self.importer)
importer.filename.return_value = '1970-01-03.name.pdf'
with self.assertRaises(exceptions.Error) as ex:
filepath = archive.filepath(importer, path.abspath('test.pdf'))
self.assertRegex(ex.exception.message, r'contains [\w\s]+ date')
def test_filepath_sep_in_name(self):
importer = mock.MagicMock(wraps=self.importer)
importer.filename.return_value = f'dir{os.sep:}name.pdf'
with self.assertRaises(exceptions.Error) as ex:
filepath = archive.filepath(importer, path.abspath('test.pdf'))
self.assertRegex(ex.exception.message, r'contains path separator')
def test_filepath_no_filename(self):
filepath = archive.filepath(self.importer, path.abspath('test.pdf'))
self.assertEqual(filepath, 'Assets/Tests/1970-01-01.test.pdf')
def test_filepath(self):
importer = mock.MagicMock(wraps=self.importer)
importer.filename.return_value = 'foo.csv'
filepath = archive.filepath(importer, path.abspath('test.pdf'))
self.assertEqual(filepath, 'Assets/Tests/1970-01-01.foo.csv')
def _archive(ctx, src, destination, dry_run, overwrite, failfast):
"""Archive documents.
Walk the SRC list of files or directories and move each file
identified by one of the configured importers in a directory
hierarchy mirroring the structure of the accounts associated to
the documents and with a file name composed by the document date
and document name returned by the importer.
Documents are moved to their filing location only when no errors
are encountered processing all the input files. Documents in the
destination directory are not overwritten, unless the --force
option is used. When the directory hierarchy root is not
specified with the --destination DIR options, it is assumed to be
directory in which the ingest script is located.
"""
# If the output directory is not specified, move the files at the
# root where the import script is located. Providing this default
# seems better than using a required option.
if destination is None:
import __main__
destination = os.path.dirname(os.path.abspath(__main__.__file__))
log = utils.logger()
errors = exceptions.ExceptionsTrap(log)
renames = []
for filename in _walk(src, log):
with errors:
importer = identify.identify(ctx.importers, filename)
if not importer:
log('') # Newline.
continue
# Signal processing of this document.
log(' ...', nl=False)
destpath = archive.filepath(importer, filename)
# Prepend destination directory path.
destpath = os.path.join(destination, destpath)
# Check for destination filename collisions.
collisions = [src for src, dst in renames if dst == destpath]
if collisions:
raise exceptions.Error('Collision in destination file path.',
destpath)
# Check if the destination file already exists.
if not overwrite and os.path.exists(destpath):
raise exceptions.Error('Destination file already exists.',
destpath)
renames.append((filename, destpath))
log(' OK', fg='green')
log(f' {destpath:}')
if failfast and errors:
break
# If there are any errors, stop here.
if errors:
log('# Errors detected: documents will not be filed.')
sys.exit(1)
if not dry_run:
for src, dst in renames:
archive.move(src, dst)
def test_filepath_no_date(self):
importer = mock.MagicMock(wraps=self.importer)
importer.date.return_value = None
with mock.patch('os.path.getmtime', return_value=86401):
filepath = archive.filepath(importer, path.abspath('test.pdf'))
self.assertEqual(filepath, 'Assets/Tests/1970-01-02.test.pdf')