def test__parse_multi_layer_file__is_output_all_unicode():
doclist = parse.parse_multi_layer_file(join(TESTDIR,
'sample_enron_email.'))
for docdict in doclist:
docdict['content']['attachment']['filename'] = u''
docdict['rawbody'] = u''
for item in flatten_handle_all(docdict, dictvalues=True):
if isinstance(item, _STRINGTYPES):
assert(isinstance(item, unicode))
def test__parse_large_email_with_large_mix_of_files():
doclist = parse.parse_multi_layer_file(MINIEMLWithAttch)
for prt in doclist:
if utils.file_exts(prt['filename']).lower() in parse.OKEXT:
assert(20 < len(prt['content']['body']))
def test__parse_multi_layer_file():
cached = utils.load_json(join(TESTDIR, 'sample_parsed_email.json'),
mode='r')
prsd = parse.parse_multi_layer_file('sample_enron_email.')
new = json.loads(json.dumps(prsd))
assert(new[0]['content']['body'] == cached[0]['content']['body'])
