def parse_json():
parser = EmailReplyParser(language='en')
with open('english.json', 'rb') as fl:
messages = json.load(fl)
parsed = []
for text in messages:
soup = BeautifulSoup(text, 'lxml')
text = soup.getText('\n')
text = parser.parse_reply(text)
parsed.append(text)
import code
code.interact(local=locals())
def verify():
parser = EmailReplyParser(language='fi')
texts = json.load(open('test/emails/emails.json'))
texts = list(filter(lambda d: type(d) == str, texts))
parsed = []
for text in texts:
print('-'*100)
soup = BeautifulSoup(text, 'lxml')
text = soup.getText('\n')
text = parser.parse_reply(text)
parsed.append(text)
print(text)
def parse_df():
parser = EmailReplyParser(language='en')
path = 'test/emails/zipwrotetest.csv'
df = pd.read_csv(path)
parsed = []
for text in df.sentence.values:
soup = BeautifulSoup(text, 'lxml')
text = soup.getText('\n')
text = parser.parse_reply(text)
parsed.append(text)
df = df.assign(clean=parsed)
df.to_csv(path)
import code
code.interact(local=locals())
def profile():
df = pd.read_csv('test.csv')
ground = time.time()
content = df.content.values[np.argmax([len(d) for d in df.content.values])]
start = time.time()
parser = EmailReplyParser(language='fr')
print(str(time.time() - start) + 'init parser')
start = time.time()
res = parser.parse_reply(content)
print(str(time.time() - start) + 'parse')
start = time.time()
soup = BeautifulSoup(res, 'lxml')
text = soup.getText(' ')
print(str(time.time() - start) + 'soup')
print(f'Total time: {time.time() - ground}')
def parse_text():
parser = EmailReplyParser(language='en')
with open('test/emails/caution.txt', 'r') as fl:
message = fl.read()
text = parser.parse_reply(message)
print(text)
def test_email_one_is_not_on(self):
with open('test/emails/email_one_is_not_on.txt') as email:
self.assertTrue(
"On Oct 1, 2012, at 11:55 PM, Dave Tapley wrote:" not in EmailReplyParser().parse_reply(email.read()))
def get_email(self, name):
""" Return EmailMessage instance
"""
with open('test/emails/%s.txt' % name) as f:
text = f.read()
return EmailReplyParser().read(text)
def test_sent_from_iphone(self):
with open('test/emails/email_iPhone.txt') as email:
self.assertTrue("Sent from my iPhone" not in EmailReplyParser().parse_reply(email.read()))
def test_parse_out_just_top_for_outlook_with_unusual_headers_format(self):
with open('test/emails/email_2_3.txt') as f:
self.assertEqual(
"Outlook with a reply above headers using unusual format",
EmailReplyParser().parse_reply(f.read()))
def test_parse_out_just_top_for_outlook_with_reply_directly_above_line(self):
with open('test/emails/email_2_2.txt') as f:
self.assertEqual("Outlook with a reply directly above line", EmailReplyParser().parse_reply(f.read()))
def test_parse_out_just_top_for_outlook_reply(self):
with open('test/emails/email_2_1.txt') as f:
self.assertEqual("Outlook with a reply", EmailReplyParser().parse_reply(f.read()))
def test_reply_from_gmail(self):
with open('test/emails/email_gmail.txt') as f:
self.assertEqual('This is a test for inbox replying to a github message.',
EmailReplyParser().parse_reply(f.read()))
