From 773f801141c81f7ec8ef62c9af4699744c21d40c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Bart=C5=82omiej=20Koptyra?= <rikain@e-science.pl>
Date: Fri, 7 Aug 2020 16:32:40 +0200
Subject: [PATCH] Added some emails and users handling.
---
src/anonymizer.py | 72 ++++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 71 insertions(+), 1 deletion(-)
diff --git a/src/anonymizer.py b/src/anonymizer.py
index 577638f..06c80dc 100644
--- a/src/anonymizer.py
+++ b/src/anonymizer.py
@@ -1,14 +1,84 @@
"""Implementation of anonymizer functionality."""
import re
+from string import punctuation, ascii_lowercase, ascii_uppercase
+import random
class Anonymizer:
"""Class used to edit sentences based on options."""
def __init__(self, task_options):
- self.method = task_options.get('method', 'delete')
+ self._method = task_options.get('method', 'delete')
+ self._mail_token = '[MAIL]'
+ self._user_token = '@[USER]'
def process(self):
if ctag == 'ign':
# sprawddz czy to nick a potem email
# sprawdz czy to nazwa własna jak mBank? nie wiem
+ print()
+
+ @staticmethod
+ def _get_random_chatacter(upper=False):
+ return random.choice(ascii_uppercase) \
+ if upper else random.choice(ascii_lowercase)
+
+ def _generate_pseudo_email(self, email):
+ new_mail = []
+ it = iter(email)
+ top_domain_len = len(email) - email.rfind('.')
+ for char in it:
+ if char == '@':
+ new_mail.append(char)
+ break
+ elif char in punctuation:
+ new_mail.append(char)
+ else:
+ new_mail.append(self._get_random_chatacter(char.isupper()))
+ for char in it:
+ if char == '.':
+ if len(list(it)) == top_domain_len:
+ new_mail.append(char)
+ break
+ new_mail.append(char)
+ elif char in punctuation:
+ new_mail.append(char)
+ else:
+ new_mail.append(self._get_random_chatacter(char.isupper()))
+ for char in it:
+ new_mail.append(char)
+ return ''.join(new_mail)
+
+ def _generate_pseudo_user(self, user):
+ it = iter(user)
+ new_user = []
+ new_user.append(next(it))
+ for char in it:
+ if char in punctuation:
+ new_user.append(char)
+ else:
+ new_user.append(self._get_random_chatacter(char.isupper()))
+ return ''.join(new_user)
+
+ def _anonoymize_email(self, token):
+ """Handles removal/changing of emails addresses."""
+ email_regex = r'[\w\.-]+@[\w\.-]+\.\w{2,4}'
+ if self._method == 'delete':
+ token = re.sub(email_regex, r'', token)
+ elif self._method == 'tag':
+ token = re.sub(email_regex, self._mail_token, token)
+ elif self._method == 'pseudo':
+ token = self._generate_pseudo_email(token)
+ return token
+
+ def _anonoymize_user(self, token):
+ """Handles removal/change of users."""
+ mention_regex = r'\B\@([\w\-]+)'
+ if self._method == 'delete':
+ token = re.sub(mention_regex, r'', token)
+ elif self._method == 'tag':
+ token = re.sub(mention_regex, self._user_token, token)
+ elif self._method == 'pseudo':
+ token = self._generate_pseudo_user(token)
+ return token
+
--
GitLab