Add link unshortener

helpers.py

@@ -3,6 +3,7 @@
 from collections import namedtuple
 from datetime import datetime
 from termcolor import colored
+import requests
 
 Response = namedtuple('Response', 'command_status message')
 
@@ -41,3 +42,32 @@ def only_blacklists_changed(diff):
     files_changed = diff.split()
     non_blacklist_files = [f for f in files_changed if f not in blacklist_files]
     return not bool(non_blacklist_files)
+
+
+# FAIR WARNING: Sending HEAD requests to resolve a shortened link is generally okay - there aren't
+# as many exploits that work on just HEAD responses. If you specify sending a GET request, you
+# acknowledge that this will fetch the full, potentially unsafe response from the shortener.
+def unshorten_link(url, request_type='HEAD', explicitly_ignore_security_warning=False):
+    requesters = {
+        'GET': requests.get,
+        'HEAD': requests.head
+    }
+    if request_type not in requesters:
+        raise KeyError('Unavailable request_type {}'.format(request_type))
+    if request_type == 'GET' and not explicitly_ignore_security_warning:
+        raise SecurityError('Potentially unsafe request type GET not acknowledged')
+
+    requester = requesters[request_type]
+    response_code = 301
+    headers = {'User-Agent': 'SmokeDetector/git (+https://github.com/Charcoal-SE/SmokeDetector)'}
+    while response_code in [301, 302, 303, 307, 308]:
+        res = requester(url, headers=headers)
+        response_code = res.status_code
+        if 'Location' in res.headers:
+            url = res.headers['Location']
+
+    return url
+
+
+class SecurityError(Exception):
+    pass

test/test_helpers.py

@@ -0,0 +1,12 @@
+import pytest
+import helpers
+
+
+@pytest.mark.parametrize('shortened, original', [
+    ('https://goo.gl/kAb9rz', 'https://charcoal-se.org/'),
+    ('https://bit.ly/2jhMbxn', 'https://charcoal-se.org/'),
+    ('https://tinyurl.com/y7ba4pth', 'https://charcoal-se.org/'),
+    ('https://tiny.cc/gsbbpy', 'https://charcoal-se.org/')
+])
+def test_unshorten_link(shortened, original):
+    assert helpers.unshorten_link(shortened) == original