facebook email: handle other datetime string formats

fixes https://console.cloud.google.com/errors/CKORxvuphMyeIw

granary/facebook.py

@@ -57,6 +57,7 @@
 import xml.sax.saxutils
 
 from bs4 import BeautifulSoup
+import dateutil.parser
 import mf2util
 from oauth_dropins.webutil import util
 
@@ -1669,12 +1670,7 @@ def email_to_object(cls, html):
     if not comment:
       return None
 
-    # example email date/time string: 'December 14 at 12:35 PM'
-    published = datetime.strptime(when.get_text(strip=True), '%B %d at %I:%M %p')\
-                        .replace(year=now_fn().year)
-
     obj = {
-      'published': published.isoformat(util.T),
       'author': {
         'objectType': 'person',
         'displayName': name,
@@ -1685,6 +1681,17 @@ def email_to_object(cls, html):
       'to': [{'objectType':'group', 'alias':'@public'}],
     }
 
+    # try to parse datetime string. examples seen in the wild:
+    #   December 14 at 12:35 PM
+    #   5 July at 21:50
+    when = when.get_text(strip=True)
+    try:
+      parsed = dateutil.parser.parse(when, default=now_fn())
+      obj['published'] = parsed.isoformat(util.T)
+    except (ValueError, OverflowError):
+      logging.warning("Couldn't parse datetime string %r", when, exc_info=True)
+
+    # extract Facebook post ID from URL
     url_parts = urllib.parse.urlparse(resp_url)
     path = url_parts.path.strip('/').split('/')
     url_params = urllib.parse.parse_qs(url_parts.query)

granary/tests/test_facebook.py

@@ -3152,6 +3152,28 @@ def test_email_to_object_comment_different_text(self):
     email = COMMENT_EMAIL_USERNAME.replace('commented on your', 'commented on')
     self.assert_equals(EMAIL_COMMENT_OBJ_USERNAME, self.fb.email_to_object(email))
 
+  def test_email_to_object_comment_different_datetime_format(self):
+    """https://console.cloud.google.com/errors/CKORxvuphMyeIw
+    """
+    facebook.now_fn().AndReturn(datetime(1999, 1, 1))
+    self.mox.ReplayAll()
+
+    email = COMMENT_EMAIL_USERNAME.replace('December 14 at 12:35 PM',
+                                           '14 December at 12:35')
+    self.assert_equals(EMAIL_COMMENT_OBJ_USERNAME, self.fb.email_to_object(email))
+
+  def test_email_to_object_comment_bad_datetime(self):
+    """https://console.cloud.google.com/errors/CKORxvuphMyeIw
+    """
+    facebook.now_fn().AndReturn(datetime(1999, 1, 1))
+    self.mox.ReplayAll()
+
+    email = COMMENT_EMAIL_USERNAME.replace('December 14 at 12:35 PM',
+                                           'asdf 29 qwert')
+    expected = copy.deepcopy(EMAIL_COMMENT_OBJ_USERNAME)
+    del expected['published']
+    self.assert_equals(expected, self.fb.email_to_object(email))
+
   def test_email_to_object_like(self):
     facebook.now_fn().AndReturn(datetime(1999, 1, 1))
     self.mox.ReplayAll()

requirements.txt

@@ -9,6 +9,7 @@ mf2py>=1.1.2
 mf2util>=0.5.0
 mox3>=0.24.0
 oauth-dropins>=1.14
+python-dateutil
 requests>=2.10.0
 requests-toolbelt>=0.6.2
 urllib3>=1.14

setup.py

@@ -53,6 +53,7 @@ def __init__(self, *args, **kwargs):
           'mf2py>=1.1.2',
           'mf2util>=0.5.0',
           'oauth-dropins>=1.14',
+          'python-dateutil',
           'requests-toolbelt>=0.6.2',
           'requests>=2.10.0',
           'urllib3>=1.14',