fix issue #15 by change regexp on _MULTI_QUOTE_HDR_REGEX

email_reply_parser/__init__.py

@@ -37,10 +37,10 @@ class EmailMessage(object):
     """
 
     SIG_REGEX = re.compile(r'(--|__|-\w)|(^Sent from my (\w+\s*){1,3})')
-    QUOTE_HDR_REGEX = re.compile('On.*wrote:$')
+    QUOTE_HDR_REGEX = re.compile('On\s((?!\sOn\s).)+wrote:\s*$')
     QUOTED_REGEX = re.compile(r'(>+)')
     HEADER_REGEX = re.compile(r'^\*?(From|Sent|To|Subject):\*? .+')
-    _MULTI_QUOTE_HDR_REGEX = r'(?!On.*On\s.+?wrote:)(On\s(.+?)wrote:)'
+    _MULTI_QUOTE_HDR_REGEX =r'(\bOn\s((?!\bOn\b).){2,100}wrote:)'
     MULTI_QUOTE_HDR_REGEX = re.compile(_MULTI_QUOTE_HDR_REGEX, re.DOTALL | re.MULTILINE)
     MULTI_QUOTE_HDR_REGEX_MULTILINE = re.compile(_MULTI_QUOTE_HDR_REGEX, re.DOTALL)
 

test/emails/email_issue_15.txt

@@ -0,0 +1,15 @@
+And this is a response to the test response.
+On function On  wrote:
+
+On Fri, Jan 5, 2018 at 12:34 PM, Adam Taylor <sampleaddress@example.com>
+wrote: 
+
+> And this is a test response.
+>
+> On Fri, Jan 5, 2018 at 12:34 PM, Adam Taylor <sampleaddress@example.com>
+> wrote:
+>
+>> This is just a test.
+>>
+>
+>

test/test_email_reply_parser.py

@@ -7,6 +7,8 @@
 
 sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
 from email_reply_parser import EmailReplyParser
+from email_reply_parser import EmailMessage    
+
 
 
 class EmailMessageTest(unittest.TestCase):
@@ -196,6 +198,40 @@ def get_email(self, name):
             text = f.read()
         return EmailReplyParser.read(text)
 
-
+    def test_issue_15(self):              
+        message = self.get_email("email_issue_15")
+        self.assertEqual('And this is a response to the test response.\nOn function On  wrote:', message.reply)
+
+    def test_multi_quater_reg(self):
+        test_string='On Fri, Jan 5, 2018 at 12:39 PM, Adam Taylor <sampleaddress@example.com>\n\
+            wrote: > And this is a test response.\n\
+            >\n\
+            > On Fri, Jan 5, 2018 at 12:34 PM, Adam Taylor <sampleaddress@example.com>\n\
+            > wrote:'
+        wrong_result='On Fri, Jan 5, 2018 at 12:34 PM, Adam Taylor <sampleaddress@example.com>\n\
+            > wrote:'
+        right_result='On Fri, Jan 5, 2018 at 12:39 PM, Adam Taylor <sampleaddress@example.com>\n\
+            wrote:'
+        #the original reg,find the second reply which is 'at 12:34 PM'
+        _MULTI_QUOTE_HDR_REGEX =r'(?!On.*On\s.+?wrote:)(On\s(.+?)wrote:)'
+        MULTI_QUOTE_HDR_REGEX = re.compile(_MULTI_QUOTE_HDR_REGEX, re.DOTALL | re.MULTILINE)
+        is_multi_quote_header = MULTI_QUOTE_HDR_REGEX.search(test_string)
+        self.assertEqual(wrong_result,is_multi_quote_header.groups()[0])
+
+        #the new reg,find the first reply 
+        # r'(\bOn\s((?!\bOn\b).){2,100}wrote:)'
+        NEW_MULTI_QUOTE_HDR_REGEX=re.compile(EmailMessage._MULTI_QUOTE_HDR_REGEX, re.DOTALL | re.MULTILINE)
+        self.assertEqual(right_result,NEW_MULTI_QUOTE_HDR_REGEX.search(test_string).groups()[0])
+
+        #test greedy case, don't get 'On functiOn On wrote:\n'
+        greedy_string=' '.join(['On functiOn On wrote:\n',test_string])
+        self.assertEqual(wrong_result,MULTI_QUOTE_HDR_REGEX.search(greedy_string).groups()[0])
+        self.assertEqual(right_result,NEW_MULTI_QUOTE_HDR_REGEX.search(greedy_string).groups()[0])
+
+        #test On, On
+        greedy_string_more=' '.join([' On functiOn with an other On, On wrote: ',test_string])
+        self.assertEqual(wrong_result,MULTI_QUOTE_HDR_REGEX.search(greedy_string_more).groups()[0])
+        self.assertEqual(right_result,NEW_MULTI_QUOTE_HDR_REGEX.search(greedy_string_more).groups()[0])
+
 if __name__ == '__main__':
     unittest.main()