Merge pull request #38 from rawouter/master

Fix for #37

datefinder.py

@@ -41,6 +41,7 @@ class DateFinder(object):
             \:
             (?P<minutes>\d{{1,2}})
             (\:(?<seconds>\d{{1,2}}))?
+            ([\.\,](?<microseconds>\d{{1,6}}))?
             \s*
             (?P<time_periods>{time_periods})?
             \s*
@@ -196,24 +197,29 @@ def _add_tzinfo(self, datetime_obj, tz_string):
         return datetime_obj.replace(tzinfo=tzinfo_match)
 
     def parse_date_string(self, date_string, captures):
-        # replace tokens that are problematic for dateutil
-        date_string, tz_string = self._find_and_replace(date_string, captures)
-
-        ## One last sweep after removing
-        date_string = date_string.strip(self.STRIP_CHARS)
-        ## Match strings must be at least 3 characters long
-        ## < 3 tends to be garbage
-        if len(date_string) < 3:
-            return None
-
+        # For well formatted string, we can already let dateutils parse them
+        # otherwise self._find_and_replace method might corrupt them
         try:
-            logger.debug('Parsing {0} with dateutil'.format(date_string))
             as_dt = parser.parse(date_string, default=self.base_date)
-        except Exception as e:
-            logger.debug(e)
-            as_dt = None
-        if tz_string:
-            as_dt = self._add_tzinfo(as_dt, tz_string)
+        except ValueError:
+            # replace tokens that are problematic for dateutil
+            date_string, tz_string = self._find_and_replace(date_string, captures)
+
+            ## One last sweep after removing
+            date_string = date_string.strip(self.STRIP_CHARS)
+            ## Match strings must be at least 3 characters long
+            ## < 3 tends to be garbage
+            if len(date_string) < 3:
+                return None
+
+            try:
+                logger.debug('Parsing {0} with dateutil'.format(date_string))
+                as_dt = parser.parse(date_string, default=self.base_date)
+            except Exception as e:
+                logger.debug(e)
+                as_dt = None
+            if tz_string:
+                as_dt = self._add_tzinfo(as_dt, tz_string)
         return as_dt
 
     def extract_date_strings(self, text, strict=False):
@@ -237,7 +243,6 @@ def extract_date_strings(self, text, strict=False):
             months = captures.get('months')
             timezones = captures.get('timezones')
             delimiters = captures.get('delimiters')
-            time = captures.get('time)')
             time_periods = captures.get('time_periods')
             extra_tokens = captures.get('extra_tokens')
 

tests/test_find_dates.py

@@ -44,6 +44,10 @@
         datetime(1994, 10, 27),
         datetime(1995,  6,  1)
     ]),
+    # Z dates with and without millis, from https://github.com/akoumjian/datefinder/issues/37
+    ("2017-02-03T09:04:08.001Z", datetime(2017, 2, 3, 9, 4, 8, 1000, tzinfo=pytz.utc)),
+    ("2017-02-03T09:04:08,00123Z", datetime(2017, 2, 3, 9, 4, 8, 1230, tzinfo=pytz.utc)),
+    ("2017-02-03T09:04:08Z", datetime(2017, 2, 3, 9, 4, 8, tzinfo=pytz.utc)),
 ])
 def test_find_date_strings(input_text, expected_date):
     if isinstance(expected_date,list):
@@ -53,4 +57,4 @@ def test_find_date_strings(input_text, expected_date):
         return_date = None
         for return_date in datefinder.find_dates(input_text):
             assert return_date == expected_date
-        assert return_date is not None # handles dates that were never matched
+        assert return_date is not None, 'Did not find date for test line: "{}"'.format(input_text) # handles dates that were never matched