From 1c491217bff7dc2c55d9ec9a0bcd2c9402119b8d Mon Sep 17 00:00:00 2001
From: Christopher Corley <cscorley@ua.edu>
Date: Mon, 24 Dec 2012 12:57:01 -0600
Subject: Add fuzzy parsing that additionally returns the skipped substrings.

---
 dateutil/parser.py | 40 ++++++++++++++++++++++++++++++++++------
 test.py            | 16 +++++++++++++---
 2 files changed, 47 insertions(+), 9 deletions(-)

diff --git a/dateutil/parser.py b/dateutil/parser.py
index a2604a3..aef8362 100644
--- a/dateutil/parser.py
+++ b/dateutil/parser.py
@@ -174,7 +174,7 @@ class parserinfo(object):
     # m from a.m/p.m, t from ISO T separator
     JUMP = [" ", ".", ",", ";", "-", "/", "'",
             "at", "on", "and", "ad", "m", "t", "of",
-            "st", "nd", "rd", "th"] 
+            "st", "nd", "rd", "th"]
 
     WEEKDAYS = [("Mon", "Monday"),
                 ("Tue", "Tuesday"),
@@ -305,7 +305,10 @@ class parser(object):
         if not default:
             default = datetime.datetime.now().replace(hour=0, minute=0,
                                                       second=0, microsecond=0)
-        res = self._parse(timestr, **kwargs)
+
+
+        res, skipped_tokens = self._parse(timestr, **kwargs)
+
         if res is None:
             raise ValueError("unknown string format")
         repl = {}
@@ -339,6 +342,10 @@ class parser(object):
                 ret = ret.replace(tzinfo=tz.tzutc())
             elif res.tzoffset:
                 ret = ret.replace(tzinfo=tz.tzoffset(res.tzname, res.tzoffset))
+
+        if skipped_tokens:
+            return ret, skipped_tokens
+
         return ret
 
     class _result(_resultbase):
@@ -346,7 +353,10 @@ class parser(object):
                      "hour", "minute", "second", "microsecond",
                      "tzname", "tzoffset"]
 
-    def _parse(self, timestr, dayfirst=None, yearfirst=None, fuzzy=False):
+    def _parse(self, timestr, dayfirst=None, yearfirst=None, fuzzy=False, fuzzy_with_tokens=False):
+        if fuzzy_with_tokens:
+            fuzzy = True
+
         info = self.info
         if dayfirst is None:
             dayfirst = info.dayfirst
@@ -354,6 +364,13 @@ class parser(object):
             yearfirst = info.yearfirst
         res = self._result()
         l = _timelex.split(timestr)
+
+
+        # keep up with the last token skipped so we can recombine
+        # consecutively skipped tokens (-2 for when i begins at 0).
+        last_skipped_token_i = -2
+        skipped_tokens = list()
+
         try:
 
             # year/month/day list
@@ -387,7 +404,7 @@ class parser(object):
                             res.minute = int(s[2:])
                     elif len_li == 6 or (len_li > 6 and l[i-1].find('.') == 6):
                         # YYMMDD or HHMMSS[.ss]
-                        s = l[i-1] 
+                        s = l[i-1]
                         if not ymd and l[i-1].find('.') == -1:
                             ymd.append(info.convertyear(int(s[:2])))
                             ymd.append(int(s[2:4]))
@@ -636,6 +653,13 @@ class parser(object):
                 if not (info.jump(l[i]) or fuzzy):
                     return None
 
+                if last_skipped_token_i == i - 1:
+                    # recombine the tokens
+                    skipped_tokens[-1] += l[i]
+                else:
+                    # just append
+                    skipped_tokens.append(l[i])
+                last_skipped_token_i = i
                 i += 1
 
             # Process year/month/day
@@ -705,7 +729,11 @@ class parser(object):
 
         if not info.validate(res):
             return None
-        return res
+
+        if fuzzy_with_tokens:
+            return res, tuple(skipped_tokens)
+
+        return res, None
 
 DEFAULTPARSER = parser()
 def parse(timestr, parserinfo=None, **kwargs):
@@ -888,7 +916,7 @@ class _tzparser(object):
 
         except (IndexError, ValueError, AssertionError):
             return None
-        
+
         return res
 
 
diff --git a/test.py b/test.py
index 80a7776..4ec2c1f 100755
--- a/test.py
+++ b/test.py
@@ -3447,7 +3447,7 @@ class ParserTest(unittest.TestCase):
     def testLongMonth(self):
         self.assertEqual(parse("October", default=self.default),
                          datetime(2003, 10, 25))
-        
+
     def testZeroYear(self):
         self.assertEqual(parse("31-Dec-00", default=self.default),
                          datetime(2000, 12, 31))
@@ -3458,6 +3458,16 @@ class ParserTest(unittest.TestCase):
         self.assertEqual(parse(s, fuzzy=True),
                          datetime(2003, 9, 25, 10, 49, 41,
                                   tzinfo=self.brsttz))
+    def testFuzzyWithTokens(self):
+        s = "Today is 25 of September of 2003, exactly " \
+            "at 10:49:41 with timezone -03:00."
+        self.assertEqual(parse(s, fuzzy_with_tokens=True),
+                (datetime(2003, 9, 25, 10, 49, 41,
+                    tzinfo=self.brsttz),
+                    ('Today is ', 'of ', ', exactly at ',
+                        ' with timezone ', '.')
+                    )
+                )
 
     def testExtraSpace(self):
         self.assertEqual(parse("  July   4 ,  1976   12:01:02   am  "),
@@ -3907,7 +3917,7 @@ END:VTIMEZONE
     def testStrCmp1(self):
         self.assertEqual(tzstr("EST5EDT"),
                          tzstr("EST5EDT4,M4.1.0/02:00:00,M10-5-0/02:00"))
-        
+
     def testStrCmp2(self):
         self.assertEqual(tzstr("EST5EDT"),
                          tzstr("EST5EDT,4,1,0,7200,10,-1,0,7200,3600"))
@@ -3930,7 +3940,7 @@ END:VTIMEZONE
         tz = tzfile(BytesIO(base64.decodestring(self.TZFILE_EST5EDT)))
         self.assertEqual(datetime(2003, 4, 6, 1, 59, tzinfo=tz).tzname(), "EST")
         self.assertEqual(datetime(2003, 4, 6, 2, 00, tzinfo=tz).tzname(), "EDT")
-        
+
     def testFileEnd1(self):
         tz = tzfile(BytesIO(base64.decodestring(self.TZFILE_EST5EDT)))
         self.assertEqual(datetime(2003, 10, 26, 0, 59, tzinfo=tz).tzname(), "EDT")
-- 
cgit v1.2.1