Merge pull request #9701 from williamhogman/feature-dashed-quarters

ENH Make year and quarter dash-separatable

Author: jreback

doc/source/whatsnew/v0.16.1.txt

@@ -32,6 +32,9 @@ API changes
 
 
 
+- Add support for separating years and quarters using dashes, for
+  example 2014-Q1.  (:issue:`9688`)
+
 .. _whatsnew_0161.performance:
 
 Performance Improvements

pandas/tseries/tests/test_offsets.py

@@ -21,7 +21,7 @@
 from pandas import Series
 from pandas.tseries.frequencies import _offset_map
 from pandas.tseries.index import _to_m8, DatetimeIndex, _daterange_cache, date_range
-from pandas.tseries.tools import parse_time_string
+from pandas.tseries.tools import parse_time_string, DateParseError
 import pandas.tseries.offsets as offsets
 
 from pandas.io.pickle import read_pickle
@@ -3010,12 +3010,33 @@ def test_get_offset():
                                     (name, expected, offset))
 
 
-def test_parse_time_string():
-    (date, parsed, reso) = parse_time_string('4Q1984')
-    (date_lower, parsed_lower, reso_lower) = parse_time_string('4q1984')
-    assert date == date_lower
-    assert parsed == parsed_lower
-    assert reso == reso_lower
+class TestParseTimeString(tm.TestCase):
+
+    def test_parse_time_string(self):
+        (date, parsed, reso) = parse_time_string('4Q1984')
+        (date_lower, parsed_lower, reso_lower) = parse_time_string('4q1984')
+        self.assertEqual(date, date_lower)
+        self.assertEqual(parsed, parsed_lower)
+        self.assertEqual(reso, reso_lower)
+
+    def test_parse_time_quarter_w_dash(self):
+        # https://github.com/pydata/pandas/issue/9688
+        pairs = [
+            ('1988-Q2', '1988Q2'),
+            ('2Q-1988', '2Q1988'),
+        ]
+
+        for dashed, normal in pairs:
+            (date_dash, parsed_dash, reso_dash) = parse_time_string(dashed)
+            (date, parsed, reso) = parse_time_string(normal)
+
+            self.assertEqual(date_dash, date)
+            self.assertEqual(parsed_dash, parsed)
+            self.assertEqual(reso_dash, reso)
+
+        self.assertRaises(DateParseError, parse_time_string, "-2Q1992")
+        self.assertRaises(DateParseError, parse_time_string, "2-Q1992")
+        self.assertRaises(DateParseError, parse_time_string, "4-4Q1992")
 
 
 def test_get_standard_freq():

pandas/tseries/tools.py

@@ -383,10 +383,10 @@ def calc_with_mask(carg,mask):
     return None
 
 # patterns for quarters like '4Q2005', '05Q1'
-qpat1full = re.compile(r'(\d)Q(\d\d\d\d)')
-qpat2full = re.compile(r'(\d\d\d\d)Q(\d)')
-qpat1 = re.compile(r'(\d)Q(\d\d)')
-qpat2 = re.compile(r'(\d\d)Q(\d)')
+qpat1full = re.compile(r'(\d)Q-?(\d\d\d\d)')
+qpat2full = re.compile(r'(\d\d\d\d)-?Q(\d)')
+qpat1 = re.compile(r'(\d)Q-?(\d\d)')
+qpat2 = re.compile(r'(\d\d)-?Q(\d)')
 ypat = re.compile(r'(\d\d\d\d)$')
 has_time = re.compile('(.+)([\s]|T)+(.+)')
 
@@ -424,18 +424,18 @@ def parse_time_string(arg, freq=None, dayfirst=None, yearfirst=None):
                                         second=0, microsecond=0)
 
     # special handling for possibilities eg, 2Q2005, 2Q05, 2005Q1, 05Q1
-    if len(arg) in [4, 6]:
+    if len(arg) in [4, 5, 6, 7]:
         m = ypat.match(arg)
         if m:
             ret = default.replace(year=int(m.group(1)))
             return ret, ret, 'year'
 
         add_century = False
-        if len(arg) == 4:
+        if len(arg) > 5:
+            qpats = [(qpat1full, 1), (qpat2full, 0)]
+        else:
             add_century = True
             qpats = [(qpat1, 1), (qpat2, 0)]
-        else:
-            qpats = [(qpat1full, 1), (qpat2full, 0)]
 
         for pat, yfirst in qpats:
             qparse = pat.match(arg)