tests for handling "2 poł XIX w." dates + utf8 check

diff --git a/librarian/dcparser.py b/librarian/dcparser.py
index bcee932..079c434 100644 (file)
--- a/librarian/dcparser.py
+++ b/librarian/dcparser.py
@@ -65,7 +65,8 @@ class Person(object):
 def as_date(text):
     try:
         # check out the "N. poł X w." syntax
-        m = re.match(u"([12]) *poł[.]? ([MCDXVI]+) .*[.]?", text)
+        if isinstance(text, str): text = text.decode("utf-8")
+        m = re.match(u"([12]) *poł[.]? ([MCDXVI]+) *w[.]?", text)
         if m:
             half = int(m.groups()[0])
             century = roman_to_int(str(m.groups()[1]))

diff --git a/tests/test_dcparser.py b/tests/test_dcparser.py
index ee29bc9..a0845a0 100644 (file)
--- a/tests/test_dcparser.py
+++ b/tests/test_dcparser.py
@@ -9,7 +9,7 @@ from nose.tools import *
 from os.path import splitext
 from tests.utils import get_all_fixtures
 import codecs
-
+from datetime import date
 
 def check_dcparser(xml_file, result_file):
     xml = file(xml_file).read()
@@ -46,3 +46,7 @@ def test_serialize():
     for fixture in get_all_fixtures('dcparser', '*.xml'):
         yield check_serialize, fixture
 
+def test_as_date():
+    assert_equals(dcparser.as_date(u"2010-10-03"), date(2010, 10, 03))
+    assert_equals(dcparser.as_date("2011"), date(2011, 1, 1))
+    assert_equals(dcparser.as_date("2 poł. XIX w."), date(1950, 1, 1))