fixes for dates etc

[librarian.git] / librarian / dcparser.py

diff --git a/librarian/dcparser.py b/librarian/dcparser.py
index abee11b..7418f70 100644 (file)
--- a/librarian/dcparser.py
+++ b/librarian/dcparser.py
@@ -6,6 +6,8 @@
 from xml.parsers.expat import ExpatError
 from datetime import date
 import time
+import re
+from librarian.util import roman_to_int
 
 from librarian import (ValidationError, NoDublinCore, ParseError, DCNS, RDFNS,
                        WLURI)
@@ -62,10 +64,24 @@ class Person(object):
 
 def as_date(text):
     try:
-        try:
-            t = time.strptime(text, '%Y-%m-%d')
-        except ValueError:
-            t = time.strptime(text, '%Y')
+        # check out the "N. poł X w." syntax
+        if isinstance(text, str): text = text.decode("utf-8")
+        m = re.match(u"(?:([12]) *poł[.]? )?([MCDXVI]+) *w[.]?", text)
+        if m:
+            
+            half = m.groups()[0]
+            if half is not None: 
+                half = int(half)
+            else: 
+                half = 1
+            century = roman_to_int(str(m.groups()[1]))
+            t = ((century*100 + (half-1)*50), 1, 1)
+        else:
+            text = re.sub(r"(po|ok[.]?) *", "", text)
+            try:
+                t = time.strptime(text, '%Y-%m-%d')
+            except ValueError:
+                t = time.strptime(re.split(r'[-/]', text)[0], '%Y')
         return date(t[0], t[1], t[2])
     except ValueError, e:
         raise ValueError("Unrecognized date format. Try YYYY-MM-DD or YYYY.")
@@ -173,6 +189,7 @@ class WorkInfo(object):
             as_person, salias='technical_editor', multiple=True, default=[]),
         Field( DCNS('contributor.funding'), 'funders',
             salias='funder', multiple=True, default=[]),
+        Field( DCNS('contributor.thanks'), 'thanks', required=False),
 
         Field( DCNS('date'), 'created_at', as_date),
         Field( DCNS('date.pd'), 'released_to_public_domain_at', as_date, required=False),