Handle all date formats in a primitive way (always take the beginning of period)

[librarian.git] / librarian / dcparser.py

diff --git a/librarian/dcparser.py b/librarian/dcparser.py
index eddd8e5..1df1e5a 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)
@@ -61,11 +63,44 @@ class Person(object):
         return 'Person(last_name=%r, first_names=*%r)' % (self.last_name, self.first_names)
 
 def as_date(text):
+    """Dates for digitization of pictures. It seems we need the following:
+ranges:                '1350-1450',
+centuries:     "XVIII w.'
+half centuries/decades: '2 poł. XVIII w.', 'XVII w., l. 20'
+later-then: 'po 1450'
+circa 'ok. 1813-1814', 'ok.1876-ok.1886
+turn: 1893/1894
+for now we will translate this to some single date losing information of course.
+    """
     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")
+        century_format = u"(?:([12]) *poł[.]? +)?([MCDXVI]+) *w[.,]*(?: *l[.]? *([0-9]+))?"
+        vague_format = u"(?:po *|ok. *)([0-9]+)"
+
+        m = re.match(century_format, text)
+        m2 = re.match(vague_format, text)
+        if m:
+            half = m.group(1)
+            decade = m.group(3)
+            century = roman_to_int(str(m.group(2)))
+            if half is not None:
+                if decade is not None:
+                    raise ValueError("Bad date format. Cannot specify both half and decade of century")
+                half = int(half)
+                t = ((century*100 + (half-1)*50), 1, 1)
+            else:
+                decade = int(decade or 0)
+                t = ((century*100 + decade), 1, 1)
+        elif m2:
+            year = int(m2.group(1))
+            t = (year, 1, 1)
+        else:
+            try:
+                t = time.strptime(text, '%Y-%m-%d')
+            except ValueError:
+                t = time.strptime(text, '%Y')
+
         return date(t[0], t[1], t[2])
     except ValueError, e:
         raise ValueError("Unrecognized date format. Try YYYY-MM-DD or YYYY.")
@@ -171,6 +206,9 @@ class WorkInfo(object):
             as_person, salias='editor', multiple=True, default=[]),
         Field( DCNS('contributor.technical_editor'), 'technical_editors',
             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),