Fix a bug in fix_acti2com_cppm (sublitted date).

modules/harvest_tools/__init__.py

@@ -16,7 +16,7 @@ from base import (DRY_RUN,
 
 from automaton import Automaton
 from articles import Articles
-from checkandfix import CheckAndFix
+from checkandfix import CheckAndFix, MONTHS
 from exception import CheckException, ToolException
 from factory import build_harvester_tool, get_harvester_tool
 from msg import Msg

scripts/fix_acti2com_cppm.py

@@ -35,8 +35,17 @@
 import re
 
 
+from harvest_tools import MONTHS
+
+
+reg1 = r'(\d{1,2}) ([A-Z][a-z]{2}) (\d{4})'
+reg2 = r'(\d{1,2})-\d{1,2} ([A-Z][a-z]{2}) (\d{4})'
+reg3 = r'(\d{1,2}) ([A-Z][a-z]{2}) - \d{1,2} [A-Z][a-z]{2} (\d{4})'
+REG_CONF_DATES = re.compile(r'%s|%s|%s' % (reg1, reg2, reg3))
+
 REG_ORIGIN = re.compile("https?://([a-z\.]+)/record/(\d+)")
 
+
 if __name__ == "__main__":
 
     import sys
@@ -70,15 +79,22 @@ if __name__ == "__main__":
         data["id_publishers"] = 1
         data["pages"] = ""
         data["volume"] = ""
-        data["submitted"] = ""
         data["publication_url"] = ""
         data["preprint"] = ""
 
         # the year is the one of the conference
-        conference_dates = data["conference_dates"]
-        if conference_dates:
-            data["year"] = int(data["conference_dates"][-4:])
-
+        # submitted date is when the conference start
+        match = REG_CONF_DATES.match(data["conference_dates"])
+        if match:
+            offset = 0
+            for i in xrange(3):
+                if match.group(1+i*3) is not None:
+                    offset = i*3
+                    break
+
+            data["year"] = match.group(offset+3)
+            month = MONTHS[match.group(offset+2)]
+            data["submitted"] = "%s-%s-%s" % (match.group(offset+3), month, match.group(offset+1))
         else:
             print "No conferences dates", data["id"]