Move the function recover_oai in the CheckAndFix class.

modules/harvest_tools/automaton.py

@@ -6,7 +6,7 @@ import re
 import traceback
 
 
-from base import MSG_FIX_ORIGIN, MSG_IN_DB, recover_oai, ToolException
+from base import MSG_FIX_ORIGIN, MSG_IN_DB, ToolException
 from gluon.storage import Storage
 from invenio_tools import (CheckAndFix,
                            InvenioStore,
@@ -280,6 +280,8 @@ class Automaton(object):
             print "check record"
 
         try:
+            self.check.recover_oai(record, self.harvester.host)
+
             if self.check.is_bad_oai_used(record):
                 self.logs[-1].idle(MSG_IN_DB, record.year())
                 return False
@@ -528,22 +530,13 @@ class Automaton(object):
                                  record_id=record.id(),
                                  title=record.title()))
 
-            # the OAI is not defined -- recover it
-            oai = record.oai()
-            if oai is None:
-                recover_oai(record, self.harvester.host)
-
-            # the OAI is not well --recover it
-            if not REG_OAI.match(oai):
-                recover_oai(record, self.harvester.host)
-
             # check that the record is well formed
             # repair non-conformity as far as possible
             if not self.check_record(record):
                 continue
 
             if self.dbg:
-                print "start loading in the database"
+                print "insert record in the database"
 
             # insert the record in the database
             self.insert_record(record)

modules/harvest_tools/base.py

@@ -14,10 +14,6 @@ MSG_FIX_ORIGIN = "Fixed the origin field"
 MSG_IN_DB = "Already in the database"
 MSG_LOAD = "Load in the database"
 
-MSG_INVALID_HOST = "Invalid host"
-
-OAI_INVENIO = "oai:%s:%s"
-
 
 def family_name_fr(full_name):
     """Extract the family name when the full name is encoded as C{J. Doe}.
@@ -145,28 +141,5 @@ def learn_my_authors(db,
         db.my_authors[row.id] = dict(authors=', '.join(database_authors))
 
 
-def recover_oai(record, host):
-    """Helper function to recover the OAI identifier when it is not defined
-    or not well form.
-
-    @type record: Record
-    @param record:
-
-    """
-    if host == "cds.cern.ch":
-        field, subfield = u"0248", "a"
-
-    elif host == "inspirehep.net":
-        field, subfield = u"909CO", "o"
-
-    else:
-        raise ValueError(MSG_INVALID_HOST)
-
-    if field not in record:
-        record[field] = dict()
-
-    record[field][subfield] = OAI_INVENIO % (host, record.id())
-
-
 class ToolException(Exception): pass
 

modules/invenio_tools/checkandfix.py

@@ -51,6 +51,8 @@ MONTHS = {u'Jan':'01',
           u'Nov':'11',
           u'Dec':'12'}
 
+MSG_INVALID_HOST = "Invalid host"
+
 MSG_NO_AUTHOR = "Reject no author(s)"
 MSG_NO_COUNTRY = "Reject invalid country"
 MSG_NO_CONF_DATE = "Reject no conference date"
@@ -71,6 +73,8 @@ MSG_WELL_FORMED_DATE = "Reject submission date is not well formed"
 
 MSG_WELL_FORMED_EDITOR = "Reject editor is not well formed"
 
+OAI_INVENIO = "oai:%s:%s"
+
 REG_COLLABORATION = re.compile(regex.REG_COLLABORATION)
 REG_CONF_DATES_1 = re.compile("(\d+) *-? *(\d+) *([A-Z][a-z]{2}) *(\d{4})")
 
@@ -740,6 +744,35 @@ class CheckAndFix(object):
                     self._repair_paper_reference(record)
                     return
 
+    def recover_oai(self, record, host):
+        """Recover the OAI identifier when it is not defined
+        or not well form.
+
+        @type record: Record
+        @param record:
+
+        @type host: unicode
+        @param host: either cds.cern.ch or inspirehep.net
+
+        """
+        oai = record.oai()
+        if oai is not None and REG_OAI.match(oai):
+            return
+
+        if host == "cds.cern.ch":
+            field, subfield = u"0248", "a"
+
+        elif host == "inspirehep.net":
+            field, subfield = u"909CO", "o"
+
+        else:
+            raise ValueError(MSG_INVALID_HOST)
+
+        if field not in record:
+            record[field] = dict()
+
+        record[field][subfield] = OAI_INVENIO % (host, record.id())
+
     def submitted(self, record):
         """Standardize the submitted date as YYYY-MM or YYYY-MM-DD.
         Look for alternative when it is not defined.