Improved method Record.oai.

modules/invenio_tools/checkandfix.py

@@ -525,57 +525,33 @@ class CheckAndFix(object):
         @raise CheckException:
 
         """
-        # the location of the OAI information depends on the store
-        # CDS: (248, a) or INSPIREHEP: (909CO, o)
-        if "0248" in record:
-            field, subfield = "0248", "a"
-
-        elif "909CO" in record:
-            field, subfield = "909CO", "o"
-
-        else:
+        # check that the OAI is defined
+        value = record.oai()
+        if not value:
             raise CheckException(MSG_NO_OAI)
 
-        myid = record.id()
-
-        # Clean OAI information.
-        # in some case OAI is a list,e.g when two records were entered
-        # for the same entry but one deleted.
-        # Select the OAI corresponding to the current ID.
-        if isinstance(record[field], list):
-            val = ''
-            for di in record[field]:
-                if di[subfield].endswith(myid):
-                    val = di
-                    break
-
-            if val:
-                record[field] = val
-            else:
-                raise CheckException(MSG_NO_OAI)
-
         # check that the OAI is well formed
-        m = REG_OAI.match(record[field][subfield])
-        if not m:
+        match = REG_OAI.match(value)
+        if not match:
                 raise CheckException(MSG_WELL_FORMED_OAI)
 
         # The id in the OAI field might be different from the record id.
         # In INVENIO there is a mechanism to redirect to the correct one
         #
         # The fix depend on the content of the database
-        if m.group(2) != myid:
+        if match.group(2) != myid:
             db = current.globalenv['db']
 
             # The record OAI is already used in the database. Do nothing
-            oai_url = OAI_URL % (m.group(1), m.group(2))
+            oai_url = OAI_URL % (match.group(1), match.group(2))
             if get_id(db.publications, origin=oai_url):
                 return
 
             # The OAI based on the record id is already used in the database.
             # Modify the record OAI
-            oai_url = OAI_URL % (m.group(1), myid)
+            oai_url = OAI_URL % (match.group(1), myid)
             if get_id(db.publications, origin=oai_url):
-                record[field][subfield] = OAI_INVENIO % (m.group(1), myid)
+                record[field][subfield] = OAI_INVENIO % (match.group(1), myid)
 
     def paper_reference(self, record):
         """Check that editor, page, volume and paper year are defined

modules/invenio_tools/iterrecord.py

@@ -244,7 +244,7 @@ class IterRecord(object):
         return di
 
     def _is_not_xml(self, xml):
-        """C{True} when the C{xml} sting is well formed.
+        """C{True} when the C{xml} string is not well formed.
 
         @type xml: unicode
         @param xml:

modules/invenio_tools/record.py

@@ -6,6 +6,7 @@ import pprint
 
 
 from base import OAI_URL, REG_OAI
+from exception import RecordException
 
 
 class Record(dict):
@@ -106,22 +107,10 @@ class Record(dict):
             self.__host = u''
             return self.__host
 
-        if isinstance(val, unicode):
-            match = REG_OAI.match(val)
-            if match:
-                self.__host = match.group(1)
-                return self.__host
-
-        # NOTE: in few case we can have a list
-        # see cds.cern.ch/record/1513204
-        # [u'oai:cds.cern.ch:1513204', u'oai:cds.cern.ch:1512766']
-        if isinstance(val, list):
-            for elt in val:
-                if elt.endswith(self.id()):
-                    match = REG_OAI.match(elt)
-                    if match:
-                        self.__host = match.group(1)
-                        return self.__host
+        match = REG_OAI.match(val)
+        if match:
+            self.__host = match.group(1)
+            return self.__host
 
         return u''
 
@@ -138,16 +127,35 @@ class Record(dict):
         """The Open Archive Initiative identifier
 
         @rtype: unicode
-        @return: the pattern of the string is "oai:host:record_id"
+        @return: the pattern of the string is "oai:host:record_id".
+        It is en empty string when not defined
 
         """
         # the location of the oai information depends on the store
         # cds: (248, a), inspirehep: (909C0, o)
         if u"0248" in self:
-            return self._get(u"0248", "a")
+            field, subfield = u"0248", "a"
 
         elif u"909CO" in self:
-            return self._get(u"909CO", "o")
+            field, subfield = u"909CO", "o"
+
+        else:
+            return u""
+
+        value = self._get(field, subfield)
+
+        # in some case OAI is a list, e.g when two records were entered
+        # for the same entry but one deleted.
+        # Select the OAI corresponding to the current ID.
+        if isinstance(value, list):
+            myid = self.id()
+            for el in value:
+                if el.endswith(myid):
+                    return el
+
+            return u""
+
+        return value
 
     def oai_url(self):
         """The Open Archive Initiative URL
@@ -157,10 +165,15 @@ class Record(dict):
         "http://inspirehep.net/record/123456" or an empty string
         when it is not defined
 
+
         """
         val = self.oai()
-        m = REG_OAI.match(val)
-        return OAI_URL % (m.group(1), m.group(2))
+        match = REG_OAI.match(val)
+
+        if match:
+            return OAI_URL % (match.group(1), match.group(2))
+
+        return u""
 
     def sysno(self):
         return self._get(u"970", "a")

tests/harvester/Record/test_non_conformities.py

@@ -8,9 +8,10 @@ Protection are add in the record method to correct them
 from invenio_tools import load_record
 
 
-def test_protection_host():
+def test_protection_oai():
+    """['oai:cds.cern.ch:1513204', 'oai:cds.cern.ch:1512766']"""
     record = load_record('cds.cern.ch', 1513204)
-    assert record.oai() == ['oai:cds.cern.ch:1513204', 'oai:cds.cern.ch:1512766']
+    assert record.oai() == 'oai:cds.cern.ch:1513204'
     assert record.host() == "cds.cern.ch"