Migrate Automaton.

modules/harvest_tools/checkandfix.py

@@ -469,7 +469,7 @@ class CheckAndFix(object):
         value = record.oai()
         match = REG_OAI.match(value)
 
-        if match.group(2) != record.id():
+        if int(match.group(2)) != record.id():
             db = self.db
 
             # a record with the bad OAI exists in the database

modules/harvest_tools/msg.py

@@ -129,7 +129,7 @@ class Msg(Storage):
                             self.synonym = record.paper_editor()
 
         if year is None and record is not None:
-            year = record.year()
+            year = record.submitted()
 
         self._set(txt, year, translate)
 

tests/basis/test_09_Automaton.py

+# -*- coding: utf-8 -*-
+"""test_08_Automaton
+
+    * Test Automaton methods
+
+"""
+import pytest
+from gluon import current
+from harvest_tools.automaton import Automaton, ToolException
+from harvest_tools.msgcollection import MsgCollection
+from invenio_tools.inveniostore import InvenioStore
+
+
+@pytest.fixture(scope="module")
+def svc():
+    # article for lhcb and forthe current year
+    return Automaton(current.db,
+                     7,
+                     8,
+                     "article",
+                     2,
+                     year_start=current.request.now.year,
+                     dry_run=True,
+                     debug=True)
+
+
+def test_process_recid(svc):
+    """Test the deepest method to retrieve a record.
+
+    """
+    # mimic high level stage process_collection and process_url
+    collection = u"LHCb Papers"
+
+    svc.harvester.host = u"cds.cern.ch"
+    svc.harvester.collections = collection
+    svc.store = InvenioStore()
+
+    ctitle = "LHCb / article / %s" % collection
+    svc.collection_logs.append(MsgCollection(title=ctitle))
+
+    # get a list of ids
+    kwargs = svc._search_parameters(collection)
+    recids = svc.store.get_ids(**kwargs)
+    assert len(recids) > 0
+
+    # try with the oldest one
+    svc.dbg = True
+    recid = recids[-1]
+    assert svc.process_recid(recid) is None
+
+    assert len(svc.logs) == 1
+    assert svc.logs[-1].action is None
+    assert svc.logs[-1].txt is None
+
+    # reset
+    svc.collection_logs = []
+    svc.logs = []
+    del svc.harvester.host
+    del svc.harvester.collections
+    del svc.store
+
+
+def test_process_collection(svc):
+
+    # mimic the previous stage process_url
+    collection = u"LHCb Papers"
+
+    svc.harvester.host = u"cds.cern.ch"
+    svc.harvester.collections = collection
+    svc.store = InvenioStore()
+
+    # do it
+    svc.dbg = True
+    assert svc.process_collection(u"LHCb Papers") is None
+
+    # reset
+    svc.collection_logs = []
+    svc.logs = []
+    del svc.harvester.host
+    del svc.harvester.collections
+    del svc.store
+
+
+def test_process_url_cds(svc):
+    svc.dbg = True
+    assert svc.process_url("cds.cern.ch", u"LHCb Papers") is None
+
+
+def test_process_url_ins(svc):
+    svc.dbg = True
+    assert svc.process_url(
+        "inspirehep.net", "find cn lhcb and tc p and not tc c") is None