Migrate Thesis.

modules/harvest_tools/thesis.py

@@ -8,7 +8,6 @@ import traceback
 from .automaton import Automaton
 from .base import MSG_CRASH, MSG_LOAD
 from .checkandfix import CheckException
-from invenio_tools import RecordThesis
 from plugin_dbui import get_id, UNDEF_ID
 
 
@@ -23,20 +22,25 @@ class Thesis(Automaton):
         """Check the content of the thesis in order to fix non conformities.
 
         Args:
-            record (RecordThesis): record describing a thesis.
+            record (RecordThesis):
+                record describing a thesis.
 
         Returns:
-            bool: ``False`` when a non conformity is found and can not be
-            corrected.
+            bool:
+                ``False`` when a non conformity is found and can not be
+                corrected.
 
         """
         if not Automaton.check_record(self, record):
             return False
 
+        if self.dbg:
+            print("check thesis record")
+
         try:
             self.check.is_thesis(record)
             self.check.submitted(record)
-            self.check.year(record)
+
             self.check.format_universities(record)
 
             self.check.format_authors(record, fmt="F. Last")
@@ -51,23 +55,18 @@ class Thesis(Automaton):
             print(traceback.format_exc())
             return False
 
-        if self.dbg:
-            print("check thesis record")
-
-        if isinstance(record, RecordThesis):
-            return True
-
-        self.logs[-1].reject(MSG_NO_THESIS, record=record)
-        return False
+        return True
 
     def insert_record(self, record):
         """Insert a thesis in the database.
 
         Args:
-            record (RecordThesis): record describing a thesis.
+            record (RecordThesis):
+                record describing a thesis.
 
         Returns:
-            int: one when the record is inserted / updated in the database
+            int:
+                one when the record is inserted / updated in the database
                 zero otherwise.
 
         """

tests/scan/test_Thesis.py

+# -*- coding: utf-8 -*-
+"""test_Thesis
+
+    * Harvester is Thesis
+    * Store is cds.cern.ch
+    * LHCb PHD for the current year
+    * Check that all error messages are expected
+
+"""
+import pytest
+from gluon import current
+from harvest_tools.thesis import Thesis
+from harvest_tools.factory import build_harvester_tool
+from test_tools import messages
+
+
+@pytest.fixture(scope="module")
+def harvester_messages():
+    return messages()
+
+
+def test_lhcb_acl(harvester_messages):
+    """Harvest LHCb thesis for a given year.
+
+    This test is useful to:
+        * debug an harvester
+        * profile its performance to see where the time is spent.
+        * compare different implementation to measure improvements.
+        * ...
+
+    """
+    # These parameter only make sense if you are inserting record in database
+    # Select the current year in order to test different case
+    db = current.db
+    id_team = 7         # LHCb
+    id_project = 8      # LHCb
+    id_category = 16    # PHD
+    year = current.request.now.year
+
+    # build the harvester
+    tool = build_harvester_tool(
+        db,
+        id_team,
+        id_project,
+        "theses",
+        id_category,
+        year_start=str(year),
+        year_end="",
+        dry_run=True,
+        debug=True)
+
+    assert isinstance(tool, Thesis)
+
+    # run the harvester
+    tool.process_url("cds.cern.ch", "LHCb Theses")
+
+    # analyse the log
+    # Number of article cannot be check since it evolve within a year
+    # Only test that there are no unexpected messages
+
+    msgs = set([el.txt for el in tool.logs])
+    assert msgs.issubset(harvester_messages)