Replace the function fix_amu by the method CheckAndFix.format_universities.

languages/fr-fr.py

@@ -297,8 +297,8 @@
 'insert new %s': 'insert new %s',
 'install': 'installé',
 'Institute': 'Institut',
+'Institute identifier in inspirehep.net.': 'Identifiant du laboratoire dans inspirehep.net.',
 'Institute number associated to CPPM authors': "Numéro de l'Institut associé aux auteurs du CPPM",
-'Institute identifier in inspirehep.net.': "Identifiant du laboratoire dans inspirehep.net.",
 'Invalid': 'Non conforme',
 "Invalid database table '%s'": "Invalid database table '%s'",
 'Invalid email': 'Invalid email',
@@ -486,6 +486,7 @@
 'Reject no %s authors': "Rejeté pas d'autheur(s) du %s",
 'Reject no author(s)': "Rejeté pas d'autheur(s)",
 'Reject no authors': "Rejeté pas d'auteurs",
+'Reject no authors of my institute': "Rejeté pas d'auteurs de mon laboratoire",
 'Reject no conference information': "Rejeté pas d'information sur la conférence",
 'Reject no CPPM authors': "Rejeté pas d'auteurs du CPPM",
 'Reject no OAI identifier': "Rejeté pas d'identifiant OAI",

modules/harvest_tools/__init__.py

@@ -5,7 +5,6 @@ and to push them in the database.
 """
 from base import (DRY_RUN,
                   family_name_fr,
-                  fix_amu,
                   format_author_fr,
                   ToolException)
 

modules/harvest_tools/base.py

@@ -25,29 +25,6 @@ def family_name_fr(full_name):
     return full_name[full_name.find(' ') + 1:]
 
 
-def fix_amu(record):
-    """Fix the name of the C{Aix Marseille University}
-
-    @type record: L{Record}
-
-    @rtype: unicode
-    @return: the university names separated by comma.
-
-    """
-    universities = record.these_universities()
-    for idx in range(len(universities)):
-        if re.search(current.app.reg_institute, universities[idx]):
-
-            year = re.search(r"(\d\d\d\d)", record.these_defense()).group(1)
-            if int(year) < 2012:
-                universities[idx] = \
-                    u"Université de la Méditerrannée Aix-Marseille II"
-            else:
-                universities[idx] = u"Aix Marseille Université"
-
-    return ', '.join(universities)
-
-
 def format_author_fr(name):
     """Format the author name according to French typographic rules,
     I{i.e.} C{J.-P. Doe}.

modules/harvest_tools/thesis.py

@@ -6,7 +6,7 @@ import re
 import traceback
 
 
-from base import family_name_fr, fix_amu, MSG_CRASH, MSG_LOAD
+from base import family_name_fr, MSG_CRASH, MSG_LOAD
 from invenio_tools import CheckException
 from publicationstool import PublicationsTool
 from plugin_dbui import get_id, UNDEF_ID
@@ -39,7 +39,7 @@ class Thesis(PublicationsTool):
         id_category = get_id(db.categories, code='PHD')
         oai_url = record.oai_url()
         title = record.title()
-        universities = fix_amu(record)
+        universities = ', '.join(record.these_universities())
 
         # extract the year from the defense date
         # this approach seems the most reliable
@@ -99,6 +99,7 @@ class Thesis(PublicationsTool):
             self.check.oai(record)
             self.check.submitted(record)
             self.check.year(record)
+            self.check.format_universities(record)
 
         except CheckException as e:
             self.logs[-1].reject(e, record.year())

modules/invenio_tools/checkandfix.py

@@ -47,7 +47,7 @@ MSG_NO_AUTHOR = "Reject no author(s)"
 MSG_NO_CONF = "Reject no conference information"
 MSG_NO_COUNTRY = "Reject invalid country"
 MSG_NO_DATE = "Reject no submission date"
-MSG_NO_MY_AUTHOR = "Reject no %s authors"
+MSG_NO_MY_AUTHOR = "Reject no authors of my institute"
 MSG_NO_OAI = "Reject no OAI identifier"
 MSG_NO_REF = "Reject incomplete paper reference"
 MSG_NO_YEAR = "Reject no publication year"
@@ -74,6 +74,8 @@ REG_CONF_DATES_2 = re.compile("(\d+) *([A-Z][a-z]{2}) *-? *(\d+) *([A-Z][a-z]{2}
 REG_CONF_DATES = re.compile(regex.REG_CONF_DATES)
 REG_SUBMITTED = re.compile(regex.REG_SUBMITTED)
 
+UNIVERSITY = "University"
+
 
 def load_record(host, record_id):
     """Helper function to load a single record from an invenio store.
@@ -482,6 +484,57 @@ class CheckAndFix(object):
                 record["773"][i]["p"] = editor
                 record["773"][i]["v"] = volume
 
+    def format_universities(self, record):
+        """Format the name of the university for PhD:
+
+            - Fix the name of Aix-Marseille University
+            - Replace U. by University
+
+        @type record: L{Record}
+        @param record:
+
+        """
+        # protection
+        if not record.is_thesis():
+            return
+
+        # CPPM: fix the name of Aix-Marseille university
+        if current.app.inspirehep_institute_id == 902989:
+
+            year = REG_YEAR.search(record.these_defense()).group(1)
+            if int(year) < 2012:
+                university = u"Université de la Méditerrannée Aix-Marseille II"
+            else:
+                university = u"Aix Marseille Université"
+
+            if u'502' in record and "b" in record[u'502']:
+                if isinstance(record[u'502']['b'], unicode):
+                    if "Marseille" in record[u'502']['b']:
+                        record[u'502']['b'] = university
+
+                elif isinstance(record[u'502']['b'], list):
+                    for i in xrange(len(record[u'502']['b'])):
+                        if "Marseille" in record[u'502']['b'][i]:
+                            record[u'502']['b'][i] = university
+
+        # Other: replace U. by University
+        else:
+            university = current.T(UNIVERSITY).decode("utf8")
+
+            if u'502' in record and "b" in record[u'502']:
+                if isinstance(record[u'502']['b'], unicode):
+                    value = record[u'502']['b']
+                    if "U." in value:
+                        value = value.replace('U.', university)
+                        record[u'502']['b'] = value
+
+                elif isinstance(record[u'502']['b'], list):
+                    for i in xrange(len(record[u'502']['b'])):
+                        value = record[u'502']['b'][i]
+                        if "U." in value:
+                            value = value.replace('U.', university)
+                            record[u'502']['b'][i] = value
+
     def my_authors(self, record, reference=[], cmpFct=None):
         """Check that authors of my institutes signed the record.
         Fill the meta data record.my_authors.
@@ -533,7 +586,7 @@ class CheckAndFix(object):
             record.my_authors = s
             return
 
-        raise CheckException(MSG_NO_MY_AUTHOR % reg_institute.encode("utf8"))
+        raise CheckException(MSG_NO_MY_AUTHOR)
 
     def oai(self, record):
         """Check that the OAI field is defined and well formed.

modules/invenio_tools/record.py

@@ -535,7 +535,7 @@ class Record(dict):
         return False
 
     def is_thesis(self):
-        """C{True} when the record corresponf to a thesis.
+        """C{True} when the record corresponds to a thesis.
 
         @rtype: bool
         @return:

tests/harvester/CheckAndFix/test_phd_cds1394605_fix.py

+# -*- coding: utf-8 -*-
+"""PHD
+
+    http://cds.cern.ch/record/1394605
+
+    Search for New Physics in the $B^0_s \to J/\psi\phi$ decay channel at LHCb
+    B. Khanji
+    16 Sep 2011
+
+Corrections are applied to the record.
+
+Note:
+    * Only the first author is defined
+    * Record submitted is not defined
+    * The year is not defined
+
+"""
+import pytest
+
+
+from harvest_tools import format_author_fr
+from invenio_tools import CheckAndFix, load_record
+
+
+@pytest.fixture(scope="module")
+def record():
+    rec = load_record('cds.cern.ch', 1394605)
+    svc = CheckAndFix()
+
+    svc.authors(rec)
+    svc.format_authors(rec, format_author_fr)
+    svc.my_authors(rec)
+    svc.submitted(rec)
+    svc.year(rec)
+
+    svc.format_universities(rec)
+
+    return rec
+
+
+def test_authors(record):
+    assert record.authors() == "B. Khanji"
+
+
+def test_first_author(record):
+    assert record.first_author() == "B. Khanji"
+
+
+def test_these_defense(record):
+    assert record.these_defense() == "16 Sep 2011"
+
+
+def test_these_directors(record):
+    assert record.these_directors() == "R. Le Gac, O. Leroy"
+
+
+def test_is_thesis(record):
+    assert record.is_thesis()
+
+
+def test_submitted(record):
+    assert record.submitted() == ['2011-09-16']
+
+
+def test_these_universities(record):
+    assert record.these_universities() == [u"Université de la Méditerrannée Aix-Marseille II"]
+
+
+def test_year(record):
+    assert record.year() == "2011"

tests/harvester/CheckAndFix/test_phd_cds1642541_fix.py

+# -*- coding: utf-8 -*-
+"""PHD
+
+    http://cds.cern.ch/record/1642541
+
+    Searches for B meson decays to purely leptonic final states
+    M. Perrin-Terrin
+    12 Jul 2013
+
+Corrections are applied to the record.
+
+Note:
+    * Only the first author is defined
+    * Record submitted is not defined
+    * The year is not defined
+
+"""
+import pytest
+
+
+from harvest_tools import format_author_fr
+from invenio_tools import CheckAndFix, load_record
+
+
+@pytest.fixture(scope="module")
+def record():
+    rec = load_record('cds.cern.ch', 1642541)
+    svc = CheckAndFix()
+
+    svc.authors(rec)
+    svc.format_authors(rec, format_author_fr)
+    svc.my_authors(rec)
+    svc.submitted(rec)
+    svc.year(rec)
+
+    svc.format_universities(rec)
+
+    return rec
+
+
+def test_authors(record):
+    assert record.authors() == "M. Perrin-Terrin"
+
+
+def test_first_author(record):
+    assert record.first_author() == "M. Perrin-Terrin"
+
+
+def test_these_defense(record):
+    assert record.these_defense() == "23 Sep 2013"
+
+
+def test_these_directors(record):
+    assert record.these_directors() == "G. Mancinelli, R. Le Gac"
+
+
+def test_is_thesis(record):
+    assert record.is_thesis()
+
+
+def test_submitted(record):
+    assert record.submitted() == ['2013-09-23']
+
+
+def test_these_universities(record):
+    assert record.these_universities() == [u"Aix Marseille Université"]
+
+
+def test_year(record):
+    assert record.year() == "2013"

tests/harvester/CheckAndFix/test_phd_cds2015250_fix.py

+# -*- coding: utf-8 -*-
+"""PHD
+
+    http://cds.cern.ch/record/2015250
+
+
+
+"""
+import pytest
+
+from gluon import current
+from harvest_tools import format_author_fr
+from invenio_tools import CheckAndFix, load_record
+
+
+def test_these_universities():
+    record = load_record('cds.cern.ch', 2015250)
+
+    current.app.inspirehep_institute_id = 9999
+
+    svc = CheckAndFix()
+    svc.format_universities(record)
+
+    assert record.these_universities() == [u"Milan Bicocca Université"]