Adapt base method of CheckAndFix

modules/harvest_tools/checkandfix.py

@@ -517,6 +517,9 @@ class CheckAndFix(object):
     def is_oai(self, record):
         """``True`` when the OAI is not defined in the record.
 
+        Note:
+            make sense only for record from cds.cern.ch or old.inspirehep.net
+
         Args:
             record (RecordPubli): record describing a publication.
 
@@ -526,6 +529,10 @@ class CheckAndFix(object):
         """
         self.logger.debug(f"{T6}check is oai")
 
+        # make no sense for record from new inspirehep.net (March 2020)
+        if record.host() == "inspirehep.net":
+            return True
+
         # field / subfield depends on the store
         test = ("oai" in record and "value" in record["oai"]) or \
                ("FIXME_OAI" in record and "id" in record["FIXME_OAI"])
@@ -780,12 +787,12 @@ class CheckAndFix(object):
         record.reformat_authors(fmt_rescue)
 
         if sort:
-            authors = (record["authors"][["last_name", "fmt_name"]]
+            authors = (record.df_authors[["last_name", "fmt_name"]]
                        .sort_values(by="last_name")
                        .fmt_name)
 
         else:
-            authors = (record["authors"].fmt_name
+            authors = (record.df_authors.fmt_name
                        .sort_index())
 
         # go back to the origin formatting

tests/basis/test_11_CheckAndFix_base.py

@@ -23,6 +23,11 @@ def reccds():
     return load_record("cds.cern.ch", 1951625)
 
 
+@pytest.fixture(scope="module")
+def recins():
+    return load_record("inspirehep.net", 1319638, shelf="literature")
+
+
 @pytest.fixture(scope="module")
 def svc():
     return CheckAndFix()
@@ -32,18 +37,23 @@ def test_constructor_11001(svc):
     assert svc.reg_institute == \
         r"(^|\|)Marseille, CPPM($|\|)|" \
         r"(^|\|)CPPM, Marseille($|\|)|" \
-        r"(^|\|)Centre de Physique des Particules de Marseille \(CPPM\)($|\|)"
+        r"(^|\|)Centre de Physique des Particules de Marseille \(CPPM\)($|\|)|" \
+        r"(^|\|)Aix Marseille Univ, CNRS/IN2P3, CPPM, Marseille, France($|\|)"
 
 
-def test_is_oai_cds_11002(svc, reccds):
+# ............................................................................
+#
+# cds.cern.ch record
+#
+def test_is_oai_cds_11010(svc, reccds):
     assert svc.is_oai(reccds)
 
 
-def test_is_bad_oai_cds_11003(svc, reccds):
+def test_is_bad_oai_cds_11011(svc, reccds):
     assert not svc.is_bad_oai_used(reccds)
 
 
-def test_temporary_record_cds_11004(svc, reccds):
+def test_temporary_record_cds_11012(svc, reccds):
 
     assert svc.temporary_record(reccds) is None
 
@@ -61,11 +71,11 @@ def test_temporary_record_cds_11004(svc, reccds):
         svc.temporary_record(recins)
 
 
-def test_authors_cds_11005(svc, reccds):
+def test_authors_cds_11013(svc, reccds):
     assert svc.authors(reccds) is None
 
 
-def test__get_author_rescue_list_cds_11006(svc, reccds):
+def test__get_author_rescue_list_cds_11014(svc, reccds):
     assert svc._get_author_rescue_list(reccds, 8, 7) == ["C. Adrover",
                                                          "S. Akar",
                                                          "E. Aslanides",
@@ -83,7 +93,7 @@ def test__get_author_rescue_list_cds_11006(svc, reccds):
                                                          "A. Tsaregorodtsev"]
 
 
-def test_my_affiliation_cds_11007(svc, reccds):
+def test_my_affiliation_cds_11015(svc, reccds):
     assert svc.my_affiliation(reccds, 8, 7) == "Marseille, CPPM"
 
     # a paper from NA62 -- no CPPM author
@@ -92,11 +102,66 @@ def test_my_affiliation_cds_11007(svc, reccds):
         svc.my_affiliation(recna62, id_project=8, id_team=7)
 
 
-def test_search_synonym_11008():
+def test_collaboration_cds_11016(svc, reccds):
+    assert svc.collaboration(reccds) is None
+
+
+# ............................................................................
+#
+# inspirehep.net record (March 2020 onward)
+#
+def test_is_oai_ins_11020(svc, recins):
+    assert svc.is_oai(recins)
+
+
+def test_is_bad_oai_ins_11021(svc, recins):
+    assert not svc.is_bad_oai_used(recins)
+
+
+def test_authors_ins_11023(svc, recins):
+    assert svc.authors(recins) is None
+
+
+def test__get_author_rescue_list_ins_11024(svc, recins):
+    assert svc._get_author_rescue_list(recins, 8, 7) == ["C. Adrover",
+                                                         "S. Akar",
+                                                         "E. Aslanides",
+                                                         "J. Cogan",
+                                                         "W. Kanso",
+                                                         "R. Le Gac",
+                                                         "O. Leroy",
+                                                         "G. Mancinelli",
+                                                         "E. Maurice",
+                                                         "A. Morda",
+                                                         "A. Mordà",
+                                                         "M. Perrin-Terrin",
+                                                         "M. Sapunov",
+                                                         "J. Serrano",
+                                                         "A. Tsaregorodtsev"]
+
+
+def test_my_affiliation_ins_11025(svc, recins):
+    assert svc.my_affiliation(recins, 8, 7) == "Marseille, CPPM"
+
+    # a paper from NA62 -- no CPPM author
+    recna62 = load_record("cds.cern.ch", 1434415)
+    with pytest.raises(CheckException):
+        svc.my_affiliation(recna62, id_project=8, id_team=7)
+
+
+def test_collaboration_ins_11026(svc, recins):
+    assert svc.collaboration(recins) is None
+
+
+# ............................................................................
+#
+# others
+#
+def test_search_synonym_11030():
     db = current.db
 
     # collaboration ANTARES, TANAMI (defined as synonym in the db)
-    record = load_record("inspirehep.net", 1342250)
+    record = load_record("inspirehep.net", 1342250, shelf="literature")
 
     colid = search_synonym(
         db.collaborations,
@@ -106,7 +171,7 @@ def test_search_synonym_11008():
     assert colid == 2
 
     # collaboration = ANTARES (defined as synonym in the db))
-    record = load_record("inspirehep.net", 718872)
+    record = load_record("inspirehep.net", 718872, shelf="literature")
 
     colid = search_synonym(
         db.collaborations,
@@ -115,6 +180,3 @@ def test_search_synonym_11008():
 
     assert colid == 2
 
-
-def test_collaboration_cds_11009(svc, reccds):
-    assert svc.collaboration(reccds) is None