Polish and improve check and fix for RecordCdsThesis

modules/harvest_tools/thesis.py

@@ -32,7 +32,7 @@ class Thesis(Automaton):
         """
         self.logger.debug(f"{T4}check and fix record (thesis)")
 
-        if not isinstance(record, (RecordCdsThesis, RecordHepThesis)):
+        if record.subtype() == "thesis":
             self.logs[-1].reject(MSG_NOT_THESIS, record)
             return False
 

modules/store_tools/recordcdspubli.py

@@ -833,7 +833,7 @@ class RecordCdsPubli(dict, AuthorsMixin, PublicationInfoMixin):
 
         Returns:
             str:
-                * "articles", "preprint", "proceeding", "note" or "report"
+                * articles, preprint, proceeding, note, report ot thesis
                 * empty string when it is not defined
 
         """
@@ -841,13 +841,16 @@ class RecordCdsPubli(dict, AuthorsMixin, PublicationInfoMixin):
         if collection is None:
             return ""
 
+        if not isinstance(collection, list):
+            collection = [collection]
+
         lst = [dct.get("primary", "").lower() for dct in collection]
 
         if "conferencepaper" in lst:
             return "proceeding"
 
         # order matter since we have (preprint+note)
-        for val in ("article", "note", "report", "preprint"):
+        for val in ("article", "note", "report", "thesis", "preprint"):
             if val in lst:
                 return val
 

modules/store_tools/recordcdsthesis.py

@@ -2,12 +2,15 @@
 
 """
 from .base import (AFF_CPPM,
+                   MSG_WELL_FORMED_DATE,
                    REG_YEAR,
+                   REG_DATE_YYYYMM,
                    THESIS_DIR,
                    UNIVERSITY,
                    UNIV_AIX_MARSEILLE,
                    UNIV_MARSEILLE,
                    T6)
+from .exception import CheckException
 from filters import CLEAN_THESIS_DEFENSE
 from gluon import current
 from store_tools.recordcdspubli import RecordCdsPubli
@@ -63,6 +66,37 @@ class RecordCdsThesis(RecordCdsPubli):
 
         return li
 
+    def check_submitted_date(self):
+        """Check that submitted date is either ``YYYY-MM`` or ``YYYY-MM-DD``.
+
+        Raises:
+            CheckException::
+                * the date is not well formed
+
+        """
+        self.logger.debug(f"{T6}check submitted date")
+
+        xdate = self.submitted()
+        if REG_DATE_YYYYMM.match(xdate):
+            return
+
+        # recover by using the defense date
+        val = self.these_defense()
+        if REG_DATE_YYYYMM.match(val):
+            if "prepublication" in self:
+
+                prepublication = self["prepublication"]
+                if isinstance(prepublication, list):
+                    prepublication[0]["date"] = val
+                else:
+                    prepublication["date"] = val
+
+            else:
+                self["prepublication"] = {"date": val}
+
+        else:
+            raise CheckException(MSG_WELL_FORMED_DATE)
+
     def format_universities(self):
         """Format the name of the university for PhD:
 
@@ -86,6 +120,9 @@ class RecordCdsThesis(RecordCdsPubli):
                 val = (UNIV_MARSEILLE if year < 2012 else UNIV_AIX_MARSEILLE)
                 values = values.replace(AFF_CPPM, val)
 
+        if "Marseille U., Luminy" in values:
+            values = values.replace("Marseille U., Luminy", UNIV_MARSEILLE)
+
         # Other -- replace U. by University
         university = current.T(UNIVERSITY).decode("utf8")
         values = values.replace('U.', university)

modules/store_tools/recordheppubli.py

@@ -530,7 +530,7 @@ class RecordHepPubli(dict, AuthorsMixin, PublicationInfoMixin):
 
         Returns:
             str:
-                * "articles", "preprint", "proceeding", "note" or "report"
+                * articles, preprint, proceeding, note, report or thesis
                 * empty string when it is not defined
 
         """

modules/store_tools/recordhepthesis.py

@@ -2,11 +2,14 @@
 
 """
 from .base import (AFF_CPPM,
+                   MSG_WELL_FORMED_DATE,
+                   REG_DATE_YYYYMM,
                    REG_YEAR,
                    UNIVERSITY,
                    UNIV_AIX_MARSEILLE,
                    UNIV_MARSEILLE,
                    T6)
+from .exception import CheckException
 from gluon import current
 from filters import CLEAN_THESIS_DEFENSE
 from .recordheppubli import RecordHepPubli
@@ -58,6 +61,28 @@ class RecordHepThesis(RecordHepPubli):
 
         return li
 
+    def check_submitted_date(self):
+        """Check that submitted date is either ``YYYY-MM`` or ``YYYY-MM-DD``.
+
+        Raises:
+            CheckException::
+                * the date is not well formed
+
+        """
+        self.logger.debug(f"{T6}check submitted date")
+
+        xdate = self.submitted()
+        if REG_DATE_YYYYMM.match(xdate):
+            return
+
+        # recover by using the defense date
+        val = self.these_defense()
+        if REG_DATE_YYYYMM.match(val):
+            self["thesis_info"]["defense_date"] = val
+
+        else:
+            raise CheckException(MSG_WELL_FORMED_DATE)
+
     def format_universities(self):
         """Format the name of the university for PhD:
 
@@ -81,6 +106,9 @@ class RecordHepThesis(RecordHepPubli):
                 val = (UNIV_MARSEILLE if year < 2012 else UNIV_AIX_MARSEILLE)
                 values = values.replace(AFF_CPPM, val)
 
+        if "Marseille U., Luminy" in values:
+            values = values.replace("Marseille U., Luminy", UNIV_MARSEILLE)
+
         # Other -- replace U. by University
         university = current.T(UNIVERSITY).decode("utf8")
         values = values.replace('U.', university)

tests/basis/test_08_RecordCdsThesis.py

@@ -53,3 +53,10 @@ def test_format_universities_cds_08006(record):
     record.format_universities()
     assert record.these_universities() == \
         "Shandong Université & Aix Marseille Université"
+
+
+def test_format_universities_cds_08007():
+    record = load_record("cds.cern.ch", 1394605)
+    record.format_universities()
+    assert record.these_universities() == \
+        "Université de la Méditerrannée Aix-Marseille II"

tests/basis/test_09_RecordHepThesis.py

@@ -50,7 +50,15 @@ def test_these_universities_ins_09005(record):
     assert record.these_universities() == "Shandong U. & Marseille, CPPM"
 
 
-def test_format_universities_cds_09006(record):
+def test_format_universities_ins_09006(record):
     record.format_universities()
     assert record.these_universities() == \
         "Shandong Université & Aix Marseille Université"
+
+
+def test_format_universities_ins_08007():
+    record = load_record("inspirehep.net", 1088032, shelf="literature")
+    record.format_universities()
+    assert record.these_universities() == \
+        "Université de la Méditerrannée Aix-Marseille II"
+

tests/basis/test_13_check_and_fix_article_cds.py

 """test_13_check_and_fix_article_cds
 
-    cds.cern.ch 2242641: Phys. Rev. D 95 (2017) 052005
-    (same as inspirehep.net article 15009922)
+    https://cds.cern.ch/record/2242641
+    (same as https://inspirehep.net/api/literature/15009922)
 
     Test individual method of check and fix process for article:
 

tests/basis/test_14_check_fix_article_ins.py

 """test_14_check_and_fix_article_ins
 
-    inspirehep.net article 15009922: Phys. Rev. D 95 (2017) 052005
-    (same as cds.cern.ch 2242641)
+    https://inspirehep.net article/api/literature/15009922
+    (same as https://cds.cern.ch/record/2242641)
 
     Test individual method of check and fix process for article:
 

tests/basis/test_15_check_and_fix_proceeding_cds.py

 """test_15_check_and_fix_proceeding_cds
 
-    cds.cern.ch 1411352
-    (same as inspirehep.net 1089237)
+    https://cds.cern.ch/record/1411352
+    (same as https://inspirehep.net/api/literature/1089237)
 
-    Test individual method of check and fix process for article:
+    Test individual method of check and fix process for proceeding:
 
         * publication is a published proceeding
         * is with authors form my institute

tests/basis/test_16_check_and_fix_proceeding_ins.py

 """test_16_check_and_fix_proceeding_ins
 
-    inspirehep.net 1089237
-    (same as cds.cern.ch 1411352)
+    https://inspirehep.net/api/literature/1089237
+    (same as https://cds.cern.ch/record/1411352)
 
     Test individual method of check and fix process for article:
 

tests/basis/test_17_check_and_fix_thesis.py

-"""test_16_CheckAndFix_thesis
-
-    * Test CheckAndFix methods for thesis.
-      Use the one talk in cds.cern.ch
-
-        - is_thesis
-        - submitted
-        - format_universities
-        - format_authors (already test with article)
-        - get_my_authors (already test with article)
-
-"""
-import pytest
-
-from harvest_tools.checkandfix import CheckAndFix
-from store_tools import load_record
-
-
-@pytest.fixture(scope="module")
-def reccds():
-    return load_record("cds.cern.ch", 1394605)
-
-
-@pytest.fixture(scope="module")
-def svc():
-    return CheckAndFix()
-
-
-def test_is_thesis_cds_17001(svc, reccds):
-    assert svc.is_thesis(reccds) is None
-
-
-def test_submitted_cds_17002(svc, reccds):
-
-    assert reccds.these_defense() == "2011"
-    assert reccds.submitted() == ""
-
-    svc.submitted(reccds)
-    assert reccds.submitted() == "2011-11"
-
-
-def test_format_universities_cds_17003(svc, reccds):
-
-    # Khanji en 2011 (Université de la Méditerrannée)
-    assert reccds["dissertation_note"]["university"] == \
-        "Marseille U., Luminy"
-
-    svc.format_universities(reccds)
-    assert reccds["dissertation_note"]["university"] == \
-        "Université de la Méditerrannée Aix-Marseille II"
-
-    # Chen en 2013 (Aix marseille Université)
-    reccds2 = load_record("cds.cern.ch", 1632177)
-
-    assert reccds2["dissertation_note"]["university"] == \
-        "Shandong U. & Marseille, CPPM"
-
-    svc.format_universities(reccds2)
-    assert reccds2["dissertation_note"]["university"] == \
-        "Shandong Université & Aix Marseille Université"

tests/basis/test_17_check_and_fix_thesis_cds.py

+"""test_17_check_and_fix_thesis_cds
+
+    https://cds.cern.ch/record/1394605
+    (same as https://inspirehep.net/api/literature/10888032)
+
+    Test individual method of check and fix process for thesis:
+
+        * is with authors form my institute
+        * standardise name of collaboration
+        * format authors according to my format
+        * extract authors form my institute signing the publication
+        * is submitted date well formed
+
+"""
+import pytest
+
+from gluon import current
+from harvest_tools import get_rex_institute
+from store_tools import CheckException, load_record
+
+
+@pytest.fixture(scope="module")
+def record():
+    return load_record("cds.cern.ch", 1394605)
+
+
+def test_subtype_17001(record):
+    assert record.subtype() == "thesis"
+
+
+# ............................................................................
+#
+# Check and fix implemented in the RecordCdsConf
+#
+def test_check_authors_17010(record):
+    # raise CheckException in case of problem
+    assert record.check_authors() is None
+
+
+def test_check_my_affiliation_17011(record):
+    # raise CheckException in case of problem
+    rex_institute = get_rex_institute(current.db, current.app)
+    assert record.check_my_affiliation(rex_institute) is None
+
+
+def test_check_collaboration_17012(record):
+    assert record.collaboration() == ""
+    record.check_collaboration(current.db)
+    assert record.collaboration() == ""
+
+
+def test_check_format_authors_17013(record):
+    authors = record.authors_as_list()
+
+    assert len(authors) == 1
+    assert authors[0] == "Khanji, Basem"
+
+    record.check_format_authors(fmt="F. Last")
+    authors = record.authors_as_list()
+
+    assert authors[0] == "B. Khanji"
+
+
+def test_extract_my_authors_17014(record):
+    # raise CheckException in case of problem
+    rex_institute = get_rex_institute(current.db, current.app)
+    assert record.my_authors is None
+
+    record.check_format_authors(fmt="F. Last")
+    assert record.extract_my_authors(rex_institute, sep="|", sort=True) is None
+
+    my_authors = record.my_authors
+    assert my_authors == "B. Khanji"
+
+
+def test_check_submitted_date_17015(record):
+    # raise CheckException in case of problem
+    assert record.submitted() == ""
+    with pytest.raises(CheckException):
+        record.check_submitted_date()