Upgrade to handle thesis supervisor in RecordHepThesis and ReportCdsThesis

modules/store_tools/authorsmixin.py

@@ -72,8 +72,6 @@ class AuthorsMixin(object):
                 +---------------+--------------------------------+
                 | last_name     | family name                    |
                 +---------------+--------------------------------+
-                | role          | equal to dir. for phd director |
-                +---------------+--------------------------------+
 
     """
 

modules/store_tools/recordcdspubli.py

@@ -229,8 +229,6 @@ class RecordCdsPubli(dict, AuthorsMixin, PublicationInfoMixin):
             +---------------+--------------------------------+
             | last_name     | family name                    |
             +---------------+--------------------------------+
-            | relator_name  | equal to dir. for phd director |
-            +---------------+--------------------------------+
 
         Note:
             After running this method, the attribute ``df_authors`` is defined.
@@ -252,7 +250,7 @@ class RecordCdsPubli(dict, AuthorsMixin, PublicationInfoMixin):
         data = self["authors"]
         data = (data if isinstance(data, list) else [data])
 
-        df = DataFrame(data)
+        df = (DataFrame(data).assign(role=""))
 
         # drop useless columns
         refcols = ["affiliation",

modules/store_tools/recordcdsthesis.py

@@ -155,27 +155,29 @@ class RecordCdsThesis(RecordCdsPubli):
         """
         return self._get("dissertation_note", "diploma")
 
-    def these_directors(self, sep=", "):
+    def these_directors(self, sep=", ", fmt="F. Last"):
         """The list of director(s)
 
+        Note:
+            Supervisor was defined in the field ``relator_name``
+            but is not available anymore. Therefore this method is a dummy one
+            which is kept to preserve the record interface.
+
+        Args:
+            sep (str):
+                separator between names
+
+            fmt (str):
+                format the name of supervisor
+                Possible values are "F. Last" and "Last F."
+
         Returns:
             str:
                 * Names are separated by the ``sep`` argument.
                 * Empty string when it is not defined.
 
         """
-        # for a thesis, the author field 700 field contains
-        # names of the director as well as the name of authors
-        df = self.df_authors
-
-        if "relator_name" in df:
-            query = df.relator_name == THESIS_DIR
-            df = df.loc[query]
-
-            return (sep.join(df.fmt_name) if len(df) > 0 else "")
-
-        else:
-            return ""
+        return ""
 
     def these_universities(self):
         """The university(ies) delivering the thesis diploma.

modules/store_tools/recordheppubli.py

@@ -89,8 +89,6 @@ class RecordHepPubli(dict, AuthorsMixin, PublicationInfoMixin):
             +---------------+--------------------------------+
             | last_name     | family name                    |
             +---------------+--------------------------------+
-            | role          | equal to dir. for phd director |
-            +---------------+--------------------------------+
 
         Note:
             After running this method, the attribute ``df_authors`` is defined.
@@ -107,8 +105,7 @@ class RecordHepPubli(dict, AuthorsMixin, PublicationInfoMixin):
                     "first_name",
                     "fmt_name",
                     "full_name",
-                    "last_name",
-                    "role"]
+                    "last_name"]
             self.df_authors = DataFrame([[""] * len(cols)], columns=cols)
             return
 
@@ -119,8 +116,9 @@ class RecordHepPubli(dict, AuthorsMixin, PublicationInfoMixin):
             if "affiliations" in author:
                 affiliations = [elt["value"] for elt in author["affiliations"]]
 
-            role = \
-                (author["inspire_roles"] if "inspire_roles" in author else [])
+            # remove thesis supervisor
+            if len(author.get("inspire_roles", [])) > 0:
+                continue
 
             full_name = author["full_name"]
             idx = full_name.find(",")
@@ -131,8 +129,7 @@ class RecordHepPubli(dict, AuthorsMixin, PublicationInfoMixin):
                    "first_name": first_name.strip(),
                    "fmt_name": full_name,
                    "full_name": full_name,
-                   "last_name": last_name.strip(),
-                   "role": ", ".join(role)}
+                   "last_name": last_name.strip()}
 
             data.append(dct)
 

modules/store_tools/recordhepthesis.py

 """ store_tools.recordhepthesis
 
 """
+from .authorsmixin import to_initial
 from .base import (AFF_CPPM,
                    MSG_WELL_FORMED_DATE,
                    REG_DATE_YYYYMM,
@@ -23,44 +24,6 @@ class RecordHepThesis(RecordHepPubli):
 
     """
 
-    def authors_as_list(self, sort=False):
-        """The list of author(s) signing the publication.
-
-        Note:
-            supersede the base class since the authors field contains
-            the author as well as names of director.
-
-        Args:
-            sort (bool): sort authors by first name when true.
-
-        Returns:
-            list:
-                the list is empty when authors are not defined.
-
-        """
-        # for a thesis, the authors field contains names of author
-        # as well as those of directors. The latter have to be removed.
-        df = self.df_authors
-
-        if "role" in df:
-            df = df[df.role.str.len() == 0]
-
-        if sort:
-            li = (df[["last_name", "fmt_name"]]
-                  .sort_values(by="last_name")
-                  .fmt_name
-                  .tolist())
-
-        else:
-            li = (df.fmt_name
-                  .sort_index()
-                  .tolist())
-
-        if len(li) == 1 and li[0] == "":
-            li = []
-
-        return li
-
     def check_submitted_date(self):
         """Check that submitted date is either ``YYYY-MM`` or ``YYYY-MM-DD``.
 
@@ -78,7 +41,7 @@ class RecordHepThesis(RecordHepPubli):
         # recover by using the defense date
         val = self.these_defense()
         if REG_DATE_YYYYMM.match(val):
-            self["thesis_info"]["defense_date"] = val
+            self["preprint_date"] = val
 
         else:
             raise CheckException(MSG_WELL_FORMED_DATE)
@@ -138,27 +101,44 @@ class RecordHepThesis(RecordHepPubli):
         """
         return self.get("thesis_info", {}).get("degree_type", "")
 
-    def these_directors(self, sep=", "):
+    def these_directors(self, sep=", ", fmt="F. Last"):
         """The list of director(s)
 
+        Args:
+            sep (str):
+                separator between names
+
+            fmt (str):
+                format the name of supervisor
+                Possible values are "F. Last" and "Last F."
+
         Returns:
             str:
                 * Names are separated by the ``sep`` argument.
                 * Empty string when it is not defined.
 
         """
-        # for a thesis, the author field 700 field contains
-        # names of the director as well as the name of authors
-        df = self.df_authors
+        lst = []
+        # supervisor are in the list of author with a role equal to supervisor
+        for author in self.get("authors", []):
 
-        if "role" in df:
-            query = df.role.str.len() > 0
-            df = df.loc[query]
+            if len(author.get("inspire_roles", [])) == 0:
+                continue
 
-            return (sep.join(df.fmt_name) if len(df) > 0 else "")
+            full_name = author.get("full_name")
 
-        else:
-            return ""
+            # name of supervisor is encoded ad "Last, First"
+            if fmt in ("F. Last", "Last F."):
+                last, first = full_name.split(",")
+                first = to_initial(first.strip())
+
+                full_name = (
+                    f"{last} {first}" if fmt == "Last F."
+                    else f"{first} {last}")
+
+            lst.append(full_name)
+
+        return sep.join(lst)
 
     def these_universities(self):
         """The university(ies) delivering the thesis diploma.

tests/basis/test_09_RecordHepThesis.py

@@ -42,8 +42,7 @@ def test_these_level_ins_09003(record):
 
 
 def test_these_directors_ins_09004(record):
-    assert record.these_directors(sep=u"|") == \
-        "He, Mao|Monnier, Emmanuel|Zhu, Chengguang"
+    assert record.these_directors(sep=u"|") == "M. He|E. Monnier|C. Zhu"
 
 
 def test_these_universities_ins_09005(record):

tests/basis/test_18_check_and_fix_thesis_ins.py

+"""test_18_check_and_fix_thesis_ins
+
+    https://inspirehep.net/api/literature/10888032
+    (same as https://cds.cern.ch/record/1394605)
+
+    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 load_record
+
+
+@pytest.fixture(scope="module")
+def record():
+    return load_record("inspirehep.net", 1088032, shelf="literature")
+
+
+def test_subtype_18001(record):
+    assert record.subtype() == "thesis"
+
+
+# ............................................................................
+#
+# Check and fix implemented in the RecordCdsConf
+#
+def test_check_authors_18010(record):
+    # raise CheckException in case of problem
+    assert record.check_authors() is None
+
+
+def test_check_my_affiliation_18011(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_18012(record):
+    assert record.collaboration() == ""
+    record.check_collaboration(current.db)
+    assert record.collaboration() == ""
+
+
+def test_check_format_authors_18013(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_18014(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_18015(record):
+    # raise CheckException in case of problem
+    assert record.submitted() == ""
+    record.check_submitted_date()
+    assert record.submitted() == "2011-09-16"