Add RecordHepThesis and Test_09_RecordHepThesis

modules/store_tools/__init__.py

@@ -31,6 +31,7 @@ from .recordconf import RecordConf
 from .recordhepconf import RecordHepConf
 from .recordhepinst import RecordHepInst
 from .recordheppubli import RecordHepPubli
+from .recordhepthesis import RecordHepThesis
 from .recordpubli import RecordPubli
 from .recordthesis import RecordThesis
 

modules/store_tools/factory.py

@@ -20,6 +20,7 @@ from .recordconf import RecordConf
 from .recordhepconf import RecordHepConf
 from .recordheppubli import RecordHepPubli
 from .recordhepinst import RecordHepInst
+from .recordhepthesis import RecordHepThesis
 from .recordpubli import RecordPubli
 from .recordthesis import RecordThesis
 
@@ -216,8 +217,8 @@ def build_record(recjson, shelf=None):
         elif shelf == "institutions":
             upcast_record = RecordHepInst(recjson)
 
-#         elif is_thesis(recjson) and shelf == "literature":
-#             upcast_record = RecordHepThesis(recjson)
+        elif is_thesis(recjson) and shelf == "literature":
+            upcast_record = RecordHepThesis(recjson)
 
         elif shelf == "literature":
             upcast_record = RecordHepPubli(recjson)

modules/store_tools/recordheppubli.py

@@ -43,7 +43,7 @@ class RecordHepPubli(RecordHep, PluginAuthors, PluginPublicationInfo):
             +---------------+--------------------------------+
             | last_name     | family name                    |
             +---------------+--------------------------------+
-            | relator_name  | equal to dir. for phd director |
+            | role          | equal to dir. for phd director |
             +---------------+--------------------------------+
 
         Note:
@@ -59,7 +59,8 @@ class RecordHepPubli(RecordHep, PluginAuthors, PluginPublicationInfo):
                     "first_name",
                     "fmt_name",
                     "full_name",
-                    "last_name"]
+                    "last_name",
+                    "role"]
             self.df_authors = DataFrame([[""] * len(cols)], columns=cols)
             return
 
@@ -70,6 +71,9 @@ class RecordHepPubli(RecordHep, PluginAuthors, PluginPublicationInfo):
             if "affiliations" in author:
                 affiliations = [elt["value"] for elt in author["affiliations"]]
 
+            role = \
+                (author["inspire_roles"] if "inspire_roles" in author else [])
+
             full_name = author["full_name"]
             last_name, first_name = full_name.split(",")
 
@@ -77,7 +81,8 @@ class RecordHepPubli(RecordHep, PluginAuthors, PluginPublicationInfo):
                    "first_name": first_name.strip(),
                    "fmt_name": full_name,
                    "full_name": full_name,
-                   "last_name": last_name.strip()}
+                   "last_name": last_name.strip(),
+                   "role": ", ".join(role)}
 
             data.append(dct)
 

modules/store_tools/recordhepthesis.py

+""" store_tools.recordhepthesis
+
+"""
+from filters import CLEAN_THESIS_DEFENSE
+from .recordheppubli import RecordHepPubli
+
+
+class RecordHepThesis(RecordHepPubli):
+    """Thesis from inspirehep.net version 2.
+
+    Schema for publication is documented here:
+    https://inspire-schemas.readthedocs.io/en/latest/schemas/
+
+    """
+
+    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 these_defense(self):
+        """The defence date for a master/phd thesis.
+
+        Returns:
+            str:
+                * 10 Dec 2012
+                * The filter CLEAN_THESIS_DEFENSE is applied.
+
+        """
+        val = self.get("thesis_info", {}).get("defense_date", "")
+        return CLEAN_THESIS_DEFENSE(val)
+
+    def these_level(self):
+        """The level of the thesis.
+
+        Returns:
+            str:
+                * The value is ``master`` or ``phd``.
+                * Empty string when it is not defined
+
+        """
+        return self.get("thesis_info", {}).get("degree_type", "")
+
+    def these_directors(self, sep=", "):
+        """The list of director(s)
+
+        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 "role" in df:
+            query = df.role.str.len() > 0
+            df = df.loc[query]
+
+            return (sep.join(df.fmt_name) if len(df) > 0 else "")
+
+        else:
+            return ""
+
+    def these_universities(self):
+        """The university(ies) delivering the thesis diploma.
+
+        Returns:
+            str:
+                - empty when university(ies) is not defined
+                - several university are separated by ``&`` character.
+
+        """
+        lst = self.get("thesis_info", {}).get("institutions", [])
+        val = [dct.get("name", "") for dct in lst]
+        return " & ".join(val)

tests/basis/test_02_factory_tools.py

@@ -19,6 +19,7 @@ from store_tools.recordconf import RecordConf
 from store_tools.recordhepconf import RecordHepConf
 from store_tools.recordhepinst import RecordHepInst
 from store_tools.recordheppubli import RecordHepPubli
+from store_tools.recordhepthesis import RecordHepThesis
 from store_tools.recordpubli import RecordPubli
 from store_tools.recordthesis import RecordThesis
 
@@ -156,7 +157,7 @@ def test_institute_ins_02010():
 #
 # Article, ...
 #
-def test_article_cds_02011():
+def test_article_cds_02021():
     """Precision luminosity measurements at LHCb"""
 
     store = build_store("cds.cern.ch")
@@ -170,7 +171,7 @@ def test_article_cds_02011():
     assert isinstance(record, RecordPubli)
 
 
-def test_article_inspirehep_02012():
+def test_article_inspirehep_02022():
     """Precision luminosity measurements at LHCb"""
 
     store = build_store("inspirehep.net", shelf="literature")
@@ -188,7 +189,7 @@ def test_article_inspirehep_02012():
 #
 # Thesis
 #
-def test_thesis_cds_02013():
+def test_thesis_cds_02030():
 
     store = build_store("cds.cern.ch")
     recjson = store.get_record(1632177)
@@ -201,11 +202,24 @@ def test_thesis_cds_02013():
     assert isinstance(record, RecordThesis)
 
 
+def test_thesis_ins_02031():
+
+    store = build_store("inspirehep.net", shelf="literature")
+    recjson = store.get_record(1296381)
+
+    assert not is_conference(recjson)
+    assert not is_institute(recjson)
+    assert is_thesis(recjson)
+
+    record = build_record(recjson, shelf="literature")
+    assert isinstance(record, RecordHepThesis)
+
+
 # ............................................................................
 #
 # Author manipulation
 #
-def test_to_initial_02014():
+def test_to_initial_02040():
 
     assert to_initial("Albert") == "A."
     assert to_initial("Antonio Augusto") == "A. A."

tests/basis/test_05_RecordHepPubli.py

@@ -46,7 +46,8 @@ def test_constructor_ins_05002(record):
                "first_name",
                "fmt_name",
                "full_name",
-               "last_name"]
+               "last_name",
+               "role"]
 
     assert len(authors.columns.difference(refcols)) == 0
     assert len(authors) == 704

tests/basis/test_08_RecordThesis.py

@@ -3,6 +3,7 @@
 Test specific methods of the RecordThesis class for::
 
     https://cds.cern.ch/record/1632177
+    (same as https://inspirehep.net/api/literature/1296381)
 
     Measurement of $t\bar{t}$ spin correlation and W polarization with
     the $t\bar{t}$  di-leptonic channel events at ATLAS

tests/basis/test_09_RecordHepThesis.py

+"""test_09_RecordHepThesis
+
+Test specific methods of the RecordThesis class for::
+
+    https://inspirehep.net/api/literature/1296381
+    (same as https://cds.cern.ch/record/1632177)
+
+    Measurement of $t\bar{t}$ spin correlation and W polarization with
+    the $t\bar{t}$  di-leptonic channel events at ATLAS
+    L. Chen
+    10 Dec 2013
+
+No corrections are applied to the record.
+Allow to test the brute force decoding with its mistakes.
+
+Note:
+    * Only the first author is defined
+    * Record submitted is not defined
+    * The year is not defined
+
+"""
+import pytest
+
+from store_tools import load_record
+
+
+@pytest.fixture(scope="module")
+def record():
+    return load_record("inspirehep.net", 1296381, shelf="literature")
+
+
+def test_authors_as_list_ins_09001(record):
+    assert record.authors_as_list() == [u'Chen, Liming']
+
+
+def test_these_defense_ins_09002(record):
+    assert record.these_defense() == "2013-12-10"
+
+
+def test_these_level_ins_09003(record):
+    assert record.these_level() == "phd"
+
+
+def test_these_directors_ins_09004(record):
+    assert record.these_directors(sep=u"|") == \
+        "He, Mao|Monnier, Emmanuel|Zhu, Chengguang"
+
+
+def test_these_universities_ins_09005(record):
+    assert record.these_universities() == "Shandong U. & Marseille, CPPM"