Update Automatons and controller to remove depency to format_author_fr.

controllers/harvest.py

@@ -10,8 +10,6 @@ from harvest_tools import (build_harvester_tool,
                            CheckAndFix,
                            CheckException,
                            DRY_RUN,
-                           format_author_fr,
-                           family_name_fr,
                            search_synonym,
                            ToolException)
 from invenio_tools import (load_record,
@@ -158,12 +156,12 @@ def edit_insert():
         # authors
         try:
             check.authors(record)
-            check.format_authors(record, format_author_fr)
+            check.format_authors(record, fmt="F. Last")
 
             check.my_affiliation(
                 record, selector.id_projects, selector.id_teams)
 
-            check.get_my_authors(record, cmpFct=family_name_fr)
+            check.get_my_authors(record, sort=True)
 
         except CheckException:
             pass

docs/api/generated/harvest_tools.base.format_author_fr.rst

-harvest_tools.base.format_author_fr
-===================================
-
-.. currentmodule:: harvest_tools.base
-
-.. autofunction:: format_author_fr
\ No newline at end of file

docs/api/harvester.rst

@@ -50,7 +50,6 @@ Helper functions
        :toctree: generated/
 
         ~base.family_name_fr
-        ~base.format_author_fr
         ~base.learn_my_authors
         ~base.search_synonym
 

modules/harvest_tools/__init__.py

@@ -11,7 +11,6 @@ from base import (DRY_RUN,
                   MSG_NO_ENTRY,
                   MSG_TOOMANY_SYNONYM,
                   family_name_fr,
-                  format_author_fr,
                   search_synonym)
 
 from automaton import Automaton

modules/harvest_tools/articles.py

@@ -6,9 +6,7 @@ import traceback
 
 
 from automaton import Automaton
-from base import (family_name_fr,
-                  format_author_fr,
-                  learn_my_authors,
+from base import (learn_my_authors,
                   MSG_CRASH,
                   MSG_FIX_ORIGIN,
                   MSG_IN_DB,
@@ -63,8 +61,8 @@ class Articles(Automaton):
             self.check.submitted(record)
             self.check.year(record)
 
-            self.check.format_authors(record, format_author_fr)
-            self.check.get_my_authors(record, family_name_fr)
+            self.check.format_authors(record, fmt="F. Last")
+            self.check.get_my_authors(record, sort=True)
 
         except CheckException as e:
             self.logs[-1].reject(e, record=record)

modules/harvest_tools/base.py

@@ -3,7 +3,6 @@
 
 """
 from exception import ToolException
-from invenio_tools import REG_AUTHOR
 from plugin_dbui import get_id, UNDEF_ID
 
 
@@ -31,59 +30,6 @@ def family_name_fr(full_name):
     return full_name[full_name.find(' ') + 1:]
 
 
-def format_author_fr(name):
-    """Format the author name according to French typographic rules.
-
-    Note:
-        The name stays unchanged when the formatting failed.
-
-    Args:
-        name (unicode): full name. Possible patterns are:
-
-            * ``Family, L``
-            * ``Family, L.``
-            * ``Family, P L``
-            * ``Family, P.L.``
-            * ``Family, P.-L.``
-            * ``Family, M -H``
-            * ``Family Name, J``
-            * ``Family-Name, J``
-            * ``Family, F Name``
-            * ``Family, First``
-            * ...
-
-    Returns:
-        unicode: the author name encode as ``J.-P. Doe``.
-
-    """
-    # protection
-    if name == '' or name is None:
-        return name
-
-    # decode the name Family, First + many variant
-    # group(1) family name
-    # group(2) first name
-    # group(3) second part of the first name
-    match = REG_AUTHOR.match(name)
-
-    # reformat the name as L. Family
-    # or keep it as it is
-    if match:
-        if match.group(3):
-            tpl = (match.group(2)[0], match.group(3)[0], match.group(1))
-            result = '%s.-%s. %s' % tpl
-
-        else:
-            result = '%s. %s' % (match.group(2)[0], match.group(1))
-    else:
-        result = name
-
-    # avoid author name in upper case (R. LE FOO --> R. Le Foo)
-    result = result.title()
-
-    return result
-
-
 def learn_my_authors(db,
                      authors=None,
                      id_project=None,

modules/harvest_tools/notes.py

@@ -6,7 +6,7 @@ import traceback
 
 
 from automaton import Automaton
-from base import family_name_fr, format_author_fr, MSG_CRASH, MSG_LOAD
+from base import MSG_CRASH, MSG_LOAD
 from checkandfix import CheckException
 from plugin_dbui import UNDEF_ID
 
@@ -36,8 +36,8 @@ class Notes(Automaton):
             self.check.submitted(record)
             self.check.year(record)
 
-            self.check.format_authors(record, format_author_fr)
-            self.check.get_my_authors(record, family_name_fr)
+            self.check.format_authors(record, fmt="F. Last")
+            self.check.get_my_authors(record, sort=True)
 
         except CheckException as e:
             self.logs[-1].reject(e, record=record)

modules/harvest_tools/preprints.py

@@ -6,7 +6,7 @@ import traceback
 
 
 from automaton import Automaton
-from base import family_name_fr, format_author_fr, MSG_CRASH, MSG_LOAD
+from base import MSG_CRASH, MSG_LOAD
 from checkandfix import CheckException
 from invenio_tools import RecordConf, RecordThesis
 from plugin_dbui import UNDEF_ID
@@ -59,8 +59,8 @@ class Preprints(Automaton):
             self.check.submitted(record)
             self.check.year(record)
 
-            self.check.format_authors(record, format_author_fr)
-            self.check.get_my_authors(record, family_name_fr)
+            self.check.format_authors(record, fmt="F. Last")
+            self.check.get_my_authors(record, sort=True)
 
         except CheckException as e:
             self.logs[-1].reject(e, record=record)

modules/harvest_tools/proceedings.py

@@ -6,7 +6,7 @@ import traceback
 
 
 from automaton import Automaton
-from base import family_name_fr, format_author_fr, MSG_CRASH, MSG_LOAD
+from base import MSG_CRASH, MSG_LOAD
 from checkandfix import CheckException
 from plugin_dbui import UNDEF_ID
 
@@ -45,8 +45,8 @@ class Proceedings(Automaton):
             self.check.publisher(record)
             self.check.paper_reference(record)
 
-            self.check.format_authors(record, format_author_fr)
-            self.check.get_my_authors(record, family_name_fr)
+            self.check.format_authors(record, fmt="F. Last")
+            self.check.get_my_authors(record, sort=True)
 
         except CheckException as e:
             self.logs[-1].reject(e, record=record)

modules/harvest_tools/reports.py

@@ -6,7 +6,7 @@ import traceback
 
 
 from automaton import Automaton
-from base import family_name_fr, format_author_fr, MSG_CRASH, MSG_LOAD
+from base import MSG_CRASH, MSG_LOAD
 from checkandfix import CheckException
 from plugin_dbui import get_id, UNDEF_ID, UNKNOWN
 
@@ -43,8 +43,8 @@ class Reports(Automaton):
             self.check.submitted(record)
             self.check.year(record)
 
-            self.check.format_authors(record, format_author_fr)
-            self.check.get_my_authors(record, family_name_fr)
+            self.check.format_authors(record, fmt="F. Last")
+            self.check.get_my_authors(record, sort=True)
 
         except CheckException as e:
             self.logs[-1].reject(e, record=record)
@@ -80,9 +80,7 @@ class Reports(Automaton):
 
         # allow undefined institute authors
         try:
-            self.check.my_authors(record,
-                                  reference=self._my_author_list(record),
-                                  cmpFct=family_name_fr)
+            self.check.get_my_authors(record, sort=True)
             authors_institute = record.my_authors
 
         except CheckException:

modules/harvest_tools/talks.py

@@ -6,7 +6,7 @@ import traceback
 
 
 from automaton import Automaton
-from base import family_name_fr, format_author_fr, MSG_CRASH, MSG_LOAD
+from base import MSG_CRASH, MSG_LOAD
 from checkandfix import CheckException
 from plugin_dbui import UNDEF_ID
 
@@ -40,8 +40,8 @@ class Talks(Automaton):
             self.check.submitted(record)
             self.check.year(record)
 
-            self.check.format_authors(record, format_author_fr)
-            self.check.get_my_authors(record, family_name_fr)
+            self.check.format_authors(record, fmt="F. Last")
+            self.check.get_my_authors(record, sort=True)
 
         except CheckException as e:
             self.logs[-1].reject(e, record=record)

modules/harvest_tools/thesis.py

@@ -7,7 +7,7 @@ import traceback
 
 
 from automaton import Automaton
-from base import family_name_fr, format_author_fr, MSG_CRASH, MSG_LOAD
+from base import MSG_CRASH, MSG_LOAD
 from checkandfix import CheckException
 from invenio_tools import RecordThesis
 from plugin_dbui import get_id, UNDEF_ID
@@ -40,8 +40,8 @@ class Thesis(Automaton):
             self.check.year(record)
             self.check.format_universities(record)
 
-            self.check.format_authors(record, format_author_fr)
-            self.check.get_my_authors(record, family_name_fr)
+            self.check.format_authors(record, fmt="F. Last")
+            self.check.get_my_authors(record, sort=True)
 
         except CheckException as e:
             self.logs[-1].reject(e, record=record)

tests/harvest_tools/test_harvest_tools_functions.py

@@ -6,7 +6,7 @@ import pytest
 
 
 from gluon import current
-from harvest_tools import format_author_fr, search_synonym, ToolException
+from harvest_tools import search_synonym, ToolException
 from invenio_tools import load_record
 
 
@@ -20,36 +20,13 @@ def test_author_type():
         assert type(author) == unicode
 
 
-@pytest.mark.parametrize("value, expected", [
-    (u"Aaij, Roel", u"R. Aaij"),
-    (u"Le Gac, Renaud", u"R. Le Gac"),
-    (u"Le Gac, R.", u"R. Le Gac"),
-    (u"Le Gac, R", u"R. Le Gac"),
-    (u"Perrin-Terrin, Mathieu", u"M. Perrin-Terrin"),
-    (u"Perrin-Terrin, M.", u"M. Perrin-Terrin"),
-    (u"Perrin-Terrin, M", u"M. Perrin-Terrin"),
-    (u"Schune, Marie-Hélène", u"M.-H. Schune"),
-    (u"Schune, Marie-Helene", u"M.-H. Schune"),
-    (u"Schune, Marie Helene", u"M.-H. Schune"),
-    (u"Schune, M.H.", u"M.-H. Schune"),
-    (u"Schune, M.-H.", u"M.-H. Schune"),
-    (u"Schune, M. -H.", u"M.-H. Schune"),
-    (u"Schune, M-H", u"M.-H. Schune"),
-    (u"Schune, M -H", u"M.-H. Schune")
-])
-def test_format_author(value, expected):
-    print value
-    assert type(value) == unicode
-    assert format_author_fr(value) == expected
-
-
 def test_search_synonym():
     db = current.db
 
-    # collaboration ANTARES, TANAMI (should not be defined as a synonym)
+    # collaboration ANTARES, TANAMI (defined as synonym in the db)
     record = load_record("inspirehep.net", 1342250)
-    with pytest.raises(ToolException):
-        search_synonym(db.collaborations, "collaboration", record.collaboration())
+    colid = search_synonym(db.collaborations, "collaboration", record.collaboration())
+    assert colid == 2
 
     # collaboration = ANTARES (defined as synonym in the db))
     record = load_record("inspirehep.net", 718872)