Merge pull request #123 from benureau/fix/common_strings

Fix/common strings

Fix the regression on translating month abbreviations into full month names in bibtex files.

pubs/endecoder.py

@@ -113,14 +113,10 @@ class EnDecoder(object):
     def decode_bibdata(self, bibdata):
         """"""
         try:
-            try:
-                entries = bp.bparser.BibTexParser(
-                    bibdata, homogenize_fields=True,
-                    customization=customizations).get_entry_dict()
-            except TypeError:
-                entries = bp.bparser.BibTexParser(
-                    bibdata,
-                    customization=customizations).get_entry_dict()
+            entries = bp.bparser.BibTexParser(
+                bibdata, common_strings=True,
+                customization=customizations,
+                homogenize_fields=True).get_entry_dict()
 
             # Remove id from bibtexparser attribute which is stored as citekey
             for e in entries:

setup.py

@@ -25,7 +25,7 @@ setup(
             ],
         },
 
-    install_requires=['pyyaml', 'bibtexparser>=1.0.1', 'python-dateutil', 'requests',
+    install_requires=['pyyaml', 'bibtexparser>=1.0', 'python-dateutil', 'requests',
                       'configobj',
                       'beautifulsoup4'],  # to be made optional?
     tests_require=['pyfakefs>=2.7'],

tests/str_fixtures.py

@@ -69,6 +69,14 @@ bibtex_no_citekey = """@Manual{,
 }
 """
 
+bibtex_month= """@inproceedings{Goyal2017,
+ author = {Goyal, Anirudh and Sordoni, Alessandro and C{\^{o}}t{\'{e}}, Marc-Alexandre and Ke, Nan Rosemary and Bengio, Yoshua},
+ title = {Z-Forcing: Training Stochastic Recurrent Networks},
+ year = {2017},
+ month = dec,
+ pages = {6716--6726},
+}
+"""
 
 sample_conf = """
 [main]

tests/test_endecoder.py

@@ -9,7 +9,7 @@ from pubs import endecoder
 from pubs.p3 import ustr
 
 from fixtures import dummy_metadata
-from str_fixtures import bibtex_raw0, metadata_raw0, turing_bib
+from str_fixtures import bibtex_raw0, metadata_raw0, turing_bib, bibtex_month
 
 
 def compare_yaml_str(s1, s2):
@@ -35,6 +35,7 @@ class TestEnDecode(unittest.TestCase):
         self.assertIsInstance(data, ustr)
 
     def test_endecode_bibtex(self):
+        """Test that multiple encode/decode step preserve data"""
         decoder = endecoder.EnDecoder()
         entry = decoder.decode_bibdata(bibtex_raw0)
 
@@ -51,6 +52,14 @@ class TestEnDecode(unittest.TestCase):
 
         self.assertEqual(bibraw1, bibraw2)
 
+    def test_endecode_bibtex(self):
+        """Test if `month=dec` is correctly recognized and transformed into
+        `month={December}`"""
+        decoder = endecoder.EnDecoder()
+        entry = decoder.decode_bibdata(bibtex_month)['Goyal2017']
+
+        self.assertEqual(entry['month'], 'December')
+
     def test_endecode_bibtex_editor(self):
         decoder = endecoder.EnDecoder()
         entry = decoder.decode_bibdata(turing_bib)