support for non-standard bibtex types, fix #210 #218

pubs/endecoder.py

@@ -119,18 +119,18 @@ class EnDecoder(object):
                 keyword for keyword in entry['keyword'])
         return entry
 
-    def decode_bibdata(self, bibdata):
+    def decode_bibdata(self, bibstr):
         """Decodes bibdata from string.
 
         If the decoding fails, returns a BibDecodingError.
         """
-        if len(bibdata) == 0:
+        if len(bibstr) == 0:
             error_msg = 'parsing error: the provided string has length zero.'
-            raise self.BibDecodingError(error_msg, bibdata)
+            raise self.BibDecodingError(error_msg, bibstr)
         try:
             entries = bp.bparser.BibTexParser(
-                bibdata, common_strings=True, customization=customizations,
-                homogenize_fields=True).get_entry_dict()
+                bibstr, common_strings=True, customization=customizations,
+                homogenize_fields=True, ignore_nonstandard_types=False).get_entry_dict()
             # Remove id from bibtexparser attribute which is stored as citekey
             for e in entries:
                 entries[e].pop(BP_ID_KEY)
@@ -147,13 +147,13 @@ class EnDecoder(object):
                 return entries
             else:
                 raise self.BibDecodingError(('no valid entry found in the provided data: '
-                                            ' {}').format(bibdata), bibdata)
+                                            ' {}').format(bibstr), bibstr)
         except (pyparsing.ParseException, pyparsing.ParseSyntaxException) as e:
             error_msg = self._format_parsing_error(e)
-            raise self.BibDecodingError(error_msg, bibdata)
+            raise self.BibDecodingError(error_msg, bibstr)
         except bibtexparser.bibdatabase.UndefinedString as e:
             error_msg = 'parsing error: undefined string in provided data: {}'.format(e)
-            raise self.BibDecodingError(error_msg, bibdata)
+            raise self.BibDecodingError(error_msg, bibstr)
 
     @classmethod
     def _format_parsing_error(cls, e):

tests/data_non_standard/non_standard_collection.bib

@@ -0,0 +1,9 @@
+@collection{Geometric_phases,
+ title = {Geometric phases in physics},
+ editor = {Shapere, Alfred and Wilczek, Frank},
+ year = {1989},
+ series = {Advanced Series in Mathematical Physics},
+ volume = {5},
+ publisher = {World Scientific},
+ isbn = {9789971506216}
+}

tests/data_non_standard/non_standard_software.bib

@@ -0,0 +1,7 @@
+@software{hadoop,
+  author = {{Apache Software Foundation}},
+  title = {Hadoop},
+  url = {https://hadoop.apache.org},
+  version = {0.20.2},
+  date = {2010-02-19},
+}

tests/fake_env.py

@@ -43,8 +43,8 @@ def capture(f, verbose=False):
     """
     def newf(*args, **kwargs):
         old_stderr, old_stdout = sys.stderr, sys.stdout
-        sys.stdout = _fake_stdio(additional_out=old_stderr if verbose else None)
-        sys.stderr = _fake_stdio(additional_out=old_stderr if False else None)
+        sys.stdout = _fake_stdio(additional_out=old_stdout if verbose else None)
+        sys.stderr = _fake_stdio(additional_out=old_stderr if verbose else None)
         try:
             return f(*args, **kwargs), _get_fake_stdio_ucontent(sys.stdout), _get_fake_stdio_ucontent(sys.stderr)
         finally:

tests/test_usecase.py

@@ -1091,6 +1091,22 @@ class TestUsecase(DataCommandTestCase):
         actual = self.execute_cmds(cmds, capture_output=True)
         self.assertEqual(correct, actual)
 
+    def test_add_non_standard(self):
+        """Test that non-standard bibtex are correctly added"""
+        self.fs.add_real_directory(os.path.join(self.rootpath, 'data_non_standard'), read_only=False)
+        correct = ['Initializing pubs in /pubs\n',
+                   'added to pubs:\n[Geometric_phases]  "Geometric phases in physics" (1989) \n',
+                   'added to pubs:\n[hadoop] Foundation, Apache Software "Hadoop" \n',
+                  ]
+        cmds = ['pubs init -p /pubs',
+                'pubs add data_non_standard/non_standard_collection.bib',
+                'pubs add data_non_standard/non_standard_software.bib',
+                # 'pubs list',
+               ]
+        actual = self.execute_cmds(cmds, capture_output=True)
+        self.assertEqual(correct, actual)
+
+
     @mock.patch('pubs.apis.requests.get', side_effect=mock_requests.mock_requests_get)
     def test_readme(self, reqget):
         """Test that the readme example work."""
@@ -1099,7 +1115,7 @@ class TestUsecase(DataCommandTestCase):
         self.fs.add_real_file(os.path.join(self.rootpath, 'data/pagerank.pdf'), target_path='data/Knuth1995.pdf')
 
         cmds = ['pubs init',
-                'pubs import data/collection.bib',
+                'pubs import data/three_articles.bib',
                 'pubs add data/pagerank.bib -d data/pagerank.pdf',
                 #'pubs add -D 10.1007/s00422-012-0514-6 -d data/pagerank.pdf',
                 'pubs add -I 978-0822324669 -d data/oyama2000the.pdf',