Better handle utf-8 citekeys

Utf-8 citekeys generate errors in bibtexparser. Until this is fixed,
this commit replace the stack-trace by a clear error message.

related: #28

pubs/bibstruct.py

@@ -24,13 +24,15 @@ def str2citekey(s):
 def check_citekey(citekey):
     # TODO This is not the right way to test that (17/12/2012)
     if ustr(citekey) != str2citekey(citekey):
-        raise ValueError("Invalid citekey: %s" % citekey)
+        raise ValueError(u"Invalid `{}` citekey; ".format(citekey) +
+                         u"utf-8 citekeys are not supported yet.\n"
+                         u"See https://github.com/pubs/pubs/issues/28 for details.")
 
 def verify_bibdata(bibdata):
     if bibdata is None or len(bibdata) == 0:
-        raise ValueError('no valid bibdata')
+        raise ValueError(u"no valid bibdata")
     if len(bibdata) > 1:
-        raise ValueError('ambiguous: multiple entries in the bibdata.')
+        raise ValueError(u"ambiguous: multiple entries in the bibdata.")
 
 def get_entry(bibdata):
     verify_bibdata(bibdata)
@@ -59,7 +61,7 @@ def generate_citekey(bibdata):
         first_author = entry[author_key][0]
     except KeyError:
         raise ValueError(
-                'No author or editor defined: cannot generate a citekey.')
+                u"No author or editor defined: cannot generate a citekey.")
     try:
         year = entry['year']
     except KeyError:

pubs/commands/add_cmd.py

@@ -106,7 +106,11 @@ def command(conf, args):
         ui.error('citekey already exist {}.'.format(citekey))
         ui.exit(1)
 
-    p = paper.Paper.from_bibentry(bibentry, citekey=citekey)
+    try:
+        p = paper.Paper.from_bibentry(bibentry, citekey=citekey)
+    except Exception as e:
+        ui.error(e.args[0])
+        ui.exit(1)
 
     # tags
 

pubs/commands/import_cmd.py

@@ -77,10 +77,10 @@ def command(conf, args):
         try:
             p = papers[k]
             if isinstance(p, Exception):
-                ui.error('Could not load entry for citekey {}.'.format(k))
+                ui.error(u'Could not load entry for citekey {}.'.format(k))
             else:
                 rp.push_paper(p)
-                ui.info('{} imported.'.format(color.dye_out(p.citekey, 'citekey')))
+                ui.info(u'{} imported.'.format(color.dye_out(p.citekey, 'citekey')))
                 docfile = bibstruct.extract_docfile(p.bibdata)
                 if docfile is None:
                     ui.warning("No file for {}.".format(p.citekey))
@@ -88,6 +88,6 @@ def command(conf, args):
                     rp.push_doc(p.citekey, docfile, copy=copy)
                     #FIXME should move the file if configured to do so.
         except KeyError:
-            ui.error('No entry found for citekey {}.'.format(k))
+            ui.error(u'No entry found for citekey {}.'.format(k))
         except IOError as e:
             ui.error(e.message)

pubs/uis.py

@@ -60,15 +60,15 @@ class PrintUI(object):
 
     def info(self, message, **kwargs):
         kwargs['file'] = self._stderr
-        print('{}: {}'.format(color.dye_err('info', 'ok'), message), **kwargs)
+        print(u'{}: {}'.format(color.dye_err('info', 'ok'), message), **kwargs)
 
     def warning(self, message, **kwargs):
         kwargs['file'] = self._stderr
-        print('{}: {}'.format(color.dye_err('warning', 'warning'), message), **kwargs)
+        print(u'{}: {}'.format(color.dye_err('warning', 'warning'), message), **kwargs)
 
     def error(self, message, **kwargs):
         kwargs['file'] = self._stderr
-        print('{}: {}'.format(color.dye_err('error', 'error'), message), **kwargs)
+        print(u'{}: {}'.format(color.dye_err('error', 'error'), message), **kwargs)
 
     def exit(self, error_code=1):
         sys.exit(error_code)
@@ -109,10 +109,10 @@ class InputUI(PrintUI):
         if len(set(option_chars)) != len(option_chars): # duplicate chars, char choices are deactivated. #FIXME: should only deactivate ambiguous chars
             option_chars = []
 
-        option_str = '/'.join(["{}{}".format(color.dye_out(c, 'bold'), s[1:])
+        option_str = u'/'.join(["{}{}".format(color.dye_out(c, 'bold'), s[1:])
-                                for c, s in zip(displayed_chars, options)])
+                               for c, s in zip(displayed_chars, options)])
 
-        self.message('{}: {} {}: '.format(color.dye_err('prompt', 'warning'), question, option_str), end='')
+        self.message(u'{}: {} {}: '.format(color.dye_err('prompt', 'warning'), question, option_str), end='')
         while True:
             answer = self.input()
             if answer is None or answer == '':
@@ -126,7 +126,7 @@ class InputUI(PrintUI):
                         return option_chars.index(answer.lower())
                     except ValueError:
                         pass
-            self.message('Incorrect option.', option_str)
+            self.message(u'Incorrect option.', option_str)
 
 
     def input_choice(self, options, option_chars, default=None, question=''):
@@ -159,7 +159,7 @@ class InputUI(PrintUI):
                     return option_chars.index(answer.lower())
                 except ValueError:
                     pass
-            self.message('Incorrect option.', option_str)
+            self.message(u'Incorrect option.', option_str)
 
     def input_yn(self, question='', default='y'):
         d = 0 if default in (True, 'y', 'yes') else 1

tests/data/utf8.bib

@@ -0,0 +1,8 @@
+@book{hausdorff1949grundzüge,
+    author = {Hausdorff, Felix},
+    title = {Grundzüge der Mengenlehre},
+    publisher = {Chelsea Pub. Co},
+    year = {1949},
+    address = {New York},
+    isbn = {978-0-8284-0061-9}
+}

tests/test_usecase.py

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 from __future__ import print_function, unicode_literals
 
 import unittest
@@ -168,6 +169,16 @@ class TestAdd(DataCommandTestCase):
         bib_dir = self.fs['os'].path.join(self.default_pubs_dir, 'bib')
         self.assertEqual(set(self.fs['os'].listdir(bib_dir)), {'CustomCitekey.bib'})
 
+    def test_add_utf8_citekey(self):
+        err = ("error: Invalid `hausdorff1949grundzüge` citekey; "
+               "utf-8 citekeys are not supported yet.\n"
+               "See https://github.com/pubs/pubs/issues/28 for details.") # actually not checked
+        cmds = ['pubs init',
+                ('pubs add /data/utf8.bib', [], '', err),
+               ]
+        with self.assertRaises(SystemExit):
+            self.execute_cmds(cmds)
+
     def test_add_doc_nocopy_does_not_copy(self):
         cmds = ['pubs init',
                 'pubs add /data/pagerank.bib --link -d /data/pagerank.pdf',