Fix catalog updating with fuzzy matches. Closes #82.

diff --git a/ChangeLog b/ChangeLog
index 87ed7f96a9a60c98d3714810d09f572d2499f56b..5e2d545da79f8ce1337a3bde0a3d72adb98e7773 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -3,7 +3,9 @@ http://svn.edgewall.org/repos/babel/tags/0.9.2/
 (?, from branches/stable/0.9.x)
 
  * Fixed catalogs' charset values not being recognized (ticket #66).
-       
+ * Fixed fuzzy matching when updating message catalogs (ticket #82).
+
+
 Version 0.9.1
 http://svn.edgewall.org/repos/babel/tags/0.9.1/
 (Sep 7 2007, from branches/stable/0.9.x)

diff --git a/babel/messages/catalog.py b/babel/messages/catalog.py
index 3c0046862c4e0b91b37f678eb931c4aa23bcc937..0de66461653bd642a5ddec7bed721b939d0fbc6a 100644 (file)
--- a/babel/messages/catalog.py
+++ b/babel/messages/catalog.py
@@ -231,11 +231,11 @@ class Catalog(object):
 
     >>> catalog = Catalog(project='Foobar', version='1.0',
     ...                   copyright_holder='Foo Company')
-    >>> print catalog.header_comment
+    >>> print catalog.header_comment #doctest: +ELLIPSIS
     # Translations template for Foobar.
-    # Copyright (C) 2007 Foo Company
+    # Copyright (C) ... Foo Company
     # This file is distributed under the same license as the Foobar project.
-    # FIRST AUTHOR <EMAIL@ADDRESS>, 2007.
+    # FIRST AUTHOR <EMAIL@ADDRESS>, ....
     #
 
     The header can also be set from a string. Any known upper-case variables
@@ -611,17 +611,30 @@ class Catalog(object):
         :param no_fuzzy_matching: whether to use fuzzy matching of message IDs
         """
         messages = self._messages
+        remaining = messages.copy()
         self._messages = odict()
 
+        # Prepare for fuzzy matching
+        fuzzy_candidates = []
+        if not no_fuzzy_matching:
+            fuzzy_candidates = [
+                self._key_for(msgid) for msgid in messages
+                if msgid and messages[msgid].string
+            ]
+        fuzzy_matches = set()
+
         def _merge(message, oldkey, newkey):
             fuzzy = False
-            oldmsg = messages.pop(oldkey)
             if oldkey != newkey:
                 fuzzy = True
+                fuzzy_matches.add(oldkey)
+                oldmsg = messages.get(oldkey)
                 if isinstance(oldmsg.id, basestring):
                     message.previous_id = [oldmsg.id]
                 else:
                     message.previous_id = list(oldmsg.id)
+            else:
+                oldmsg = remaining.pop(oldkey)
             message.string = oldmsg.string
             if isinstance(message.id, (list, tuple)):
                 if not isinstance(message.string, (list, tuple)):
@@ -649,14 +662,17 @@ class Catalog(object):
                     if no_fuzzy_matching is False:
                         # do some fuzzy matching with difflib
                         matches = get_close_matches(key.lower().strip(),
-                            [self._key_for(msgid) for msgid in messages], 1)
+                                                    fuzzy_candidates, 1)
                         if matches:
                             _merge(message, matches[0], key)
                             continue
 
                     self[message.id] = message
 
-        self.obsolete = messages
+        self.obsolete = odict()
+        for msgid in remaining:
+            if no_fuzzy_matching or msgid not in fuzzy_matches:
+                self.obsolete[msgid] = remaining[msgid]
 
     def _key_for(self, id):
         """The key for a message is just the singular ID even for pluralizable

diff --git a/babel/messages/tests/catalog.py b/babel/messages/tests/catalog.py
index 5dd5b5529cda48405d5026a48fe893aea4f87236..f9ade90fee8d1ee9d6267be51f42b01310c856c0 100644 (file)
--- a/babel/messages/tests/catalog.py
+++ b/babel/messages/tests/catalog.py
@@ -131,6 +131,40 @@ class CatalogTestCase(unittest.TestCase):
         self.assertEqual('Voh', cat['foo'].string)
         self.assertEqual(True, cat['foo'].fuzzy)
 
+    def test_update_fuzzy_matching_no_msgstr(self):
+        cat = catalog.Catalog()
+        cat.add('fo', '')
+        tmpl = catalog.Catalog()
+        tmpl.add('fo')
+        tmpl.add('foo')
+        cat.update(tmpl)
+        assert 'fo' in cat
+        assert 'foo' in cat
+
+        self.assertEqual('', cat['fo'].string)
+        self.assertEqual(False, cat['fo'].fuzzy)
+        self.assertEqual(None, cat['foo'].string)
+        self.assertEqual(False, cat['foo'].fuzzy)
+
+    def test_update_fuzzy_matching_no_cascading(self):
+        cat = catalog.Catalog()
+        cat.add('fo', 'Voh')
+        cat.add('foo', 'Vohe')
+        tmpl = catalog.Catalog()
+        tmpl.add('fo')
+        tmpl.add('foo')
+        tmpl.add('fooo')
+        cat.update(tmpl)
+        assert 'fo' in cat
+        assert 'foo' in cat
+
+        self.assertEqual('Voh', cat['fo'].string)
+        self.assertEqual(False, cat['fo'].fuzzy)
+        self.assertEqual('Vohe', cat['foo'].string)
+        self.assertEqual(False, cat['foo'].fuzzy)
+        self.assertEqual('Vohe', cat['fooo'].string)
+        self.assertEqual(True, cat['fooo'].fuzzy)
+
     def test_update_without_fuzzy_matching(self):
         cat = catalog.Catalog()
         cat.add('fo', 'Voh')