Ticket #7027: fix_tag_translation-refactored-re-strict_gettext_matching_7027.diff

django/utils/text.py

@@ -197 +197 @@
     return str(ustring_re.sub(fix, s))
 javascript_quote = allow_lazy(javascript_quote, unicode)
 
-smart_split_re = re.compile('("(?:[^"\\\\]*(?:\\\\.[^"\\\\]*)*)"|\'(?:[^\'\\\\]*(?:\\\\.[^\'\\\\]*)*)\'|[^\\s]+)')
+# There are three match blocks (A|B|C). A and B have identical structure,
+# differing only by the matched quotation marks. These blocks catch anything
+# in quotes as a single match group, optionally surrounded by `_()`.
+# C is the uninteresting, trivial case -- it matches any non-empty sequence
+# of non-space characters.
+#
+# Block A should be read as follows, where `` mark string literals and
+# NG stands for "non-grouping".
+#
+# Block B should be read the same way, replacing `'` for `"`.
+#
+# Reading of block A:
+#
+# (?:_\()? -- NG optional match ugettext start marker `_(` in the beginning,
+#             NG needed as `_(` is a digraph
+#
+# `"` -- match string start marker
+#
+# (?:[^"\\]|\\.)+ -- NG match for a non-empty sequence of either
+#                    * any character except `"` or `\`,
+#                    * or digraph `\` followed by any single character
+#
+# `"` -- match string end marker
+#
+# \)? -- match optional ugettext `)` in the end
+#
+# This one won't let `" ")` and `\(" "` through.
+smart_split_re = re.compile(
+        r'((?:_\()"(?:[^"\\]|\\.)+"\)' # block A, for _("foo bar")
+        r'|"(?:[^"\\]|\\.)+"' # block A, for "foo bar"
+        r"|(?:_\()'(?:[^'\\]|\\.)+'\)" # block B, for _('foo bar')
+        r"|'(?:[^'\\]|\\.)+'" # block B, for 'foo bar'
+        r"|[^\s]+)" # block C, anything other without whitespace
+)
 def smart_split(text):
     r"""
     Generator that splits a string by spaces, leaving quoted phrases together.
     Supports both single and double quotes, and supports escaping quotes with
     backslashes. In the output, strings will keep their initial and trailing
-    quote marks.
+    quote marks. Also, gettext markers '_(', ')' are preserved.
 
     >>> list(smart_split(r'This is "a person\'s" test.'))
     [u'This', u'is', u'"a person\\\'s"', u'test.']
-        >>> list(smart_split(r"Another 'person\'s' test.")) 
+        >>> list(smart_split(r"Another 'person\'s' test."))
         [u'Another', u"'person's'", u'test.']
-        >>> list(smart_split(r'A "\"funky\" style" test.')) 
+        >>> list(smart_split(r'A "\"funky\" style" test.'))
         [u'A', u'""funky" style"', u'test.']
+    >>> list(smart_split(' _("my quoted string") '))
+    [u'_("my quoted string")']
+    >>> list(smart_split(" _('my quoted string') "))
+    [u"_('my quoted string')"]
     """
     text = force_unicode(text)
     for bit in smart_split_re.finditer(text):
         bit = bit.group(0)
-        if bit[0] == '"' and bit[-1] == '"':
-            yield '"' + bit[1:-1].replace('\\"', '"').replace('\\\\', '\\') + '"'
-        elif bit[0] == "'" and bit[-1] == "'":
-            yield "'" + bit[1:-1].replace("\\'", "'").replace("\\\\", "\\") + "'"
+        prefix, suffix = '', ''
+        start, end = 1, -1
+        if bit[0:2] == '_(' and bit[-1] == ')':
+            prefix, suffix = '_(', ')'
+            start, end = 3, -2
+        if (bit[0] == '"' and bit[-1] == '"'
+                or bit[0:3] == '_("' and bit[-2:] == '")'):
+            yield '%s"%s"%s' % (prefix,
+                    bit[start:end].replace(r'\"', '"').replace(r'\\', '\\'),
+                    suffix)
+        elif (bit[0] == "'" and bit[-1] == "'"
+                or bit[0:3] == "_('" and bit[-2:] == "')"):
+            yield "%s'%s'%s" % (prefix,
+                    bit[start:end].replace(r"\'", "'").replace(r'\\', '\\'),
+                    suffix)
         else:
             yield bit
 smart_split = allow_lazy(smart_split, unicode)

tests/regressiontests/text/tests.py

@@ -15 +15 @@
 [u'"a', u"'one"]
 >>> print list(smart_split(r'''all friends' tests'''))[1]
 friends'
+>>> list(smart_split(' _("my quoted string") '))
+[u'_("my quoted string")']
+>>> list(smart_split(" _('my quoted string') "))
+[u"_('my quoted string')"]
+>>> print list(smart_split(" _('my \"quoted\" string') "))[0]
+_('my "quoted" string')
 
 ### urlquote #############################################################
 >>> from django.utils.http import urlquote, urlquote_plus