Treat fullwidth space characters defined in Unicode as valid whitespace.

Patch by Manabu TERADA.

src/Products/ZCTextIndex/QueryParser.py

@@ -94,6 +94,11 @@ _tokenizer_regex = re.compile(r"""
     )
 """, re.VERBOSE)
 
+# Use unicode regex to treat fullwidth space characters defined in Unicode
+# as valid whitespace.
+_tokenizer_unicode_regex = re.compile(
+    _tokenizer_regex.pattern, _tokenizer_regex.flags|re.UNICODE)
+
 class QueryParser:
 
     implements(IQueryParser)
@@ -109,6 +114,12 @@ class QueryParser:
 
     def parseQuery(self, query):
         # Lexical analysis.
+        try:
+            # Try to use unicode and treat fullwidth whitespace as valid one.
+            if not isinstance(query, unicode):
+                query = query.decode('utf-8')
+            tokens = _tokenizer_unicode_regex.findall(query)
+        except UnicodeDecodeError:
             tokens = _tokenizer_regex.findall(query)
         self._tokens = tokens
         # classify tokens

src/Products/ZCTextIndex/tests/testQueryParser.py

@@ -210,6 +210,18 @@ class TestQueryParser(TestQueryParserBase):
         self.expect("foo* bar", AndNode([GlobNode("foo*"),
                                          AtomNode("bar")]))
 
+    def test024(self):
+        # Split by UTF-8 fullwidth space
+        from Products.ZCTextIndex.ParseTree import AndNode
+        from Products.ZCTextIndex.ParseTree import AtomNode
+        self.expect("foo\xe3\x80\x80bar", AndNode([AtomNode("foo"), AtomNode("bar")]))
+
+    def test025(self):
+        # Split by Unicode fullwidth space
+        from Products.ZCTextIndex.ParseTree import AndNode
+        from Products.ZCTextIndex.ParseTree import AtomNode
+        self.expect(u"foo\u3000bar", AndNode([AtomNode(u"foo"), AtomNode(u"bar")]))
+
     def test101(self):
         self.failure("")