[Jython-checkins] SF.net SVN: jython:[6995] trunk/jython

[<am...@us...>]

Revision: 6995
          http://jython.svn.sourceforge.net/jython/?rev=6995&view=rev
Author:   amak
Date:     2010-04-01 12:07:25 +0000 (Thu, 01 Apr 2010)

Log Message:
-----------
Some unittests for unicode usage with xml.dom.pulldom

Modified Paths:
--------------
    trunk/jython/Lib/xml/sax/drivers2/drv_javasax.py
    trunk/jython/NEWS

Added Paths:
-----------
    trunk/jython/Lib/test/test_pulldom.py

Added: trunk/jython/Lib/test/test_pulldom.py
===================================================================
--- trunk/jython/Lib/test/test_pulldom.py	                        (rev 0)
+++ trunk/jython/Lib/test/test_pulldom.py	2010-04-01 12:07:25 UTC (rev 6995)
@@ -0,0 +1,77 @@
+"""
+	This is not an exhaustive pulldom test suite.
+	Instead, it is a place to put jython-specific tests,
+	relating to bugs like this one, for example.
+
+	"xml.dom.Node.data returns bytestrings of decoded unicode"
+	http://bugs.jython.org/issue1583
+
+	amak.
+"""
+
+import StringIO
+import unittest
+from xml.dom import pulldom
+from test import test_support
+
+class UnicodeTests(unittest.TestCase):
+
+    testDoc = """\
+<?xml version="1.0" encoding="ascii"?>
+<document>
+    <p>Some greek: &#x391;&#x392;&#x393;&#x394;&#x395;</p>
+    <greek attrs="&#x396;&#x397;&#x398;&#x399;&#x39a;"/>
+    <?greek &#x39b;&#x39c;&#x39d;&#x39e;&#x39f;?>
+</document>
+"""
+
+    def setUp(self):
+        self.testFile = StringIO.StringIO(self.testDoc)
+
+    def testTextNodes(self):
+        text = []
+        for event, node in pulldom.parse(self.testFile):
+            if event == pulldom.CHARACTERS:
+                text.append(node.data)
+        try:
+            result = u"".join(text)
+            self.failUnlessEqual(repr(result), r"u'\n    Some greek: \u0391\u0392\u0393\u0394\u0395\n    \n    \n'")
+        except Exception, x:
+            self.fail("Unexpected exception joining text pieces: %s" % str(x))
+
+    def testAttributes(self):
+        attrText = []
+        for event, node in pulldom.parse(self.testFile):
+            if event == pulldom.START_ELEMENT:
+                for attrIx in range(node.attributes.length):
+                    attrText.append(node.attributes.item(attrIx).value)
+        try:
+            result = u"".join(attrText)
+            self.failUnlessEqual(repr(result), r"u'\u0396\u0397\u0398\u0399\u039a'")
+        except Exception, x:
+            self.fail("Unexpected exception joining attribute text pieces: %s" % str(x))
+
+    def testProcessingInstruction(self):
+        piText = []
+        for event, node in pulldom.parse(self.testFile):
+            if event == pulldom.PROCESSING_INSTRUCTION:
+                piText.append(node.data)
+        try:
+            result = u"".join(piText)
+            # Weird how the repr for PI data is different from text and char data.
+            # Still, the whole xml.dom.* and xml.sax.* hierarchy is rather a 
+            # labyrinthine mess under jython, mostly because it's so old, and
+            # yet survived through major evolutionary changes in both jython and java.
+            self.failUnlessEqual(repr(result), r"u'&#x39b;&#x39c;&#x39d;&#x39e;&#x39f;'")
+        except Exception, x:
+            self.fail("Unexpected exception joining pi data pieces: %s" % str(x))
+
+def test_main():
+    tests = [
+        UnicodeTests,
+    ]
+    suites = [unittest.makeSuite(klass, 'test') for klass in tests]
+    test_support.run_suite(unittest.TestSuite(suites))
+
+if __name__ == "__main__":
+    test_main()

Modified: trunk/jython/Lib/xml/sax/drivers2/drv_javasax.py
===================================================================
--- trunk/jython/Lib/xml/sax/drivers2/drv_javasax.py	2010-04-01 07:41:40 UTC (rev 6994)
+++ trunk/jython/Lib/xml/sax/drivers2/drv_javasax.py	2010-04-01 12:07:25 UTC (rev 6995)
@@ -188,7 +188,7 @@
         self._cont_handler.characters(String(char, start, len).getBytes('utf-8').tostring().decode('utf-8'))
 
     def ignorableWhitespace(self, char, start, len):
-        self._cont_handler.ignorableWhitespace(str(String(char, start, len)))
+        self._cont_handler.ignorableWhitespace(String(char, start, len).getBytes('utf-8').tostring().decode('utf-8'))
 
     def endElement(self, uri, lname, qname):
         if self._namespaces:

Modified: trunk/jython/NEWS
===================================================================
--- trunk/jython/NEWS	2010-04-01 07:41:40 UTC (rev 6994)
+++ trunk/jython/NEWS	2010-04-01 12:07:25 UTC (rev 6995)
@@ -2,6 +2,7 @@
 
 Jython 2.5.2a1
   Bugs Fixed
+    - [ 1583 ] xml.dom.Node.data returns bytestrings of decoded unicode
     - [ 1225 ] socket.getservbyname/port() not yet supported
     - [ 1532 ] Cannot use docstring when defining class
     - [ 1530 ] BoolOp in multiple assign causes VerifyError


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.