[Jython-checkins] SF.net SVN: jython:[7070] trunk/jython

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

Revision: 7070
          http://jython.svn.sourceforge.net/jython/?rev=7070&view=rev
Author:   amak
Date:     2010-06-25 14:22:08 +0000 (Fri, 25 Jun 2010)

Log Message:
-----------
Fix for bug 1614: minidom chunks the character input on multi-line values

Modified Paths:
--------------
    trunk/jython/Lib/test/test_minidom.py
    trunk/jython/Lib/xml/dom/minidom.py
    trunk/jython/NEWS

Modified: trunk/jython/Lib/test/test_minidom.py
===================================================================
--- trunk/jython/Lib/test/test_minidom.py	2010-06-25 13:16:13 UTC (rev 7069)
+++ trunk/jython/Lib/test/test_minidom.py	2010-06-25 14:22:08 UTC (rev 7070)
@@ -523,6 +523,29 @@
             "testNormalize -- single empty node removed")
     doc.unlink()
 
+def testNormalizedAfterLoad():
+    """
+    Introduced this test on jython because 
+    1. Cpython guarantees, by the use of xml.dom.expatbuilder, 
+       that all text nodes are normalized after loading.
+    2. Jython has no expat, and thus uses xml.dom.pulldom.parse 
+       (which uses any java SAX2 compliant parser), and which makes 
+       no guarantees about text node normalization.
+    Thus we have to check if text nodes are normalized after a parse.
+    See this bug for further information
+    minidom chunks the character input on multi-line values
+    http://bugs.jython.org/issue1614
+    """
+    num_lines = 2
+    # Up to 16K lines should be enough to guarantee failure without normalization
+    while num_lines <= 2**14:
+        doc_content = "\n".join( ("Line %d" % i for i in xrange(num_lines)) )
+        doc_text = "<document>%s</document>" % doc_content
+        dom = parseString(doc_text)
+        node_content = dom.getElementsByTagName("document")[0].childNodes[0].nodeValue
+        confirm(node_content == doc_content, "testNormalizedAfterLoad")
+        num_lines *= 2
+
 def testSiblings():
     doc = parseString("<doc><?pi?>text?<elm/></doc>")
     root = doc.documentElement

Modified: trunk/jython/Lib/xml/dom/minidom.py
===================================================================
--- trunk/jython/Lib/xml/dom/minidom.py	2010-06-25 13:16:13 UTC (rev 7069)
+++ trunk/jython/Lib/xml/dom/minidom.py	2010-06-25 14:22:08 UTC (rev 7070)
@@ -1908,6 +1908,7 @@
     toktype, rootNode = events.getEvent()
     events.expandNode(rootNode)
     events.clear()
+    rootNode.normalize()
     return rootNode
 
 def parse(file, parser=None, bufsize=None):

Modified: trunk/jython/NEWS
===================================================================
--- trunk/jython/NEWS	2010-06-25 13:16:13 UTC (rev 7069)
+++ trunk/jython/NEWS	2010-06-25 14:22:08 UTC (rev 7070)
@@ -2,6 +2,7 @@
 
 Jython 2.5.2a1
   Bugs Fixed
+    - [ 1614 ] minidom chunks the character input on multi-line values
     - [ 1615 ] Can't invoke Java method that takes a variable number of arguments with zero arguments
     - [ 1605 ] float preference over PyComplex as arg to __call__ breaks logic
     - [ 1586 ] weakref reference count leak when kwargs are used


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