Issue #11159: SAX parser now supports unicode file names.

2 files changed, 27 insertions, 6 deletions

diff --git a/Lib/xml/sax/expatreader.py b/Lib/xml/sax/expatreader.py
index 92a79c1..9de3e72 100644
--- a/Lib/xml/sax/expatreader.py
+++ b/Lib/xml/sax/expatreader.py
@@ -108,7 +108,10 @@ class ExpatParser(xmlreader.IncrementalParser, xmlreader.Locator):
 
     def prepareParser(self, source):
         if source.getSystemId() is not None:
-            self._parser.SetBase(source.getSystemId())
+            base = source.getSystemId()
+            if isinstance(base, unicode):
+                base = base.encode('utf-8')
+            self._parser.SetBase(base)
 
     # Redefined setContentHandler to allow changing handlers during parsing
 
diff --git a/Lib/xml/sax/saxutils.py b/Lib/xml/sax/saxutils.py
index 97d65d8..7989713 100644
--- a/Lib/xml/sax/saxutils.py
+++ b/Lib/xml/sax/saxutils.py
@@ -4,6 +4,7 @@ convenience of application and driver writers.
 """
 
 import os, urlparse, urllib, types
+import sys
 import handler
 import xmlreader
 
@@ -293,14 +294,31 @@ def prepare_input_source(source, base = ""):
             source.setSystemId(f.name)
 
     if source.getByteStream() is None:
-        sysid = source.getSystemId()
-        basehead = os.path.dirname(os.path.normpath(base))
-        sysidfilename = os.path.join(basehead, sysid)
-        if os.path.isfile(sysidfilename):
+        try:
+            sysid = source.getSystemId()
+            basehead = os.path.dirname(os.path.normpath(base))
+            encoding = sys.getfilesystemencoding()
+            if isinstance(sysid, unicode):
+                if not isinstance(basehead, unicode):
+                    try:
+                        basehead = basehead.decode(encoding)
+                    except UnicodeDecodeError:
+                        sysid = sysid.encode(encoding)
+            else:
+                if isinstance(basehead, unicode):
+                    try:
+                        sysid = sysid.decode(encoding)
+                    except UnicodeDecodeError:
+                        basehead = basehead.encode(encoding)
+            sysidfilename = os.path.join(basehead, sysid)
+            isfile = os.path.isfile(sysidfilename)
+        except UnicodeError:
+            isfile = False
+        if isfile:
             source.setSystemId(sysidfilename)
             f = open(sysidfilename, "rb")
         else:
-            source.setSystemId(urlparse.urljoin(base, sysid))
+            source.setSystemId(urlparse.urljoin(base, source.getSystemId()))
             f = urllib.urlopen(source.getSystemId())
 
         source.setByteStream(f)