Fix html.py to work with Python 3.x

Fix up running in non-UTF-8 environments too.

Signed-off-by: David Woodhouse <dwmw2@infradead.org>

configure.ac

@@ -936,18 +936,8 @@ AC_CHECK_HEADER([endian.h],
             [AC_DEFINE([ENDIAN_HDR], [<sys/isa_defs.h>])])])])
 
 build_www=yes
-AC_PATH_PROGS(PYTHON, [python2 python], [], $PATH:/bin:/usr/bin)
-if (test -n "${ac_cv_path_PYTHON}"); then
-   AC_MSG_CHECKING([that python is version 2.x])
-   if $PYTHON --version 2>&1 | grep "Python 2\." > /dev/null; then
-      AC_MSG_RESULT([yes])
-      AC_SUBST(PYTHON, ${ac_cv_path_PYTHON})
-   else
-      AC_MSG_RESULT([no])
-      AC_MSG_NOTICE([Python is not v2.x; not building HTML pages])
-      build_www=no
-   fi
-else
+AC_PATH_PROGS(PYTHON, [python3 python2 python], [], $PATH:/bin:/usr/bin)
+if test -z "${ac_cv_path_PYTHON}"; then
    AC_MSG_NOTICE([Python not found; not building HTML pages])
    build_www=no
 fi

www/html.py

@@ -1,4 +1,4 @@
-#!/usr/bin/env python
+#!/usr/bin/env python3
 #
 # Simple XML to HTML converter.
 # 
@@ -19,21 +19,24 @@
 import socket
 import time
 import xml.sax
-import commands
 import codecs
 
-reload(sys)
-sys.setdefaultencoding('utf-8')
+if sys.version_info >= (3,0):
+        sys.stdout = codecs.getwriter("utf-8")(sys.stdout.detach())
+else:
+        reload(sys)
+        sys.setdefaultencoding('utf-8')
+        sys.stdout = codecs.getwriter("utf-8")(sys.stdout)
 
 lookupdir = ''
 
 # Print the usage information
 def usage():
-	print "USAGE:"
-	print "html.py <-f -h file.xml>"
-	print "  -d DIR    use DIR as base directory for opening files"
-	print "  -f        write output to file.html (default is stdout)"
-	print "  -h        help"
+	print ("USAGE:")
+	print ("html.py <-f -h file.xml>")
+	print ("  -d DIR    use DIR as base directory for opening files")
+	print ("  -f        write output to file.html (default is stdout)")
+	print ("  -h        help")
 	return
 
 
@@ -98,9 +101,9 @@ def startElement(self, name, attrs):
 			return
 		elif name == "INCLUDE":
 			try:
-				fd = open(attrs.get('file'), 'r')
+				fd = codecs.open(attrs.get('file'), 'r', 'utf-8')
 			except:
-				fd = open(lookupdir + attrs.get('file'), 'r')
+				fd = codecs.open(lookupdir + attrs.get('file'), 'r', 'utf-8')
 			lines = fd.readlines()
 			fd.close()
 			for line in lines:
@@ -127,15 +130,15 @@ def startElement(self, name, attrs):
 
 		elif name == "br":
 			writeHtml("<br")
-			if attrs.getLength > 0:
+			if attrs.getLength() > 0:
 				names = attrs.getNames()
 				for name in names:
 					writeHtml(" " + name + "=\"" + attrs.get(name) + "\"")
 			writeHtml(" />")
 
 		else:
 			writeHtml("<" + name)
-			if attrs.getLength > 0:
+			if attrs.getLength() > 0:
 				names = attrs.getNames()
 				for name in names:
 					writeHtml(" " + name + "=\"" + attrs.get(name) + "\"")
@@ -198,9 +201,9 @@ def parseConfig(file):
 	parser.setErrorHandler(eh)
 
 	try:
-		fd = open(file, 'r')
+		fd = codecs.open(file, 'r', 'utf-8')
 	except:
-		fd = open(lookupdir + file, 'r')
+		fd = codecs.open(lookupdir + file, 'r', 'utf-8')
 
 	# Parse the file
 	parser.parse(fd)
@@ -214,10 +217,10 @@ def parseConfig(file):
 
 try:
 	(options, arguments) = getopt.getopt(sys.argv[1:],'fhd:')
-except getopt.GetoptError, ex:
+except getopt.GetoptError as ex:
 	print
-	print "ERROR:"
-	print ex.msg
+	print ("ERROR:")
+	print (ex.msg)
 	usage()
 	sys.exit(1)
 	pass
@@ -240,7 +243,7 @@ def parseConfig(file):
 replace[idx:] = [lookupdir]
 
 if not arguments:
-	print "No source file specified"
+	print ("No source file specified")
 	usage()
 	sys.exit(1)
 	pass