Final scour 0.18: more fixes to XML serialization (wellformedness), some unit tests, update to package script to use zip file and to package the inkscape extension

package.sh

@@ -1,6 +1,6 @@
 #!/bin/bash
 SCOURVER="0.18"
 cd ..
-tar cvf scour/tarballs/scour-$SCOURVER.tar scour/scour.py scour/svg_regex.py scour/LICENSE scour/NOTICE scour/README.txt scour/release-notes.html
+zip scour/tarballs/scour-$SCOURVER.zip scour/scour.py scour/svg_regex.py scour/LICENSE scour/NOTICE scour/README.txt scour/release-notes.html
-gzip scour/tarballs/scour-$SCOURVER.tar
 cd scour
+zip tarballs/scour-inkscape-extension-$SCOURVER.zip scour.inx scour.inkscape.py scour.py svg_regex.py

release-notes.html

@@ -13,11 +13,11 @@
 	<header>
 		<h2><a href="#0.18">Version 0.18</a></h2>
 	</header>
-	<p>Aug 5th, 2009</p>
+	<p>Aug 9th, 2009</p>
 	<ul>
 		<li>Remove attributes of gradients if they contain default values</li>
 		<li>Reduce bezier/quadratic (c/q) segments to their shorthand equivalents (s/t)</li>
-		<li>Custom XML serialization such that id/xml:id is printed first (Thanks to Richard Hutch for the suggestion)</li>
+		<li>Move to a custom XML serialization such that id/xml:id is printed first (Thanks to Richard Hutch for the suggestion)</li>
 		<li>Added --indent option to specify indentation type (default='space', other options: 'none', 'tab')</li>
 	</ul>
 </section>

scour.py

@@ -1876,6 +1876,23 @@ def remapNamespacePrefix(node, oldprefix, newprefix):
 	for child in node.childNodes :
 		remapNamespacePrefix(child, oldprefix, newprefix)	
 
+def makeWellFormed(str):
+	newstr = str
+	
+	# encode & as & ( must do this first so that < does not become < )
+	if str.find('&') != -1:
+		newstr = str.replace('&', '&')
+			
+	# encode < as &lt;
+	if str.find("<") != -1:
+		newstr = str.replace('<', '&lt;')
+		
+	# encode > as &gt; (TODO: is this necessary?)
+	if str.find('>') != -1:
+		newstr = str.replace('>', '&gt;')
+	
+	return newstr
+
 # hand-rolled serialization function that has the following benefits:
 # - pretty printing
 # - somewhat judicious use of whitespace
@@ -1912,7 +1929,9 @@ def serializeXML(element, options, ind = 0):
 		if attr.nodeValue.find('"') != -1:
 			quot = "'"
 
-		outString += ' ' + attr.nodeName + '=' + quot + attr.nodeValue + quot
+		attrValue = makeWellFormed( attr.nodeValue )
+					
+		outString += ' ' + attr.nodeName + '=' + quot + attrValue + quot
 	
 	# if no children, self-close
 	children = element.childNodes
@@ -1930,9 +1949,9 @@ def serializeXML(element, options, ind = 0):
 				# trim it only in the case of not being a child of an element
 				# where whitespace might be important
 				if element.nodeName in ["text", "tspan", "textPath", "tref", "title", "desc", "textArea"]:
-					outString += child.nodeValue
+					outString += makeWellFormed(child.nodeValue)
 				else:
-					outString += child.nodeValue.strip()
+					outString += makeWellFormed(child.nodeValue.strip())
 			# CDATA node
 			elif child.nodeType == 4:
 				outString += '<![CDATA[' + child.nodeValue + ']]>'

scra.py

@@ -87,6 +87,7 @@ class ScourOptions:
 	embed_rasters = False
 	keep_editor_data = False
 	strip_xml_prolog = False
+	indent_type = "space"
 
 # params are the form elements (if a checkbox is unchecked it will not be present)
 def fetch(req, indoc,**params):

testscour.py

@@ -833,6 +833,30 @@ class CDATAInXml(unittest.TestCase):
 </svg>''',
 			'Improperly serialized the cdata unit tests')
 
+class WellFormedXMLLesserThanInAttrValue(unittest.TestCase):
+	def runTest(self):
+		wellformed = scour.scourString(open('unittests/xml-well-formed.svg').read())
+		self.assert_( wellformed.find('unicode="&lt;"') != -1,
+			"Improperly serialized &lt; in attribute value")
+
+class WellFormedXMLAmpersandInAttrValue(unittest.TestCase):
+	def runTest(self):
+		wellformed = scour.scourString(open('unittests/xml-well-formed.svg').read())
+		self.assert_( wellformed.find('unicode="&amp;"') != -1,
+			'Improperly serialized &amp; in attribute value' )
+
+class WellFormedXMLLesserThanInTextContent(unittest.TestCase):
+	def runTest(self):
+		wellformed = scour.scourString(open('unittests/xml-well-formed.svg').read())
+		self.assert_( wellformed.find('<title>2 &lt; 5</title>') != -1,
+			'Improperly serialized &lt; in text content')
+
+class WellFormedXMLAmpersandInTextContent(unittest.TestCase):
+	def runTest(self):
+		wellformed = scour.scourString(open('unittests/xml-well-formed.svg').read())
+		self.assert_( wellformed.find('<desc>Peanut Butter &amp; Jelly</desc>') != -1,
+			'Improperly serialized &amp; in text content')
+			
 # TODO; write a test for embedding rasters
 # TODO: write a test for --disable-embed-rasters
 # TODO: write tests for --keep-editor-data

unittests/xml-well-formed.svg

@@ -0,0 +1,9 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg xmlns="http://www.w3.org/2000/svg">
+	<!-- &lt;Jack &amp; Jill&gt; -->
+	<title>2 &lt; 5</title>
+	<desc>Peanut Butter &amp; Jelly</desc>
+	<glyph id="lt" unicode="&lt;"/>
+	<glyph id="amp" unicode="&amp;"/>
+	<text x="50" y="50">&#x389;TML &#x26; CSS</text>
+</svg>