From e0aacf646d6f59218b5ba0e5b449e15fe556840f Mon Sep 17 00:00:00 2001
From: JSCHILL1 <jschill1@Lithium-3.local>
Date: Thu, 24 Sep 2009 10:37:09 -0500
Subject: [PATCH] Fix XML serialization when xml:space=preserve

---
 scour.py                           | 15 ++++++++++++---
 testscour.py                       | 12 +++++++++++-
 unittests/whitespace-important.svg |  4 ++++
 3 files changed, 27 insertions(+), 4 deletions(-)
 create mode 100644 unittests/whitespace-important.svg

diff --git a/scour.py b/scour.py
index aac653e..a4dbf7d 100755
--- a/scour.py
+++ b/scour.py
@@ -38,6 +38,7 @@
 # + prevent elements from being stripped if they are referenced in a <style> element
 # + only move common attributes and remove unused attributes after removing duplicate gradients
 # + only move common attributes to parent if the parent contains non-whitespace text nodes
+# + do not pretty-print elements if whitespace is important (xml:space="preserve")
 # - TODO: fix the removal of comment elements (between <?xml?> and <svg>)
 #   (for instance, filter, marker, pattern) - need a crude CSS parser
 # - add an option to remove ids if they match the Inkscape-style of IDs
@@ -2039,6 +2040,7 @@ def serializeXML(element, options, ind = 0):
 	I=''
 	if options.indent_type == 'tab': I='\t'
 	elif options.indent_type == 'space': I=' '
+	preserveWhitespace = False
 	
 	outString = (I * ind) + '<' + element.nodeName
 
@@ -2073,6 +2075,9 @@ def serializeXML(element, options, ind = 0):
 		if attr.namespaceURI == 'http://www.w3.org/2000/xmlns/' and attr.nodeName.find('xmlns') == -1:
 			outString += 'xmlns:'
 		outString += attr.nodeName + '=' + quot + attrValue + quot
+		
+		if attr.nodeName == 'xml:space' and attrValue == 'preserve':
+			preserveWhitespace = True
 	
 	# if no children, self-close
 	children = element.childNodes
@@ -2083,13 +2088,17 @@ def serializeXML(element, options, ind = 0):
 		for child in element.childNodes:
 			# element node
 			if child.nodeType == 1:
-				outString += '\n' + serializeXML(child, options, indent + 1)
-				onNewLine = True
+				if preserveWhitespace:
+					outString += serializeXML(child, options, 0)
+				else:
+					outString += '\n' + serializeXML(child, options, indent + 1)
+					onNewLine = True
 			# text node
 			elif child.nodeType == 3:
 				# 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"]:
+				if element.nodeName in ["text", "tspan", "textPath", "tref", "title", "desc", "textArea", 
+										"flowRoot", "flowDiv", "flowSpan", "flowPara", "flowRegion"]:
 					outString += makeWellFormed(child.nodeValue)
 				else:
 					outString += makeWellFormed(child.nodeValue.strip())
diff --git a/testscour.py b/testscour.py
index b7f1898..c7017b7 100755
--- a/testscour.py
+++ b/testscour.py
@@ -926,7 +926,17 @@ class DoNotRemoveGradientsWhenReferencedInStyleCss(unittest.TestCase):
 		grads = scour.scourXmlFile('unittests/css-reference.svg').getElementsByTagNameNS(SVGNS, 'linearGradient')
 		self.assertEquals( grads.length, 2,
 			'Gradients removed when referenced in CSS')
-		
+
+class DoNotPrettyPrintWhenWhitespacePreserved(unittest.TestCase):
+	def runTest(self):
+		self.assertEquals( scour.scourString(open('unittests/whitespace-important.svg').read()), 
+			'''<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg xmlns="http://www.w3.org/2000/svg">
+ <text xml:space="preserve">This is some <tspan font-style="italic">messed-up</tspan> markup</text>
+</svg>''',
+			'Whitespace not preserved')
+
+
 # TODO; write a test for embedding rasters
 # TODO: write a test for --disable-embed-rasters
 # TODO: write tests for --keep-editor-data
diff --git a/unittests/whitespace-important.svg b/unittests/whitespace-important.svg
new file mode 100644
index 0000000..e829569
--- /dev/null
+++ b/unittests/whitespace-important.svg
@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg xmlns="http://www.w3.org/2000/svg">
+ <text xml:space="preserve">This is some <tspan font-style="italic">messed-up</tspan> markup</text>
+</svg>
\ No newline at end of file