Remove uselessly nested groups, fixed up test script

scour.py

@@ -28,8 +28,6 @@
 # TODO: Adapt this script into an Inkscape python plugin
 #
 # * Specify a limit to the precision of all positional elements.
-# * Clean up XML Elements
-#  * Collapse multiple redundent groups
 # * Clean up Definitions
 #  * Remove duplicate gradient stops
 #  * Collapse duplicate gradient definitions
@@ -48,12 +46,23 @@
 #  * Collapse all group based transformations
 
 # Next Up:
-# - Remove unnecessary nested <g> elements
+# + Remove unnecessary nested <g> elements
-# - Remove duplicate gradient stops
+# - Remove duplicate gradient stops (same offset, stop-color, stop-opacity)
 # - Convert all colors to #RRGGBB format
+# - Reduce #RRGGBB format to #RGB format when possible
 # - rework command-line argument processing so that options are configurable
 # - remove unreferenced patterns? https://bugs.edge.launchpad.net/ubuntu/+source/human-icon-theme/+bug/361667/
 
+# Some notes to not forget:
+# - removing unreferenced IDs loses some semantic information
+# - removing empty nested groups also potentially loses some semantic information
+#   (i.e. the following button:
+#     <g>
+#       <rect .../>
+#       <text .../>
+#     </g>
+#   will be flattened)
+
 
 # necessary to get true division
 from __future__ import division
@@ -255,7 +264,7 @@ def removeNamespacedAttributes(node, namespaces):
 		
 		# now recurse for children
 		for child in node.childNodes:
-			removeNamespacedAttributes(child, namespaces)
+			num += removeNamespacedAttributes(child, namespaces)
 	return num
 	
 def removeNamespacedElements(node, namespaces):
@@ -275,10 +284,34 @@ def removeNamespacedElements(node, namespaces):
 		
 		# now recurse for children
 		for child in node.childNodes:
-			removeNamespacedElements(child, namespaces)
+			num += removeNamespacedElements(child, namespaces)
 	return num
 
-# TODO: create a class for a SVGLength type (including value and unit)
+# this walks further and further down the tree, removing groups
+# which do not have any attributes and promoting their children
+# up one level
+def removeNestedGroups(node):
+	global numElemsRemoved
+	num = 0
+	
+	groupsToRemove = []
+	for child in node.childNodes:
+		if child.nodeName == 'g' and child.namespaceURI == NS['SVG'] and len(child.attributes) == 0:
+			groupsToRemove.append(child)
+
+	for g in groupsToRemove:
+		while g.childNodes.length > 0:
+			g.parentNode.insertBefore(g.firstChild, g)
+		g.parentNode.removeChild(g)
+		numElemsRemoved += 1
+		num += 1
+
+	# now recurse for children
+	for child in node.childNodes:
+		if child.nodeType == 1:
+			num += removeNestedGroups(child)
+		
+	return num
 
 coord = re.compile("\\-?\\d+\\.?\\d*")
 scinumber = re.compile("[\\-\\+]?(\\d*\\.?)?\\d+[eE][\\-\\+]?\\d+")
@@ -629,13 +662,17 @@ def scourString(in_string):
 				numElemsRemoved += 1
 
 	# remove unreferenced gradients/patterns outside of defs
-	removeUnreferencedElements(doc)
+	while removeUnreferencedElements(doc) > 0:
+		pass
+	
+	while removeNestedGroups(doc.documentElement) > 0:
+		pass
 
 	# clean path data
 	for elem in doc.documentElement.getElementsByTagNameNS(NS['SVG'], 'path') :
 		cleanPath(elem)
 
-	# convert rasters refereces to base64-encoded strings 
+	# convert rasters references to base64-encoded strings 
 	for elem in doc.documentElement.getElementsByTagNameNS(NS['SVG'], 'image') :
 		embedRasters(elem)		
 

testscour.py

@@ -21,6 +21,8 @@ import unittest
 import scour
 import xml.dom.minidom
 
+SVGNS = 'http://www.w3.org/2000/svg'
+
 # I couldn't figure out how to get ElementTree to work with the following XPath 
 # "//*[namespace-uri()='http://example.com']"
 # so I decided to use minidom and this helper function that performs a test on a given node 
@@ -95,6 +97,55 @@ class NoAdobeXPathElements(unittest.TestCase):
 			lambda e: e.namespaceURI != 'http://ns.adobe.com/XPath/1.0/'), False,
 			'Found Adobe XPath elements' )
 
+class DoNotRemoveMetadataWithOnlyText(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/metadata-with-text.svg')
+		self.assertEquals(len(doc.getElementsByTagNameNS(SVGNS, 'metadata')), 1,
+			'Removed metadata element with only text child' )
+
+class RemoveEmptyMetadataElement(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/empty-metadata.svg')
+		self.assertEquals(len(doc.getElementsByTagNameNS(SVGNS, 'metadata')), 0,
+			'Did not remove empty metadata element' )
+
+class RemoveEmptyGElements(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/empty-g.svg')
+		self.assertEquals(len(doc.getElementsByTagNameNS(SVGNS, 'g')), 1,
+			'Did not remove empty g element' )
+
+class RemoveUnreferencedPattern(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/unreferenced-pattern.svg')
+		self.assertEquals(len(doc.getElementsByTagNameNS(SVGNS, 'pattern')), 0,
+			'Unreferenced pattern not removed' )
+
+class RemoveUnreferencedLinearGradient(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/unreferenced-linearGradient.svg')
+		self.assertEquals(len(doc.getElementsByTagNameNS(SVGNS, 'linearGradient')), 0,
+			'Unreferenced linearGradient not removed' )
+
+class RemoveUnreferencedRadialGradient(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/unreferenced-radialGradient.svg')
+		self.assertEquals(len(doc.getElementsByTagNameNS(SVGNS, 'radialradient')), 0,
+			'Unreferenced radialGradient not removed' )
+
+class RemoveUselessNestedGroups(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/nested-useless-groups.svg')
+		self.assertEquals(len(doc.getElementsByTagNameNS(SVGNS, 'g')), 1,
+			'Useless nested groups not removed' )
+
+# These tests will fail at present
+class RemoveDuplicateGradientStops(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/duplicate-gradient-stops.svg')
+		self.assertEquals(len(doc.getElementsByTagNameNS(SVGNS, 'stop')), 3,
+			'Duplicate gradient stops not removed' )
+
 #class NoInkscapeAttributes(unittest.TestCase):
 #	def runTest(self):
 #		self.assertNotEquals(walkTree(scour.scourXmlFile('unittests/inkscape.svg').documentElement,
@@ -102,54 +153,6 @@ class NoAdobeXPathElements(unittest.TestCase):
 #			False, 
 #			'Found Inkscape attributes')
 
-class DoNotRemoveMetadataWithOnlyText(unittest.TestCase):
-	def runTest(self):
-		doc = scour.scourXmlFile('unittests/metadata-with-text.svg')
-		self.assertEquals(len(doc.getElementsByTagNameNS('http://www.w3.org/2000/svg', 'metadata')), 1,
-			'Removed metadata element with only text child' )
-
-class RemoveEmptyMetadataElement(unittest.TestCase):
-	def runTest(self):
-		doc = scour.scourXmlFile('unittests/empty-metadata.svg')
-		self.assertEquals(len(doc.getElementsByTagNameNS('http://www.w3.org/2000/svg', 'metadata')), 0,
-			'Did not remove empty metadata element' )
-
-class RemoveEmptyGElements(unittest.TestCase):
-	def runTest(self):
-		doc = scour.scourXmlFile('unittests/empty-g.svg')
-		self.assertEquals(len(doc.getElementsByTagNameNS('http://www.w3.org/2000/svg', 'g')), 1,
-			'Did not remove empty g element' )
-
-class RemoveUnreferencedPattern(unittest.TestCase):
-	def runTest(self):
-		doc = scour.scourXmlFile('unittests/unreferenced-pattern.svg')
-		self.assertEquals(len(doc.getElementsByTagNameNS('http://www.w3.org/2000/svg', 'pattern')), 0,
-			'Unreferenced pattern not removed' )
-
-class RemoveUnreferencedLinearGradient(unittest.TestCase):
-	def runTest(self):
-		doc = scour.scourXmlFile('unittests/unreferenced-linearGradient.svg')
-		self.assertEquals(len(doc.getElementsByTagNameNS('http://www.w3.org/2000/svg', 'linearGradient')), 0,
-			'Unreferenced linearGradient not removed' )
-
-class RemoveUnreferencedRadialGradient(unittest.TestCase):
-	def runTest(self):
-		doc = scour.scourXmlFile('unittests/unreferenced-radialGradient.svg')
-		self.assertEquals(len(doc.getElementsByTagNameNS('http://www.w3.org/2000/svg', 'radialradient')), 0,
-			'Unreferenced radialGradient not removed' )
-
-# These tests will fail at present
-class RemoveUselessNestedGroups(unittest.TestCase):
-	def runTest(self):
-		doc = scour.scourXmlFile('unittests/nested-useless-groups.svg')
-		self.assertEquals(len(doc.getElementsByTagNameNS('http://wwww.w3.org/2000/svg', 'g')), 1,
-			'Useless nested groups not removed' )
-
-class RemoveDuplicateGradientStops(unittest.TestCase):
-	def runTest(self):
-		doc = scour.scourXmlFile('unittests/duplicate-gradient-stops.svg')
-		self.assertEquals(len(doc.getElementsByTagNameNS('http://www.w3.org/2000/svg', 'stop')), 3,
-			'Duplicate gradient stops not removed' )
 
 if __name__ == '__main__':
     unittest.main()

unittests/empty-g.svg

@@ -1,6 +1,6 @@
 <svg xmlns="http://www.w3.org/2000/svg">
   <g/>
-  <g>
+  <g transform="translate(10,10)">
     <rect width="300" height="200" fill="green" />
   </g>
 </svg>