From 404c013e5f00457a657ebf43d724f21d0b3a7b5b Mon Sep 17 00:00:00 2001
From: Cynthia Gauthier <cynthia@jolteon>
Date: Fri, 2 Jul 2010 05:35:31 -0400
Subject: [PATCH] Implement the feature described in bug 598976: Create a <g>
 with the common attributes of a run of elements if there are 3 or more
 elements in the run.

---
 scour.inkscape.py               |   3 +
 scour.inx                       |   8 ++-
 scour.py                        | 124 ++++++++++++++++++++++++++++++--
 testscour.py                    |  22 ++++++
 unittests/group-creation.svg    |   6 ++
 unittests/group-no-creation.svg |   6 ++
 6 files changed, 162 insertions(+), 7 deletions(-)
 create mode 100644 unittests/group-creation.svg
 create mode 100644 unittests/group-no-creation.svg

diff --git a/scour.inkscape.py b/scour.inkscape.py
index ad3bfcd..ebbae81 100755
--- a/scour.inkscape.py
+++ b/scour.inkscape.py
@@ -19,6 +19,9 @@ class ScourInkscape (inkex.Effect):
         self.OptionParser.add_option("--group-collapsing", type="inkbool",
             action="store", dest="group_collapse", default=True,
             help="won't collapse <g> elements")
+        self.OptionParser.add_option("--create-groups", type="inkbool",
+            action="store", dest="group_create", default=False,
+            help="create <g> elements for runs of elements with identical attributes")
         self.OptionParser.add_option("--enable-id-stripping", type="inkbool",
             action="store", dest="strip_ids", default=False,
             help="remove all un-referenced ID attributes")
diff --git a/scour.inx b/scour.inx
index 4254882..26d9b62 100644
--- a/scour.inx
+++ b/scour.inx
@@ -10,6 +10,7 @@
             <param name="simplify-colors"  type="boolean" _gui-text="Shorten color values">true</param>
             <param name="style-to-xml"  type="boolean" _gui-text="Convert CSS attributes to XML attributes">true</param>
             <param name="group-collapsing"  type="boolean" _gui-text="Group collapsing">true</param>
+            <param name="create-groups"  type="boolean" _gui-text="Create groups for similar attributes">true</param>
             <param name="enable-id-stripping"  type="boolean" _gui-text="Remove unused ID names for elements">false</param>
             <param name="shorten-ids"  type="boolean" _gui-text="Shorten IDs">false</param>
             <param name="embed-rasters"  type="boolean" _gui-text="Embed rasters">true</param>
@@ -18,7 +19,7 @@
             <param name="enable-comment-stripping"  type="boolean" _gui-text="Remove comments">false</param>
             <param name="renderer-workaround"  type="boolean" _gui-text="Work around renderer bugs">false</param>
             <param name="enable-viewboxing" type="boolean" _gui-text="Enable viewboxing">false</param>
-            <param name="strip-xml-prolog"  type="boolean" _gui-text="Remove the <?xml?> declaration">false</param>
+            <param name="strip-xml-prolog"  type="boolean" _gui-text="Remove the &lt;?xml?&gt; declaration">false</param>
             <param name="set-precision"  type="int" _gui-text="Number of significant digits for coords">5</param>
       	    <param name="indent" type="enum" _gui-text="XML indentation (pretty-printing)">
                 <_item value="space">Space</_item>
@@ -29,8 +30,9 @@
         <page name="Help" _gui-text="Help">
             <_param name="instructions" type="description" xml:space="preserve">This extension optimizes the SVG file according to the following options:
     * Shorten color names: convert all colors to #RRGGBB or #RGB format.
-    * Convert CSS attributes to XML attributes: convert styles from <style> tags and inline style="" declarations into XML attributes.
-    * Group collapsing: removes useless <g> elements, promoting their contents up one level. Requires "Remove unused ID names for elements" to be set.
+    * Convert CSS attributes to XML attributes: convert styles from &lt;style&gt; tags and inline style="" declarations into XML attributes.
+    * Group collapsing: removes useless &lt;g&gt; elements, promoting their contents up one level. Requires "Remove unused ID names for elements" to be set.
+    * Create groups for similar attributes: create &lt;g&gt; elements for runs of elements having at least one attribute in common (e.g. fill color, stroke opacity, ...).
     * Remove unused ID names for elements: remove all unreferenced ID attributes.
     * Shorten IDs: reduce the length of all ID attributes, assigning the shortest to the most-referenced elements. For instance, #linearGradient5621, referenced 100 times, can become #a.
     * Embed rasters: embed raster images as base64-encoded data URLs.
diff --git a/scour.py b/scour.py
index b2aae96..8e0bc3c 100755
--- a/scour.py
+++ b/scour.py
@@ -853,6 +853,112 @@ def moveCommonAttributesToParentGroup(elem):
 	num += (len(childElements)-1) * len(commonAttrs)
 	return num
 
+def createGroupsForCommonAttributes(elem):
+	"""
+	Creates <g> elements to contain runs of 3 or more
+	consecutive child elements having at least one common attribute.
+	
+	Common attributes are not promoted to the <g> by this function.
+	This is handled by moveCommonAttributesToParentGroup.
+	
+	If all children have a common attribute, an extra <g> is not created.
+	
+	This function acts recursively on the given element.
+	"""
+	num = 0
+	global numElemsRemoved
+	
+	# TODO perhaps all of the Presentation attributes in http://www.w3.org/TR/SVG/struct.html#GElement
+	# could be added here
+	# Cyn: These attributes are the same as in moveAttributesToParentGroup, and must always be
+	for curAttr in ['clip-rule',
+				'display-align', 
+				'fill', 'fill-opacity', 'fill-rule', 
+				'font', 'font-family', 'font-size', 'font-size-adjust', 'font-stretch',
+				'font-style', 'font-variant', 'font-weight',
+				'letter-spacing',
+				'pointer-events', 'shape-rendering',
+				'stroke', 'stroke-dasharray', 'stroke-dashoffset', 'stroke-linecap', 'stroke-linejoin',
+				'stroke-miterlimit', 'stroke-opacity', 'stroke-width',
+				'text-anchor', 'text-decoration', 'text-rendering', 'visibility', 
+				'word-spacing', 'writing-mode']:
+		# Iterate through the children in reverse order, so item(i) for
+		# items we have yet to visit still returns the correct nodes.
+		curChild = elem.childNodes.length - 1
+		while curChild >= 0:
+			childNode = elem.childNodes.item(curChild)
+			
+			if childNode.nodeType == 1 and childNode.getAttribute(curAttr) != '':
+				# We're in a possible run! Track the value and run length.
+				value = childNode.getAttribute(curAttr)
+				runStart, runEnd = curChild, curChild
+				# Run elements includes only element tags, no whitespace/comments/etc.
+				# Later, we calculate a run length which includes these.
+				runElements = 1
+				
+				# Backtrack to get all the nodes having the same
+				# attribute value, preserving any nodes in-between.
+				while runStart > 0:
+					nextNode = elem.childNodes.item(runStart - 1)
+					if nextNode.nodeType == 1:
+						if nextNode.getAttribute(curAttr) != value: break
+						else:
+							runElements += 1
+							runStart -= 1
+					else: runStart -= 1
+				
+				if runElements >= 3:
+					# Include whitespace/comment/etc. nodes in the run.
+					while runEnd < elem.childNodes.length - 1:
+						if elem.childNodes.item(runEnd + 1).nodeType == 1: break
+						else: runEnd += 1
+					
+					runLength = runEnd - runStart + 1
+					if runLength == elem.childNodes.length: # Every child has this
+						# If the current parent is a <g> already,
+						if elem.nodeName == 'g' and elem.namespaceURI == NS['SVG']:
+							# do not act altogether on this attribute; all the
+							# children have it in common.
+							# Let moveCommonAttributesToParentGroup do it.
+							curChild = -1
+							continue
+						# otherwise, it might be an <svg> element, and
+						# even if all children have the same attribute value,
+						# it's going to be worth making the <g> since
+						# <svg> doesn't support attributes like 'stroke'.
+						# Fall through.
+					
+					# Create a <g> element from scratch.
+					# We need the Document for this.
+					document = elem.parentNode
+					while document.parentNode is not None:
+						document = document.parentNode
+					group = document.createElementNS(NS['SVG'], 'g')
+					# Move the run of elements to the group.
+					for dummy in xrange(runLength):
+						node = elem.childNodes.item(runStart)
+						elem.removeChild(node)
+						group.appendChild(node)
+					# Include the group in elem's children.
+					if elem.childNodes.length == runStart:
+						elem.appendChild(group)
+					else:
+						elem.insertBefore(group, elem.childNodes.item(runStart))
+					num += 1
+					curChild = runStart - 1
+					numElemsRemoved -= 1
+				else:
+					curChild -= 1
+			else:
+				curChild -= 1
+	
+	# each child gets the same treatment, recursively
+	for childNode in elem.childNodes:
+		if childNode.nodeType == 1:
+			num += createGroupsForCommonAttributes(childNode)
+
+	return num
+
 def removeUnusedAttributesOnParent(elem):
 	"""
 	This recursively calls this function on all children of the element passed in,
@@ -2527,10 +2633,6 @@ def scourString(in_string, options=None):
 			identifiedElements = findElementsWithId(doc.documentElement)
 			referencedIDs = findReferencedElements(doc.documentElement)
 			bContinueLooping = (removeUnreferencedIDs(referencedIDs, identifiedElements) > 0)
-	
-	if options.group_collapse:
-		while removeNestedGroups(doc.documentElement) > 0:
-			pass
 
 	while removeDuplicateGradientStops(doc) > 0:
 		pass
@@ -2543,6 +2645,11 @@ def scourString(in_string, options=None):
 	while removeDuplicateGradients(doc) > 0:
 		pass
 	
+	# create <g> elements if there are runs of elements with the same attributes.
+	# this MUST be before moveCommonAttributesToParentGroup.
+	if options.group_create:
+		createGroupsForCommonAttributes(doc.documentElement)
+	
 	# move common attributes to parent group
 	# NOTE: the if the <svg> element's immediate children
 	# all have the same value for an attribute, it must not
@@ -2553,6 +2660,12 @@ def scourString(in_string, options=None):
 	
 	# remove unused attributes from parent
 	numAttrsRemoved += removeUnusedAttributesOnParent(doc.documentElement)
+	
+	# Collapse groups LAST, because we've created groups. If done before
+	# moveAttributesToParentGroup, empty <g>'s may remain.
+	if options.group_collapse:
+		while removeNestedGroups(doc.documentElement) > 0:
+			pass
 
 	# remove unnecessary closing point of polygons and scour points
 	for polygon in doc.documentElement.getElementsByTagName('polygon') :
@@ -2665,6 +2778,9 @@ _options_parser.add_option("--disable-style-to-xml",
 _options_parser.add_option("--disable-group-collapsing",
 	action="store_false", dest="group_collapse", default=True,
 	help="won't collapse <g> elements")
+_options_parser.add_option("--create-groups",
+	action="store_true", dest="group_create", default=False,
+	help="create <g> elements for runs of elements with identical attributes")
 _options_parser.add_option("--enable-id-stripping",
 	action="store_true", dest="strip_ids", default=False,
 	help="remove all un-referenced ID attributes")
diff --git a/testscour.py b/testscour.py
index 1213951..bd70c40 100755
--- a/testscour.py
+++ b/testscour.py
@@ -51,6 +51,7 @@ class ScourOptions:
 	shorten_ids = False
 	strip_comments = False
 	remove_metadata = False
+	group_create = False
 
 class NoInkscapeElements(unittest.TestCase):
 	def runTest(self):
@@ -1075,6 +1076,27 @@ class MustKeepGInSwitch2(unittest.TestCase):
 		self.assertEquals(doc.getElementsByTagName('g').length, 1,
 			'Erroneously removed a <g> in a <switch>')
 
+class GroupCreation(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/group-creation.svg',
+			scour.parse_args(['--create-groups'])[0])
+		self.assertEquals(doc.getElementsByTagName('g').length, 1,
+			'Did not create a <g> for a run of elements having similar attributes')
+
+class GroupCreationForInheritableAttributesOnly(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/group-creation.svg',
+			scour.parse_args(['--create-groups'])[0])
+		self.assertEquals(doc.getElementsByTagName('g').item(0).getAttribute('y'), '',
+			'Promoted the uninheritable attribute y to a <g>')
+
+class GroupNoCreation(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/group-no-creation.svg',
+			scour.parse_args(['--create-groups'])[0])
+		self.assertEquals(doc.getElementsByTagName('g').length, 0,
+			'Created a <g> for a run of elements having dissimilar attributes')
+
 
 # TODO: write tests for --enable-viewboxing
 # TODO; write a test for embedding rasters
diff --git a/unittests/group-creation.svg b/unittests/group-creation.svg
new file mode 100644
index 0000000..96776c0
--- /dev/null
+++ b/unittests/group-creation.svg
@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg xmlns="http://www.w3.org/2000/svg">
+    <rect fill="red" stroke="blue" x="0" y="0" width="4" height="4" />
+    <rect fill="red" stroke="blue" x="8" y="0" width="4" height="4" />
+    <rect fill="red" stroke="blue" x="16" y="0" width="4" height="4" />
+</svg>
diff --git a/unittests/group-no-creation.svg b/unittests/group-no-creation.svg
new file mode 100644
index 0000000..bea6419
--- /dev/null
+++ b/unittests/group-no-creation.svg
@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg xmlns="http://www.w3.org/2000/svg">
+    <rect fill="green" stroke="blue" x="0" y="0" width="4" height="4" />
+    <rect fill="yellow" stroke="red" x="8" y="0" width="4" height="4" />
+    <rect fill="blue" stroke="red" x="16" y="0" width="4" height="4" />
+</svg>