From b28ae6ec8d4667ac54084d252fb7cc937051b94f Mon Sep 17 00:00:00 2001
From: JSCHILL1 <jschill1@Lithium-3.local>
Date: Wed, 29 Jul 2009 12:27:51 -0500
Subject: [PATCH] Remove trailing zeros from path coordinates.  Use scientific
 notation in path coords if shorter.  Scour polygon coordinates just like path
 coordinates.  Added tests

---
 release-notes.html                            | 13 +++
 scour.py                                      | 96 +++++++++++++------
 testscour.py                                  | 33 +++++--
 unittests/path-precision.svg                  |  3 +-
 unittests/path-truncate-zeros-calc.svg        |  4 +
 ...ate-zeroes.svg => path-truncate-zeros.svg} |  0
 unittests/path-use-scientific-notation.svg    |  4 +
 unittests/polygon-coord.svg                   |  4 +
 unittests/polygon.svg                         |  1 +
 9 files changed, 119 insertions(+), 39 deletions(-)
 create mode 100644 unittests/path-truncate-zeros-calc.svg
 rename unittests/{path-truncate-zeroes.svg => path-truncate-zeros.svg} (100%)
 create mode 100644 unittests/path-use-scientific-notation.svg
 create mode 100644 unittests/polygon-coord.svg

diff --git a/release-notes.html b/release-notes.html
index e6ea75a..b44eb46 100644
--- a/release-notes.html
+++ b/release-notes.html
@@ -9,6 +9,19 @@
 
 <p>Copyright 2009, Jeff Schiller</p>
 
+<section id="0.16">
+	<header>
+		<h2><a href="#0.16">Version 0.16</a></h2>
+	</header>
+	<p>July 29th, 2009</p>
+	<ul>
+		<li>Fix <a href="https://bugs.launchpad.net/scour/+bug/401628">Bug 401628</a>: Keep namespace declarations when using --keep-editor-data (Thanks YoNoSoyTu!)</li>
+		<li>Remove trailing zeros after decimal places for all path coordinates</li>
+		<li>Use scientific notation in path coordinates if that representation is shorter</li>
+		<li>Scour polygon coordinates just like path coordinates</li>
+	</ul>
+</section>
+
 <section id="0.15">
 	<header>
 		<h2><a href="#0.15">Version 0.15</a></h2>
diff --git a/scour.py b/scour.py
index aab3244..1681751 100755
--- a/scour.py
+++ b/scour.py
@@ -39,10 +39,14 @@
 
 # Next Up:
 # + add option to keep inkscape, adobe, sodipodi elements and attributes
-# - ensure a really good understanding of prec vs. quantize and what I want --set-precision to do
+# + if any path coordinate has decimal places, remove any trailing zeros
+# + use scientific notation is shorter in path coordinates
+# + scour polygon coordinates just like path coordinates
+# - scour polyline coordinates just like path coordinates
+# - if after reducing precision we have duplicate path segments, then remove the duplicates and
+#   leave it as a straight line segment
 # - enable the precision argument to affect all numbers: polygon points, lengths, coordinates
 # - remove id if it matches the Inkscape-style of IDs (also provide a switch to disable this)
-# - convert polygons/polylines to path? (actually the change in semantics may not be worth the marginal savings)
 # - prevent elements from being stripped if they are referenced in a <style> element
 #   (for instance, filter, marker, pattern) - need a crude CSS parser
 # - Remove any unused glyphs from font elements?
@@ -69,7 +73,7 @@ except ImportError:
 	Decimal = FixedPoint	
 
 APP = 'scour'
-VER = '0.15'
+VER = '0.16'
 COPYRIGHT = 'Copyright Jeff Schiller, 2009'
 
 NS = { 	'SVG': 		'http://www.w3.org/2000/svg', 
@@ -956,9 +960,9 @@ def cleanPath(element) :
 			# one or more tuples, each containing two numbers
 			nums = []
 			for t in dataset:
-				# convert to a Decimal and ensure precision
-				nums.append(Decimal(str(t[0])) * Decimal(1))
-				nums.append(Decimal(str(t[1])) * Decimal(1))
+				# convert to a Decimal
+				nums.append(Decimal(str(t[0])))
+				nums.append(Decimal(str(t[1])))
 					
 			# only create this segment if it is not empty
 			if nums:
@@ -1339,7 +1343,7 @@ def parseListOfPoints(s):
 	"""
 		Parse string into a list of points.
 	
-		Returns a list of (x,y) tuples where x and y are strings
+		Returns a list of containing an even number of coordinate strings
 	"""
 	
 	# (wsp)? comma-or-wsp-separated coordinate pairs (wsp)?
@@ -1356,7 +1360,8 @@ def parseListOfPoints(s):
 		
 		# if the coordinates were not unitless, return empty
 		if x.units != Unit.NONE or y.units != Unit.NONE: return []
-		points.append( (str(x.value),str(y.value)) )
+		points.append( str(x.value) )
+		points.append( str(y.value) )
 		i += 2
 	
 	return points
@@ -1368,41 +1373,70 @@ def cleanPolygon(elem):
 	global numPointsRemovedFromPolygon
 	
 	pts = parseListOfPoints(elem.getAttribute('points'))
-	N = len(pts)
+	N = len(pts)/2
 	if N >= 2:		
-		(startx,starty) = (pts[0][0],pts[0][1])
-		(endx,endy) = (pts[N-1][0],pts[N-1][1])
+		(startx,starty) = (pts[0],pts[0])
+		(endx,endy) = (pts[len(pts)-2],pts[len(pts)-1])
 		if startx == endx and starty == endy:
-			str = ''
-			for pt in pts[:-1]:
-				str += (pt[0] + ',' + pt[1] + ' ')
-			elem.setAttribute('points', str[:-1])
+			pts = pts[:-2]
 			numPointsRemovedFromPolygon += 1
+		
+	elem.setAttribute('points', scourCoordinates(pts))
 	
 def serializePath(pathObj):
 	"""
-		Reserializes the path data with some cleanups:
-			- removes scientific notation (exponents)
-			- removes all trailing zeros after the decimal
-			- removes extraneous whitespace
-			- adds commas between values in a subcommand if required
+		Reserializes the path data with some cleanups.
 	"""
 	pathStr = ""
 	for (cmd,data) in pathObj:
 		pathStr += cmd
-		if data != None:
-			c = 0
-			for coord in data:
-				# if coord can be an integer without loss of precision, go for it
-				if int(coord) == coord: pathStr += str(int(coord))
-				else: pathStr += str(coord)
-				
-				# only need the comma if the next number is non-negative
-				if c < len(data)-1 and data[c+1] >= 0:
-					pathStr += ','
-				c += 1
+		pathStr += scourCoordinates(data)
 	return pathStr
 
+def scourCoordinates(data):
+	"""
+		Serializes coordinate data with some cleanups:
+			- removes all trailing zeros after the decimal
+			- integerize coordinates if possible
+			- removes extraneous whitespace
+			- adds commas between values in a subcommand if required
+	"""
+	coordsStr = ""
+	if data != None:
+		c = 0
+		for coord in data:
+			if int(coord) == coord: coord = Decimal(str(int(coord)))
+				
+			# Decimal.trim() is available in Python 2.6+ to trim trailing zeros
+			try:
+				coord = coord.trim()
+			except AttributeError:
+				# trim it ourselves
+				s = str(coord)
+				dec = s.find('.')
+				if dec != -1:
+					while s[-1] == '0':
+						s = s[:-1]
+				coord = Decimal(s)
+
+			# reduce to the proper number of digits
+			coord = coord * Decimal(1)
+
+			# Decimal.normalize() will uses scientific notation - if that
+			# string is smaller, then use it
+			normd = coord.normalize()
+			if len(str(normd)) < len(str(coord)):
+				coord = normd
+
+			# finally add the coordinate to the path string
+			coordsStr += str(coord)
+			
+			# only need the comma if the next number is non-negative
+			if c < len(data)-1 and Decimal(data[c+1]) >= 0:
+				coordsStr += ','
+			c += 1
+	return coordsStr
+
 def embedRasters(element, options) :
 	"""
 		Converts raster references to inline images.
diff --git a/testscour.py b/testscour.py
index 3af701f..228c630 100755
--- a/testscour.py
+++ b/testscour.py
@@ -470,19 +470,33 @@ class DoNotCollapseMultiplyReferencedGradients(unittest.TestCase):
 		self.assertNotEquals(len(doc.getElementsByTagNameNS(SVGNS, 'linearGradient')), 0,
 			'Multiply-referenced linear gradient collapsed' )
 
-class RemoveTrailingZeroesFromPath(unittest.TestCase):
+class RemoveTrailingZerosFromPath(unittest.TestCase):
 	def runTest(self):
-		doc = scour.scourXmlFile('unittests/path-truncate-zeroes.svg')
+		doc = scour.scourXmlFile('unittests/path-truncate-zeros.svg')
 		path = doc.getElementsByTagNameNS(SVGNS, 'path')[0].getAttribute('d')
 		self.assertEquals(path[:4] == 'M300' and path[4] != '.', True,
 			'Trailing zeros not removed from path data' )
 
+class RemoveTrailingZerosFromPathAfterCalculation(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/path-truncate-zeros-calc.svg')
+		path = doc.getElementsByTagNameNS(SVGNS, 'path')[0].getAttribute('d')
+		self.assertEquals(path, 'M5.81,0h0.1',
+			'Trailing zeros not removed from path data after calculation' )
+
 class RemoveDelimiterBeforeNegativeCoordsInPath(unittest.TestCase):
 	def runTest(self):
-		doc = scour.scourXmlFile('unittests/path-truncate-zeroes.svg')
+		doc = scour.scourXmlFile('unittests/path-truncate-zeros.svg')
 		path = doc.getElementsByTagNameNS(SVGNS, 'path')[0].getAttribute('d')
 		self.assertEquals(path[4], '-', 
 			'Delimiters not removed before negative coordinates in path data' )
+			
+class UseScientificNotationToShortenCoordsInPath(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/path-use-scientific-notation.svg')
+		path = doc.getElementsByTagNameNS(SVGNS, 'path')[0].getAttribute('d')
+		self.assertEquals(path, 'M1E+4,0',
+			'Not using scientific notation for path coord when representation is shorter')
 
 class ConvertAbsoluteToRelativePathCommands(unittest.TestCase):
 	def runTest(self):
@@ -506,7 +520,7 @@ class LimitPrecisionInPathData(unittest.TestCase):
 	def runTest(self):
 		doc = scour.scourXmlFile('unittests/path-precision.svg')
 		path = svg_parser.parse(doc.getElementsByTagNameNS(SVGNS, 'path')[0].getAttribute('d'))
-		self.assertEquals(path[1][1][0], 100.001,
+		self.assertEquals(path[1][1][0], 100.01,
 			'Not correctly limiting precision on path data' )
 
 class RemoveEmptyLineSegmentsFromPath(unittest.TestCase):
@@ -616,14 +630,20 @@ class ConvertStraightCurvesToLines(unittest.TestCase):
 class RemoveUnnecessaryPolgonEndPoint(unittest.TestCase):
 	def runTest(self):
 		p = scour.scourXmlFile('unittests/polygon.svg').getElementsByTagNameNS(SVGNS, 'polygon')[0]
-		self.assertEquals(p.getAttribute('points'), '50,50 150,50 150,150 50,150',
+		self.assertEquals(p.getAttribute('points'), '50,50,150,50,150,150,50,150',
 			'Unnecessary polygon end point not removed' )
 
 class DoNotRemovePolgonLastPoint(unittest.TestCase):
 	def runTest(self):
 		p = scour.scourXmlFile('unittests/polygon.svg').getElementsByTagNameNS(SVGNS, 'polygon')[1]
-		self.assertEquals(p.getAttribute('points'), '200,50 300,50 300,150 200,150',
+		self.assertEquals(p.getAttribute('points'), '200,50,300,50,300,150,200,150',
 			'Last point of polygon removed' )
+			
+class ScourPolygonCoordinates(unittest.TestCase):
+	def runTest(self):
+		p = scour.scourXmlFile('unittests/polygon-coord.svg').getElementsByTagNameNS(SVGNS, 'polygon')[0]
+		self.assertEquals(p.getAttribute('points'), '1E+4-50',
+			'Polygon coordinates not scoured')
 
 class DoNotRemoveGroupsWithIDsInDefs(unittest.TestCase):
 	def runTest(self):
@@ -637,7 +657,6 @@ class AlwaysKeepClosePathSegments(unittest.TestCase):
 		self.assertEquals(p.getAttribute('d'), 'M10,10h100v100h-100z',
 			'Path with closepath not preserved')
 
-# TODO: write tests for --set-precision for path data, for polygon data, for attributes
 # 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/path-precision.svg b/unittests/path-precision.svg
index 1eb2bfe..8e1e267 100644
--- a/unittests/path-precision.svg
+++ b/unittests/path-precision.svg
@@ -1,3 +1,4 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
 <svg viewBox="0 0 400 400" xmlns="http://www.w3.org/2000/svg" version="1.1">
-  <path d="M 100.0000001 99.9999999 h100.001 v123456789.123456789 h-100 z" fill="red" />
+  <path d="M 100.0000001 99.9999999 h100.01 v123456789.123456789 h-100 z" fill="red" />
 </svg>
diff --git a/unittests/path-truncate-zeros-calc.svg b/unittests/path-truncate-zeros-calc.svg
new file mode 100644
index 0000000..c889fff
--- /dev/null
+++ b/unittests/path-truncate-zeros-calc.svg
@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="5 0 3 3">
+  <path stroke="blue" stroke-width="0.01" fill="none" d="M5.81,0 H5.91000" />
+</svg>
diff --git a/unittests/path-truncate-zeroes.svg b/unittests/path-truncate-zeros.svg
similarity index 100%
rename from unittests/path-truncate-zeroes.svg
rename to unittests/path-truncate-zeros.svg
diff --git a/unittests/path-use-scientific-notation.svg b/unittests/path-use-scientific-notation.svg
new file mode 100644
index 0000000..afbbf05
--- /dev/null
+++ b/unittests/path-use-scientific-notation.svg
@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="5 0 3 3">
+  <path d="M10000,0" />
+</svg>
diff --git a/unittests/polygon-coord.svg b/unittests/polygon-coord.svg
new file mode 100644
index 0000000..dbd5927
--- /dev/null
+++ b/unittests/polygon-coord.svg
@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg xmlns="http://www.w3.org/2000/svg">
+	<polygon fill="blue" points="10000,-50" />
+</svg>
\ No newline at end of file
diff --git a/unittests/polygon.svg b/unittests/polygon.svg
index b75aeee..f756050 100644
--- a/unittests/polygon.svg
+++ b/unittests/polygon.svg
@@ -1,3 +1,4 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
 <svg xmlns="http://www.w3.org/2000/svg">
 	<polygon fill="blue" points="50,50 150,50 150,150 50,150 +5e1,500.00e-1" />
 	<polygon fill="green" points="200,50 300,50 300,150 200,150" />