#!/usr/bin/python

"""
Author: dysbulic <dys@dhappy.org>
Date: June 2008

Graphs generated by R contain paths with tens of thousands of elements which kills firefox

This is a program to split those pathes into component parts. Additionally it collapses
collinear lines into a single line.
"""

import sys, re
import xml.dom.minidom
from xml.dom.minidom import Node
from cStringIO import StringIO

if len(sys.argv) <= 1:
    print __doc__
    sys.exit(1)

file = sys.argv[1]
doc = xml.dom.minidom.parse(file)

allLinesRegex = re.compile("^( *([LM]) *(\\d+\.?\\d*)[, ](\\d+\.?\\d*) *)*")
lineRegex = re.compile("([LM]) (\\d+\.?\\d*) (\\d+\.?\\d*)")

def addPath(node, path, style):
    pathNode = doc.createElement("path")
    pathNode.setAttribute("d", path)
    pathNode.setAttribute("style", style)
    node.appendChild(pathNode)
    node.appendChild(doc.createTextNode("\n"))

def compressPathes(node):
    """Process a SVG document and merge certain sequential pathes that start and end at the same point"""
    index = 0
    while index < node.childNodes.length:
        child = node.childNodes[index]
        index += 1
        if child.nodeType == Node.ELEMENT_NODE:
            if child.nodeName == "g":
                compressPathes(child)
            elif child.nodeName == "path":
                path = child.getAttribute("d")
                
                if len(path) > 5000 and allLinesRegex.match(path):
                    style = child.getAttribute("style")
                    parent = child.parentNode
                    parent.removeChild(child)
                    points = [None, None]
                    outPath = ""

                    for match in lineRegex.findall(path):
                        point = [float(match[1]), float(match[2])]

                        # There is only one M at the beginning
                        if match[0] == "M":
                            outPath += "M%.3f,%.3f" % (point[0], point[1])
                            continue

                        # Fill the points array before continuing
                        for i in range(0, len(points)):
                            if points[i] is None:
                                points[i] = point
                                point = None
                        if point is None:
                            continue

                        # Skip intermediate collinear points
                        if ((points[1][1] - points[0][1]) * (point[0] - points[0][0]) -
                            (points[1][0] - points[0][0]) * (point[1] - points[0][1])) == 0:
                            points[1] = point
                            continue
                            
                        outPath += " L%.3f,%.3f" % (points[0][0], points[0][1])

                        if len(outPath) > 4000:
                            addPath(parent, outPath, style)
                            outPath = "M %.3f,%.3f" % (points[0][0], points[0][1])

                        # Shift points back
                        points[0] = points[1]
                        points[1] = None
                        
                    # Print any remaining path elements
                    outPath += " L%.3f,%.3f" % (points[0][0], points[0][1])
                    if points[1] is not None:
                        outPath += " L%.3f,%.3f" % (points[1][0], points[1][1])
                    addPath(parent, outPath, style)

compressPathes(doc.documentElement)
print doc.documentElement.toxml()