"""
This program converts SVG files to raw Unicode strings and back again
It was written for a twitter image encoding competition on Stack Overflow
http://stackoverflow.com/questions/891643/twitter-image-encoding-challenge

Copyright: SpliFF
License: Public Domain
"""

import re

HEADER = """<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
<svg width="%dpt" height="%dpt" viewBox="0 0 %d %d" version="1.1" xmlns="http://www.w3.org/2000/svg">
"""
FOOTER = """</svg>"""
PATH = """<path fill="#%s" d=" M %s Z" />"""

RE_WIDTH_HEIGHT = re.compile(r'<svg width="([0-9]+)pt" height="([0-9]+)pt"')
RE_FILL_COLOR = re.compile(r'fill="#([0-9A-Fa-f]+)"')
RE_POINTS = re.compile(r' d="([^"]+)"')

POINT_MAX = 0xD700 # 55040, the maximum unicode value (with some buffer) before we need to exclude reserved values (slower/harder)
SVG_PATH_START = unichr(POINT_MAX + 1)
SVG_PATH_COMMANDS = ('M','Z','C','L','H','V','Q','S','A')
SVG_PATH_COMMANDS_ENCODED = tuple([POINT_MAX + ord(c) for c in SVG_PATH_COMMANDS])

def encode(svg):
    """Takes SVG and encodes to compact unicode string"""
    #print svg
    width, height = RE_WIDTH_HEIGHT.findall(svg)[0]
    output = [unichr(int(width)),unichr(int(height))]
    line_no = 3
    lines = svg.strip().split('\n')
    num_lines = len(lines)
    for line in lines[line_no:num_lines-1]: #between header and footer
        
        line_no += 1
        #print 'line %d:'%line_no, line

        output += SVG_PATH_START # mark line start

        # fill color
        color = RE_FILL_COLOR.findall(line)[0]
        short_color = [color[0],color[2],color[4]] #r g b
        print 'color:', short_color
        output += short_color
        
        # path
        points = RE_POINTS.findall(line)[0]
        for token in points.split(' '):
            
            if token.find('.') != -1: # bezier point
                point = int(round(float(token))) #round the float
                output += unichr(point)
            elif token in ('M','Z'): # start and end path
                pass
            elif token in SVG_PATH_COMMANDS: # commands
                output += unichr(POINT_MAX + ord(token)) # move to reserved range
            else: # joins
                output += token

    print 'CHARACTERS:', len(output)
    return u''.join(output)

def decode(data):
    """Takes compact unicode string and encodes to SVG"""
    # header
    width = ord(data[0])
    height = ord(data[1])
    print "width: %d height: %d" % (width,height)
    output = [HEADER % (width,height,width,height)]
    # paths
    for path in data.split(SVG_PATH_START)[1:]:
        #fill color
        color = (path[0]*2) + (path[1]*2) + (path[2]*2)
        print 'color:', color
        
        #points
        points = []
        for point in path[3:]:
            if ord(point) > POINT_MAX: # commands
                points.append(unichr(ord(point) - POINT_MAX))
            else: # points
                points.append('%d.00'%ord(point))

        output.append(PATH % (color, ' '.join(points)))
        
    output.append(FOOTER)
    return u'\n'.join(output)


if __name__ == '__main__':
    
    #encode to unicode
    data_file = file(r"so-logo.svg")
    svg_data = data_file.read()
    encoded_data = encode(svg_data)
    print repr(encoded_data)
    
    #decode to svg
    new_svg_data = decode(encoded_data)
    print new_svg_data