Mercator to rectangular conversion

#! python2
# http://pastebin.com/mZ35nEKz
# http://ge.tt/4fWcPA91/v/0

import png, sys, os
from math import pi, log, tan

address = eval(os.path.normpath(str(sys.argv)))[-1]
#print 'address', address
datain = png.Reader(address).read()
#f = datain = png.Reader('/media/ToshikHD/Dropbox/Programming/Python/Maps/RasterMecatorToRectangle/TestMercator.out.png').read()
bitmap = list(datain[2])
planes = datain[3]['planes']
size = datain[3]['size']
deg = pi/180

def peek (x, y):
    global planes, bitmap
    #if grayscale == True: pass
    return (bitmap[y][x*planes],bitmap[y][x*planes+1],bitmap[y][x*planes+2])

sizeout = (size[0],size[0]//2)
bitmapout = range(sizeout[1])
for i in range(sizeout[1]):
    bitmapout[i]=()
    lat = (1-(i+0.5)/sizeout[1]*2)*pi/2
    latmer = log (tan (pi/4 + lat/2))
    imer = size[1]//2 - latmer/(pi/2) * sizeout[1]/2 - 0.5
    if imer < 0: imer = 0
    if imer > size[1]-2: imer = size[1]-2
    #print 'i', i, 'imer', imer, 'lat', lat/deg, 'latmer', latmer/deg
    w0 = abs (int (imer+1) - imer)
    w1 = abs (int (imer) - imer)
    for j in range(sizeout[0]):
        m0 = peek (j, int (imer))
        m1 = peek (j, int (imer+1))
        m = (min(int(m0[0]*w0+m1[0]*w1+0.5),255), min(int(m0[1]*w0+m1[1]*w1+0.5),255), min(int(m0[2]*w0+m1[2]*w1+0.5),255))
        bitmapout[i] += m

f = open('.rect.png'.join(address.split('.png')), 'wb')
#f = open('/media/ToshikHD/Dropbox/Programming/Python/Maps/RasterMecatorToRectangle/TestMercator.out.png', 'wb')

w = png.Writer(size[0],size[0]//2)
w.write(f, bitmapout)
f.close()

description = """
Mercator to equirectangular projection projection converter

Hi. I have noticed a common mistake among the worldbuilders here. Often they choose the equirectangular projection for the base map of their conworld. (Equirectangular projection is a rectangle 360°×180°, with X and Y coordinates directly corresponding to the degrees of longitude and latitude.) This projection is quite good to work with, and it can be feed into G.Projector to produce many other neat projections, but the trouble is it isn't the best choice when you're *designing* a world, because nearly nobody can produce plausible distorsions of it (the horizontal stretching in the polar zones).

Here is one of the examples: http://i.imgur.com/miWaWpU.png At first glance it looks okay. But let's see how the north pole looks like. That' how: http://i.imgur.com/m0PZem2.png It has an unnatural stretching of land shapes towards the pole.

In fact, for shaping the continents you should rather use one of the conformal projections, with Mercator projection being most well known of them. They are good because they preserve the shapes (at the expense of conformity of scale), so you don't have to worry about further distorsions, and a little round island on your map will remain a round island on the globe (probably not as little though).

So, I decided to make a program for converting Mercator maps into equirectangular projection. I'm not a good programmer, but it seems to working. It's written in Python 2, and here is its code: http://pastebin.com/mZ35nEKz , and here you can download a compiled version for Windows: http://ge.tt/4fWcPA91/v/0

It can be used to convert full-color (not grayscale) PNG files. To run the EXE program from command prompt you can type something like "Mercator2Rectangle.exe MyMap.png" or simply drag-and-drop the map file to the program icon. If everything was done correctly, there will appear yet one file, named something like "MyMap.rect.png".

The program presumes that your Mercator map is symmetrical, i.e. the the equator runs through its center. The rectangle for Mercator map normally should be higher than for the equirectangular map — I suggest to use a square.

So, ultimately this program is supposed to convert something like this: http://i.imgur.com/fEbRnjG.png
Into something like this: http://i.imgur.com/M4zothB.png
"""