source: trollforge/dsp-trollforge/MovableType/Blogpost.py @ 212

#!/usr/bin/env python
#
# blogpost.py:
#   Post a comment on any Drupal/MovableType blog entry. Also defeats stupid
#   SCode "security" check that is so annoying when using a text browser.
#
# Usage:
#   blogpost.py "story_url" "nick" "valid@email.com" "comment"
#
# Copyright: (c) 2004 Sam Hocevar <sam@zoy.org>
#   This program is free software; you can redistribute it and/or
#   modify it under the terms of the Do What The Fuck You Want To
#   Public License as published by Banlu Kemiyatorn. See
#   http://sam.zoy.org/projects/COPYING.WTFPL for more details.
#
# Changelog:
#   2004/12/26: first release
#               tested on http://www.parablog.com/mt/mt-comments.cgi?entry_id=876
#   2004/12/26: defeat captchas with uneven bottom lines
#               tested on http://ramk.net/archives/000118.html
#   2004/12/27: work on Drupal instead of only MovableType
#               tested on http://james.seng.cc/archives/000145.html
#   2004/12/27: do not send empty POST data
#               fixes http://underscorebleach.net/content/jotsheet/2004/10/combat_mt_comment_spam_captcha
#

import gd, urllib, urllib2, sys, re, time

# Debug stuff?
debug = 1

# Behave nicely, like MSIE
user_agent = "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322)"
referer = ""
cookie = ""

# Guess a glyph from its bounding box in the picture. This function is very,
# very naive. But since mt-scode is very, very naive as well, it's more than
# enough.
def guess_glyph(im, xmin, xmax):
    (w,h) = im.size()
    # Find ymin/ymax bounding box
    ymin = -1
    ymax = -1
    for y in range(h):
        found = 0
        for x in range(xmin, xmax):
            p = im.getPixel((x,y))
            if p == 1:
                found = 1
                break
        if found:
            if ymin == -1:
                ymin = y
            else:
                ymax = y + 1
    # Compute simple glyph checksum
    values = [162, 131, 150, 139, 155, 159, 181, 90, 180, 170]
    count = 0
    for x in range(xmin, xmax):
        for y in range(ymin, ymax):
            count += im.getPixel((x,y)) * (y - ymin)
    if count not in values:
        return "?"
    return values.index(count)

# Remove background stuff from an scode picture
def remove_background(im):
    (w,h) = im.size()
    # Stats on first 3 lines to determine background colours
    stats = [0, 0, 0, 0]
    fg = -1
    bg = -1
    for y in range(3):
        for x in range(w):
            p = im.getPixel((x,y))
            if p < 4:
                stats[p] += 1
    for y in range(h):
        for x in range(w):
            p = im.getPixel((x,y))
            if not stats[p]:
                fg = p
    # Unify background colours
    for y in range(h):
        for x in range(w):
            p = im.getPixel((x,y))
            if p == fg:
                im.setPixel((x,y),1)
            else:
                im.setPixel((x,y),0)

# Decode an scode picture from its URL
def sdecode(url):
    if debug:
        print "getting", url
    r = urllib2.Request(url)
    r.add_header('User-Agent', user_agent)
    r.add_header('Referer', referer)
    if cookie != "":
        r.add_header('Cookie', cookie)
    f = urllib2.build_opener().open(r)
    im = gd.image(f, "png")
    (w,h) = im.size()
    f.close()
    chars = [' ', '.', ':', '@']
    for y in range(h):
        line = ""
        for x in range(w):
            p = im.getPixel((x,y))
            line += chars[p]
        if debug:
            print line
    remove_background(im)
    # Read character cells
    result = ""
    incell = 0
    for x in range(w):
        found = 0
        for y in range(h):
            p = im.getPixel((x,y))
            if p == 1:
                found = 1
                break
        if found and not incell:
            incell = 1
            xmin = x
        if not found and incell:
            incell = 0
            result += str(guess_glyph(im, xmin, x))
    return result

def mt_post(url, nick, email, comment, data):
    global referer, cookie
    re_base = re.compile('.*<base.*href="([^"]*)".*\n')
    re_startform = re.compile('.*action="([^"]*comment[^"]*)".*\n')
    re_endform = re.compile('.*</form>.*\n')
    re_captcha = re.compile('.*src="([^"]*(scode|captcha)[^"]*)".*\n')
    re_captchatext = re.compile('.*name="([^"]*(scode|captcha)[^"]*)".*\n')
    re_hidden = re.compile('.*type="hidden".*name="([^"]*)".*value="([^"]*)".*\n')
    re_submit = re.compile('.*type="submit".*name="([^"]*)".*value="([^"]*)".*\n')
    if debug:
        print "getting", url
    if data:
        r = urllib2.Request(url, urllib.urlencode(data))
    else:
        r = urllib2.Request(url)
    r.add_header('User-Agent', user_agent)
    if referer != url:
        r.add_header('Referer', referer)
    if cookie != "":
        r.add_header('Cookie', cookie)
    referer = url
    f = urllib2.build_opener().open(r)
    info = f.info()
    if info.has_key('set-cookie'):
        cookie = info['set-cookie'].split(';')[0]
        if debug:
            print "got cookie!"
    else:
        if debug:
            print "no cookie!"
    inform = 0
    canpost = 0
    base = ""
    scode = ""
    captchatext = "scode"
    while True:
        l = f.readline()
        if not l:
            break
        # Detect base URL
        m = re_base.match(l)
        if m:
            base = m.group(1)
        # Detect start of form
        m = re_startform.match(l)
        if m:
            target = m.group(1)
            inform = 1
        if not inform:
            continue
        # Detect end of form
        m = re_endform.match(l)
        if m:
            inform = 0
            continue
        # Detect a captcha image
        m = re_captcha.match(l)
        if m:
            captcha = base + m.group(1)
            scode = sdecode(captcha)
            if debug:
                print "found captcha", captcha, "decoded to", scode
        # Detect a captcha entry
        m = re_captchatext.match(l)
        if m:
            captchatext = m.group(1)
        # Detect a hidden input
        m = re_hidden.match(l)
        if m:
            if debug:
                print 'hidden item "' + m.group(1) + '" value "' + m.group(2) + '"'
            data[m.group(1)] = m.group(2)
        # Detect a submit button
        m = re_submit.match(l)
        if m:
            if debug:
                print 'submit item "' + m.group(1) + '" value "' + m.group(2) + '"'
            if re.match(".*[pP]ost.*", m.group(1) + " " + m.group(2)):
                canpost = 1
            if re.match(".*[pP]review.*", m.group(1) + " " + m.group(2)):
                preview = 1
                if not canpost:
                    data[m.group(1)] = m.group(2)
                elif debug:
                    print "ignoring because we have a post button"
            else:
                data[m.group(1)] = m.group(2)
    f.close()
    if scode != "":
        data[captchatext] = scode
    data['author'] = nick
    data['edit[name]'] = nick
    data['email'] = email
    data['edit[mail]'] = email
    data['homepage'] = ""
    data['edit[homepage]'] = ""
    data['text'] = comment
    data['edit[comment]'] = comment
    url = base + target
    if not canpost:
        mt_post(url, nick, email, comment, data)
    else:
        if debug:
            print "getting", url
        r = urllib2.Request(url, urllib.urlencode(data))
        r.add_header('User-Agent', user_agent)
        r.add_header('Referer', referer)
        if cookie != "":
            r.add_header('Cookie', cookie)
        f = urllib2.build_opener().open(r)
        while True:
            l = f.readline()
            if not l:
                break
            #sys.stdout.write(l)
        f.close()

mt_post(sys.argv[1], sys.argv[2], sys.argv[3], sys.argv[4], {})