#!/usr/bin/python
from config import directory

from glob import glob
import os, re, time, sgmllib

spamwords =  [
  'adipex',
  'ambien',
  'casino',
  'cialis',
  'clenbuterol',
  'dianabol',
  'hydrocodone',
  'lesbian',
  'levitra',
  'phentermine',
  'propecia',
  'tramadol', 
  'ugg',
  'valium',
  'viagra',
  'vicodin',
  'xanax',
  'xenical',
]

class link(sgmllib.SGMLParser):
  def __init__(self, body):
    sgmllib.SGMLParser.__init__(self)
    self.title=None
    self.className=None
    self.href=None
    self.feed(body)
  def start_a(self, attrs):
    attrs = dict(attrs)
    if attrs.has_key('title'): 
      self.title=attrs['title']
      if attrs.has_key('href'): self.href=attrs['href']
      if attrs.has_key('class'): self.className=attrs['class']

def registered(baseurl):
  from ConfigParser import ConfigParser
  for reg in glob(directory.registry+'*'):
    parser = ConfigParser()
    parser.read(reg)
    if parser.get('owner','id') == baseurl: return True

def warn(entry, baseip, baseurl, basetext):
  warnings=[]
  rank = spamrank(baseip, baseurl, basetext)
  if rank > 2: 
    warnings.append('block')
  elif rank > 1:
    rate = spamrank(baseip, baseurl, basetext, rateonly=1)
    if rate > 0:
      warnings.append('rate')
    if rank > rate*2:
      warnings.append('spam')
  elif rank > 0: # or baseurl.find('intertwingly') >= 0:
    warnings.append('captcha')
  try:
    if (time.time()-os.stat(directory.data+entry+'.txt').st_mtime) > 30*86400:
      warnings.append('age')
  except:
    pass
  return warnings

def captcha():
  import md5, os, random, re
  dict='/usr/share/dict/words'

  # remove any expired captchas
  expired = time.time()-3600.0
  for name in glob(os.path.join(directory.captcha,'*')):
    try:
      if os.stat(name).st_mtime < expired:
        os.remove(name)
    except:
      pass

  # create a new captcha
  w6=[word for word in open(dict).read().split() if len(word)==6]
  word ='-'
  while not re.match("^[a-z]*$",word):
    word=w6[random.randrange(0,len(w6))]
  cmd='convert -size 160x45 xc:transparent -pointsize 36 -fill darkred ' + \
    '''-draw "text 20,30 '%s'" %s%s.png'''
  hash = md5.new(word).hexdigest().strip()
  os.system(cmd % (word,directory.captcha,hash))

  return (word,hash)

def captcha_valid(word):
  import os,md5
  hash = md5.new(word).hexdigest().strip()+'.png'
  if os.path.exists(directory.captcha+hash):
    try:
      os.unlink(directory.captcha+hash)
    except:
      pass
    return True

def spamrank(baseip, baseurl, basetext, rateonly=0):
  credits, debits = spamtally(baseip, baseurl, basetext, rateonly)
  return max(len(debits) - len(credits), 0)

def spamtally(baseip, baseurl, basetext, rateonly=0):
  # if baseurl and baseurl.find('intertwingly.net')>=0: return [[], []]
  # if baseurl and baseurl.find('diveintomark')>=0: return 0
  # if baseurl and baseurl.find('redmonk.com/sogrady')>=0: return 0
  # if baseurl and baseurl.find('torrez.us')>=0: return 0
  # if baseurl and baseurl.find('2pauls.com')>=0: return 99
  # if baseurl and baseurl.find('subasta.pl')>=0: return 99
  # if baseurl and baseurl.find('einemillioneurohomepage.de')>=0: return 99

  # overall parameters
  now=time.mktime(time.gmtime())
  today=now-86400.0
  three_days=now-3*86400.0
  fifteen_minutes=now-300.0
  recent=5

  debits=[]
  credits=[]
  cmt_re = re.compile('\d+-(\d+)\.cmt$')

  if baseurl and baseurl.find('uggprovide.com')>=0:
    debits += 3*['spamword']

  # gather up recent comments
  comments=[]
  for file in glob(directory.data+"*.cmt"):
    try:
      mtime=int((cmt_re.findall(file) or [os.stat(file).st_mtime])[0])
      if mtime>=today:
        comments += [[mtime,file]]
        if mtime>=fifteen_minutes: recent+=1
    except:
      pass # removed as spam?
  # if recent>20: debits += 10*['recent']
  if len(comments)>100: debits += 100*['comments']
  comments.sort()
  comments=comments[-recent:]

  # add in up 72 hours of spam
  for file in glob(directory.spam+"*.cmt"):
    mtime=os.stat(file).st_mtime
    if mtime>=three_days:
      comments += [[mtime,file]]
  
  # look for similarities
  for file in [file for (mtime,file) in comments]:
    data=open(file).read().decode('utf-8')
    if baseurl and link(data).href==baseurl: 
      # multiple posts with the same URL
      debits.append('same-url')
      if link(data).className == 'openid': baseip='openid'
    elif baseip and link(data).title==baseip:
      # multiple posts from the same ip address
      debits.append('same-ip')
    elif '\n'.join(data.split('\n')[1:-1]) == basetext:
      # multiple posts of the same content
      import math # give a time based boost to this count
      debits.append('same-content')
      debits += int(math.sqrt((time.time()-mtime)/240))*['content-rate']

  if rateonly: return credits, debits

  try:
    if type(basetext) == str: basetext = basetext.decode('utf-8')
  except:
    debits.append('encoding')
    basetext = basetext.decode('iso-8859-1')

  if basetext.find(' style="position: absolute; left: -')>=0:
    # lame attempt to hide
    debits.append('position')
  if basetext.find('" target="_blank&gt;"')>=0:
    # lame popup window
    debits.append('popup')
    if len(basetext.split('" target="_blank&gt;"'))>7:
      debits.append('multi-popup')
  if basetext and baseurl and basetext.find(baseurl)>=0:
    # trying too hard at self-promotion
    debits.append('self-promotion')
  if basetext and max([basetext.find(word) for word in spamwords])>=0:
    # druggies and gamblers
    debits.append('spamword')
  if basetext.find('[/URL]')>=0:
    # lame UBB code
    debits.append('UBB')
    if len(basetext.split('[/URL]'))>7:
      debits.append('multi-UBB')
  if basetext.find('[/url]')>=0:
    # lame UBB code
    debits.append('ubb')
    if len(basetext.split('[/url]'))>7:
      debits.append('multi-ubb')
  if basetext.find('[link]')>=0:
    # naked urls dropped into comments
    debits.append('link')
    if len(basetext.split('[link]'))>7:
      debits.append('multi-link')
  if basetext.find(' HREF=')>=0:
    # lame html
    debits.append('HREF')
    if len(basetext.split(' HREF='))>7:
      debits.append('multi-HREF')
  if basetext.find('&lt;a href=http')>=0:
    # more lame html
    debits.append('unquoted-href')
    if len(basetext.split('&lt;a href=http'))>7:
      debits.append('multi-unquoted-href')

  if len(debits) > 0 and baseip:
    import community
    if baseip == 'openid':
      # openid bonus
      credits.append('openid')
      if len(debits) > len(credits) and registered(baseurl):
        credits.append('registered')
    elif not community.find(baseip, baseurl):
      # stranger
      debits.append('stranger')

  return credits, debits

def spammer(text):
  basetext='\n'.join(text.split('\n')[:-1])
  sig = link(text)
  baseip=sig.title
  baseurl=sig.href
  if sig.className == 'openid': baseip=sig.className
  if spamrank(baseip, baseurl, basetext) > 2:
    tokens = spamtally(baseip, baseurl, basetext)
    import nonce
    nonce.generate(time.time()+86400, repr(['throttle traceback',tokens,text]))
    return True

if __name__ == '__main__':
  # directory.data = '/home/rubys/web/intertwingly.net/spam/'
  import sys
  print spammer('<a title="%s"/>' % sys.argv[1])