#!/usr/bin/env python2.4
import sys
import feedparser
reload(sys)
sys.setdefaultencoding('utf-8')

encodings = [ 'windows-1251',    'cp1251',     'koi8-r',    'koi8-u',    ]
encunicodes = [ 'utf-8', 'utf8',  'utf-16',  'utf16',   'ucs-2',    ]
badlangs = [ 'zh' ]

encbonusValue = 1000
langbonusValue = -1000
rules = {}

Debug = False
def bonus(reason, value):
    ## XXX Hack for russian letters count
    length = 40 + reason.count('\xd0') + reason.count('\xd1')
    if Debug:
        print "  %-*s | %10d" % ( length, reason, value)
    return value
def encBonus(enc):
    if enc in encodings:
        return bonus('Good encoding (%s)' % enc, encbonusValue)
    elif enc in encunicodes:
        return bonus('Unicode encoding (%s), no bonuses' % enc, 0)
    else:
        return reason('Bad encoding (%s)' % enc, -encbonusValue)

def langBonus(lang):
    rb = 0
    for pattern in badlangs:
        if lang.count(pattern) > 0:
            rb += bonus('Bonus for language pattern "%s" (lang="%s")' % (pattern, lang), langbonusValue)
    return rb
        
def getRules():
    global rules
    for rule in [x.strip().split('>>',2) for x in open('rules.txt').readlines() if len(x) > 1]:
        if len(rule) != 2:
            continue
        (rule, weight) = (rule[1].strip(), int(rule[0]))
        rule = rule [ rule.index('"') + 1 : rule.rindex('"') ]
        rules[rule] = weight

def main(args):
    global Debug
    for url in args:
        if url == '-d':
            Debug = True
            continue
        if Debug:
            print "> Parsing %s" % url
        fl = feedparser.parse(url)
        points = encBonus(fl.encoding.lower())
        try:
            points += langBonus(fl.feed.language.lower())
        except:
            pass
        text  = '\n'.join( [ item.summary_detail.value for item in fl.entries if item.has_key('summary_detail')])
        text += '\n'.join( [ item.title_detail.value   for item in fl.entries if item.has_key('title_detail'  )])
        for key in rules.keys():
            cnt = text.count(key)
            increment = cnt * rules[key]
            if increment != 0 :
                points += increment
                points += bonus("for key %s (%d nums)" % (key, cnt), increment)
        if points < 0:
            print "Feed: %s not in Russian (weight: %d)" % (url, points)
        else:
            print "Feed: %s is good (weight: %d)" % (url, points)

if __name__ == '__main__':
    getRules()
    main(sys.argv[1:])