Парсинг HTML через bash

X=`grep -m3 -B1 37 statlist | grep -v 37 | tail -n 1 | grep -m1 -oE map.php[0-9a-zA-Z\-_\.?&=]*" | cut -d "=" -f2 | cut -d "&" -f1`
Y=`grep -m3 -B1 37 statlist | grep -v 37 | tail -n 1 | grep -m1 -oE "map.php[0-9a-zA-Z\-_\.?&=]*" | cut -d "&" -f2 | cut -d "=" -f2`
OBJECT_ID=`grep -m3 -B1 37 statlist | grep -v 37 | tail -n 1 | grep -m1 -oE "object.php[0-9a-zA-Z\-_\.?&=]*" | cut -d "=" -f2 | uniq`

 cat evil_spider.py 
#!/usr/bin/env python
#-*- coding: utf-8 -*-

from HTMLParser import HTMLParser
import sys
import re

class tobacco_parser(HTMLParser):

    def __init__(self,what='Купить',dollar=' \$37 '):
        HTMLParser.__init__(self)
        self.result = []
        pat1 = r'%s' % what
        pat2 = r'Донской особенный'
        pat3 = r'%s' % dollar
        self.what_match = re.compile(pat1)
        self.dollar_match = re.compile(pat3)
        self.don_match = re.compile(pat2)
        self.pars_table = 0
        self.pars_dollar = 0
        self.price_ok = 0
        self.buf_row = []

    def handle_starttag(self, tag, attrs):
        self.check_for_table_tag(tag,0)
        self.search_for_table(tag,0)
        self.row_analys_tag(tag,attrs,0)
        #print self.pars_table
        #print self.pars_dollar

    def handle_endtag(self, tag):
        self.check_for_table_tag(tag,1)
        self.search_for_table(tag,1)
        self.row_analys_tag(tag,None,1)
        #print self.pars_table
        #print self.pars_dollar

    def handle_data(self, data):
        self.check_for_table(data)
        self.row_analys_data(data)
        #print self.pars_table
        #print self.pars_dollar

    def check_for_table(self,data):
        if(self.pars_table >= 4): return

        if(self.pars_table < 1 and self.what_match.search(data)):
            self.pars_table = 1
        if(self.pars_table == 2 and self.don_match.search(data)):
            self.pars_table = 3

    def check_for_table_tag(self,tag,end_tag=0):
        if(self.pars_table >= 4): return

        if(self.pars_table == 1 and end_tag == 0 and tag == 'b'):
            self.pars_table = 2
        elif(self.pars_table == 3 and end_tag == 1 and tag == 'b'):
            self.pars_table = 4
        else:
            self.pars_table = 0

    def search_for_table(self,tag,end_tag=0):
        if(self.pars_table < 4 or self.pars_table > 5): return

        if(self.pars_table == 4 and end_tag == 0 and tag == 'table'):
            self.pars_table = 5
        elif(self.pars_table == 5 and end_tag == 1 and tag == 'table'):
            self.pars_table = 6

    def row_analys_tag(self,tag,attrs,end_tag=0):
        if(self.pars_table != 5): return
        if(self.pars_dollar < 1 and end_tag == 0 and tag == 'tr'):
            self.pars_dollar = 1
            
            self.buf_row = []
        elif(self.pars_dollar > 0 and end_tag == 1 and tag == 'tr'):
            self.pars_dollar = 0
            if(self.price_ok):
                self.result.append(self.buf_row[:])
                #print self.buf_row
                self.price_ok = 0
            
        elif(self.pars_dollar == 1 and end_tag == 0 and tag == 'td'):
            self.pars_dollar = 2
        elif(self.pars_dollar > 1 and end_tag == 1 and tag == 'td'):
            self.pars_dollar = 1
             
        elif(self.pars_dollar == 2 and end_tag == 0 and tag == 'a'):
            #self.buf_row.append(attrs[:])
            for i in attrs:
                if(i[0] == 'href'):
                    self.buf_row.append(i[1])

    def row_analys_data(self,data):
        if(self.pars_table != 5): return
        if(self.pars_dollar == 2):
            if(self.dollar_match.match(data)):
                self.price_ok = 1
                #print data


        

f = open(sys.argv[1])
page = f.read()
f.close()

tobacco = tobacco_parser('Купить',r' \$35 ')
tobacco.feed(page)
print "Купить по 35"
print tobacco.result
print '----------------------'
tobacco.close()

tobacco = tobacco_parser('Продать',r' \$37 ')
tobacco.feed(page)
print "Продать по 37"
print tobacco.result
print '----------------------'
tobacco.close()

$ ./evil_spider.py sec_test.html 
Купить по 35
[['/map.php?sx=47&sy=52', '/object.php?id=79872', '/object.php?id=79872']]
----------------------
Продать по 37
[[], ['/map.php?sx=47&sy=49', '/object.php?id=80046', '/object.php?id=80046'], ['/map.php?sx=52&sy=52', '/object.php?id=73941', '/object.php?id=73941'], ['/map.php?sx=50&sy=51', '/object.php?id=66317', '/object.php?id=66317'], ['/map.php?sx=51&sy=48', '/object.php?id=1256', '/object.php?id=1256']]
----------------------