#!/usr/bin/python
import os
import re
from urllib import urlopen
DEBUG = 1
# TO DO:
# Keep track of what has already been sent.
# INPUTS:
subscribers='dvlink@gmail.com'
location = 'wch'
search_term = 'buick enclave'
#search_term = 'phone'
#search_term = 'guitar'
#search_term = 'ipod'
#search_term = 'macbook'
href_base = "http://newyork.craigslist.org"
url='%s/search/sss/%s?query=%s' % (href_base, location, search_term)
email_filename = 'email_body.txt'
# GO TO TOWN:
#x = re.compile('.*')
print 'url:', url
page = urlopen(url).read()
ps = page.split('')
report = ""
fp = open(email_filename, 'w')
fp.write('\n\n')
for p in ps:
m = re.match(r'.*
(.*) - (.*)', p)
if m and len(m.groups()) == 3:
mon_day = m.group(1)
href = m.group(2)
desc = m.group(3)
fp.write('%s - %s
\n' % \
(mon_day, href_base, href, desc))
fp.write('
/n\n')
fp.close()
if DEBUG:
print 'search_term:', search_term;
print 'subscribers:', subscribers;
print 'email_filename:', email_filename
cmd = 'sendmail -s "CraigsList Scrape for %s" %s < %s' % \
(search_term, subscribers, email_filename)
print 'cmd:', cmd
os.system(cmd)