Simple Django one-off view that pulls down a listing of Craigslist ads from all Craigslist domains. Ain't exactly pretty, but it works. It's not fast by web standards, but it sure beats searching manually.

views.py

import urllib2,os,cookielib
from urlparse import urlparse
from BeautifulSoup import BeautifulSoup
from django.core.cache import cache
from django.http import HttpResponse
import multiprocessing
m=multiprocessing.Manager()
links=m.list()
hosts=m.Queue()

COOKIEFILEBASE = '/tmp/cookie.tmp.'


#Fetch a url.  Simple enough, maybe.  But we also check against a cache
#backend, if available.  And we honor cookies.  Cache the results for 12
#hours, so we don't smear anyone's servers unnecessarily, because we're
#not looking for any time-sensitive data.
def fetch_url(request,url):
    server_name=urlparse(url).netloc
    COOKIEFILE=COOKIEFILEBASE + server_name
    headers={
        'Host':server_name,
        'User-Agent':request.META['HTTP_USER_AGENT']
        }
    cache_key = url
    if cache.get(cache_key):
        return cache.get(cache_key)
    cj = cookielib.LWPCookieJar()
    urlopen = urllib2.urlopen
    Request = urllib2.Request
    if os.path.isfile(COOKIEFILE):
        cj.load(COOKIEFILE)
    opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cj))
    urllib2.install_opener(opener)
    req=Request(url,None,headers)
    cj.save(COOKIEFILE)
    response=urlopen(req).read()
    cache.set(cache_key,response,60*60*12)
    return response

#Called for each process created in the view function.
#Pulls a queued hostname, runs a standardized Craigslist search on that
#hostname, and adds an HTML anchor to the list of output in links.
#This particular Craigslist search is for 'django' with the 
#'telecommuting' flag set.
def worker_thread(request):
    while True:
        try:
            host=hosts.get_nowait()
        except:
            print "thread done"
            return
        url='http://%s/search/jjj?query=django&srchType=A&addOne=telecommuting'%(host,)
        html=fetch_url(request,url)
        try:
            soup=BeautifulSoup(html)
        except:
            print html
        for p in soup.findAll('blockquote')[-1].findAll('p'):
            links.append('<a href="%s">%s - %s</a>'%(p.a['href'],p.text,host))
        hosts.task_done()
        print host

#Actual view that is mapped to a url.
#Builds a list of sites based on the craigslist front page.
#Pushes them all into a queue, and starts off 10 processes working
#that queue
def find_all_postings(request):
    url='http://www.craigslist.org/about/sites'
    soup=BeautifulSoup(fetch_url(request,url))
    for li in soup.find(attrs={'class':'colmask'}).findAll('li'):
        hosts.put(urlparse(li.a['href']).netloc)
    for index in range(0,10):
        #If you'll notice, these two lines (other than some variables) are 
        #the only two we need to make this a multi-process app
        thread=multiprocessing.Process(target=worker_thread, args=(request,))
        thread.start()
    hosts.join()
    sorted_links = sorted(links, key=lambda x: BeautifulSoup(x).text)
    return HttpResponse('<br/>'.join(sorted_links))