def run():
sites = get_all_sites()
try:
sites.remove("newyork")
sites.remove("chicago")
except:
pass
cl = CLWatcher()
print "Sites:", sites
for site in sites:
print "Processing for", site
cl.setSite(site)
cl.getHousingPosts()
cl.getEvents()
cl.save()
import json, time
from craigslist import CraigslistHousing, CraigslistEvents
from craigslist import get_all_sites
print "sites:", get_all_sites()
#DEFAULT_OUTPUT_PATH = "../Viewer/data/craigslist_data.json"
DEFAULT_OUTPUT_PATH = "craigslist_data.json"
class CLWatcher:
def __init__(self, site="sfbay", autoSave=True, limit=500):
self.recs = []
self.site = site
self.limit = limit
self.autoSave = autoSave
def setSite(self, site):
self.site = site
def save(self, path=None):
if path == None:
path = DEFAULT_OUTPUT_PATH
print "Saving to", path
t0 = time.time()
obj = {'type': 'craigslist',
't': time.time(),
'records': self.recs
}
json.dump(obj, file(path, 'w'), indent=3, sort_keys=True)
