camp-hack.py

from __future__ import print_function
from lxml.html.soupparser import fromstring
import requests
from datetime import datetime, timedelta
from collections import defaultdict


start_date = datetime(2012, 8, 6)
end_date = start_date + timedelta(days=60)

days = """Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
Sunday"""

day_labels = {}
for i, d in enumerate(days.split('\n')):
    day_labels[i] = d

url = 'http://www.reserveamerica.com/campsiteCalendar.do?page=calendar&contractCode=CA&parkId=120030&calarvdate={caldate}&sitepage=true&startIdx={page_index}&sitefilter=STANDARD'
s = requests.session()
# s.config['keep_alive'] = False
calendar_start = start_date
calendar_page = 1
availability = defaultdict(list)
while calendar_start < end_date:
    days = []
    for i in range(14):
        days.append(calendar_start + timedelta(days=i))
    calendar_start_str = calendar_start.strftime("%m/%d/%Y")
    page_index = 0
    while page_index < 125:
        page_url = url.format(caldate=calendar_start_str,
                page_index=page_index)
        # print(page_url)
        page_request = s.get(page_url, prefetch=True)
        page = page_request.content
        tree = fromstring(page)
        sites = []
        availables = []
        availables = tree.cssselect('td.status.a')
        # print(len(availables))

        for item in availables:
            row = item.getparent()
            site_id = row[0].text_content().split('\n')[0]
            if site_id not in sites:
                sites.append(site_id)
                for i, cell in enumerate(row[2:]):
                    if cell not in availables:
                        continue
                    cell_date = days[i]
                    if cell_date.weekday() in [4, 5]:
                        # print("booya -- site: {} on {}".format(site_id, cell_date))
                        availability[cell_date].append(site_id)
        page_index += 25
    calendar_start = calendar_start + timedelta(days=14)
for d in sorted(availability.keys()):
    print(day_labels[d.weekday()], d, availability[d])

output.txt

Friday 2012-08-24 00:00:00 ['119']
Friday 2012-09-07 00:00:00 ['113', '040', '066']
Friday 2012-09-14 00:00:00 ['017', '055', '099', '038', '080']
Friday 2012-09-21 00:00:00 ['127', '112', '020', 'GANZ', '114']
Friday 2012-09-28 00:00:00 ['079', '008', '009', '020', '026', '040', '114', '039']
Saturday 2012-09-29 00:00:00 ['009', '020', '026']
Friday 2012-10-05 00:00:00 ['024', '131', '031', '035', '036', '037', '063', '066', '111', '009', '026', '039', '040', '080']
Saturday 2012-10-06 00:00:00 ['131', '008', '009', '026', '040', '080']
Friday 2012-10-12 00:00:00 ['125', '127', '128', '129', '130', '131', '064', '065', '073', '074', '094', '102', '104', '008', '080', '126', '099']
Saturday 2012-10-13 00:00:00 ['125', '127', '128', '129', '130', '131', '080', '126']

requirements.txt

BeautifulSoup==3.2.1
Pygments==1.5
bpython==0.11
cssselect==0.7.1
distribute==0.6.27
flake8==1.4
lxml==2.3.5
pudb==2012.2.1
pyquery==1.2.1
python-dateutil==2.1
-e git+https://github.com/kennethreitz/requests.git@fc6ddfd78a26afe926c0b5430807d2bbe6d7c874#egg=requests-dev
six==1.1.0
urwid==1.0.2
wsgiref==0.1.2