def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(
ur'^http://\w+\.ganji\.com(|/|/fang1/?|/fang1/\d+x\.htm|/fang1/tuiguang[-]\d+.htm|/fang1/f\d+/?)$',
re.U | re.I)
self.arch_pattern = re.compile(
ur'[房户](\s| )*型[^：:]*[：:]([^\d]*)(\d[^<>\s]+)[\s<]',
re.U | re.I)
def run(T):
t = 0.
bot = Robot()
tau = 0.0005
while t < T:
print bot.q
bot.log.append(bot.q)
bot.next_pos(tau)
t += tau
speedup = 0.1
show('jump.html', bot.log, tau/speedup)
def is_valid_url(self, url):
#print url
ans = Robot.is_valid_url(self, url) and self.url_pattern.match(
url) != None and not url.startswith(
'http://www.ganji.com/fang1') and not url.startswith(
'http://help.ganji.com') and not url.startswith(
'http://club.ganji.com')
#print ans
return ans
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(
ur'^http://[\w\d\.]+\.soufun\.com(|/|/house/?|/chuzu/[\d_]+\.htm)$',
re.U | re.I)
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(ur'^http://\w+\.ganji\.com(|/|/fang1/?|/fang1/\d+x\.htm|/fang1/tuiguang[-]\d+.htm|/fang1/f\d+/?)$', re.U | re.I)
self.arch_pattern = re.compile(ur'[房户](\s| )*型[^：:]*[：:]([^\d]*)(\d[^<>\s]+)[\s<]', re.U | re.I)
def is_valid_url(self, url):
#print url
ans = Robot.is_valid_url(self, url) and self.url_pattern.match(url) != None and not url.startswith('http://www.ganji.com/fang1') and not url.startswith('http://help.ganji.com') and not url.startswith('http://club.ganji.com')
#print ans
return ans
def get_address(self, page):
ans = Robot.get_address(self, page)
if ans == None:
ans = Robot.get_district(self, page)
return ans
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(
ur'^http://\w+.anjuke.com/(rental/(p\d+)?|prop/rent/\d+)/?$')
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.arch_pattern = re.compile(ur'[房户](\s| )*型[^：:]*[：:]\s*(<b>)(\d[^<\s]+)[<\s]', re.U | re.I)
self.url_pattern = re.compile(ur'^http://\w+\.esf\.focus\.cn(|/|/zufang/?|/zufang/\d+\.html|/zufang/zj0m0h0jz0zx0zlfs0bq0fbr0p\d+px0t0jd0bx0pic0q0b0xl0zd0xq0st1s0tx1/?)$', re.U | re.I)
def is_valid_url(self, url):
return Robot.is_valid_url(self, url) and self.url_pattern.match(url) != None
def get_address(self, page):
ans = Robot.get_address(self, page)
if ans == None:
ans = Robot.get_district(self, page)
return ans
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(ur'^http://\w+.anjuke.com/(rental/(p\d+)?|prop/rent/\d+)/?$')
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(ur'^(http://\w+\.zufang\.sina\.com\.cn(|/|/detail/\d+/?|/house/(n\d+)?/?)|http://\w+\.esf\.sina\.com\.cn/?)$', re.U | re.I)
def is_valid_url(self, url):
return Robot.is_valid_url(self, url) and self.url_pattern.match(url) != None
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(
ur'http://\w+\.haozu\.com/(listing/(p%d+)?|rental/broker/n/\d+)?/?$'
)
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(ur'http://\w+\.58\.com/zufang/(pn\d+/?|\d+x.shtml)?$', re.U | re.I)
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(ur'http://\w+\.haozu\.com/(listing/(p%d+)?|rental/broker/n/\d+)?/?$')
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(ur'^http://[\w\d\.]+\.soufun\.com(|/|/house/?|/chuzu/[\d_]+\.htm)$', re.U | re.I)
def __init__(self, root, charset):
Robot.__init__(self, root, charset)
self.url_pattern = re.compile(
ur'^(http://\w+\.zufang\.sina\.com\.cn(|/|/detail/\d+/?|/house/(n\d+)?/?)|http://\w+\.esf\.sina\.com\.cn/?)$',
re.U | re.I)
