def __init__(self, key_word=None, limit=1000, proxy=None):
self.base_url = "http://www.baidu.com/s?wd=%40{query}&pn={page_no}"
self.engine_name = "baidu"
self.counter_step = 10 #item number,10 items each page
self.proxy = None # no proxy need for baidu
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
self.proxy)
def __init__(self, key_word=None, limit=None, proxy=None):
self.base_url = "http://cn.bing.com/search?q={query}&go=&count=50&FORM=QBHL&qs=n&first={page_no}"
self.engine_name = "bing"
self.counter_step = 50
self.proxy = None
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
self.proxy)
def __init__(self, key_word=None, limit=None, proxy=None):
self.base_url = "http://www.dogpile.com/search/web?q={query}&qsi={page_no}"
self.engine_name = "dogpile"
self.counter_step = 10
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
proxy)
self.proxy = None
def __init__(self, key_word=None, limit=None, proxy=None):
self.base_url = "http://www.yandex.com/search?text={query}&numdoc=50&lr={page_no}"
self.engine_name = "yandex"
self.counter_step = 50
self.proxy = None
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
self.proxy)
def __init__(self, key_word=None, limit=None, proxy=None):
self.base_url = "http://search.yahoo.com//search?p={query}&b={page_no}&pz=10"
self.engine_name = "yahoo"
self.counter_step = 10
self.proxy = None
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
self.proxy)
def __init__(self, key_word=None, limit=None, proxy=None):
self.base_url = "http://www.exalead.com/search/web/results/?q={query}&&elements_per_page=50&start_index=={page_no}"
self.engine_name = "exalead"
self.counter_step = 50
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
proxy)
self.proxy = None # no proxy need for baidu
def __init__(self, start, depth = 1, num_iters = 100):
search.__init__(self, start)
self.depth = depth
self.num_iters = num_iters
self.num_iters_taken = 0
def __init__(self, key_word=None, limit=None, proxy=None):
self.base_url = "http://www.ask.com/web?q={query}&pu=100&page={page_no}"
self.engine_name = "ask"
self.counter_step = 1 #page_no ,
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
proxy) #need proxy
def __init__(self, key_word=None, limit=None, proxy=None):
self.base_url = "https://google.com/search?q={query}&btnG=Search&hl=en-US&biw=&bih=&gbv=1&start={page_no}&filter=0"
self.engine_name = "google"
self.counter_step = 10
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
proxy)
def __init__(self, key_word=None, limit=None, proxy=None):
self.base_url = "http://www.google.com/search?num=100&q={query}&start={page_no}&hl=en&meta="
self.engine_name = "google"
self.counter_step = 50
search.__init__(self, self.base_url, self.engine_name, key_word, limit,
proxy)
def __init__(self, start, num_iters = 100):
search.__init__(self, start)
self.num_iters = num_iters
self.num_iters_taken = 0
