def tycFromPage():
idbloom = getBloom()
homeUrl = 'http://www.tianyancha.com/'
driver = getWebDriver()
driver.get(homeUrl)
def getQichachaInvestDigests():
idbloom = getBloom()
conn, csor = getConnCsor()
csor.execute('select uid from com_invest')
ids = csor.fetchall()
[idbloom.add(mid[0]) for mid in ids]
# if ids[0][0] in idbloom:
print 'load exists ids ok'
return idbloom
def getQichachaDigests():
idbloom = loadBloomFromFile('qichachaUIDs')
if idbloom:
print 'load bloom from file succ, no need load from db'
# return idbloom
else:
print 'no dump bloom file, load from db'
idbloom = getBloom(2000 * 10000)
# idbloom = getBloom()
conn, csor = getConnCsor()
csor.execute('select id from com_base_copy')
# csor.execute('select id from com_base_copy limit 10')
ids = csor.fetchall()
[idbloom.add(mid[0]) for mid in ids]
# if ids[0][0] in idbloom:
print 'load exists ids ok, generate dump bloom file'
dumpBloomToFile(idbloom, fileName='qichachaUIDs')
return idbloom
