def getCrawlNoRssRequestLength():
try:
http = HttpRequest()
url = requst_norss_length_url
response = http.setUrl(url).setBody({}).encrypt([]).post()
res = json.loads(response)["data"]
if res == "null":
res = None
except Exception, e:
logging.info("-----%s-----" % e)
return None
def getCrawlNoRssRequestLength():
try:
http = HttpRequest()
url = requst_norss_length_url
response = http.setUrl(url).setBody({}).encrypt([]).post()
res = json.loads(response)['data']
if res == 'null':
res = None
except Exception, e:
logging.info("-----%s-----" % e)
return None
def syncLastMd5(params):
try:
http = HttpRequest()
url = sync_last_md5_url
response = http.setUrl(url).setBody(params).encrypt([]).post()
res = json.loads(response)['data']
if res == 'null':
res = None
except Exception, e:
print e
logging.info("-----%s-----" % e)
return None
def getCrawlRssRequest(params={}):
try:
http = HttpRequest()
url = request_rss_url
response = http.setUrl(url).setBody(params).encrypt([]).post()
res = json.loads(response)['data']
if res == 'null':
res = None
except Exception, e:
print e
logging.info("-----%s-----" % e)
return None
def getCrawlRssRequest(params={}):
try:
http = HttpRequest()
url = request_rss_url
response = http.setUrl(url).setBody(params).encrypt([]).post()
res = json.loads(response)["data"]
if res == "null":
res = None
except Exception, e:
print e
logging.info("-----%s-----" % e)
return None
def syncLastMd5(params):
try:
http = HttpRequest()
url = sync_last_md5_url
response = http.setUrl(url).setBody(params).encrypt([]).post()
res = json.loads(response)["data"]
if res == "null":
res = None
except Exception, e:
print e
logging.info("-----%s-----" % e)
return None
def requstDistinct(hashCode):
try:
http = HttpRequest()
url = requst_distinct_url
hashCode = ",".join(hashCode)
body = {'field': hashCode}
encryptFields = []
response = http.setUrl(url).setBody(body).encrypt(encryptFields).post()
res = json.loads(response)['data']
if not res:
return []
return res
except Exception, e:
res = []
logging.info('-----------%s-------' % e)
return res
def requstDistinct(hashCode):
try:
http = HttpRequest()
url = requst_distinct_url
hashCode = ",".join(hashCode)
body = {"field": hashCode}
encryptFields = []
response = http.setUrl(url).setBody(body).encrypt(encryptFields).post()
res = json.loads(response)["data"]
if not res:
return []
return res
except Exception, e:
res = []
logging.info("-----------%s-------" % e)
return res
def retrieve_data(key, code):
try:
logging.info("get daily data of %s" % (key+code))
s = re.compile(r",(\d+)").sub(r"\1", HttpRequest(DAILY_PRICE_URL % (key+code)).get())
s = re.compile(r"(\d+)\_(\d+)\_(\d+)").sub(r"\1\2\3", s)
self.data[code] = json.loads(s)["record"]
except:
pass
def syncCrawlInfos(dataList):
try:
http = HttpRequest()
http.setTimeout(900)
url = sync_crawl_infos_url
sqlList = json.dumps(dataList)
body = {"sql": sqlList, "checksum": toMd5(sqlList)}
encryptFields = []
headerDict = {"Content-Encoding": "gzip", "Accept-Encoding": "gzip"}
response = http.setUrl(url).setBody(body).setHeader(headerDict).encrypt(encryptFields).post()
res = json.loads(response)["data"]
if not res:
return []
return res
except Exception, e:
res = []
logging.info("-----------%s-------" % e, True)
return res
def syncCrawlInfos(dataList):
try:
http = HttpRequest()
http.setTimeout(900)
url = sync_crawl_infos_url
sqlList = json.dumps(dataList)
body = {'sql': sqlList, 'checksum': toMd5(sqlList)}
encryptFields = []
headerDict = {'Content-Encoding': 'gzip', 'Accept-Encoding': "gzip"}
response = http.setUrl(url).setBody(body).setHeader(
headerDict).encrypt(encryptFields).post()
res = json.loads(response)['data']
if not res:
return []
return res
except Exception, e:
res = []
logging.info('-----------%s-------' % e, True)
return res
def __init__(self):
self.code.clear()
bs = BeautifulSoup(HttpRequest("http://quote.eastmoney.com/stocklist.html").get(), "lxml")
for a in bs.find('div', id='quotesearch').find_all('a'):
try:
rg = re.match(".*(sh|sz)([0-9]{6})\.html", a['href']).groups()
self.code[rg[0]].add(rg[1])
except:
pass
for key in self.code:
logging.info("%s: %d" % (key, len(self.code[key])))
