def searchRange(casperScript, dep, arr, dateRange, retryTimes):
depTime = datetime.datetime.today()
d = datetime.timedelta(days=1)
for i in range(dateRange):
depTime = depTime + d
ret = ER_SUCC
for j in range(retryTimes):
ret = searchOne(casperScript, dep, arr, depTime.strftime("%Y-%m-%d"))
if ret == ER_SUCC:
L.info("{} -> {} {}", dep[0], arr[0], depTime.strftime("%Y-%m-%d"))
break
if ER_SUCC != ret:
L.error("retry {} times, {} -> {} {} failed", retryTimes, dep[0], arr[0], depTime.strftime("%Y-%m-%d"))
def searchRange(casperScript, dep, arr, dateRange, retryTimes):
depTime = datetime.datetime.today()
d = datetime.timedelta(days=1)
for i in range(dateRange):
depTime = depTime + d
ret = ER_SUCC
for j in range(retryTimes):
ret = searchOne(casperScript, dep, arr,
depTime.strftime("%Y-%m-%d"))
if ret == ER_SUCC:
L.info("{} -> {} {}", dep[0], arr[0],
depTime.strftime("%Y-%m-%d"))
break
if ER_SUCC != ret:
L.error("retry {} times, {} -> {} {} failed", retryTimes, dep[0],
arr[0], depTime.strftime("%Y-%m-%d"))
def crawlAllAirlinesWithRetry(self, retryTime):
self.crawlAllAirlines()
lstCur = self.lstFailAirline
lstNext = []
for i in range(1, retryTime + 1):
if len(lstCur) == 0:
return ER_SUCC
for it in lstCur:
L.info("retry[{}] {} -> {}", i, it[0][0], it[1][0])
startDate = datetime.datetime.today() + datetime.timedelta(days=1)
if ER_SUCC != self.crawlOneAirline(it[0], it[1], startDate):
lstNext.append(it)
lstCur = lstNext
for it in lstCur:
L.error("{} -> {} retry {} times failed", it[0][0], it[1][0], retryTime)
def crawlOneAirline(self, depInfo, arrInfo, startDate):
curDateTime = time.localtime(time.time())
queryDate = time.strftime('%Y-%m-%d', curDateTime)
queryTime = time.strftime('%H:%M:%S', curDateTime)
urlBase = "http://flight.qunar.com/twell/flight/farecast.jsp?departureCity={}&arrivalCity={}&nextNDays=0&departureDate={}&searchType=OnewayFlight&searchLangs=zh&locale=zh&serverIP=twell4&allowOld=true&queryID=127.0.0.1%3A1c1ea29%3A113aed2be0b%3A-7bfb&dayNum={}&pageNum=0"
url = urlBase.format(depInfo[1], arrInfo[1], startDate, self.dateRange)
# print url
try:
r = requests.get(url, timeout=10)
except Exception as e:
L.error(e)
L.error("{} -> {} timeout, url={}", depInfo[0], arrInfo[0], url)
return ER_REQUEST_TIMEOUT
if r.status_code != 200:
L.error("{} -> {} failed, url={}", depInfo[0], arrInfo[0], url)
return ER_RESPONSE_FAIL
L.info("{} -> {}", depInfo[0], arrInfo[0])
bs = BeautifulSoup(r.text, 'lxml-xml')
resultData = bs.find('ResultData')
for airline in resultData.children:
if airline.name == 'lowestPrice':
d = airline.attrs
allAttrs = ('date', 'code', 'depTime', 'arrTime', 'carrier',
'vendorName', 'price')
attrOK = True
for attr in allAttrs:
if attr not in d:
attrOK = False
break
if not attrOK:
continue
info = FlightLowestPriceInfo(queryDate, queryTime, depInfo[0],
arrInfo[0],
[d[x] for x in allAttrs])
# print info.asRec()
self.dbHandle.insertOneRec(info)
return ER_SUCC
def crawlAllAirlinesWithRetry(self, retryTime):
self.crawlAllAirlines()
lstCur = self.lstFailAirline
lstNext = []
for i in range(1, retryTime + 1):
if len(lstCur) == 0:
return ER_SUCC
for it in lstCur:
L.info("retry[{}] {} -> {}", i, it[0][0], it[1][0])
startDate = datetime.datetime.today() + datetime.timedelta(
days=1)
if ER_SUCC != self.crawlOneAirline(it[0], it[1], startDate):
lstNext.append(it)
lstCur = lstNext
for it in lstCur:
L.error("{} -> {} retry {} times failed", it[0][0], it[1][0],
retryTime)
def crawlOneAirline(self, depInfo, arrInfo, startDate):
curDateTime = time.localtime(time.time())
queryDate = time.strftime('%Y-%m-%d', curDateTime)
queryTime = time.strftime('%H:%M:%S', curDateTime)
urlBase = "http://flight.qunar.com/twell/flight/farecast.jsp?departureCity={}&arrivalCity={}&nextNDays=0&departureDate={}&searchType=OnewayFlight&searchLangs=zh&locale=zh&serverIP=twell4&allowOld=true&queryID=127.0.0.1%3A1c1ea29%3A113aed2be0b%3A-7bfb&dayNum={}&pageNum=0"
url = urlBase.format(depInfo[1], arrInfo[1], startDate, self.dateRange)
# print url
try:
r = requests.get(url, timeout=10)
except Exception as e:
L.error(e)
L.error("{} -> {} timeout, url={}", depInfo[0], arrInfo[0], url)
return ER_REQUEST_TIMEOUT
if r.status_code != 200:
L.error("{} -> {} failed, url={}", depInfo[0], arrInfo[0], url)
return ER_RESPONSE_FAIL
L.info("{} -> {}", depInfo[0], arrInfo[0])
bs = BeautifulSoup(r.text, 'lxml-xml')
resultData = bs.find('ResultData')
for airline in resultData.children:
if airline.name == 'lowestPrice':
d = airline.attrs
allAttrs = ('date', 'code', 'depTime', 'arrTime', 'carrier', 'vendorName', 'price')
attrOK = True
for attr in allAttrs:
if attr not in d:
attrOK = False
break
if not attrOK:
continue
info = FlightLowestPriceInfo(queryDate, queryTime, depInfo[0], arrInfo[0], [d[x] for x in allAttrs])
# print info.asRec()
self.dbHandle.insertOneRec(info)
return ER_SUCC