def getInputDaysStr(sortedDays,nowDay,baseDir,N=7):
baseFiles=getFiles(baseDir)
baseFileDict={}
for f in baseFiles:
baseFileDict[f]=1
if len(sortedDays) > 0:
lastDay=TimeUtil.addDay(sortedDays[-1],1)
if not baseDir.endswith("/"):
baseDir=baseDir+"/"
inputDir=""
while lastDay <= nowDay:
if len(inputDir) > 0:
inputDir+=","
newDir=baseDir+lastDay
if newDir not in baseFileDict:
sys.stderr.write(newDir+"is not exists!\n")
continue
inputDir+=newDir
lastDay=TimeUtil.addDay(lastDay,1)
else:
inputDir=""
for i in range(N):
if len(inputDir) > 0:
inputDir+=","
newDir=baseDir+datetime.strftime(date.today()-timedelta(days=i+1),"%Y-%m-%d")
if newDir not in baseFileDict:
sys.stderr.write(newDir+"is not exists!\n")
continue
inputDir+=newDir
return inputDir
def getInputDaysStr(sortedDays, nowDay, baseDir, N=7):
baseFiles = getFiles(baseDir)
baseFileDict = {}
for f in baseFiles:
baseFileDict[f] = 1
if len(sortedDays) > 0:
lastDay = TimeUtil.addDay(sortedDays[-1], 1)
if not baseDir.endswith("/"):
baseDir = baseDir + "/"
inputDir = ""
while lastDay <= nowDay:
if len(inputDir) > 0:
inputDir += ","
newDir = baseDir + lastDay
if newDir not in baseFileDict:
sys.stderr.write(newDir + "is not exists!\n")
continue
inputDir += newDir
lastDay = TimeUtil.addDay(lastDay, 1)
else:
inputDir = ""
for i in range(N):
if len(inputDir) > 0:
inputDir += ","
newDir = baseDir + datetime.strftime(
date.today() - timedelta(days=i + 1), "%Y-%m-%d")
if newDir not in baseFileDict:
sys.stderr.write(newDir + "is not exists!\n")
continue
inputDir += newDir
return inputDir
def getLastDay(hdfsDir, N=7):
sortedDays = getDays(hdfsDir)
if len(sortedDays) > 0:
lastDay = TimeUtil.addDay(sortedDays[-1], 1)
return lastDay
else:
return datetime.strftime(date.today() - timedelta(days=N), "%Y-%m-%d")
def getLastDay(hdfsDir,N=7): sortedDays=getDays(hdfsDir) if len(sortedDays) > 0: lastDay=TimeUtil.addDay(sortedDays[-1],1) return lastDay else: return datetime.strftime(date.today()-timedelta(days=N),"%Y-%m-%d")
def readMall(day,end):
db=DB()
db.connect("haodou_mall")
cursor=db.execute("select og.UserId,og.GoodsId,max(ol.CreateTime),ol.OrderId from OrderBase as ob,OrderGoods as og, OrderLog as ol where og.OrderId=ob.OrderId and ob.OrderId=ol.OrderId and ob.OrderStatus=70 and ol.CreateTime >='"+day+" 00:00:00' and ol.CreateTime <= '"+end+" 23:59:59' group by og.UserId,og.GoodsId,ol.OrderId;")
ret=cursor.fetchall()
for r in ret:
#sys.stdout.write(str(r[2])+"\n")
nutPrint("uid-"+str(r[0]),"buy","goods-%d"%(r[1]),str(int(time.mktime(r[2].timetuple()))))
if __name__=="__main__":
if len(sys.argv) >= 4 and sys.argv[1] == "acc":
hdfsDir=sys.argv[2]
N=int(sys.argv[3])
sortedDays=hdfsFile.getDays(hdfsDir)
if len(sortedDays) > 0:
lastDay=TimeUtil.addDay(sortedDays[-1],1)
else:
lastDay=datetime.strftime(date.today()-timedelta(days=N),"%Y-%m-%d")
end=datetime.strftime(date.today()-timedelta(days=1),"%Y-%m-%d")
sys.stderr.write("lastDay for dbNut:"+lastDay+"\n")
sys.stderr.write("end for dbNut:"+end+"\n")
readComment(lastDay,end)
readMall(lastDay,end)
elif len(sys.argv) >= 3:
readComment(sys.argv[1],sys.argv[2])
readMall(sys.argv[1],sys.argv[2])
elif len(sys.argv) >= 2:
N=int(sys.argv[1])
end=datetime.strftime(date.today()-timedelta(days=1),"%Y-%m-%d")
day=datetime.strftime(date.today()-timedelta(days=N),"%Y-%m-%d")
print N,day,end
import sys
sys.path.append("../util")
import TimeUtil
print TimeUtil.addDay(sys.argv[1],int(sys.argv[2]))
import sys
sys.path.append("../util")
import TimeUtil
print TimeUtil.addDay(sys.argv[1], int(sys.argv[2]))
