//1. create database and table
db1 = database(,VALUE,1000..2000) //  tenantId
db2  = database(, VALUE, 2022.01.01..2022.12.30) // DATE

// TSDB for iot 
dbNoise = database("dfs://NoiseDB",COMPO,[db1,db2], engine="TSDB", atomic="CHUNK") 
// OLAP 
dbNoiseOLAP = database("dfs://NoiseDBOLAP",COMPO,[db1,db2], atomic="CHUNK") 

create table "dfs://NoiseDB"."noise"(
    tenantId INT,
    deviceId INT,
    soundPressureLevel INT,
    soundPowerLevel DOUBLE,
    ts TIMESTAMP,
    date DATE
)
partitioned by tenantId,date
sortColumns=[`deviceId,`ts]

// OLAP
create table "dfs://NoiseDBOLAP"."noise"(
    tenantId INT,
    deviceId INT,
    soundPressureLevel INT,
    soundPowerLevel DOUBLE,
    ts TIMESTAMP,
    date DATE
)
partitioned by tenantId,date


//2.  prepare data
def genData(tenantNum, nodeNum, frequency, date){
	totalRecords =  frequency * tenantNum * nodeNum
	tenantId = rand(1001..(1000+tenantNum), totalRecords)
	deviceId = rand(10001..(10000+nodeNum), totalRecords)
	soundPressureLevel = rand(180, totalRecords)
	soundPowerLevel = rand(10000.0, totalRecords)
	startTs= rand(timestamp(date)..(timestamp(date) + 60 * 60 * 1000), 1)[0]
	endTs = startTs +frequency*100
	ts = rand(startTs..endTs, totalRecords) 
	date1 =take(date, totalRecords)
	t = table(tenantId, deviceId, soundPressureLevel, soundPowerLevel, ts, date1)
	pt = loadTable(database("dfs://NoiseDB"),"noise")
	pt.append!(t)
}

def genDataByday(d){
	print(now() + " start to generate data of " + d)
	genData(100, 100, 10000,d)
	print(now() + "successfully generated data of " + d)
}

for(d in 2022.01.01..2022.01.12){	
	genDataByday(d)
}


select count(*) from  loadTable(database("dfs://NoiseDB"),"noise") where date between 2022.01.01:2022.01.12
select sum(diskUsage)/1024/1024/1024 as GB from getTabletsMeta(tableName="noise",top=-1, diskUsage=true)


// before testing compact manually
chunkIds = exec chunkId from getChunksMeta() where type=1
for (x in chunkIds) {
  triggerTSDBCompaction(x)
}

//clear os cache,run it on linux
//sudo bash -c "echo 3 > /proc/sys/vm/drop_caches"  

//3. time series search case 
//3.1 query the  last 100 record of a node/device 
noise = loadTable(database("dfs://NoiseDB"),"noise")
tenantIdRand=1055
deviceIdRand=10067
dateRand=2022.01.01

timer(10) select * from noise 
where date=dateRand and tenantId=tenantIdRand and deviceId=deviceIdRand
order by ts asc
limit 100


//3.2 query the last state of a  node/device 
tenantIdRand=rand(1001..1100, 100)[0]
dateRand=2022.01.01

timer(10) select * from noise 
where date=dateRand and tenantId=tenantIdRand 
context by deviceId
csort ts desc
limit 1



//3.3 query the statis of a node/device  in  5 min
tenantIdRand=1055
deviceIdRand=10067
dateRand=2022.01.01
ts1=exec max(ts) from  noise where date = dateRand and tenantId = tenantIdRand and deviceId = deviceIdRand
ts2 = ts1 - 5 * 60 * 1000

timer(10)  select
	 min(ts) as startTs
	,max(ts) as endTs
	,max(soundPressureLevel)
	,avg(soundPressureLevel)
	, max(soundPowerLevel) 
	,avg(soundPowerLevel) 
from noise
where date=dateRand and tenantId=tenantIdRand and deviceId=deviceIdRand and ts between ts2:ts1
group by tenantId,deviceId


//3.4 query a whole day data of a device
tenantIdRand=1055
deviceIdRand=10067
dateRand=2022.01.01

timer(10) select *
from noise
where date=2022.01.01 and tenantId=tenantIdRand and deviceId=deviceIdRand
order by ts


//clear os cache,run it on linux
//sudo bash -c "echo 3 > /proc/sys/vm/drop_caches"  

//clearAllCache()

// 4 compare with OLAP
noiseOLAP = loadTable("dfs://NoiseDBOLAP","noise")
noiseOLAP.append!(select * from noise where date=2022.01.01)

// 4.1 case1
noiseOLAP = loadTable("dfs://NoiseDBOLAP","noise")
tenantIdRand=1055
deviceIdRand=10067
dateRand=2022.01.01
timer(10) select * from noiseOLAP 
where date=dateRand and tenantId=tenantIdRand and deviceId=deviceIdRand
order by ts asc
limit 100


//4.2 query the last state of a  node/device 
tenantIdRand=rand(1001..1100, 100)[0]
dateRand=2022.01.01

timer(10) select * from noiseOLAP 
where date=dateRand and tenantId=tenantIdRand 
context by deviceId
csort ts desc
limit 1



//4.3 query the statis of a node/device  in  5 min
tenantIdRand=1055
deviceIdRand=10067
dateRand=2022.01.01
ts1=exec max(ts) from  noiseOLAP where date = dateRand and tenantId = tenantIdRand and deviceId = deviceIdRand
ts2 = ts1 - 5 * 60 * 1000

timer(10)  select
	 min(ts) as startTs
	,max(ts) as endTs
	,max(soundPressureLevel)
	,avg(soundPressureLevel)
	, max(soundPowerLevel) 
	,avg(soundPowerLevel) 
from noiseOLAP
where date=dateRand and tenantId=tenantIdRand and deviceId=deviceIdRand and ts between ts2:ts1
group by tenantId,deviceId


//4.4 query a whole day data of a device
tenantIdRand=1055
deviceIdRand=10067
dateRand=2022.01.01

timer(10) select *
from noiseOLAP
where date=2022.01.01 and tenantId=tenantIdRand and deviceId=deviceIdRand
order by ts



// 5 export data
t = select * from noise where date=2022.01.01
saveText(t, "/data/noise20220101.csv")



// 6 notice
// on the community version or with limited resource , split the  data  into small slices,and manually gc the cache
// example1
def genDataByday(d){
	print(now() + " start to generate data of " + d)
	genData(100, 100, 3000,d)
	flushTSDBCache()
	sleep(1000)
	genData(100, 100, 3000, d)
	flushTSDBCache()
	sleep(1000)
	genData(100, 100, 4000, d)
	flushTSDBCache()
	sleep(1000)
	print(now() + "successfully generated data of " + d)
}

// example2
noiseOLAP = loadTable("dfs://NoiseDBOLAP","noise")
noiseOLAP.append!(select * from noise where date=2022.01.01 and tenantId between 1001:1033)
purgeCacheEngine()
sleep(3000)
noiseOLAP.append!(select * from noise where date=2022.01.01 and tenantId between 1034:1066)
purgeCacheEngine()
sleep(3000)
noiseOLAP.append!(select * from noise where date=2022.01.01 and tenantId between 1067:1100)
purgeCacheEngine()
sleep(3000)