#### Script to facilitate processing of MOD06 data

setwd("/nobackupp1/awilso10/mod35")

library(rgdal)

library(raster)

library(RSQLite)

verbose=T

## get MODLAND tile information

tb=read.table("http://landweb.nascom.nasa.gov/developers/sn_tiles/sn_bound_10deg.txt",skip=6,nrows=648,header=T)

tb$tile=paste("h",sprintf("%02d",tb$ih),"v",sprintf("%02d",tb$iv),sep="")

tb=tb[tb$lon_min!=-999,]

save(tb,file="modlandTiles.Rdata")

load("modlandTiles.Rdata")

## delete temporary log file that can grow to GB

system("rm /nobackupp1/awilso10/software/heg/TOOLKIT_MTD/runtime/LogStatus")

tile="h11v08"  # Venezuela

#tile="h11v07"  # Venezuela coast

#tile="h09v04"  # Oregon

tile="h21v09"  #Kenya

### list of tiles to process

tiles=c("h11v08","h21v09","h08v04","h09v04","h08v05","h09v05","h20v11","h31v11")

tiles=c("h10v08","h11v08","h12v08","h10v07","h11v07","h12v07")  # South America

## subset to MODLAND tiles

  modlandtiles=system("ls -r /nobackupp4/datapool/modis/MOD11A1.005/2010* | grep hdf$ | cut -c18-23 | sort | uniq - ",intern=T)

 tb$land=tb$tile%in%modlandtiles

tiles=tb$tile[tb$land]

## subset tile corner matrix to tiles selected above

tile_bb=tb[tb$tile%in%tiles,]

### get list of files to process

datadir="/nobackupp4/datapool/modis/MOD35_L2.006/"

outdir="daily/" #paste("daily/",tile,sep="")

##find swaths in region from sqlite database for the specified date/tile

## path to swath database

db="/nobackupp4/pvotava/DB/export/swath_geo.sql.sqlite3.db"

con=dbConnect("SQLite", dbname = db)

fs=do.call(rbind.data.frame,lapply(1:nrow(tile_bb),function(i){

  d=dbGetQuery(con,paste("SELECT * from swath_geo6

            WHERE east>=",tile_bb$lon_min[i]," AND

                  west<=",tile_bb$lon_max[i]," AND

                  north>=",tile_bb$lat_min[i]," AND

                  south<=",tile_bb$lat_max[i])

    )

  d$tile=tile_bb$tile[i]

  print(paste("Finished tile",tile_bb$tile[i]))

  return(d)

}))

con=dbDisconnect(con)

fs$id=substr(fs$id,7,19)

### Identify which swaths are available in the datapool

swaths=data.frame(path=list.files(datadir,pattern=paste("hdf$"),recursive=T,full=T),stringsAsFactors=F)  #all swaths in data pool

swaths$id=substr(basename(swaths$path),10,22)

fs$exists=fs$id%in%swaths$id 

fs$path=swaths$path[match(fs$id,swaths$id)]

if(verbose) print(paste("###############",nrow(fs)," swath IDs recieved from database"))

## get all unique dates

fs$dateid=format(as.Date(paste(fs$year,fs$day,sep=""),"%Y%j"),"%Y%m%d")

alldates=unique(fs$dateid[fs$exists])

#### Generate submission file

startdate="2000-03-01"

stopdate="2011-12-31"

## just 2005

startdate="2005-01-01"

stopdate="2005-12-31"

alldates=format(seq(as.Date(startdate),as.Date(stopdate),1),"%Y%m%d")

proclist=expand.grid(date=alldates,tile=tiles)

proclist$year=substr(proclist$date,1,4)

## identify tile-dates with no available swaths

avail=unique(cbind.data.frame(tile=fs$tile,date=fs$dateid)[fs$exists, ])

proclist$avail=paste(proclist$tile,proclist$date,sep="_")%in%paste(avail$tile,avail$date,sep="_")

## identify which have been completed

fdone=data.frame(path=system("ssh lou 'find MOD35/daily -name \"*.nc\"' ",intern=T))

#fdone=data.frame(path=list.files(outdir,pattern="nc$",recursive=T))

fdone$date=substr(basename(as.character(fdone$path)),14,21)

fdone$tile=substr(basename(as.character(fdone$path)),7,12)

proclist$done=paste(proclist$tile,proclist$date,sep="_")%in%substr(basename(as.character(fdone$path)),7,21)

### report on what has already been processed

print(paste(sum(!proclist$done)," out of ",nrow(proclist)," (",round(100*sum(!proclist$done)/nrow(proclist),2),"%) remain"))

table(tile=proclist$tile[proclist$done],year=proclist$year[proclist$done])

script="/u/awilso10/environmental-layers/climate/procedures/MOD35_L2_process.r"

## write the table processed by mpiexec

tp=((!proclist$done)&proclist$avail)  #date-tiles to process

table(Available=proclist$avail,Completed=proclist$done)

write.table(paste("--verbose ",script," --date ",proclist$date[tp]," --verbose T --tile ",proclist$tile[tp],sep=""),

file=paste("notdone.txt",sep=""),row.names=F,col.names=F,quote=F)

### qsub script

cat(paste("

#PBS -S /bin/bash

#PBS -l select=100:ncpus=8:mpiprocs=8

##PBS -l select=20:ncpus=8:mpiprocs=8

#PBS -l walltime=5:00:00

##PBS -l walltime=2:00:00

#PBS -j n

#PBS -m be

#PBS -N mod35

#PBS -q normal

##PBS -q devel

#PBS -V

CORES=800

#CORES=160

HDIR=/u/armichae/pr/

#  source $HDIR/etc/environ.sh

  source /u/awilso10/environ.sh

  source /u/awilso10/.bashrc

IDIR=/nobackupp1/awilso10/mod35/

##WORKLIST=$HDIR/var/run/pxrRgrs/work.txt

WORKLIST=$IDIR/notdone.txt

EXE=Rscript

LOGSTDOUT=$IDIR/log/mod35_stdout

LOGSTDERR=$IDIR/log/mod35_stderr

### use mpiexec to parallelize across days

mpiexec -np $CORES pxargs -a $WORKLIST -p $EXE -v -v -v --work-analyze 1> $LOGSTDOUT 2> $LOGSTDERR

",sep=""),file=paste("mod35_qsub",sep=""))

### Check the files

system(paste("cat mod35_qsub",sep=""))

system(paste("cat notdone.txt | head",sep=""))

system(paste("cat notdone.txt | wc -l ",sep=""))

## Submit it

system(paste("qsub mod35_qsub",sep=""))

system("qstat -u awilso10")

#######################################################

### Now submit the script to generate the climatologies

tiles

ctiles=tiles#[c(1:3)]  #subset to only some tiles (for example if some aren't finished yet)?

climatescript="/pleiades/u/awilso10/environmental-layers/climate/procedures/MOD35_Climatology.r"

## check which tiles have been processed and are on lou with a filename "MOD35_[tile].nc"

cdone=data.frame(path=sapply(strsplit(basename(

                   system("ssh lou 'find MOD35/summary -name \"MOD35_h[0-9][0-9]v[0-9][0-9].nc\"' ",intern=T)),split="_"),function(x) x[2]))

cdone$tile=substr(basename(as.character(cdone$path)),1,6)

print(paste(length(ctiles[!ctiles%in%cdone$tile]),"Tiles still need to be processed: /n ",ctiles[!ctiles%in%cdone$tile]))

## write the table processed by mpiexec

write.table(paste("--verbose ",climatescript," --verbose T --tile ",ctiles[!ctiles%in%cdone$tile],sep=""),

file=paste("notdone_climate.txt",sep=""),row.names=F,col.names=F,quote=F)

## delay start until previous jobs have finished?

delay=F

## check running jobs to get JobID of job you want to wait for

system("qstat -u awilso10")

## enter JobID here:

job="881394.pbspl1.nas.nasa.gov"

### qsub script

cat(paste("

#PBS -S /bin/bash

#PBS -l select=1:ncpus=16:mem=94

#PBS -l walltime=24:00:00

#PBS -j n

#PBS -m be

#PBS -N mod35_climate

#PBS -q ldan

#PBS -V

",if(delay) paste("#PBS -W depend=afterany:",job,sep="")," 

CORES=16

HDIR=/u/armichae/pr/

  source $HDIR/etc/environ.sh

  source /pleiades/u/awilso10/environ.sh

  source /pleiades/u/awilso10/.bashrc

IDIR=/nobackupp1/awilso10/mod35/

##WORKLIST=$HDIR/var/run/pxrRgrs/work.txt

WORKLIST=$IDIR/notdone_climate.txt

EXE=Rscript

LOGSTDOUT=$IDIR/log/climatology_stdout

LOGSTDERR=$IDIR/log/climatology_stderr

### use mpiexec to parallelize across tiles

mpiexec -np $CORES pxargs -a $WORKLIST -p $EXE -v -v -v --work-analyze 1> $LOGSTDOUT 2> $LOGSTDERR

",sep=""),file=paste("mod35_climatology_qsub",sep=""))

## check files

system(paste("cat mod35_climatology_qsub",sep=""))        #qsub submission script

system(paste("cat notdone_climate.txt | head",sep=""))    #top of job file

system(paste("cat notdone_climate.txt | wc -l ",sep=""))  #number of jobs to be run

## Submit it

system(paste("qsub mod35_climatology_qsub",sep=""))

## check progress

system("qstat -u awilso10")

## start interactive job on compute node for debugging

# system("qsub -I -l walltime=2:00:00 -lselect=2:ncpus=16:model=san -q devel")

#################################################################

### copy the files back to Yale

system("ssh lou")

#scp `find MOD35/summary -name "MOD35_h[0-9][0-9]v[0-9][0-9].nc"` adamw@acrobates.eeb.yale.edu:/data/personal/adamw/projects/interp/data/modis/mod35/summary/

rsync -vv `find MOD35/summary -name "MOD35_h[0-9][0-9]v[0-9][0-9].nc"` adamw@acrobates.eeb.yale.edu:/data/personal/adamw/projects/interp/data/modis/mod35/summary/

exit