#CreateAndPopulateAmazonDirStructure_Inventory_LDFR_20230619.plx

# PROJECT
$project='inv';
$amazondir='C:\\epa\\moves\\amazon20130603';
$year=2021;
$vers='20231110';
$nonroad_counties="RepCos_2021_Inventory_20231110.txt";
#$scen='';

# BATCHSIZE -- NUMBER OF COUNTIES PER BATCH
$batchsize=1;
print "project = $project\n";
print "amazondir = $amazondir\n";
print "year = $year\n";
print "vers = $vers\n";
print "nonroad_counties = $nonroad_counties\n";
print "batchsize = $batchsize\n";

# CASE
#$case="$project-$year-$scen";  # for when the case includes a scen & year
#$case="$project";  # for when the case does not include a scen & year
$case="$project-$year";  # for when the case does not include a scen
$casevers="${case}-$vers";
$casepath="C:\\EPA_MOVES_5-19\\On-road\\$year\\AmazonStructure\\$casevers";
$casedbs="$casepath\\databases_$casevers";
$dbdir="C:\\EPA_MOVES_5-19\\On-road\\$year\\InputDatabases";
$zmhdb='2021nei_month_hour_for_nonroad_met';
$cdbdir="C:\\EPA_MOVES_5-19\\On-road\\$year\\InputDatabases\\CDBs_2021_20231108\\";

# MOVES driver output
$runspectemplate_evp="samplerunspecs\\INV_MVS4_CY${year}_JulyOnly_evap.mrs";
$runspectemplate_starts="samplerunspecs\\INV_MVS4_CY${year}_allMonths_starts.mrs";
$runspectemplate_other="samplerunspecs\\INV_MVS4_CY${year}_JanJulOnly_nonevap_nonstart.mrs";
print "cdbdir = $cdbdir\n";
print "case = $case\n";
print "casevers = $casevers\n";
print "casepath = $casepath\n";
print "casedbs = $casedbs\n";
print "dbdir = $dbdir\n";
print "zmhdb = $zmhdb\n";
print "runspectemplate_evp = $runspectemplate_evp\n";
print "runspectemplate_starts = $runspectemplate_starts\n";
print "runspectemplate_other = $runspectemplate_other\n";
`md $casepath`;
`md $casedbs`;

# COPY CASE-LEVEL DATABASES
foreach $xxx(
"$dbdir $zmhdb",
"$dbdir\\LEV lev_06_20220824",
"$dbdir\\LEV lev_09_20220824",
"$dbdir\\LEV lev_10_20220824",
"$dbdir\\LEV lev_23_20220824",
"$dbdir\\LEV lev_24_20220824",
"$dbdir\\LEV lev_25_20220824",
"$dbdir\\LEV lev_34_20220824",
"$dbdir\\LEV lev_36_20220824",
"$dbdir\\LEV lev_41_20220824",
"$dbdir\\LEV lev_42_20220824",
"$dbdir\\LEV lev_44_20220824",
"$dbdir\\LEV lev_50_20220824",
"$dbdir\\LEV lev_53_20220824",
"$dbdir\\LEV nlev_09_20220824",
"$dbdir\\LEV nlev_10_20220824",
"$dbdir\\LEV nlev_11_20220824",
"$dbdir\\LEV nlev_24_20220824",
"$dbdir\\LEV nlev_33_20220824",
"$dbdir\\LEV nlev_34_20220824",
"$dbdir\\LEV nlev_42_20220824",
"$dbdir\\LEV nlev_44_20220824",
"$dbdir\\LEV nlev_51_20220824",
){
($dbpath,$db)=split(/ /,$xxx);
print "dbpath = $dbpath\n";
print "db = $db\n";
print "Case-level DB = $db\n";
	if (-d "$casedbs\\$db") {
		print 'Line '.__LINE__.". DB $db already exists; not copied.\n";
	} else {
	# Include only mysql tables by specifying the extensions
	`xcopy $dbpath\\$db\\*.frm $casedbs\\$db /I/S/Y/D/Q`;
	`xcopy $dbpath\\$db\\*.MYD $casedbs\\$db /I/S/Y/D/Q`;
	`xcopy $dbpath\\$db\\*.MYI $casedbs\\$db /I/S/Y/D/Q`;
	}
}
open(nrcos1,"$nonroad_counties");
# COUNTY
$cntycount=0;
$batchno=0;

foreach $xxx(<nrcos1>)
{
if(substr($xxx,0,1) eq '#'){next;}
($cntyfips,$sdb,$cdb)=split(/,/,$xxx);
chomp($cdb);
$state=substr($cdb,1,2);
print "state = $state\n";

$cntycount=$cntycount+1;
print "cntyfips, sdb, cdb = $cntyfips, $sdb, $cdb\n";
# First element of batch
	print "cntycount = $cntycount\n";
	if($cntycount%$batchsize==1 or $batchsize==1){
		$batchno = $batchno+1;
		$batch=$casevers."_batch".sprintf("%04d",$batchno);
		$batchpath="$casepath\\$batch";
		$batchdbs="$batchpath\\databases";
		print "mod=1: cntycount = $cntycount\n";
		print "mod=1: cntyfips= $cntyfips\n";
		print "mod=1: batchno = $batchno\n";
		print "mod=1: batch = $batch\n";
		print "mod=1: batchpath = $batchpath\n";
		print "mod=1: batchdbs = $batchdbs\n";
		# Create folder batchpath and transfer batch-specific databases.  
		# Create folder batchdatabases and transfer batch-specific databases.  
		`md $batchpath`;
		`md $batchdbs`;
	}
	print "cntycount = $cntycount\n";
	print "cntyfips= $cntyfips\n";
	print "batchno = $batchno\n";
	print "batch = $batch\n";
	print "batchpath = $batchpath\n";
	print "batchdbs = $batchdbs\n";

# Create batch county or state level databases folder and copy CDBs or SDBs
	if($sdb ne ''){
		`xcopy $cdbdir\\$sdb $batchdbs\\$sdb /I/S/Y/D/Q`;
	}
	if($cdb ne ''){
		`xcopy $cdbdir\\$cdb $batchdbs\\$cdb /I/S/Y/D/Q`;
	}

# JOB
$jobcount=0;
foreach $xxx (
	"evp $runspectemplate_evp",
	"start $runspectemplate_starts",
	"other $runspectemplate_other",
)
{
	($runspectype,$runspectemplate)=split(/ /,$xxx);
	print "runspectype = $runspectype\n";
	print "runspectemplate = $runspectemplate\n";
	$jobcount=$jobcount+1;
if($runspectype eq 'evp'){
	$job="c".sprintf("%05d",$cntyfips)."_${year}_${runspectype}";
	$jobpath="$batchpath\\$job";
	print "job = $job\n";
	print "jobpath = $jobpath\n";
	`md $jobpath`;
	# Create jobdatabases folder and copy jobdatabases
	`md $jobpath\\databases`;
	#if($sdb ne ''){
	#	`xcopy $cdbdir\\$sdb $jobpath\\databases\\$sdb /I/S/Y/D/Q`;
	#}
	#if($cdb ne ''){
	#	`xcopy $cdbdir\\$cdb $jobpath\\databases\\$cdb /I/S/Y/D/Q`;
	#}
# EDIT RUNSPECS INTO APPROPRIATE DIRECTORY
# Input database name has been commented out for increased flexibility.
#  It is blank in the moves driver script output, and 
#  we could change the database without having to re-jar the job.
	open (in1,"$runspectemplate");
	open (out1,">$jobpath\\$job.mrs");
	while (<in1>) {
		# SET COUNTY
		if (m/geographicselection type="COUNTY"/){
			s/<geographicselection type.*$/<geographicselection type="COUNTY" key="$cntyfips" description=""\/>/;
		}
		# SET OUTPUT DB
		if (m/outputdatabase/){
			s/databasename=".*description/databasename="standardoutput" description/;
		}
		# SET CDB
		if (m/scaleinputdatabase/){
			s/databasename=".*description/databasename="$cdb" description/;
		}
		# Replace databaseselections block
		if (m/<databaseselections>/){
			print out1; 
			$indbs=1;
			# zonemonthhour
			print out1 "		<databaseselection servername=\"\" databasename=\"$zmhdb\" description=\"\"/>\n";
			# fuels
			# print out1 "		<databaseselection servername=\"\" databasename=\"$fuelsdb\" description=\"\"/>\n";
			#if($sdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$sdb\" description=\"\"/>\n";
			#}
			#if($cdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$cdb\" description=\"\"/>\n";
			#}
			# OTAQ database with Construction and Agricultural surrogates and populations
			# print out1 "		<databaseselection servername=\"\" databasename=\"$srgdb\" description=\"\"/>\n";
			# rhoffdb
			#if ($rhoffdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$rhoffdb\" description=\"\"/>\n";
			#}
			# LEV
			$lev=0;
			foreach $st(
				'06','09','10','23','24','25','34','36','41','42','44','50','53',
				){
				if($state eq $st){$lev=1;}
			}
			# NLEV
			$nlev=0;
			foreach $st(
				'09','10','11','24','33','34','42','44','51',
				){
				if($state eq $st){$nlev=1;}
			}
			if ($lev==1){
				print out1 "		<databaseselection servername=\"\" databasename=\"lev_${state}_20220824\" description=\"\"/>\n";
			}
			if ($nlev==1){
				print out1 "		<databaseselection servername=\"\" databasename=\"nlev_${state}_20220824\" description=\"\"/>\n";
			}
			next; # Go to next line without printing if still in databaseselections
		}
		if (m/<\/databaseselections>/){
			$indbs=0;
			print out1;
			next;
		}
		if ($indbs) {next;} # Go to next line without printing if still in databaseselections
		print out1;
	}
	close(in1);
	close(out1);
} # ENDIF for RunSpec Type 'evp'
if($runspectype eq 'start'){
$monthcount=0;
foreach $month(
	'1',
	'2',
	'3',
	'4',
	'5',
	'6',
	'7',
	'8',
	'9',
	'10',
	'11',
	'12',
){
	$monthcount=$monthcount+1;
	print "month = $month\n";
	$job="c".sprintf("%05d",$cntyfips)."_${year}_${month}_${runspectype}";
	$jobpath="$batchpath\\$job";
	print "job = $job\n";
	print "jobpath = $jobpath\n";
	`md $jobpath`;
	# Create jobdatabases folder and copy jobdatabases
	`md $jobpath\\databases`;
	#if($sdb ne ''){
	#	`xcopy $cdbdir\\$sdb $jobpath\\databases\\$sdb /I/S/Y/D/Q`;
	#}
	#if($cdb ne ''){
	#	`xcopy $cdbdir\\$cdb $jobpath\\databases\\$cdb /I/S/Y/D/Q`;
	#}
# EDIT RUNSPECS INTO APPROPRIATE DIRECTORY
# Input database name has been commented out for increased flexibility.
#  It is blank in the moves driver script output, and 
#  we could change the database without having to re-jar the job.
	open (in1,"$runspectemplate");
	open (out1,">$jobpath\\$job.mrs");
	while (<in1>) {
		# SET COUNTY
		if (m/geographicselection type="COUNTY"/){
			s/<geographicselection type.*$/<geographicselection type="COUNTY" key="$cntyfips" description=""\/>/;
		}
		# SET MONTH
		if (m/month id="1"/){
			s/1/$month/;
		}
		# SET OUTPUT DB
		if (m/outputdatabase/){
			s/databasename=".*description/databasename="standardoutput" description/;
		}
		# SET CDB
		if (m/scaleinputdatabase/){
			s/databasename=".*description/databasename="$cdb" description/;
		}
		# Replace databaseselections block
		if (m/<databaseselections>/){
			print out1; 
			$indbs=1;
			# zonemonthhour
			print out1 "		<databaseselection servername=\"\" databasename=\"$zmhdb\" description=\"\"/>\n";
			# fuels
			# print out1 "		<databaseselection servername=\"\" databasename=\"$fuelsdb\" description=\"\"/>\n";
			#if($sdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$sdb\" description=\"\"/>\n";
			#}
			#if($cdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$cdb\" description=\"\"/>\n";
			#}
			# OTAQ database with Construction and Agricultural surrogates and populations
			# print out1 "		<databaseselection servername=\"\" databasename=\"$srgdb\" description=\"\"/>\n";
			# rhoffdb
			#if ($rhoffdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$rhoffdb\" description=\"\"/>\n";
			#}
			# LEV
			$lev=0;
			foreach $st(
				'06','09','10','23','24','25','34','36','41','42','44','50','53',
				){
				if($state eq $st){$lev=1;}
			}
			# NLEV
			$nlev=0;
			foreach $st(
				'09','10','11','24','33','34','42','44','51',
				){
				if($state eq $st){$nlev=1;}
			}
			if ($lev==1){
				print out1 "		<databaseselection servername=\"\" databasename=\"lev_${state}_20220824\" description=\"\"/>\n";
			}
			if ($nlev==1){
				print out1 "		<databaseselection servername=\"\" databasename=\"nlev_${state}_20220824\" description=\"\"/>\n";
			}
			next; # Go to next line without printing if still in databaseselections
		}
		if (m/<\/databaseselections>/){
			$indbs=0;
			print out1;
			next;
		}
		if ($indbs) {next;} # Go to next line without printing if still in databaseselections
		print out1;
	}
	close(in1);
	close(out1);
} # END MONTH loop for RunSpec Type 'start'
} # ENDIF for RunSpec Type 'start'
if($runspectype eq 'other'){
$monthcount=0;
foreach $month(
	'1',
	'7',
){
	$monthcount=$monthcount+1;
	print "month = $month\n";
	$job="c".sprintf("%05d",$cntyfips)."_${year}_${month}_${runspectype}";
	$jobpath="$batchpath\\$job";
	print "job = $job\n";
	print "jobpath = $jobpath\n";
	`md $jobpath`;
	# Create jobdatabases folder and copy jobdatabases
	`md $jobpath\\databases`;
	#if($sdb ne ''){
	#	`xcopy $cdbdir\\$sdb $jobpath\\databases\\$sdb /I/S/Y/D/Q`;
	#}
	#if($cdb ne ''){
	#	`xcopy $cdbdir\\$cdb $jobpath\\databases\\$cdb /I/S/Y/D/Q`;
	#}
# EDIT RUNSPECS INTO APPROPRIATE DIRECTORY
# Input database name has been commented out for increased flexibility.
#  It is blank in the moves driver script output, and 
#  we could change the database without having to re-jar the job.
	open (in1,"$runspectemplate");
	open (out1,">$jobpath\\$job.mrs");
	while (<in1>) {
		# SET COUNTY
		if (m/geographicselection type="COUNTY"/){
			s/<geographicselection type.*$/<geographicselection type="COUNTY" key="$cntyfips" description=""\/>/;
		}
		# SET MONTH
		if (m/month id="1"/){
			s/1/$month/;
		}
		# SET OUTPUT DB
		if (m/outputdatabase/){
			s/databasename=".*description/databasename="standardoutput" description/;
		}
		# SET CDB
		if (m/scaleinputdatabase/){
			s/databasename=".*description/databasename="$cdb" description/;
		}
		# Replace databaseselections block
		if (m/<databaseselections>/){
			print out1; 
			$indbs=1;
			# zonemonthhour
			print out1 "		<databaseselection servername=\"\" databasename=\"$zmhdb\" description=\"\"/>\n";
			# fuels
			# print out1 "		<databaseselection servername=\"\" databasename=\"$fuelsdb\" description=\"\"/>\n";
			#if($sdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$sdb\" description=\"\"/>\n";
			#}
			#if($cdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$cdb\" description=\"\"/>\n";
			#}
			# OTAQ database with Construction and Agricultural surrogates and populations
			# print out1 "		<databaseselection servername=\"\" databasename=\"$srgdb\" description=\"\"/>\n";
			# rhoffdb
			#if ($rhoffdb ne ''){
			#	print out1 "		<databaseselection servername=\"\" databasename=\"$rhoffdb\" description=\"\"/>\n";
			#}
			# LEV
			$lev=0;
			foreach $st(
				'06','09','10','23','24','25','34','36','41','42','44','50','53',
				){
				if($state eq $st){$lev=1;}
			}
			# NLEV
			$nlev=0;
			foreach $st(
				'09','10','11','24','33','34','42','44','51',
				){
				if($state eq $st){$nlev=1;}
			}
			if ($lev==1){
				print out1 "		<databaseselection servername=\"\" databasename=\"lev_${state}_20220824\" description=\"\"/>\n";
			}
			if ($nlev==1){
				print out1 "		<databaseselection servername=\"\" databasename=\"nlev_${state}_20220824\" description=\"\"/>\n";
			}
			next; # Go to next line without printing if still in databaseselections
		}
		if (m/<\/databaseselections>/){
			$indbs=0;
			print out1;
			next;
		}
		if ($indbs) {next;} # Go to next line without printing if still in databaseselections
		print out1;
	}
	close(in1);
	close(out1);
} # END MONTH loop for RunSpec Type 'other'
} # ENDIF for RunSpec Type 'other'
} # JOB
# JARJOBS FOR EACH BATCH
open(out2,">jarjobs.bat");
print out2 "call pushd $amazondir\n";
print out2 "call setenv\n";
print out2 "call ant -Dcmd=\"jarjobs batchdir $batchpath\" job\n";
print out2 "call popd\n";
close(out2);
`call jarjobs.bat`;
} # COUNTY
close(nrcos1);