#!/usr/bin/perl
use Date::Manip;

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

# Archit Shivaprakash
# IRIS Project
# Department of Computer Science
# University of Houston, Houston TX-77204
# Email: archit@cs.uh.edu
# July 2003

# Purpose: Extracts the qacct information that is provided by SGE, HPC at UH

# Invocation syntax: Iris_Extractor.cgi Source_File_Name 
#						Destination_File_name

# Error Handling: Prints "IRIS Extractor Error!!!" to Destination file
#			and to IRIS Error Directory
#                 Program assumes that it is only invoked when both Source 
#		  	and Destination Files are present 
#		  Failure to do so will cause program to exit pre-maturely	

# Note:
# 	Assumes use in conjunction with the IRIS Tool
#       Necessary Permissions need to be set while running this script
# 	Program is tested extensively on Redhat Linux 9.0
#	Source file to script is deleted upon completion
# 	Please contact author for Information and Feedback

#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

# Command Line Arguments Retrieved
$source_file=$ARGV[0];
$destination_file=$ARGV[1];

# Redirecting STDERR to IRIS Error Directory
open(stderr ,">> ../Error/Iris_Parser.err");

# Collecting IRIS Script Execution Timestamp
# Variable Declaration
$IRIS_parse_timestamp='Unknown';
$IRIS_day_of_week='Unknown';
$IRIS_month_of_year='Unknown';
$IRIS_day_of_month=-1;
$IRIS_hour_of_day=-1;
$IRIS_minute_of_hour=-1;
$IRIS_second_of_minute=-1;
$IRIS_year=-1;

# Timestamp Computation Section
$IRIS_parse_timestamp=localtime(time);
@IRIS_current_time=split(" ",$IRIS_parse_timestamp);
$IRIS_day_of_week=shift(@IRIS_current_time);
$IRIS_month_of_year=shift(@IRIS_current_time);
$IRIS_day_of_month=shift(@IRIS_current_time);
$IRIS_temp_time=shift(@IRIS_current_time);
$IRIS_year=shift(@IRIS_current_time);
@IRIS_current_time=split(/:/,$IRIS_temp_time);
$IRIS_hour_of_day=shift(@IRIS_current_time);
$IRIS_minute_of_hour=shift(@IRIS_current_time);
$IRIS_second_of_minute=shift(@IRIS_current_time);

# Opening and Reading from Source File
open(DAT,$source_file) || die("Iris_Extractor Error!!!, $IRIS_parse_timestamp, 
			Source File Read Error!, IRIS_Extractor will Exit!\n");
@IRIS_raw_data=<DAT>;
close(DAT);

# Creation of Destination File
# The original copy of the file, if existent will be over-written
open(DAT, ">$destination_file") || die("Iris_Extractor Error!!!, $IRIS_parse_timestamp, 
			Destination File Creation Error!, IRIS_Extractor will Exit!\n");
close(DAT);

# Reopening Destination file in Append Mode for Writing
open(DAT, ">>$destination_file") || die("Iris_Extractor Error!!!, $IRIS_parse_timestamp, 
			Destination File Append Error!, IRIS_Extractor will Exit!\n");

# Variables Declaration Section
$count=0;
$temp1='Unknown';
$temp2='Unknown';
$queue_name='Unknown';
$job_owner='Unknown';
$job_name='Unknown';
$job_id='Unknown';
$execution_time='Unknown';
$job_fail_success_status=0;
$job_nodes='Unknown';
$job_cpu_time='Unknown';
$job_memory_usage='Unknown';
$qsub_timestamp='Unknown';
$qsub_day_of_week='Unknown';
$qsub_month_of_year='Unknown';
$qsub_day_of_month=-1;
$qsub_hour_of_day=-1;
$qsub_minute_of_hour=-1;
$qsub_second_of_minute=-1;
$qsub_year=-1;
$start_timestamp='Unknown';
$start_day_of_week='Unknown';
$start_month_of_year='Unknown';
$start_day_of_month=-1;
$start_hour_of_day=-1;
$start_minute_of_hour=-1;
$start_second_of_minute=-1;
$start_year=-1;
$queue_wait_weeks=-1;
$queue_wait_days=-1;
$queue_wait_hours=-1;
$queue_wait_minutes=-1;
$queue_wait_seconds=-1;

# Operation Flags Declaration Section
$job_fail_flag=0;

# IRIS_Extractor Implementation Section Begins
foreach $line_var(@IRIS_raw_data) 
{
	chop($line_var);
	@line_array=split(" ",$line_var);
	$temp1=shift(@line_array);
	
	# Extracts Queue Name
	if($temp1 eq "qname")
	{
		$queue_name=shift(@line_array);
		$count++;
	}
	
	# Extracts Job Owner
	if($temp1 eq "owner")
	{
		$job_owner=shift(@line_array);
		$count++;
	}

	# Extracts Job Name
	if($temp1 eq "jobname")
	{
		$job_name=shift(@line_array);
		$count++;
	}
	
	# Extracts Job ID
	if($temp1 eq "jobnumber")
	{
		$job_id=shift(@line_array);
		$count++;
	}

	# Extracts Execution Time in seconds
	if($temp1 eq "ru_wallclock")
	{
		$execution_time=shift(@line_array);
		$count++;
	}
	
	# Extracts Job Failure and Success information
	if($temp1 eq "failed")
	{
		$job_fail_success_status=shift(@line_array);
		if($job_fail_success_status!=0)
		{
			$job_fail_flag=1;
		}
	}

	# Extracts Number of Nodes Used	
	if($temp1 eq "slots")
	{
		$job_nodes=shift(@line_array);
		$count++;
	}
	
	# Extracts Accumulated CPU Time 
	if($temp1 eq "cpu")
	{
		$job_cpu_time=shift(@line_array);
		$count++;
	}
	
	# Extracts Accumulated Memory Usage
	if($temp1 eq "mem")
	{
		$job_memory_usage=shift(@line_array);
		$count++;
	}
	
	# Extracts Qsub Time
	if($temp1 eq "qsub_time")
	{
		$count++;
		@qsub_time_array=@line_array;
		$qsub_day_of_week=shift(@line_array);
		$qsub_month_of_year=shift(@line_array);
		$qsub_day_of_month=shift(@line_array);
		$qsub_timestamp=shift(@line_array);
		$qsub_year=shift(@line_array);
		@qsub_time=split(/:/,$qsub_timestamp);
		$qsub_hour_of_day=shift(@qsub_time);
		$qsub_minute_of_hour=shift(@qsub_time);
		$qsub_second_of_minute=shift(@qsub_time);				
	}
	
	# Extracts Job Start Time
	if($temp1 eq "start_time")
	{
		$count++;
		@start_time_array=@line_array;
		$start_day_of_week=shift(@line_array);
		$start_month_of_year=shift(@line_array);
		$start_day_of_month=shift(@line_array);
		$start_timestamp=shift(@line_array);
		$start_year=shift(@line_array);
		@start_time=split(/:/,$start_timestamp);
		$start_hour_of_day=shift(@start_time);
		$start_minute_of_hour=shift(@start_time);
		$start_second_of_minute=shift(@start_time);
	}		
	if($count>=10)
	{
		last;
	}
} # Foreach Loop Ends

# Job Wait Time Computation
$date_time1=ParseDate(\@qsub_time_array);
$date_time2=ParseDate(\@start_time_array);
$delta_time=DateCalc($date_time1,$date_time2,\$err,1);
@queue_wait_time=split(/:/,$delta_time);
shift(@queue_wait_time);
shift(@queue_wait_time);
$queue_wait_weeks=shift(@queue_wait_time);
$queue_wait_days=shift(@queue_wait_time);
$queue_wait_hours=shift(@queue_wait_time);
$queue_wait_minutes=shift(@queue_wait_time);
$queue_wait_seconds=shift(@queue_wait_time);
$queue_wait_hours=$queue_wait_hours+($queue_wait_days*24)+($queue_wait_weeks*7*24);
# Printing Parsed Content to Destination File
if($job_fail_flag==1)
{
	print DAT "IRIS_Extractor_Error\n";
	unlink $source_file;
}
if($job_fail_flag==0)
{
	print DAT "Queue_Name=$queue_name\n";
	print DAT "Job_Owner=$job_owner\n";
	print DAT "Job_Name=$job_name\n";
	print DAT "Job_ID=$job_id\n";
	print DAT "Execution_Time=$execution_time\n";
	print DAT "Job_Submission_Day_Of_Week=$qsub_day_of_week\n";
	print DAT "Job_Submission_Month_Of_Year=$qsub_month_of_year\n";
	print DAT "Job_Submission_Day_Of_Month=$qsub_day_of_month\n";
	print DAT "Job_Submission_Year=$qsub_year\n";
	print DAT "Job_Submission_Hour_Of_Day=$qsub_hour_of_day\n";
	print DAT "Job_Submission_Minute_Of_Hour=$qsub_minute_of_hour\n";
	print DAT "Job_Submission_Second_Of_Minute=$qsub_second_of_minute\n";
	print DAT "Queue_Wait_Hours=$queue_wait_hours\n";
	print DAT "Queue_Wait_Minutes=$queue_wait_minutes\n";
	print DAT "Queue_Wait_Seconds=$queue_wait_seconds\n";
	print DAT "Job_Nodes=$job_nodes\n";
	print DAT "Job_CPU_Time=$job_cpu_time\n";
	print DAT "Job_Memory_Usage=$job_memory_usage\n";
	print DAT "Current_Day_Of_Week=$IRIS_day_of_week\n";
	print DAT "Current_Month_Of_Year=$IRIS_month_of_year\n";
	print DAT "Current_Day_Of_Month=$IRIS_day_of_month\n";
	print DAT "Current_Year=$IRIS_year\n";
	print DAT "Current_Hour_Of_Day=$IRIS_hour_of_day\n";
	print DAT "Current_Minute_Of_Hour=$IRIS_minute_of_hour\n";
	print DAT "Current_Second_Of_Minute=$IRIS_second_of_minute\n";
	unlink $source_file;
}