#!/usr/bin/perl -w

#EOC
#------------------------------------------------------------------------------
#                  GEOS-Chem Global Chemical Transport Model                  !
#------------------------------------------------------------------------------
#BOP
#
# !MODULE: getRunInfo
#
# !DESCRIPTION: This script extracts information from the input.geos file
#  and prints the result. The input arguments are (1) the location of file
#  input.geos, and (2) the ID of run information to be extracted. Mapping 
#  between ID and input.geos information are is as follows: 0:MET, 1:GRID,
#  2:SIM, 3:NEST, 4:START, and 5:END.
#\\
#\\
# !USES:
#
require 5.003;                   # Need this version of Perl or newer
use English;                     # Use English language
use Carp;                        # Get detailed error messages
use strict;                      # Use "IMPLICIT NONE" syntax
#
# !PUBLIC MEMBER FUNCTIONS
#  &getRunInfo
# 
# !CALLING SEQUENCE:
#  getRunInfo RUNDIR ID
#    where RUNDIR is the path location of input.geos and ID is the 
#    ID of the run information to retrieve
#
# !REMARKS:
#  Designed for use with the root-level Makefile in a run directory when
#  using the unit test or a run directory copied from UnitTest.
#                                                                             .
#  ID value   Information returned
#  -------------------------------------------------------------------
#  0          Met field type      (geosfp, merra2, etc.             )
#  1          Horizontal grid     (4x5, 2x25, etc.                  )
#  2          Simulation type     (standard, benchmark, RnPbBe, etc.)
#  3          Nested grid option  (na, ch, as, etc.                 )
#  4          Start date & time   (in YYYYMMDDhhmm format           )
#  5          End   date & time   (in YYYYMMDDhhmm format           )
#  6          Root data directory
#  7          Start date only     (in YYYYMMDD format               )
#  8          End   date only     (in YYYYMMDD format               )
#
# !REVISION HISTORY: 
#  06 Apr 2015 - E. Lundgren - Initial version
#EOP
#------------------------------------------------------------------------------
#                  GEOS-Chem Global Chemical Transport Model                  !
#------------------------------------------------------------------------------
#BOP
#
# !IROUTINE: getRunInfo
#
# !DESCRIPTION: Extracts run information from input.geos.
#\\
#\\
# !INTERFACE:
#
sub getInfo($$) {
#
# !INPUT PARAMETERS:
#
  my ( $rundir, $id ) = @_;
#
# !CALLING SEQUENCE:
#  &getInfo( $rundir, $id );
#
# !REVISION HISTORY:
#  06 Apr 2015 - E. Lundgren - Initial version
#  27 Apr 2015 - M. Sulprizio- Now use YYYYMMDDhhmm format for start/end time
#  31 Mar 2015 - E. Lundgren - Add special case of soa_svpoa
#  22 May 2017 - R. Yantosca - For HEMCO standalone, convert YYYY-MM-DD to
#                              YYYYMMDD and hh:mm:ss to hhmmss
#  19 Sep 2017 - R. Yantosca - Add options to return start & end time in just
#                              YYYYMMDD format (e.g. for use in filenames)
#EOP
#------------------------------------------------------------------------------
#BOC
#
# !LOCAL VARIABLES:
#
  # Scalars
  my $runstr  = ""; 
  my $line    = "";
  my $runinfo = "";

  # Arrays
  my @linestrings   = ();
  my @dates         = ();
  my @times         = ();
  my @runinfoarray  = ();

  # Make sure that an input.geos file is found; otherwise print a "";
  if ( -f "$rundir/input.geos" ) {
  
    # get start run time if $id is 4 or 7, end run time is $id is 5 or 8
    if ( $id == 4 || $id == 5 || $id == 7 || $id == 8 )  {
        
      if    ( $id == 4 || $id == 7 ) { $runstr = "Start"; }
      elsif ( $id == 5 || $id == 8 ) { $runstr = "End";   }
   
      # Grep for date & time in input.geos
      $line = qx( grep "$runstr.*YYYYMMDD" $rundir/input.geos );
      chomp( $line );
  
      # Split by spaces
      @linestrings = split( ' ', $line );
  
      # Place start and end times into YYYYMMDDhhmm format (ID = 4 or 5)
      # or just YYYYMMDD format (ID = 7 or 8)
      $runinfo = "$linestrings[4]$linestrings[5]";
      if    ( $id == 4 || $id == 5 ) { $runinfo = substr( $runinfo, 0, 12 ); } 
      elsif ( $id == 7 || $id == 8 ) { $runinfo = substr( $runinfo, 0, 8  ); }

    } else {
 
      # Grep for input.geos header
      $line = qx( grep "GEOS-CHEM UNIT TEST SIMULATION" $rundir/input.geos );
      chomp( $line );
      
      # Split by spaces
      @linestrings = split( ' ', $line );
      
      # Extract the original run directory name
      $runstr = "$linestrings[4]";
      
      # Split by '_' delimiter
      @runinfoarray  = split( '_', $runstr );
      
      # If the id is less than the # of rundirstrings, then extract the 
      # run info from the rundir name. Else, set runinfo to 'n' which is 
      # used as NEST for non-nested simulations
      if ( $id < 0+@runinfoarray ) {
      
        # Special case for complexSOA_SVPOA
	if ( ( $id == 2 || $id == 3 ) && 0+@runinfoarray == 4 ) {
	  if ( $runinfoarray[3] eq "SVPOA" ) {
	    if ( $id == 2 ) { $runinfo = "complexSOA_SVPOA" } # SIM=complexSOA_SVPOA
	    if ( $id == 3 ) { $runinfo = "n" }                # NEST=n
	  } else {
	    $runinfo = "$runinfoarray[$id]"; 
	  }
        } else {
         $runinfo = "$runinfoarray[$id]"; 
        }	

      } else {
        $runinfo = "n";
      }

    }				

  } elsif ( -f "$rundir/HEMCO_sa_Config.rc" ) {

    # Hardcode simulation type to HEMCO for now
    if ( $id == 2 ) {

       $runinfo = "HEMCO"

    # Hardcode nest to "n" for now
    } elsif ( $id == 3 ) {

       $runinfo = "n"

    # Get start run time if $id is 4 or 7, end run time is $id is 5 or 8
    } elsif ( $id == 4 || $id == 5 || $id == 7 || $id == 8 ) {

      if ( -f "$rundir/HEMCO_sa_Time.rc" ) {

	if ( $id == 4 || $id == 7 ) { $runstr = "START"; }
	if ( $id == 5 || $id == 8 ) { $runstr = "END";   }

	# Grep for start and date
	$line = qx( grep "$runstr" $rundir/HEMCO_sa_Time.rc );
        chomp( $line );

        # Split by spaces
        @linestrings = split( ' ', $line );

	# Translatet YYYY-MM-DD into YYYYMMDD
	# and hh:mm:ss to hhmmss
	@dates = split( '-', $linestrings[1] );
	@times = split( ':', $linestrings[2] );

        # Return start and end times in YYYYMMDDhhmm format (ID = 4 or 5)
	# or just YYYYMMDD format (ID = 7 or 8)
	if ( $id == 7 || $id == 8 ) { 
	  $runinfo = "$dates[0]$dates[1]$dates[2]";
        } elsif ( $id == 4 || $id == 5 ) {
 	  $runinfo = "$dates[0]$dates[1]$dates[2]$times[0]$times[1]";
        }
      }

    } else {

      if ($id == 0) { $runstr = "MET:" }
      if ($id == 1) { $runstr = "RES:" }

      # Grep for input.geos header
      $line = qx( grep "$runstr" $rundir/HEMCO_sa_Config.rc );
      chomp( $line );
      
      # Split by spaces
      @linestrings = split( ' ', $line );

      # Return met or grid type
      $runinfo = "$linestrings[1]";

    }

  }

  # Print the result
  print "$runinfo";
  
  return(0)
}
#EOP
#------------------------------------------------------------------------------
#                  GEOS-Chem Global Chemical Transport Model                  !
#------------------------------------------------------------------------------
#BOP
#
# !IROUTINE: main
#
# !DESCRIPTION: Driver routine for the getRunInfo script.
#\\
#\\
# !INTERFACE:
#
sub main() {
#
# !CALLING SEQUENCE:
#  getRunInfo DIR ID
#
# !REVISION HISTORY:
#  07 Apr 2015 - E. Lundgren - Initial version
#  20 Sep 2017 - R. Yantosca - ID can now be from 0-8
#EOP
#------------------------------------------------------------------------------
#BOC
#
# !LOCAL VARIABLES:
#
  my $nArgs = scalar( @ARGV );
  my $msg   = "";

  # Exits with an error message if the ID argument is not within valid range
  if ( $nArgs >= 0 && $nArgs <= 8 ) {
    &getInfo( @ARGV ); 
  } else { 
    $msg = "getRunInfo: $ARGV[1] is an invalid value for the ID argument!";
    print "$msg"; 
    exit(1);    
  }
  return ( 0 );
}

#------------------------------------------------------------------------------

# Call the main program
main();

# Return normally
exit(0);
#EOC