#!/usr/freeware/bin/perl -w

# written by Mark Bennett and Peter Gavin

# This script generates an index of all the NetCDF files found under a
# given directory. This index has 1 line for each NetCDF file, and
# contains several fields, each separated by a comma. The fields
# are as follows:
#
# 1.  CD name
# 2.  Path to the NetCDF file, relative to the base directory
# 3.  Compressed file name (when present)
# 4.  File name
# 5.  Minimum woce_date
# 6.  Maximum woce_date
# 7.  Minimum latitude in the file
# 8.  Maximum latitude in the file
# 9.  Westmost longitude in the file
# 10. Eastmost longitude in the file
# 11.  Minimum depth
# 12.  Maximum depth 
# 	** Note: mkinv is coded to set min/max depth to 0.0. If you want to  determine a 
# 		range, the code will need to be modified.
# 13. Experiment ID (expocode)
# 14. Minimum sea temperature
# 15. Maximum sea temperature
# 16. Minimum wind speed
# 17. Maximum wind speed
# 18. Platform type and name
# 19. Existence of heading PL_HD
# 20. Existence of course PL_CRS
# 21. Existence of platform speed PL_SPD
# 22. Existence of platform relative wind direction PL_WDIR
# 23. Existence of platform relative wind speed PL_WSPD
# 24. Existence of earth relative wind direction DIR
# 25. Existence of atmospheric pressure P
# 26. Existence of air temperature T
# 27. Existence of wet-bulb temperature TW
# 28. Existence of dewpoint temperature TD
# 29. Existence of relative humidity RH
# 30. Existence of specific humidity Q
# 31. Existence of precipitation PRECIP
# 32. Existence of rain rate RRATE
# 33. Existence of atmospheric radiation RAD
# 34. Existence of weather code WX
# 35. Existence of total cloud amount TCA
# 36. Existence of low/middle cloud amount LMCA
# 37. Existence of cloud base height ZCL
# 38. Existence of low cloud type LCT
# 39. Existence of middle cloud type MCT
# 40. Existence of high cloud type HCT
#
# All times are given as minutes from 1-1-1980 00:00 UTC.

# Edit this path to the top directory in your data structure
@basedir = ();
$basedir[0] = "/Net/woce/cd3/MET/data/woce_rv"; #woce data
$basedir[1] = "/Net/woce/cd3/MET/data/woce_rels"; #non-woce data

@main::dr = ();
$main::dr[0] = "woce_rv";
$main::dr[1] = "woce_rel";
$main::dr[2] = "woce_buoy";
# This it the filename you wish to output the index into
#$indexfile = "/usr/people/bennett/Woce/data.idx";
$verbose = 0;

$switches = 0;
while ($argument = shift(@ARGV)) {
  SWITCH: {
      $argument =~ /^-o$/ && do {
	  $indexfile = shift(@ARGV);
	  $switches = 1;
	  last SWITCH;
      };
      $argument =~ /^-v$/ && do {
	  $verbose = 1;
	  last SWITCH;
      };
      $argument =~ /^-h$/ && do {
	  die "Usage: mkinv -o <output file>  -h\n  -o 	  <output file>\n     REQUIRED. Name of output file to generate.\n     Optional base directry to index.\n  -h\n     Prints this help 				  message.\n";
      };
  }
}

if($switches == 1){
use NetCDF;

open NO_DATE, ">no_date.txt";
@nc_info_datavars = ( 
				"PL_HD",
		      "PL_CRS",
		      "PL_SPD",
		      "PL_WDIR",
		      "PL_WSPD",
		      "DIR",
		      "P",
		      "T",
		      "TW",
		      "TD",
		      "RH",
		      "Q",
		      "PRECIP",
		      "RRATE",
		      "RAD",
		      "WX",
				"TCA",
            "LMCA",
            "ZCL",
            "LCT",
            "MCT",
            "HCT"
);

@nc_info_keys = ("cd_name", "file_path", "file_compressed_name", "file_name", 
		 "woce_date_min", "woce_date_max",
		 "latitude_min", "latitude_max",
		 "longitude_westmost", "longitude_eastmost",
		 "depth_min", "depth_max",
		 "EXPOCODE", "temperature_min", 
	    "temperature_max", "wind_speed_min", 
		 "wind_speed_max", "site",
		  @nc_info_datavars);


$switch_lon = 0;


########################################################################
# junk routines

sub max (@)
{
	my $wflag = 0;
	my $eflag = 0;
	$switch_lon = 0;
    my $max = shift(@_);
    foreach my $value (@_) {
		$max = $value if ($value > $max);
		if($value >= 175 and $value <= 180)
		{
			$eflag = 1;
		}
		if($value < -175 and $value > -180)
		{
			$wflag = 1;
		}
    }
		if($eflag == 1 and $wflag == 1)
		{
			$switch_lon = 1;
		}
    return  $max;
}

sub min (@)
{
    my $min = shift(@_);
	if($min < -6000)
	{
		$min = 9999;
	}
    foreach my $value (@_) {
		$min = $value if ($value > -6000 and $value < $min);
    }
    return $min;
}

sub maxp (@)
{
    my $max = shift(@_);
    foreach my $value (@_) {
	$max = $value if (($value < $max && $value > 0.0) || ($max < 0.0));
    }
    return $max;
}

sub minp (@)
{
    my $min = shift(@_);
	if($min < -6000)
	{
		$min = 9999;
	}
    foreach my $value (@_) {
		$min = $value if (($value > $min && $value < 0.0) || ($min > 0.0));
    }
    return $min;
}

sub round ($$)
{
    my $places = shift(@_);
    my $value = shift(@_);
    return (int($value*(10.0**$places) + .5))/(10.0**$places);
}

########################################################################
# read a bunch of stuff from an nc file and return it as a hash

sub read_nc_info ($$)
{
    my %info = ();
    
    my ($ncfilename, $dr) = @_;
    my $ncid = NetCDF::open($ncfilename, NetCDF::NOWRITE);
    
    my $ncndims = "";
    my $ncnvars = "";
    my $ncnatts = "";
    my $ncrecdim = "";
    
    my $ncvarid = 0;
    my @ncvar = ();
    my $ncvarname = "";
    my $ncvartype = "";
    my $ncvarndims = "";
    my @ncvardimids = ();
    my $ncvarnatts = "";
    my @ncvarstart = ();
    my @ncvarcount = ();
    my @ncvarvalue = ();
    my @varnames = ();
    my $ncatttype = "";
    my $ncattlen = "";
    my $ncattvalue = "";
    
    my $ncdimid = "";
    my $ncdimname = "";
    
    my $dim_time_size = "";
	 my @path = ();

    $info{"cd_name"} = "wocemet";

	 @path = split /\//, $ncfilename;

	if($dr eq "woce_buoy")
   {
		$path[2] = $path[1];
		$path[1] = "";
	 	$info{"file_path"} = join "/", "./data", $dr, $path[0], "";
	}
	else
	{
	 	$info{"file_path"} = join "/", "./data", $dr, $path[0], $path[1],"";
	}

    $info{"file_name"} = $path[2];
    
	 if(($dr eq "woce_rv" and ($path[0] eq "DBBH" or $path[0] eq "KCEJ")) or
		($dr eq "woce_rel"))
	 {
    	 $info{"file_compressed_name"} = $info{"file_name"} . ".gz";
	 }
	 else
	 {
    	 $info{"file_compressed_name"} = "";	 
	 }

    die "Could not open \"$ncfilename\" for reading\n"
	if ($ncid == -1);
    
    die "Could not inquire file \"$ncfilename\"\n"
	if (NetCDF::inquire($ncid, \$ncndims, $ncnvars, \$ncnatts,
			    \$ncrecdim) == -1);
    
    # get number of records
    $ncdimid = NetCDF::dimid($ncid, "time");
    die "Could not get id for dimension \"time\" in file \"$ncfilename\"\n"
	if ($ncdimid == -1);
    die "Could not inquire dimension \"time\" in file \"$ncfilename\"\n"
	if (NetCDF::diminq($ncid, $ncdimid, \$ncdimname,
			   \$dim_time_size) == -1);
    
    # read max and min latitude
    $ncvarid = NetCDF::varid($ncid, "latitude");
    die "Could not get id for variable \"latitude\" in file \"$ncfilename\"\n"
	if ($ncvarid == -1);
    @ncvarstart = ( 0 );
    @ncvarcount = ( $dim_time_size );
    @ncvarvalue = ();
    die "Could not get variable \"latitude\" in file \"$ncfilename\"\n"
	if (NetCDF::varget($ncid, $ncvarid, @ncvarstart, @ncvarcount,
			   \@ncvarvalue) == -1);
    $info{"latitude_min"} = round 2, min(@ncvarvalue);
    $info{"latitude_max"} = round 2, max(@ncvarvalue);
    
    # read max and min longitude
    $ncvarid = NetCDF::varid($ncid, "longitude");
    die "Could not get id for variable \"longitude\" in file \"$ncfilename\"\n"
	if ($ncvarid == -1);
    @ncvarstart = ( 0 );
    @ncvarcount = ( $dim_time_size );
    @ncvarvalue = ();
    die "Could not get variable \"longitude\" in file \"$ncfilename\"\n"
	if (NetCDF::varget($ncid, $ncvarid, @ncvarstart, @ncvarcount,
			   \@ncvarvalue) == -1);

	# check for ship crossing date line
    $info{"longitude_eastmost"} = round 2, max(@ncvarvalue);

	if($switch_lon eq 0)
	{	# if ship doesn't cross, use normal max/min
    $info{"longitude_westmost"} = round 2, min(@ncvarvalue);
	}
	else
	{	#ship crosses, switch values
	 print "\n** CROSSING 180 DEGREES **\n\n";
	 print POLAR "$info{\"file_path\"}  $info{\"file_name\"}\n";
    
	 $info{"longitude_westmost"} = round 2, maxp(@ncvarvalue);
    $info{"longitude_eastmost"} = round 2, minp(@ncvarvalue);
	}
    

    # read max and min temperature
	my $temperature_exist = 0;
    for ($ncvarid = 0; $ncvarid < $ncnvars; $ncvarid++) {
	die "Could not inquire variable with id $ncvarid in file \"$ncfilename\"\n"
	    if (NetCDF::varinq($ncid, $ncvarid, \$ncvarname, \$ncvartype,
			 \$ncvarndims, \@ncvardimids, \$ncvarnatts) == -1);
		push @varnames, $ncvarname if ($ncvarname =~ m/[A-Z]+/);
		if($ncvarname eq "temperature")
		{
			$temperature_exist = 1;
		}
    }
	if($temperature_exist == 1)
	{
	    $ncvarid = NetCDF::varid($ncid, "temperature");
	    @ncvarstart = ( 0 );
	    @ncvarcount = ( $dim_time_size );
	    @ncvarvalue = ();
	    die "Could not get variable \"temperature\" in file \"$ncfilename\"\n"
		if (NetCDF::varget($ncid, $ncvarid, @ncvarstart, @ncvarcount,
				   \@ncvarvalue) == -1);
	    $info{"temperature_min"} = round 2, min(@ncvarvalue);
	    $info{"temperature_max"} = round 2, max(@ncvarvalue);
	}
	else
	{
	    $info{"temperature_min"} = "";
	    $info{"temperature_max"} = "";
	}

	$info{"depth_min"} = 0;
	$info{"depth_max"} = 0;
    
	# read min and max date
	my $date_exist = 0;
    for ($ncvarid = 0; $ncvarid < $ncnvars; $ncvarid++) {
	die "Could not inquire variable with id $ncvarid in file \"$ncfilename\"\n"
	    if (NetCDF::varinq($ncid, $ncvarid, \$ncvarname, \$ncvartype,
			 \$ncvarndims, \@ncvardimids, \$ncvarnatts) == -1);
		push @varnames, $ncvarname if ($ncvarname =~ m/[A-Z]+/);
		if($ncvarname eq "woce_date")
		{
			$date_exist = 1;
		}
    }

	if($date_exist == 1)
	{
    	$ncvarid = NetCDF::varid($ncid, "woce_date");
	    die "Could not get id for variable \"woce_date\" in file \"$ncfilename\"\n"
		if ($ncvarid == -1);
   	 @ncvarstart = ( 0 );
	    @ncvarcount = ( $dim_time_size );
   	 @ncvarvalue = ();
	    die "Could not get variable \"woce_date\" in file \"$ncfilename\"\n"
		if (NetCDF::varget($ncid, $ncvarid, @ncvarstart, @ncvarcount,
			   \@ncvarvalue) == -1);
	    $info{"woce_date_min"} = min(@ncvarvalue);
   	 $info{"woce_date_max"} = max(@ncvarvalue);
	}
	else
	{
	    $info{"woce_date_min"} = 0;
   	 $info{"woce_date_max"} = 0;
		print NO_DATE "$info{\"filename\"}\n";
	}
    
	# read min and max speed
	my $speed_exist = 0;
    for ($ncvarid = 0; $ncvarid < $ncnvars; $ncvarid++) {
	die "Could not inquire variable with id $ncvarid in file \"$ncfilename\"\n"
	    if (NetCDF::varinq($ncid, $ncvarid, \$ncvarname, \$ncvartype,
			 \$ncvarndims, \@ncvardimids, \$ncvarnatts) == -1);
		push @varnames, $ncvarname if ($ncvarname =~ m/[A-Z]+/);
		if($ncvarname eq "SPD")
		{
			$speed_exist = 1;
		}
    }
	if($speed_exist == 1)
	{
	    $ncvarid = NetCDF::varid($ncid, "SPD");
	    @ncvarstart = ( 0 );
	    @ncvarcount = ( $dim_time_size );
	    @ncvarvalue = ();
	    die "Could not get variable \"wind_speed\" in file \"$ncfilename\"\n"
		if (NetCDF::varget($ncid, $ncvarid, @ncvarstart, @ncvarcount,
				   \@ncvarvalue) == -1);
	    $info{"wind_speed_min"} = round 2, min(@ncvarvalue);
	    $info{"wind_speed_max"} = round 2, max(@ncvarvalue);
	}
	else
	{
	    $info{"wind_speed_min"} = "";
	    $info{"wind_speed_max"} = "";
	}

    # read variable names
    DATAVAR: foreach my $datavar (@nc_info_datavars) {
	$info{$datavar} = 0;
	foreach my $varname (@varnames) {
	    if ($varname =~ m/$datavar([0-9]*)/) {
		$info{$datavar} = 1;
		next DATAVAR;
	    }
	}
    }
    
    # read experiment ID
    die "Could not get attribute \"EXPOCODE\" in file \"$ncfilename\"\n"
	if (NetCDF::attget($ncid, NetCDF::GLOBAL, "EXPOCODE",
			   \$ncattvalue) == -1);
    # for some reason, $ncattvalue gets an extra \0 at the end, so we
    # need to get rid of it
    $ncattvalue =~ s/\0//;
    $info{"EXPOCODE"} = $ncattvalue;
    
    # read 
    die "Could not get attribute \"site\" in file \"$ncfilename\"\n"
	if (NetCDF::attget($ncid, NetCDF::GLOBAL, "site",
			   \$ncattvalue) == -1);
    $info{"site"} = $ncattvalue;# . "_test";
#	print "$info{\"site\"} " . length $info{"site"} . " ";
	if((length $info{"site"}) lt 23)
	{
		chop $info{"site"};
	}
    
    die "Could not close file \"$ncfilename\"\n"
	if NetCDF::close($ncid);
    
    return \%info;
}

########################################################################
# convert the hash returned from the above function to a string so it
# can be saved into a file

sub nc_info_to_string (\%)
{
    *info = shift(@_);
    return join("\t", @info{@nc_info_keys});
}

########################################################################
# convert a string (from a file) into a hash like the one returned
# from the function above

sub string_to_nc_info ($)
{
    my %info = ();
    @info{@nc_info_keys} = split /,/, shift(@_);
    return \%info;
}

# process arguments

while ($argument = shift(@ARGV)) {
  SWITCH: {
      $argument =~ /^-f$/ && do {
	  $indexfile = shift(@ARGV);
	  last SWITCH;
      };
      $argument =~ /^-v$/ && do {
	  $verbose = 1;
	  last SWITCH;
      };
      $argument =~ /^-h$/ && do {
	  die "Usage: mkinv [ -f indexfile ] [ -v ] [ -h ]\n";
      };
  }
}

open POLAR, ">polar.txt";

$main::number = 0;
open INDEXFILE, ">$indexfile"
    or die "Could not open \"$indexfile\"\n";

print INDEXFILE join("\t", @nc_info_keys), "\n";

foreach $main::i (@basedir)
{
chdir $main::i;

print "Using base directory $main::i\n" if ($verbose);
print "Using index file $indexfile\n" if ($verbose);
open FINDPIPE, 'find . -type f -name "*.nc" |'
    or die "Could not run find\n";

while ($file = <FINDPIPE>) {
    $ship = "";
    chop $file;
    $file =~ s/^\.\///;
    ($ship, $version) = ($file =~ /^([A-Z0-9]+)\/[-_A-Z0-9]+\/[A-Z0-9]+\.[0-9]{9}v([0-9]{3})\.nc$/);

    if ( !defined($version)) {# && $version eq "woce" ) {
	print "Processing $main::i/$file\n" if ($verbose);
	*info = read_nc_info($file,$main::dr[$main::number]);
	print INDEXFILE nc_info_to_string(%info), "\n";
    } else {
	print "Skipping $main::i/$file\n" if ($verbose);
    }
}

++$main::number;

close FINDPIPE;

}
# Buoy Data

$main::number = 2;
$basedir = "/Net/woce/cd3/MET/data/woce_buoys";
chdir $basedir;

open FINDPIPE, 'find . -type f -name "*.nc" |'
    or die "Could not run find\n";

while ($file = <FINDPIPE>) {
    $ship = "";
    chop $file;
    $file =~ s/^\.\///;
    ($ship, $version) = ($file =~ /^([A-Z0-9]+)\/[A-Z0-9]+\.[0-9]{9}v([0-9]{3})\.nc$/);
    if (!defined($version)) {
	print "Processing $basedir/$file\n" if ($verbose);
	*info = read_nc_info($file,$main::dr[$main::number]);
	print INDEXFILE nc_info_to_string(%info), "\n";
    } else {
	print "Skipping $basedir/$file\n" if ($verbose);
    }

}

close INDEXFILE;
close FINDPIPE;
close NO_DATE;
close POLAR;
}
else
{
   die "Usage: mkinv -o <output file>  -h\n  -o 	    <output file>\n     REQUIRED. Name of output file to generate.\n      Optional base directry to index.\n  -h\n     Prints this help 			   message.\n";
}