Template:US Demographics/perlscript

1. HOW TO USE THIS SCRIPT:
2. Download all the CSV files. They will be located at
3. http://www.census.gov/popest/states/asrh/tables/
4. with filenames like SC-EST2005-03-42.csv
5. The EST2005-03 indicates which file series is being used.
6. The 42 indicates the state being referenced.
7. File formats are going to change. The "15" in
8. year 2000 get_data calls indicates that the desired data
9. is between the 15th and 16th doublequote on the line. When
10. the 2006 estimates come out, it'll probably shove everything
11. over by two doublequotes, so you would changes those 15s to 17s.
12. Put all the CSV files in the same directory with this script,
13. and execute this script. It will generate a file named
14. template.txt, which is the data which needs to be inserted
15. in Template:US Demographics

1. It's a quick-and-dirty script, which is what Perl was
2. intended for. You will want to check your data, to make
3. sure you're grabbing the right numbers from the file.
4. Otherwise, you're going to get some mighty strange tables.

1. This script is NOT subject to the GNU Free Documentation
2. License, as it expressly prohibits alteration, and this
3. script will be useless if NOT altered.
4. Script copyright by user ClairSamoht, August 2006.
5. You may download it, modify it, and execute it ONLY for
6. maintaining Wikipedia's Template:US Demographics
7. You may NOT republish or redistribute this script, or
8. use it for any purpose other than intended.
9. But then, it's an awfully simple script. If you're
10. smart enough to figure out how to modify it for other\
11. purposes, you can write the script you need from scratch
12. in less time than to modify this script.

sub pct { $inval =$_[0]; $base =$_[1];

$inval = stripcommas ($inval); $base = stripcommas ($base); if ($base < 1)

       {
       return "0.0";
       }

$portion = 100 * $inval/$base; $retval = sprintf (" %6.2f", $portion); return $retval; }

sub stripcommas { $val = $_[0]; $sc_retval = ;

@sc_parts = split (/\,/, $val); foreach $sc_pit (@sc_parts)

       {
       $sc_retval .= $sc_pit;
       }

return $sc_retval; }

sub get_data { $useline = $_[0]; $item_number = $_[1];

@parts = split (/\"/, @lines[$useline]); $data = stripcommas($parts[$item_number]); return $data; }

sub make_table { foreach $val (sort keys %DATA)

       {
       $DATA{$val} = ;

}

$DATA{'fname'} = $_[0];

open (FILE, "< $DATA{'fname'}"); @lines = <FILE>; close (FILE); $count = @lines;

print "$DATA{'fname'} has $count lines\n";

$DATA{'state'} = @lines[1]; @parts = split (/ for /, @lines[1]); $DATA{'state'} = $parts[1]; @parts = split (/: /, $DATA{'state'}); $DATA{'state'} = $parts[0];

$DATA{'tot2005'} = get_data ( 4,1);

$DATA{'white2005'} = get_data (13,1); $DATA{'black2005'} = get_data (14,1); $DATA{'AIAN2005'} = get_data (15,1); $DATA{'asian2005'} = get_data (16,1); $DATA{'NHPI2005'} = get_data (17,1);

$DATA{'nh_white2005'} = get_data (27,1); $DATA{'nh_black2005'} = get_data (28,1); $DATA{'nh_AIAN2005'} = get_data (29,1); $DATA{'nh_asian2005'} = get_data (30,1); $DATA{'nh_NHPI2005'} = get_data (31,1);

$DATA{'h_white2005'} = get_data (41,1); $DATA{'h_black2005'} = get_data (42,1); $DATA{'h_AIAN2005'} = get_data (43,1); $DATA{'h_asian2005'} = get_data (44,1); $DATA{'h_NHPI2005'} = get_data (45,1);

$DATA{'tot2000'} = get_data ( 4,15);

$DATA{'white2000'} = get_data (13,15); $DATA{'black2000'} = get_data (14,15); $DATA{'AIAN2000'} = get_data (15,15); $DATA{'asian2000'} = get_data (16,15); $DATA{'NHPI2000'} = get_data (17,15);

$DATA{'nh_white2000'} = get_data (27,15); $DATA{'nh_black2000'} = get_data (28,15); $DATA{'nh_AIAN2000'} = get_data (29,15); $DATA{'nh_asian2000'} = get_data (30,15); $DATA{'nh_NHPI2000'} = get_data (31,15);

$DATA{'h_white2000'} = get_data (41,15); $DATA{'h_black2000'} = get_data (42,15); $DATA{'h_AIAN2000'} = get_data (43,15); $DATA{'h_asian2000'} = get_data (44,15); $DATA{'h_NHPI2000'} = get_data (45,15);

if (0) { print OFILE "tot2005 $DATA{'tot2005'} \n";

print OFILE "white2005 $DATA{'white2005'} \n"; print OFILE "black2005 $DATA{'black2005'} \n"; print OFILE "AIAN2005 $DATA{'AIAN2005'} \n"; print OFILE "asian2005 $DATA{'asian2005'} \n"; print OFILE "NHPI2005 $DATA{'NHPI2005'} \n";

print OFILE "hwhite2005 $DATA{'h_white2005'} \n"; print OFILE "hblack2005 $DATA{'h_black2005'} \n"; print OFILE "hAIAN2005 $DATA{'h_AIAN2005'} \n"; print OFILE "hasian2005 $DATA{'h_asian2005'} \n"; print OFILE "hNHPI2005 $DATA{'h_NHPI2005'} \n";

print OFILE "nhwhite2005 $DATA{'nh_white2005'} \n"; print OFILE "nhblack2005 $DATA{'nh_black2005'} \n"; print OFILE "nhAIAN2005 $DATA{'nh_AIAN2005'} \n"; print OFILE "nhasian2005 $DATA{'nh_asian2005'} \n"; print OFILE "nhNHPI2005 $DATA{'nh_NHPI2005'} \n";

print OFILE "tot2000 $DATA{'tot2000'} \n";

print OFILE "white2000 $DATA{'white2000'} \n"; print OFILE "black2000 $DATA{'black2000'} \n"; print OFILE "AIAN2000 $DATA{'AIAN2000'} \n"; print OFILE "asian2000 $DATA{'asian2000'} \n"; print OFILE "NHPI2000 $DATA{'NHPI2000'} \n";

print OFILE "hwhite2000 $DATA{'h_white2000'} \n"; print OFILE "hblack2000 $DATA{'h_black2000'} \n"; print OFILE "hAIAN2000 $DATA{'h_AIAN2000'} \n"; print OFILE "hasian2000 $DATA{'h_asian2000'} \n"; print OFILE "hNHPI2000 $DATA{'h_NHPI2000'} \n";

print OFILE "nhwhite2000 $DATA{'nh_white2000'} \n"; print OFILE "nhblack2000 $DATA{'nh_black2000'} \n"; print OFILE "nhAIAN2000 $DATA{'nh_AIAN2000'} \n"; print OFILE "nhasian2000 $DATA{'nh_asian2000'} \n"; print OFILE "nhNHPI2000 $DATA{'nh_NHPI2000'} \n"; }

$template = <<EOHEADER; | state = {{US DemogTable|state|fname|= |p_white2000|p_black2000|p_AIAN2000|p_asian2000|p_NHPI2000|= |h_white2000|h_black2000|h_AIAN2000|h_asian2000|h_NHPI2000|= |p_white2005|p_black2005|p_AIAN2005|p_asian2005|p_NHPI2005|= |h_white2005|h_black2005|h_AIAN2005|h_asian2005|h_NHPI2005|= |d_white|d_black|d_AIAN|d_asian|d_NHPI|= |x_white|x_black|x_AIAN|x_asian|x_NHPI|= |h_white|h_black|h_AIAN|h_asian|h_NHPI}} EOHEADER

$state = $DATA{'state'}; $fname = $DATA{'fname'};

$p_white2000 = pct ($DATA{'white2000'}, $DATA{'tot2000'}); $p_black2000 = pct ($DATA{'black2000'}, $DATA{'tot2000'}); $p_AIAN2000 = pct ($DATA{'AIAN2000'}, $DATA{'tot2000'}); $p_asian2000 = pct ($DATA{'asian2000'}, $DATA{'tot2000'}); $p_NHPI2000 = pct ($DATA{'NHPI2000'}, $DATA{'tot2000'}); $h_white2000 = pct ($DATA{'h_white2000'}, $DATA{'tot2000'}); $h_black2000 = pct ($DATA{'h_black2000'}, $DATA{'tot2000'}); $h_AIAN2000 = pct ($DATA{'h_AIAN2000'}, $DATA{'tot2000'}); $h_asian2000 = pct ($DATA{'h_asian2000'}, $DATA{'tot2000'}); $h_NHPI2000 = pct ($DATA{'h_NHPI2000'}, $DATA{'tot2000'}); $p_white2005 = pct ($DATA{'white2005'}, $DATA{'tot2005'}); $p_black2005 = pct ($DATA{'black2005'}, $DATA{'tot2005'}); $p_AIAN2005 = pct ($DATA{'AIAN2005'}, $DATA{'tot2005'}); $p_asian2005 = pct ($DATA{'asian2005'}, $DATA{'tot2005'}); $p_NHPI2005 = pct ($DATA{'NHPI2005'}, $DATA{'tot2005'}); $h_white2005 = pct ($DATA{'h_white2005'}, $DATA{'tot2005'}); $h_black2005 = pct ($DATA{'h_black2005'}, $DATA{'tot2005'}); $h_AIAN2005 = pct ($DATA{'h_AIAN2005'}, $DATA{'tot2005'}); $h_asian2005 = pct ($DATA{'h_asian2005'}, $DATA{'tot2005'}); $h_NHPI2005 = pct ($DATA{'h_NHPI2005'}, $DATA{'tot2005'}); $d_white = pct ($DATA{'white2005'} - $DATA{'white2000'},$DATA{'white2000'}); $d_black = pct ($DATA{'black2005'} - $DATA{'black2000'},$DATA{'black2000'}); $d_AIAN = pct ($DATA{'AIAN2005'} - $DATA{'AIAN2000'}, $DATA{'AIAN2000'} ); $d_asian = pct ($DATA{'asian2005'} - $DATA{'asian2000'},$DATA{'asian2000'}); $d_NHPI = pct ($DATA{'NHPI2005'} - $DATA{'NHPI2000'}, $DATA{'NHPI2000'} ); $x_white = pct ($DATA{'nh_white2005'} - $DATA{'nh_white2000'},$DATA{'nh_white2000'}); $x_black = pct ($DATA{'nh_black2005'} - $DATA{'nh_black2000'},$DATA{'nh_black2000'}); $x_AIAN = pct ($DATA{'nh_AIAN2005'} - $DATA{'nh_AIAN2000'}, $DATA{'nh_AIAN2000'} ); $x_asian = pct ($DATA{'nh_asian2005'} - $DATA{'nh_asian2000'},$DATA{'nh_asian2000'}); $x_NHPI = pct ($DATA{'nh_NHPI2005'} - $DATA{'nh_NHPI2000'}, $DATA{'nh_NHPI2000'} ); $h_white = pct ($DATA{'h_white2005'} - $DATA{'h_white2000'},$DATA{'h_white2000'}); $h_black = pct ($DATA{'h_black2005'} - $DATA{'h_black2000'},$DATA{'h_black2000'}); $h_AIAN = pct ($DATA{'h_AIAN2005'} - $DATA{'h_AIAN2000'}, $DATA{'h_AIAN2000'} ); $h_asian = pct ($DATA{'h_asian2005'} - $DATA{'h_asian2000'},$DATA{'h_asian2000'}); $h_NHPI = pct ($DATA{'h_NHPI2005'} - $DATA{'h_NHPI2000'}, $DATA{'h_NHPI2000'} );

$fname =~ s/SC-EST2005-//i; $template =~ s/state/$state/gi; $template =~ s/fname/$fname/gi;

$template =~ s/p_white2000/$p_white2000/gi; $template =~ s/p_black2000/$p_black2000/gi; $template =~ s/p_AIAN2000/$p_AIAN2000/gi; $template =~ s/p_asian2000/$p_asian2000/gi; $template =~ s/p_NHPI2000/$p_NHPI2000/gi; $template =~ s/h_white2000/$h_white2000/gi; $template =~ s/h_black2000/$h_black2000/gi; $template =~ s/h_AIAN2000/$h_AIAN2000/gi; $template =~ s/h_asian2000/$h_asian2000/gi; $template =~ s/h_NHPI2000/$h_NHPI2000/gi; $template =~ s/p_white2005/$p_white2005/gi; $template =~ s/p_black2005/$p_black2005/gi; $template =~ s/p_AIAN2005/$p_AIAN2005/gi; $template =~ s/p_asian2005/$p_asian2005/gi; $template =~ s/p_NHPI2005/$p_NHPI2005/gi; $template =~ s/h_white2005/$h_white2005/gi; $template =~ s/h_black2005/$h_black2005/gi; $template =~ s/h_AIAN2005/$h_AIAN2005/gi; $template =~ s/h_asian2005/$h_asian2005/gi; $template =~ s/h_NHPI2005/$h_NHPI2005/gi; $template =~ s/d_white/$d_white/gi; $template =~ s/d_black/$d_black/gi; $template =~ s/d_AIAN/$d_AIAN/gi; $template =~ s/d_asian/$d_asian/gi; $template =~ s/d_NHPI/$d_NHPI/gi; $template =~ s/x_white/$x_white/gi; $template =~ s/x_black/$x_black/gi; $template =~ s/x_AIAN/$x_AIAN/gi; $template =~ s/x_asian/$x_asian/gi; $template =~ s/x_NHPI/$x_NHPI/gi; $template =~ s/h_white/$h_white/gi; $template =~ s/h_black/$h_black/gi; $template =~ s/h_AIAN/$h_AIAN/gi; $template =~ s/h_asian/$h_asian/gi; $template =~ s/h_NHPI/$h_NHPI/gi;

print OFILE $template; }

opendir(DIR, ".") or die "can't opendir $dirname: $!"; @files = readdir (DIR); closedir (DIR);

$outfile = "template.txt"; open (OFILE, "> $outfile"); print OFILE <<TOPOFILE;

EOFILE

close (OFILE);