bkt (bucketizer) initial commit

1 files changed, 82 insertions, 0 deletions

diff --git a/bkt b/bkt
new file mode 100755
index 0000000..42ea6ea
--- /dev/null
+++ b/bkt
@@ -0,0 +1,82 @@
+#!/usr/bin/perl -w
+
+use Getopt::Std;
+$Getopt::Std::STANDARD_HELP_VERSION++;
+
+($SELF = $0) =~ s,.*/,,;
+$VERSION="0.0.0";
+
+sub HELP_MESSAGE {
+	print <<EOF;
+$SELF - count repeats in input
+
+Given the following input:
+
+foo
+foo
+bar
+bar
+baz
+
+$SELF will output:
+
+bar	2	40.0%
+baz	1	20.0%
+foo	2	40.0%
+
+Usage: $SELF <options> <file> ...
+
+Options are:
+  --help
+  -h        display this help message
+  -c        show counts only (suppress percentages)
+  -p        show percentages only (suppress counts)
+  -w        remove leading and trailing whitespace from input lines
+  -W        remove ALL whitespace from input lines
+  -t        show total count
+  -e code   execute perl code for each line of input (should modify \$_,
+            make sure you quote the argument as needed by your shell)
+
+Input will be read from filenames given on the command line, or from
+standard input if none given. The input need not be sorted.
+EOF
+}
+
+sub VERSION_MESSAGE {
+	print "$SELF $VERSION\n";
+}
+
+getopts('hcpwWte:', \%opt);
+
+HELP_MESSAGE() if $opt{h};
+
+die "$SELF: can't give -c and -p options together\n"
+	if $opt{c} && $opt{p};
+
+%counts = ();
+$total = 0;
+
+while(<>) {
+	chomp;
+	if($opt{e}) {
+		eval $opt{e};
+		die "$SELF: $@" if $@;
+	}
+	s/^\s+|\s+$//g if $opt{w};
+	s/\s//g if $opt{W};
+	$counts{$_}++;
+	$total++;
+}
+
+for(sort keys %counts) {
+	print $_ . "\t";
+	print $counts{$_} . "\t" unless $opt{p};
+	printf "%.1f%%", ($counts{$_} * 100 / $total) unless $opt{c};
+	print "\n";
+}
+
+if($opt{t}) {
+	print "\n-- Total count: $total\n";
+}
+
+exit(0);