From 9306cb60c32082c5403931de0823a9fd5daa196c Mon Sep 17 00:00:00 2001
From: Jason Katz-Brown <jason@airbnb.com>
Date: Sun, 25 Aug 2013 02:17:13 -0700
Subject: Initial git commit.

---
 new_language_generation/perl_tools/analyse_gcgs.pl | 56 ++++++++++++++++++++++
 1 file changed, 56 insertions(+)
 create mode 100755 new_language_generation/perl_tools/analyse_gcgs.pl

(limited to 'new_language_generation/perl_tools/analyse_gcgs.pl')

diff --git a/new_language_generation/perl_tools/analyse_gcgs.pl b/new_language_generation/perl_tools/analyse_gcgs.pl
new file mode 100755
index 0000000..94e7cce
--- /dev/null
+++ b/new_language_generation/perl_tools/analyse_gcgs.pl
@@ -0,0 +1,56 @@
+#!/usr/bin/perl
+
+# USAGE: analyse_gcgs.pl *.gcg > gcganalysis
+
+use warnings;
+use strict;
+
+my %letter_points;
+my %letter_counts;
+
+binmode STDOUT, ':utf8';
+
+sub read_gcgs {
+	my $games_count = 0;
+	for my $arg (@ARGV) {
+		open (my $input, "<:encoding(utf8)", $arg);
+		++$games_count;
+
+		while (<$input>) {
+			chomp;
+			next if (!/^\>/);
+
+			my ($player, $rack, $position, $play, $score, $cumulativescore) = split /\s+/, $_;
+
+			next if (! defined $cumulativescore);
+			if ($play =~ /\[/) {
+				print "blank play! $play for $score\n";
+				next;
+			}
+
+			for my $letter (split /|/, $play) {
+				++$letter_counts{$letter};
+				$letter_points{$letter} += int($score);
+
+				#print "$letter now seen " . $letter_counts{$letter} . " times, scored " . $letter_points{$letter} . " points\n";
+			}
+		}
+	}
+
+	print "# Read $games_count games.\n";
+}
+
+sub spit_analysis {
+	my %points_per_turn;
+	for my $letter (keys %letter_counts) {
+		$points_per_turn{$letter} = $letter_points{$letter} / $letter_counts{$letter};
+	}
+
+	for my $letter (sort { $points_per_turn{$b} <=> $points_per_turn{$a} } keys %points_per_turn) {
+		# TODO lame that this always puts out un-bar-surrounded letters
+		print "$letter " . $points_per_turn{$letter} . "\n";
+	}
+}
+
+read_gcgs();
+spit_analysis();
-- 
cgit v1.2.3