A patch for pisg to improve its smiley parsing with a lot more smileys.

pisg_smileys.patch

From aa0d9705564580ad6059e9e9e793a25764dd3a9f Mon Sep 17 00:00:00 2001
From: Andrew <[email protected]>
Date: Mon, 9 Mar 2015 20:53:59 -0400
Subject: [PATCH] Updated smiley parsing code

---
 pisg/modules/Pisg/Parser/Logfile.pm |   51 ++++++++++++++++++++++++++++++-----
 1 file changed, 44 insertions(+), 7 deletions(-)

diff --git a/pisg/modules/Pisg/Parser/Logfile.pm b/pisg/modules/Pisg/Parser/Logfile.pm
index 63c756c..ad5d913 100644
--- a/pisg/modules/Pisg/Parser/Logfile.pm
+++ b/pisg/modules/Pisg/Parser/Logfile.pm
@@ -393,6 +393,7 @@ sub _parse_file
 
                     if ($saying !~ /[a-z]/o && $saying =~ /[A-Z]/o) {
                         # Ignore single smileys on a line. eg. '<user> :P'
+                        # TODO: Use the same smiley check as our smiley counter below
                         if ($saying !~ /^[8;:=][ ^-o]?[)pPD\}\]>]$/o) {
                             $stats->{allcaps}{$nick}++;
                             push @{ $lines->{allcaplines}{$nick} }, $line;
@@ -405,18 +406,54 @@ sub _parse_file
                     }
 
                     # Who smiles the most?
-                    my $e = '[8;:=%]'; # eyes
-                    my $n = '[-oc*^]'; # nose
-                    # smileys including asian-style (^^ ^_^' ^^; \o/)
-                    if ($saying =~ /(>?$e'?$n[\)pPD\}\]>]|[\(\{\[<]$n'?$e<?|[;:][\)pPD\}\]\>]|\([;:]|\^[_o-]*\^[';]|\\[o.]\/)/o) {
+                    my $e = '[8;:=%Xx]'; # eyes
+                    my $n = '[-oc*^v]'; # nose
+                    my $nm_a = '[-_o.~3]'; # eastern nose / mouth middles
+                    my $a_sw = '[\'";]'; # eastern face modifiers, like sweat drops
+
+                    # TODO: allow matching of lines with URLs and smileys in them
+                    my $url_regex = qr/\w+:\/\//;
+
+                    my $wholeword_start = '(?<![\w?&])'; # don't allow in middle of word or as query strings
+                    my $wholeword_end = '(?![\w])'; # prevent matching things that start `words`: (xD) will match xD, not (x
+
+                    # [\)pPD\}\]>]|[\(\{\[<]
+                    my $bi_hm = '*'; # happy mouths that work either way
+                    my $hm  = '[DPp\)\]\}>3'.$bi_hm.']'; # Directional happy mouths (LTR -->)
+                    my $rhm = '[CcL\(\[\{<'.$bi_hm.']'; # Directional happy mouths (RTL <--)
+                    my $he_a = '\^'; # happy eastern eyes
+
+                    # Happy faces, smileys including eastern-style (^^ ^_^' ^^; \o/)
+                    # Assumed here that tears are happy tears.
+                    if ($saying =~ qr/$wholeword_start(>?$e'?$n?$hm|$rhm$n?'?$e<?|($he_a)$nm_a*\g{-1}$a_sw?|\\[o.]\/)$wholeword_end/) {
                         $stats->{smiles}{$nick}++;
                         $stats->{smileys}{$1}++;
                         $stats->{smileynicks}{$1} = $nick;
                     }
 
-                    # asian frown: ;_;
-                    if ($saying =~ /($e'?$n[\(\[\\\/\{|]|[\)\]\\\/\}|]$n'?$e|[;:][\(\/]|[\)D]:|;_+;|T_+T|-[._]+-)/o and
-                        $saying !~ /\w+:\/\//o) {
+                    # Ambiguous faces. Neither happy nor sad.
+                    # There are no tears because tears + ambiguous = sad.
+                    # Ambiguous mouths would be like. :I :O :o :0 :B :F :U :u :V :v
+                    my $bi_am = 'FIOo0TUuVvn'; # ambiguous mouths
+                    my $am  = '[BF'.$bi_am.']'; # directional ambiguous mouths (LTR -->)
+                    my $ram = '['.$bi_am.']'; # directional ambiguous mouths (RTL <--)
+                    my $o_eyes = '[oO0]'; # for mismatched faces
+                    my $ae_a = '[.oO09]'; # ambiguous eastern eyes. 
+                    # 'o' is a tricky one; its meaning changes with mouth (o_o, o3o o~o) but I'm not dealing with that
+
+                    if ($saying =~ qr/$wholeword_start(>?$e$n?$am|$ram$n?$e<?|($ae_a)(?!\g{-1})$nm_a+\g{-1}$a_sw?|($o_eyes)(?!\g{-1})$nm_a+$o_eyes)$wholeword_end/) {
+                        $stats->{smileys}{$1}++;
+                        $stats->{smileynicks}{$1} = $nick;
+                    }
+
+                    # Sad faces
+                    # X x \ / L S s ( { [ < C c
+                    my $bi_sm = 'XxSs\/|\\'; # sad mouths that work either way
+                    my $sm  = '[CcL\(\[\{<'.$bi_sm.']'; # directional sad mouths (LTR -->)
+                    my $rsm = '[\)\]\}>3'.$bi_sm.']'; # directional sad mouths (RTL <--)
+                    my $se_a = '[-;TQ><]'; # sad eastern eyes
+                    # Frowny faces. >mouth< is hardcoded at end because it's weird
+                    if ($saying =~ qr/$wholeword_start(>?$e'?$n?$sm|$rsm$n?'?$e<?|($se_a)(?!\g{-1})$nm_a+\g{-1}$a_sw?|>$nm_a+<)$wholeword_end/) {
                         $stats->{frowns}{$nick}++;
                         $stats->{smileys}{$1}++;
                         $stats->{smileynicks}{$1} = $nick;
-- 
1.7.10.4