From 0c9d5cd332b3a9124f75926de5094c90b353e238 Mon Sep 17 00:00:00 2001
From: Mischa POSLAWSKY <perl@shiar.org>
Date: Sat, 22 Apr 2017 12:32:37 +0200
Subject: [PATCH] charset: generic glyph replacement from metadata

Common format with full support for partial overlaps.
---
 charset-encoding.inc.pl | 23 ++++++++++++++++--
 charset.plp             | 53 ++++++++++++++++++++---------------------
 2 files changed, 47 insertions(+), 29 deletions(-)

diff --git a/charset-encoding.inc.pl b/charset-encoding.inc.pl
index eee5aa2..71ba2ba 100644
--- a/charset-encoding.inc.pl
+++ b/charset-encoding.inc.pl
@@ -1,5 +1,6 @@
 use 5.014;
 use warnings;
+use utf8;
 
 +{
 	default    => [qw( u+0-27F utf8+realsize iso-8859-1 iso-8859-15 cp1252 cp437 cp850 )],
@@ -45,12 +46,30 @@ use warnings;
 	11 => 'thai',
 
 	'ascii'        => {inherit => ['' => '00-7F']},
-	'cp437'        => {inherit => ['cp850' => 0, 'ascii' => '00-1F+80']}, # ascii range overridden later
+	'cp437'        => {
+		inherit => ['cp850' => 0, 'ascii' => '00-1F+80'], # ascii range overridden later
+		replace => {
+			0    => ' ☺☻♥♦♣♠•◘○◙♂♀♪♫☼►◄↕‼¶§▬↨↑↓→←∟↔▲▼', # visible variants of control characters
+			0xED => 'ϕ', # non-greek usage and appearance
+		},
+	},
 	'gsm0338'      => {inherit => ['ascii' => '00-7F']},
 	'dingbats'     => {inherit => ['' => '20-7F+A0']},
 	'macdingbats'  => {inherit => ['dingbats' => '80-9F']},
 	'adobezdingbat'=> {inherit => ['MacDingbats' => '80-9F']}, # should be identical but maps to private use
-	'symbol'       => {inherit => ['' => '20-7F+A0']},
+	'symbol'       => {
+		inherit => ['' => '20-7F+A0'],
+		replace => {
+			0x60 => '│', # replace radical extender by closest unicode equivalent
+			0xBD => '⏐⎯', # arrow extenders
+			0xD2 => '®©™', # serif variants
+			0xE0 => '◊', # replace lookalike, should match AdobeSymbol
+			0xE2 => '®©™', # sans-serif variants
+			0xE6 => '⎛⎜⎝⎡⎢⎣⎧⎨⎩⎪',
+			0xF0 => '€',
+			0xF4 => '⎮⌡⎞⎟⎠⎤⎥⎦⎫⎬⎭',
+		},
+	},
 	'adobesymbol'  => {inherit => ['symbol' => '20-7F+A0', '' => '20-7F+A0']}, # minor differences, irrelevant except for different '€'
 	'wingdings'    => {inherit => ['' => '20'], setup => sub {require Encode::Wingdings}},
 	'wingdings2'   => {inherit => ['' => '20'], setup => sub {require Encode::Wingdings2}},
diff --git a/charset.plp b/charset.plp
index ed495fe..7f19bc1 100644
--- a/charset.plp
+++ b/charset.plp
@@ -164,33 +164,6 @@ sub tabinput {
 			$row{table} = Encode::decode($row{set}, pack 'C*', $row{offset} .. $row{endpoint});
 		}
 
-		if ($row{set} eq 'cp437') {
-			if ($row{offset} <= 0xED and $row{endpoint} >= 0xED) {
-				# replace phi glyph
-				substr($row{table}, 0xED - $row{offset}, 1) = 'ϕ';
-			}
-			if ($row{offset} < 0x20) {
-				# replace control characters by visible variants
-				my $sub = substr ' ☺☻♥♦♣♠•◘○◙♂♀♪♫☼►◄↕‼¶§▬↨↑↓→←∟↔▲▼', $row{offset};
-				substr($row{table}, 0, length $sub) = $sub;
-			}
-		}
-		elsif ($row{set} eq 'symbol') {
-			if ($row{offset} <= 0x60 and $row{endpoint} >= 0x60) {
-				# replace radical extender by closest unicode equivalent
-				substr($row{table}, 0x60 - $row{offset},  1) = '│';
-			}
-			if ($row{offset} <= 0xBD and $row{endpoint} >= 0xFF) {
-				substr($row{table}, 0xBD - $row{offset},  2) = '⏐⎯'; # arrow extenders
-				substr($row{table}, 0xD2 - $row{offset},  3) = '®©™'; # serif variants
-				substr($row{table}, 0xE0 - $row{offset},  1) = '◊'; # replace lookalike, should match AdobeSymbol
-				substr($row{table}, 0xE2 - $row{offset},  3) = '®©™'; # sans-serif variants
-				substr($row{table}, 0xE6 - $row{offset}, 10) = '⎛⎜⎝⎡⎢⎣⎧⎨⎩⎪';
-				substr($row{table}, 0xF0 - $row{offset},  1) = '€';
-				substr($row{table}, 0xF4 - $row{offset}, 11) = '⎮⌡⎞⎟⎠⎤⎥⎦⎫⎬⎭';
-			}
-		}
-
 		$row{endpoint} -= $row{offset};
 
 		$visible->{ascii} =  # assume common base
@@ -200,6 +173,32 @@ sub tabinput {
 		Alert("Encoding <q>$input</q> unknown");
 		return;
 	}
+
+	if (my $replace = $charset->{replace}) {
+		while (my ($offset, $sub) = each %{$replace}) {
+			$offset -= $row{offset};
+
+			my $length = length $sub;
+
+			if ($offset < 0) {
+				$offset > -$length or next; # at least one character after start
+				# trim leftmost part to start at offset
+				substr($sub, 0, -$offset) = '';
+				$length += $offset;
+				$offset = 0;
+			}
+
+			if ((my $excess = $row{endpoint} - $offset - $length + 1) < 0) {
+				$excess > -$length or next;
+				# trim rightmost part to prevent overflow
+				substr($sub, $excess) = '';
+				$length += $excess;
+			}
+
+			substr($row{table}, $offset, $length) = $sub;
+		}
+	}
+
 	push @request, \%row;
 }
 tabinput($_) for @tablist;
-- 
2.30.0