Build encoding tables from textual descriptions rather than hardcoded…

… lines of C

.bzrignore

@@ -2,3 +2,4 @@ t/test
 t/suites.h
 t/externs.h
 t/harness
+src/encoding/*.inc

Makefile

@@ -17,6 +17,9 @@ CFILES=$(wildcard src/*.c)
 OFILES=$(CFILES:.c=.o)
 HFILES=$(wildcard include/*.h)
 
+TBLFILES=$(wildcard src/encoding/*.tbl)
+INCFILES=$(TBLFILES:.tbl=.inc)
+
 HFILES_INT=$(wildcard src/*.h) $(HFILES)
 
 LIBPIECES=vterm parser encoding state input pen unicode
@@ -32,6 +35,11 @@ libvterm.so: $(addprefix src/, $(addsuffix .o, $(LIBPIECES)))
 src/%.o: src/%.c $(HFILES_INT)
 	gcc -fPIC -o $@ -c $< $(CCFLAGS)
 
+src/encoding/%.inc: src/encoding/%.tbl
+	perl -C tbl2inc_c.pl $< >$@
+
+src/encoding.o: $(INCFILES)
+
 # Need first to cancel the implict rule
 %.o: %.c
 

src/encoding.c

@@ -197,12 +197,7 @@ static int decode_table(VTermEncoding *enc, uint32_t cp[], int *cpi, int cplen,
   return 1;
 }
 
-static const struct StaticTableEncoding encoding_uk = {
-  { .decode = &decode_table },
-  {
-    ['$'] = 0x00a3,
-  }
-};
+#include "encoding/uk.inc"
 
 static struct {
   VTermEncodingType type;

src/encoding/uk.tbl

@@ -0,0 +1 @@
+2/1 = "£"

tbl2inc_c.pl

@@ -0,0 +1,30 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+
+my ( $encname ) = $ARGV[0] =~ m{/([^/.]+).tbl}
+   or die "Cannot parse encoding name out of $ARGV[0]\n";
+
+print <<"EOF";
+static const struct StaticTableEncoding encoding_$encname = {
+  { .decode = &decode_table },
+  {
+EOF
+
+while( <> ) {
+   s/\s*#.*//; # strip comment
+
+   s{^(\d+)/(\d+)}{sprintf "[0x%02x]", $1*16 + $2}e; # Convert 3/1 to [0x31]
+   s{"(.)"}{sprintf "0x%04x", ord $1}e;              # Convert "A" to 0x41
+   s{U\+}{0x};                                       # Convert U+0041 to 0x0041
+
+   s{$}{,}; # append comma
+
+   print "    $_";
+}
+
+print <<"EOF";
+  }
+};
+EOF