latin2js.c

#include <u.h>
#include <libc.h>
#include <ctype.h>
#include <bio.h>

/* See also /sys/src/cmd/aux/mklatinkbd.c */

/* Definition from /sys/src/9/port/latin1.c */
/*
 * The code makes two assumptions: strlen(ld) is 1 or 2; latintab[i].ld can be a
 * prefix of latintab[j].ld only when j<i.
 */
struct cvlist
{
	char	*ld;		/* must be seen before using this conversion */
	char	*si;		/* options for last input characters */
	Rune	*so;		/* the corresponding Rune for each si entry */
} latintab[] = {
#include "/sys/src/9/port/latin1.h"
	0,	0,		0
};

/* Do Javascript strings have the same escapes as C? */
void
cprint(Biobuf *out, char *s)
{
	for(; *s; ++s){
		switch(*s){
		case '\'':
		case '\"':
		case '\\':
			Bprint(out, "\\%c", *s);
			break;
		case '\t':
			Bprint(out, "\\t");
			break;
		default:
			Bprint(out, "%c", *s);
			break;
		}
	}
}

/* Assumes no dangerous characters in rune string. */
/* XXX Large characters will break with Javascript's use of UTF-16! */

void
main(int argc, char **argv)
{
	struct cvlist *l;
	Biobuf out;
	char *name;

	if(Binit(&out, 1, OWRITE) == Beof)
		exits("Binit");

	if(argc >= 1)
		name = argv[0];
	else
		name = "latin2js";

	Bprint(&out, "/* Generated by %s. */\n", name);
	Bprint(&out, "var Composetab = {\n");

	for(l = latintab; l->ld; ++l){
		Bprint(&out, "\t\"");
		cprint(&out, l->ld);
		Bprint(&out, "\": {from:\"");
		cprint(&out, l->si);
		Bprint(&out, "\", to:\"");
		Bprint(&out, "%S", l->so);
		Bprint(&out, "\"},\n");
	}

	Bprint(&out, "};\n");

	exits("");
}