view uptools/libcoding/ucs2_decode.c @ 926:6a0aa8d36d06

rvinterf backslash escape: introduce libprint The new helper function library named libprint is meant to replace the badly misnamed libg23, and will soon contain functions for printing all of the same kinds of GPF TST packets that are now handled in libg23. However, we are also moving safe_print_trace() from libasync to this new library, and changing it to emit our new backslash escape format.
author Mychaela Falconia <falcon@freecalypso.org>
date Tue, 23 May 2023 03:47:46 +0000
parents 30fbaa652ea5
children
line wrap: on
line source

/*
 * This library module implements the conversion of UCS2-encoded data
 * (typically received in SMS) into ASCII, ISO 8859-1 or UTF-8,
 * maintaining parallelism with the corresponding function for decoding
 * GSM7-encoded data.
 */

#include <sys/types.h>
#include <stdio.h>

ucs2_to_ascii_or_ext(inbuf, inlen, outbuf, outlenp, ascii_ext, newline_ok)
	u_char *inbuf, *outbuf;
	unsigned inlen, *outlenp;
{
	u_char *inp, *endp, *outp;
	unsigned uni;

	inp = inbuf;
	endp = inbuf + (inlen & ~1);
	outp = outbuf;
	while (inp < endp) {
		if ((endp - inp) >= 4 && (inp[0] & 0xFC) == 0xD8 &&
		    (inp[2] & 0xFC) == 0xDC) {
			uni = ((inp[0] & 3) << 18) | (inp[1] << 10) |
			      ((inp[2] & 3) << 8) | inp[3];
			inp += 4;
			uni += 0x10000;
			if (ascii_ext == 2)
				outp += emit_utf8_char(uni, outp);
			else {
				sprintf(outp, "\\U%06X", uni);
				outp += 8;
			}
			continue;
		}
		uni = (inp[0] << 8) | inp[1];
		inp += 2;
		if (uni == '\\') {
			*outp++ = '\\';
			*outp++ = '\\';
		} else if (uni == '\r') {
			*outp++ = '\\';
			*outp++ = 'r';
		} else if (uni == '\n') {
			if (newline_ok)
				*outp++ = '\n';
			else {
				*outp++ = '\\';
				*outp++ = 'n';
			}
		} else if (!is_decoded_char_ok(uni, ascii_ext)) {
			sprintf(outp, "\\u%04X", uni);
			outp += 6;
		} else if (ascii_ext == 2)
			outp += emit_utf8_char(uni, outp);
		else
			*outp++ = uni;
	}
	*outp = '\0';
	if (outlenp)
		*outlenp = outp - outbuf;
}