file/src/encoding.c

66  * ubuf, and the number of characters converted in ulen.
69 file_encoding(struct magic_set *ms, const unsigned char *buf, size_t nbytes, unichar **ubuf, size_t *ulen, const char **code, const char **code_mime, const char **type)
76 	*ulen = 0;
91 	if (looks_ascii(buf, nbytes, *ubuf, ulen)) {
92 		if (looks_utf7(buf, nbytes, *ubuf, ulen) > 0) {
93 			DPRINTF(("utf-7 %" SIZE_T_FORMAT "u\n", *ulen));
97 			DPRINTF(("ascii %" SIZE_T_FORMAT "u\n", *ulen));
101 	} else if (looks_utf8_with_BOM(buf, nbytes, *ubuf, ulen) > 0) {
102 		DPRINTF(("utf8/bom %" SIZE_T_FORMAT "u\n", *ulen));
105 	} else if (file_looks_utf8(buf, nbytes, *ubuf, ulen) > 1) {
106 		DPRINTF(("utf8 %" SIZE_T_FORMAT "u\n", *ulen));
109 	} else if ((ucs_type = looks_ucs16(buf, nbytes, *ubuf, ulen)) != 0) {
117 		DPRINTF(("ucs16 %" SIZE_T_FORMAT "u\n", *ulen));
118 	} else if (looks_latin1(buf, nbytes, *ubuf, ulen)) {
119 		DPRINTF(("latin1 %" SIZE_T_FORMAT "u\n", *ulen));
122 	} else if (looks_extended(buf, nbytes, *ubuf, ulen)) {
123 		DPRINTF(("extended %" SIZE_T_FORMAT "u\n", *ulen));
129 		if (looks_ascii(nbuf, nbytes, *ubuf, ulen)) {
130 			DPRINTF(("ebcdic %" SIZE_T_FORMAT "u\n", *ulen));
133 		} else if (looks_latin1(nbuf, nbytes, *ubuf, ulen)) {
135 			    *ulen));
232     size_t *ulen)
236 	*ulen = 0;
244 		ubuf[(*ulen)++] = buf[i];
251 looks_latin1(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
255 	*ulen = 0;
263 		ubuf[(*ulen)++] = buf[i];
271     size_t *ulen)
275 	*ulen = 0;
283 		ubuf[(*ulen)++] = buf[i];
297  * If ubuf is non-NULL on entry, text is decoded into ubuf, *ulen;
301 file_looks_utf8(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
309 		*ulen = 0;
322 				ubuf[(*ulen)++] = buf[i];
358 				ubuf[(*ulen)++] = c;
373     size_t *ulen)
376 		return file_looks_utf8(buf + 3, nbytes - 3, ubuf, ulen);
382 looks_utf7(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
391 				*ulen = 0;
402     size_t *ulen)
417 	*ulen = 0;
423 			ubuf[(*ulen)++] = buf[i + 1] + 256 * buf[i];
425 			ubuf[(*ulen)++] = buf[i] + 256 * buf[i + 1];
427 		if (ubuf[*ulen - 1] == 0xfffe)
429 		if (ubuf[*ulen - 1] < 128 &&
430 		    text_chars[(size_t)ubuf[*ulen - 1]] != T)