- state = 0;
- accum = 0U;
-
- for (i = *ii; i < ib->sz; i++) {
- cu = ib->buf[i];
- if (state) {
- if ( ! (cu & 128) || (cu & 64)) {
- /* Bad sequence header. */
- break;
- }
-
- /* Accept only legitimate bit patterns. */
-
- if (cu > 191 || cu < 128) {
- /* Bad in-sequence bits. */
- break;
- }
-
- accum |= (cu & 63) << --state * 6;
-
- if (state)
- continue;
-
- if (accum < 0x80)
- ob->buf[(*oi)++] = accum;
- else
- *oi += snprintf(ob->buf + *oi,
- 11, "\\[u%.4X]", accum);
- *ii = i + 1;
- *filenc &= ~MPARSE_LATIN1;
- return(1);
- } else {
- /*
- * Entering a UTF-8 state: if we encounter a
- * UTF-8 bitmask, calculate the expected UTF-8
- * state from it.
- */
- for (state = 0; state < 7; state++)
- if ( ! (cu & (1 << (7 - state))))
- break;
-
- /* Accept only legitimate bit patterns. */
-
- switch (state--) {
- case (4):
- if (cu <= 244 && cu >= 240) {
- accum = (cu & 7) << 18;
- continue;
- }
- /* Bad 4-sequence start bits. */
- break;
- case (3):
- if (cu <= 239 && cu >= 224) {
- accum = (cu & 15) << 12;
- continue;
- }
- /* Bad 3-sequence start bits. */
- break;
- case (2):
- if (cu <= 223 && cu >= 194) {
- accum = (cu & 31) << 6;
- continue;
- }
- /* Bad 2-sequence start bits. */
- break;
- default:
- /* Bad sequence bit mask. */
- break;
- }
- break;
- }
+ nby = 1;
+ while (nby < 5 && *cu & (1 << (7 - nby)))
+ nby++;
+
+ switch (nby) {
+ case 2:
+ accum = *cu & 0x1f;
+ if (accum < 0x02) /* Obfuscated ASCII. */
+ goto latin;
+ break;
+ case 3:
+ accum = *cu & 0x0f;
+ break;
+ case 4:
+ accum = *cu & 0x07;
+ if (accum > 0x04) /* Beyond Unicode. */
+ goto latin;
+ break;
+ default: /* Bad sequence header. */
+ goto latin;