=================================================================== RCS file: /cvs/mandoc/mandoc.c,v retrieving revision 1.50 retrieving revision 1.54 diff -u -p -r1.50 -r1.54 --- mandoc/mandoc.c 2011/05/14 16:06:09 1.50 +++ mandoc/mandoc.c 2011/07/21 15:21:13 1.54 @@ -1,4 +1,4 @@ -/* $Id: mandoc.c,v 1.50 2011/05/14 16:06:09 kristaps Exp $ */ +/* $Id: mandoc.c,v 1.54 2011/07/21 15:21:13 kristaps Exp $ */ /* * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons * Copyright (c) 2011 Ingo Schwarze @@ -125,6 +125,14 @@ mandoc_escape(const char **end, const char **start, in break; case ('['): gly = ESCAPE_SPECIAL; + /* + * Unicode escapes are defined in groff as \[uXXXX] to + * \[u10FFFF], where the contained value must be a valid + * Unicode codepoint. Here, however, only check whether + * it's not a zero-width escape. + */ + if ('u' == cp[i] && ']' != cp[i + 1]) + gly = ESCAPE_UNICODE; term = ']'; break; case ('C'): @@ -156,10 +164,6 @@ mandoc_escape(const char **end, const char **start, in if (ESCAPE_ERROR == gly) gly = ESCAPE_IGNORE; /* FALLTHROUGH */ - case ('*'): - if (ESCAPE_ERROR == gly) - gly = ESCAPE_PREDEF; - /* FALLTHROUGH */ case ('f'): if (ESCAPE_ERROR == gly) gly = ESCAPE_FONT; @@ -694,7 +698,7 @@ mandoc_getcontrol(const char *cp, int *ppos) * If the string is invalid, or is less than 0, return -1. */ int -mandoc_strntou(const char *p, size_t sz, int base) +mandoc_strntoi(const char *p, size_t sz, int base) { char buf[32]; char *ep; @@ -704,7 +708,7 @@ mandoc_strntou(const char *p, size_t sz, int base) return(-1); memcpy(buf, p, sz); - buf[sz] = '\0'; + buf[(int)sz] = '\0'; errno = 0; v = strtol(buf, &ep, base); @@ -712,11 +716,10 @@ mandoc_strntou(const char *p, size_t sz, int base) if (buf[0] == '\0' || *ep != '\0') return(-1); - if ((errno == ERANGE && - (v == LONG_MAX || v == LONG_MIN)) || - (v > INT_MAX || v < 0)) - return(-1); + if (v > INT_MAX) + v = INT_MAX; + if (v < INT_MIN) + v = INT_MIN; return((int)v); } -