=================================================================== RCS file: /cvs/mandoc/roff_escape.c,v retrieving revision 1.6 retrieving revision 1.9 diff -u -p -r1.6 -r1.9 --- mandoc/roff_escape.c 2022/06/01 23:20:26 1.6 +++ mandoc/roff_escape.c 2022/06/05 10:19:54 1.9 @@ -1,4 +1,4 @@ -/* $OpenBSD: roff_escape.c,v 1.6 2022/06/01 23:20:26 schwarze Exp $ */ +/* $OpenBSD: roff_escape.c,v 1.9 2022/06/05 10:19:54 schwarze Exp $ */ /* * Copyright (c) 2011, 2012, 2013, 2014, 2015, 2017, 2018, 2020, 2022 * Ingo Schwarze @@ -41,7 +41,8 @@ mandoc_escape(const char **rendarg, const char **rarg, int iarg, iendarg, iend; enum mandoc_esc rval; - rval = roff_escape(--*rendarg, 0, 0, NULL, &iarg, &iendarg, &iend); + rval = roff_escape(--*rendarg, 0, 0, + NULL, NULL, &iarg, &iendarg, &iend); assert(rval != ESCAPE_EXPAND); if (rarg != NULL) *rarg = *rendarg + iarg; @@ -63,20 +64,20 @@ mandoc_escape(const char **rendarg, const char **rarg, */ enum mandoc_esc roff_escape(const char *buf, const int ln, const int aesc, - int *resc, int *rarg, int *rendarg, int *rend) + int *resc, int *rnam, int *rarg, int *rendarg, int *rend) { int iesc; /* index of leading escape char */ + int inam; /* index of escape name */ int iarg; /* index beginning the argument */ int iendarg; /* index right after the argument */ int iend; /* index right after the sequence */ - int sesc, sarg, sendarg, send; /* for sub-escape */ + int sesc, snam, sarg, sendarg, send; /* for sub-escape */ int maxl; /* expected length of the argument */ int argl; /* actual length of the argument */ int c, i; /* for \[char...] parsing */ int valid_A; /* for \A parsing */ enum mandoc_esc rval; /* return value */ enum mandocerr err; /* diagnostic code */ - char esc_name; char term; /* byte terminating the argument */ /* @@ -84,21 +85,21 @@ roff_escape(const char *buf, const int ln, const int a * it only makes a difference in copy mode. */ - iesc = iarg = aesc; + iesc = inam = aesc; do { - iarg++; - } while (buf[iarg] == 'E'); + inam++; + } while (buf[inam] == 'E'); /* * Sort the following cases first by syntax category, * then by escape sequence type, and finally by ASCII code. */ - esc_name = buf[iarg]; - iendarg = iend = ++iarg; + iarg = iendarg = iend = inam + 1; maxl = INT_MAX; term = '\0'; - switch (esc_name) { + err = MANDOCERR_OK; + switch (buf[inam]) { /* Escape sequences taking no arguments at all. */ @@ -127,6 +128,7 @@ roff_escape(const char *buf, const int ln, const int a case '\0': iendarg = --iend; /* FALLTHROUGH */ + case '.': case '\\': default: iarg--; @@ -136,7 +138,6 @@ roff_escape(const char *buf, const int ln, const int a case ' ': case '\'': case '-': - case '.': case '0': case ':': case '_': @@ -269,12 +270,12 @@ roff_escape(const char *buf, const int ln, const int a if ((term == '\b' || (term == '\0' && maxl == INT_MAX)) && buf[iarg] == buf[iesc] && roff_escape(buf, ln, iendarg, - &sesc, &sarg, &sendarg, &send) == ESCAPE_EXPAND) + &sesc, &snam, &sarg, &sendarg, &send) == ESCAPE_EXPAND) goto out_sub; if (term == '\b') { - if ((esc_name == 'N' && isdigit((unsigned char)buf[iarg])) || - (esc_name == 'h' && strchr(" %&()*+-./0123456789:<=>", + if ((buf[inam] == 'N' && isdigit((unsigned char)buf[iarg])) || + (buf[inam] == 'h' && strchr(" %&()*+-./0123456789:<=>", buf[iarg]) != NULL)) { iendarg = iend = iarg + 1; rval = ESCAPE_ERROR; @@ -282,7 +283,7 @@ roff_escape(const char *buf, const int ln, const int a } term = buf[iarg++]; } else if (term == '\0' && maxl == INT_MAX) { - if (esc_name == 'n' && (buf[iarg] == '+' || buf[iarg] == '-')) + if (buf[inam] == 'n' && (buf[iarg] == '+' || buf[iarg] == '-')) iarg++; switch (buf[iarg]) { case '(': @@ -310,22 +311,26 @@ roff_escape(const char *buf, const int ln, const int a while (maxl > 0) { if (buf[iendarg] == '\0') { /* Ignore an incomplete argument except for \w. */ - if (esc_name != 'w') + if (buf[inam] != 'w') iendarg = iarg; + if (rval == ESCAPE_EXPAND) + err = MANDOCERR_ESC_BAD; + else + rval = ESCAPE_ERROR; break; } if (buf[iendarg] == term) { iend = iendarg + 1; break; } - if (esc_name == 'N' && + if (buf[inam] == 'N' && isdigit((unsigned char)buf[iendarg]) == 0) { iend = iendarg + 1; break; } if (buf[iendarg] == buf[iesc]) { switch (roff_escape(buf, ln, iendarg, - &sesc, &sarg, &sendarg, &send)) { + &sesc, &snam, &sarg, &sendarg, &send)) { case ESCAPE_EXPAND: goto out_sub; case ESCAPE_UNDEF: @@ -343,14 +348,11 @@ roff_escape(const char *buf, const int ln, const int a iend = ++iendarg; } } - if (resc != NULL && ((maxl != INT_MAX && maxl != 0) || - (term != '\0' && buf[iendarg] != term))) - mandoc_msg(MANDOCERR_ESC_BAD, ln, iesc, "%s", buf + iesc); /* Post-process depending on the content of the argument. */ argl = iendarg - iarg; - switch (esc_name) { + switch (buf[inam]) { case '*': if (resc == NULL && argl == 2 && buf[iarg] == '.' && buf[iarg + 1] == 'T') @@ -449,12 +451,15 @@ roff_escape(const char *buf, const int ln, const int a out_sub: iesc = sesc; + inam = snam; iarg = sarg; iendarg = sendarg; iend = send; rval = ESCAPE_EXPAND; out: + if (rnam != NULL) + *rnam = inam; if (rarg != NULL) *rarg = iarg; if (rendarg != NULL) @@ -478,18 +483,17 @@ out: err = MANDOCERR_ESC_UNSUPP; break; case ESCAPE_UNDEF: - if (esc_name == '\\') - return rval; - err = MANDOCERR_ESC_UNDEF; + if (buf[inam] != '\\' && buf[inam] != '.') + err = MANDOCERR_ESC_UNDEF; break; case ESCAPE_SPECIAL: - if (mchars_spec2cp(buf + iarg, argl) >= 0) - return rval; - err = MANDOCERR_ESC_BAD; + if (mchars_spec2cp(buf + iarg, argl) < 0) + err = MANDOCERR_ESC_BAD; break; default: - return rval; + break; } - mandoc_msg(err, ln, iesc, "%.*s", iend - iesc, buf + iesc); + if (err != MANDOCERR_OK) + mandoc_msg(err, ln, iesc, "%.*s", iend - iesc, buf + iesc); return rval; }