=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.208 retrieving revision 1.217 diff -u -p -r1.208 -r1.217 --- mandoc/roff.c 2014/04/20 19:40:13 1.208 +++ mandoc/roff.c 2014/07/04 16:12:08 1.217 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.208 2014/04/20 19:40:13 schwarze Exp $ */ +/* $Id: roff.c,v 1.217 2014/07/04 16:12:08 schwarze Exp $ */ /* * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons * Copyright (c) 2010-2014 Ingo Schwarze @@ -187,7 +187,7 @@ static int roff_evalstrcond(const char *, int *); static void roff_free1(struct roff *); static void roff_freereg(struct roffreg *); static void roff_freestr(struct roffkv *); -static char *roff_getname(struct roff *, char **, int, int); +static size_t roff_getname(struct roff *, char **, int, int); static int roff_getnum(const char *, int *, int *); static int roff_getop(const char *, int *, char *); static int roff_getregn(const struct roff *, @@ -200,7 +200,8 @@ static enum rofferr roff_line_ignore(ROFF_ARGS); static enum rofferr roff_nr(ROFF_ARGS); static void roff_openeqn(struct roff *, const char *, int, int, const char *); -static enum rofft roff_parse(struct roff *, const char *, int *); +static enum rofft roff_parse(struct roff *, char *, int *, + int, int); static enum rofferr roff_parsetext(char **, size_t *, int, int *); static enum rofferr roff_res(struct roff *, char **, size_t *, int, int); @@ -490,14 +491,13 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int { char ubuf[24]; /* buffer to print the number */ const char *start; /* start of the string to process */ - const char *stesc; /* start of an escape sequence ('\\') */ + char *stesc; /* start of an escape sequence ('\\') */ const char *stnam; /* start of the name, after "[(*" */ const char *cp; /* end of the name, e.g. before ']' */ const char *res; /* the string to be substituted */ char *nbuf; /* new buffer to copy bufp to */ size_t maxl; /* expected length of the escape name */ size_t naml; /* actual length of the escape name */ - size_t ressz; /* size of the replacement string */ int expand_count; /* to avoid infinite loops */ int npos; /* position in numeric expression */ int irc; /* return code from roff_evalnum() */ @@ -520,7 +520,7 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int break; if (0 == (stesc - cp) % 2) { - stesc = cp; + stesc = (char *)cp; continue; } @@ -628,21 +628,17 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int ln, (int)(stesc - *bufp), NULL); res = ""; } - ressz = strlen(res); /* Replace the escape sequence by the string. */ - *szp += ressz + 1; - nbuf = mandoc_malloc(*szp); + *stesc = '\0'; + *szp = mandoc_asprintf(&nbuf, "%s%s%s", + *bufp, res, cp) + 1; - strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1)); - strlcat(nbuf, res, *szp); - strlcat(nbuf, cp, *szp); - /* Prepare for the next replacement. */ start = nbuf + pos; - stesc = nbuf + (stesc - *bufp) + ressz; + stesc = nbuf + (stesc - *bufp) + strlen(res); free(*bufp); *bufp = nbuf; } @@ -769,7 +765,7 @@ roff_parseln(struct roff *r, int ln, char **bufp, * the compilers handle it. */ - if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) + if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos))) return(ROFF_CONT); assert(roffs[t].proc); @@ -802,29 +798,27 @@ roff_endparse(struct roff *r) * form of ".foo xxx" in the usual way. */ static enum rofft -roff_parse(struct roff *r, const char *buf, int *pos) +roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos) { + char *cp; const char *mac; size_t maclen; enum rofft t; - if ('\0' == buf[*pos] || '"' == buf[*pos] || - '\t' == buf[*pos] || ' ' == buf[*pos]) + cp = buf + *pos; + + if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp) return(ROFF_MAX); - /* We stop the macro parse at an escape, tab, space, or nil. */ + mac = cp; + maclen = roff_getname(r, &cp, ln, ppos); - mac = buf + *pos; - maclen = strcspn(mac, " \\\t\0"); - t = (r->current_string = roff_getstrn(r, mac, maclen)) ? ROFF_USERDEF : roffhash_find(mac, maclen); - *pos += (int)maclen; + if (ROFF_MAX != t) + *pos = cp - buf; - while (buf[*pos] && ' ' == buf[*pos]) - (*pos)++; - return(t); } @@ -862,7 +856,8 @@ roff_cblock(ROFF_ARGS) } if ((*bufp)[pos]) - mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL); + mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos, + ".. %s", *bufp + pos); roffnode_pop(r); roffnode_cleanscope(r); @@ -915,15 +910,16 @@ roff_ccond(struct roff *r, int ln, int ppos) static enum rofferr roff_block(ROFF_ARGS) { - int sv; - size_t sz; - char *name; + char *name, *cp; + size_t namesz; - name = NULL; + name = cp = *bufp + pos; + namesz = 0; if (ROFF_ig != tok) { - if ('\0' == (*bufp)[pos]) { - mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL); + if ('\0' == *cp) { + mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse, + ln, ppos, roffs[tok].name); return(ROFF_IGN); } @@ -934,19 +930,15 @@ roff_block(ROFF_ARGS) if (ROFF_de1 == tok) tok = ROFF_de; - if (ROFF_de == tok) - name = *bufp + pos; - else + else if (ROFF_de != tok) mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, roffs[tok].name); - while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos])) - pos++; + namesz = roff_getname(r, &cp, ln, ppos); + name[namesz] = '\0'; + } else + name = NULL; - while (isspace((unsigned char)(*bufp)[pos])) - (*bufp)[pos++] = '\0'; - } - roffnode_push(r, tok, name, ln, ppos); /* @@ -955,38 +947,23 @@ roff_block(ROFF_ARGS) * appended from roff_block_text() in multiline mode. */ - if (ROFF_de == tok) - roff_setstr(r, name, "", 0); + if (namesz && ROFF_de == tok) + roff_setstrn(&r->strtab, name, namesz, "", 0, 0); - if ('\0' == (*bufp)[pos]) + if ('\0' == *cp) return(ROFF_IGN); /* If present, process the custom end-of-line marker. */ - sv = pos; - while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos])) - pos++; + name = cp; + namesz = roff_getname(r, &cp, ln, ppos); + if (namesz) + r->last->end = mandoc_strndup(name, namesz); - /* - * Note: groff does NOT like escape characters in the input. - * Instead of detecting this, we're just going to let it fly and - * to hell with it. - */ + if ('\0' != *cp) + mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse, + ln, pos, ".%s ... %s", roffs[tok].name, cp); - assert(pos > sv); - sz = (size_t)(pos - sv); - - if (1 == sz && '.' == (*bufp)[sv]) - return(ROFF_IGN); - - r->last->end = mandoc_malloc(sz + 1); - - memcpy(r->last->end, *bufp + sv, sz); - r->last->end[(int)sz] = '\0'; - - if ((*bufp)[pos]) - mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL); - return(ROFF_IGN); } @@ -1021,7 +998,7 @@ roff_block_sub(ROFF_ARGS) i++; pos = i; - if (ROFF_MAX != roff_parse(r, *bufp, &pos)) + if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos)) return(ROFF_RERUN); return(ROFF_IGN); } @@ -1032,7 +1009,7 @@ roff_block_sub(ROFF_ARGS) * pulling it out of the hashtable. */ - t = roff_parse(r, *bufp, &pos); + t = roff_parse(r, *bufp, &pos, ln, ppos); /* * Macros other than block-end are only significant @@ -1067,7 +1044,7 @@ roff_cond_sub(ROFF_ARGS) rr = r->last->rule; roffnode_cleanscope(r); - t = roff_parse(r, *bufp, &pos); + t = roff_parse(r, *bufp, &pos, ln, ppos); /* * Fully handle known macros when they are structurally @@ -1311,7 +1288,8 @@ roff_cond(ROFF_ARGS) */ if ('\0' == (*bufp)[pos]) - mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_COND_EMPTY, r->parse, + ln, ppos, roffs[tok].name); r->last->endspan = 1; @@ -1323,29 +1301,31 @@ out: static enum rofferr roff_ds(ROFF_ARGS) { - char *name, *string; + char *string; + const char *name; + size_t namesz; /* - * A symbol is named by the first word following the macro - * invocation up to a space. Its value is anything after the - * name's trailing whitespace and optional double-quote. Thus, - * - * [.ds foo "bar " ] - * - * will have `bar " ' as its value. + * The first word is the name of the string. + * If it is empty or terminated by an escape sequence, + * abort the `ds' request without defining anything. */ - string = *bufp + pos; - name = roff_getname(r, &string, ln, pos); + name = string = *bufp + pos; if ('\0' == *name) return(ROFF_IGN); - /* Read past initial double-quote. */ + namesz = roff_getname(r, &string, ln, pos); + if ('\\' == name[namesz]) + return(ROFF_IGN); + + /* Read past the initial double-quote, if any. */ if ('"' == *string) string++; /* The rest is the value. */ - roff_setstr(r, name, string, ROFF_as == tok); + roff_setstrn(&r->strtab, name, namesz, string, strlen(string), + ROFF_as == tok); return(ROFF_IGN); } @@ -1659,14 +1639,20 @@ roff_freereg(struct roffreg *reg) static enum rofferr roff_nr(ROFF_ARGS) { - const char *key; - char *val; + char *key, *val; + size_t keysz; int iv; char sign; - val = *bufp + pos; - key = roff_getname(r, &val, ln, pos); + key = val = *bufp + pos; + if ('\0' == *key) + return(ROFF_IGN); + keysz = roff_getname(r, &val, ln, pos); + if ('\\' == key[keysz]) + return(ROFF_IGN); + key[keysz] = '\0'; + sign = *val; if ('+' == sign || '-' == sign) val++; @@ -1681,11 +1667,14 @@ static enum rofferr roff_rr(ROFF_ARGS) { struct roffreg *reg, **prev; - const char *name; - char *cp; + char *name, *cp; + size_t namesz; - cp = *bufp + pos; - name = roff_getname(r, &cp, ln, pos); + name = cp = *bufp + pos; + if ('\0' == *name) + return(ROFF_IGN); + namesz = roff_getname(r, &cp, ln, pos); + name[namesz] = '\0'; prev = &r->regtab; while (1) { @@ -1707,12 +1696,15 @@ roff_rm(ROFF_ARGS) { const char *name; char *cp; + size_t namesz; cp = *bufp + pos; while ('\0' != *cp) { - name = roff_getname(r, &cp, ln, (int)(cp - *bufp)); - if ('\0' != *name) - roff_setstr(r, name, NULL, 0); + name = cp; + namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp)); + roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0); + if ('\\' == name[namesz]) + break; } return(ROFF_IGN); } @@ -1943,7 +1935,8 @@ roff_so(ROFF_ARGS) { char *name; - mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL); + name = *bufp + pos; + mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name); /* * Handle `so'. Be EXTREMELY careful, as we shouldn't be @@ -1952,9 +1945,9 @@ roff_so(ROFF_ARGS) * or using absolute paths. */ - name = *bufp + pos; if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) { - mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL); + mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos, + ".so %s", name); return(ROFF_ERR); } @@ -1990,14 +1983,9 @@ roff_userdef(ROFF_ARGS) cp += 2; continue; } - - *szp = strlen(n1) - 3 + strlen(arg[i]) + 1; - n2 = mandoc_malloc(*szp); - - strlcpy(n2, n1, (size_t)(cp - n1 + 1)); - strlcat(n2, arg[i], *szp); - strlcat(n2, cp + 3, *szp); - + *cp = '\0'; + *szp = mandoc_asprintf(&n2, "%s%s%s", + n1, arg[i], cp + 3) + 1; cp = n2 + (cp - n1); free(n1); n1 = n2; @@ -2016,37 +2004,41 @@ roff_userdef(ROFF_ARGS) ROFF_REPARSE : ROFF_APPEND); } -static char * +static size_t roff_getname(struct roff *r, char **cpp, int ln, int pos) { char *name, *cp; + size_t namesz; name = *cpp; if ('\0' == *name) - return(name); + return(0); - /* Read until end of name. */ - for (cp = name; '\0' != *cp && ' ' != *cp; cp++) { + /* Read until end of name and terminate it with NUL. */ + for (cp = name; 1; cp++) { + if ('\0' == *cp || ' ' == *cp) { + namesz = cp - name; + break; + } if ('\\' != *cp) continue; + namesz = cp - name; + if ('{' == cp[1] || '}' == cp[1]) + break; cp++; if ('\\' == *cp) continue; mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL); - *cp = '\0'; - name = cp; + mandoc_escape((const char **)&cp, NULL, NULL); + break; } - /* Nil-terminate name. */ - if ('\0' != *cp) - *(cp++) = '\0'; - /* Read past spaces. */ while (' ' == *cp) cp++; *cpp = cp; - return(name); + return(namesz); } /* @@ -2077,7 +2069,8 @@ roff_setstrn(struct roffkv **r, const char *name, size /* Search for an existing string with the same name. */ n = *r; - while (n && strcmp(name, n->key.p)) + while (n && (namesz != n->key.sz || + strncmp(n->key.p, name, namesz))) n = n->next; if (NULL == n) {