=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.205 retrieving revision 1.225 diff -u -p -r1.205 -r1.225 --- mandoc/roff.c 2014/04/07 21:00:08 1.205 +++ mandoc/roff.c 2014/08/10 23:54:41 1.225 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.205 2014/04/07 21:00:08 schwarze Exp $ */ +/* $Id: roff.c,v 1.225 2014/08/10 23:54:41 schwarze Exp $ */ /* * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons * Copyright (c) 2010-2014 Ingo Schwarze @@ -15,10 +15,10 @@ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ -#ifdef HAVE_CONFIG_H #include "config.h" -#endif +#include + #include #include #include @@ -106,11 +106,8 @@ struct roffreg { struct roff { struct mparse *parse; /* parse point */ - int options; /* parse options */ struct roffnode *last; /* leaf of stack */ - int rstack[RSTACK_MAX]; /* stack of !`ie' rules */ - char control; /* control character */ - int rstackpos; /* position in rstack */ + int *rstack; /* stack of inverted `ie' values */ struct roffreg *regtab; /* number registers */ struct roffkv *strtab; /* user-defined strings & macros */ struct roffkv *xmbtab; /* multi-byte trans table (`tr') */ @@ -122,6 +119,10 @@ struct roff { struct eqn_node *last_eqn; /* last equation parsed */ struct eqn_node *first_eqn; /* first equation parsed */ struct eqn_node *eqn; /* current equation being parsed */ + int options; /* parse options */ + int rstacksz; /* current size limit of rstack */ + int rstackpos; /* position in rstack */ + char control; /* control character */ }; struct roffnode { @@ -137,7 +138,7 @@ struct roffnode { #define ROFF_ARGS struct roff *r, /* parse ctx */ \ enum rofft tok, /* tok of macro */ \ - char **bufp, /* input buffer */ \ + char **bufp, /* input buffer */ \ size_t *szp, /* size of input buffer */ \ int ln, /* parse line */ \ int ppos, /* original pos in buffer */ \ @@ -187,28 +188,29 @@ 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 *, const char *, size_t); static int roff_getregro(const char *name); -static const char *roff_getstrn(const struct roff *, +static const char *roff_getstrn(const struct roff *, const char *, size_t); static enum rofferr roff_it(ROFF_ARGS); 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 *, +static enum rofferr roff_res(struct roff *, char **, size_t *, int, int); static enum rofferr roff_rm(ROFF_ARGS); static enum rofferr roff_rr(ROFF_ARGS); static void roff_setstr(struct roff *, const char *, const char *, int); -static void roff_setstrn(struct roffkv **, const char *, +static void roff_setstrn(struct roffkv **, const char *, size_t, const char *, size_t, int); static enum rofferr roff_so(ROFF_ARGS); static enum rofferr roff_tr(ROFF_ARGS); @@ -315,6 +317,7 @@ static const struct predef predefs[PREDEFS_MAX] = { static int roffit_lines; /* number of lines to delay */ static char *roffit_macro; /* nil-terminated macro line */ + static void roffhash_init(void) { @@ -367,7 +370,6 @@ roffhash_find(const char *p, size_t s) return(ROFF_MAX); } - /* * Pop the current node off of the stack of roff instructions currently * pending. @@ -378,7 +380,7 @@ roffnode_pop(struct roff *r) struct roffnode *p; assert(r->last); - p = r->last; + p = r->last; r->last = r->last->parent; free(p->name); @@ -386,7 +388,6 @@ roffnode_pop(struct roff *r) free(p); } - /* * Push a roff node onto the instruction stack. This must later be * removed with roffnode_pop(). @@ -409,7 +410,6 @@ roffnode_push(struct roff *r, enum rofft tok, const ch r->last = p; } - static void roff_free1(struct roff *r) { @@ -421,32 +421,32 @@ roff_free1(struct roff *r) r->first_tbl = tbl->next; tbl_free(tbl); } - r->first_tbl = r->last_tbl = r->tbl = NULL; while (NULL != (e = r->first_eqn)) { r->first_eqn = e->next; eqn_free(e); } - r->first_eqn = r->last_eqn = r->eqn = NULL; while (r->last) roffnode_pop(r); - roff_freestr(r->strtab); - roff_freestr(r->xmbtab); + free (r->rstack); + r->rstack = NULL; + r->rstacksz = 0; + r->rstackpos = -1; - r->strtab = r->xmbtab = NULL; - roff_freereg(r->regtab); - r->regtab = NULL; + roff_freestr(r->strtab); + roff_freestr(r->xmbtab); + r->strtab = r->xmbtab = NULL; + if (r->xtab) for (i = 0; i < 128; i++) free(r->xtab[i].p); - free(r->xtab); r->xtab = NULL; } @@ -459,7 +459,6 @@ roff_reset(struct roff *r) r->control = 0; } - void roff_free(struct roff *r) { @@ -468,7 +467,6 @@ roff_free(struct roff *r) free(r); } - struct roff * roff_alloc(struct mparse *parse, int options) { @@ -478,31 +476,33 @@ roff_alloc(struct mparse *parse, int options) r->parse = parse; r->options = options; r->rstackpos = -1; - + roffhash_init(); return(r); } /* - * In the current line, expand user-defined strings ("\*") - * and references to number registers ("\n"). - * Also check the syntax of other escape sequences. + * In the current line, expand escape sequences that tend to get + * used in numerical expressions and conditional requests. + * Also check the syntax of the remaining escape sequences. */ static enum rofferr roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos) { - char ubuf[12]; /* buffer to print the number */ + 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 arg_complete; /* argument not interrupted by eol */ + char term; /* character terminating the escape */ expand_count = 0; start = *bufp + pos; @@ -521,27 +521,31 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int break; if (0 == (stesc - cp) % 2) { - stesc = cp; + stesc = (char *)cp; continue; } - /* - * Everything except user-defined strings and number - * registers is only checked, not expanded. - */ + /* Decide whether to expand or to check only. */ + term = '\0'; cp = stesc + 1; switch (*cp) { - case ('*'): + case '*': res = NULL; break; - case ('n'): + case 'B': + /* FALLTHROUGH */ + case 'w': + term = cp[1]; + /* FALLTHROUGH */ + case 'n': res = ubuf; break; default: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL)) - mandoc_msg(MANDOCERR_BADESCAPE, r->parse, - ln, (int)(stesc - *bufp), NULL); + mandoc_vmsg(MANDOCERR_ESC_BAD, + r->parse, ln, (int)(stesc - *bufp), + "%.*s", (int)(cp - stesc), stesc); continue; } @@ -557,35 +561,44 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int * Save a pointer to the name. */ - switch (*++cp) { - case ('\0'): - continue; - case ('('): - cp++; - maxl = 2; - break; - case ('['): - cp++; + if ('\0' == term) { + switch (*++cp) { + case '\0': + maxl = 0; + break; + case '(': + cp++; + maxl = 2; + break; + case '[': + cp++; + term = ']'; + maxl = 0; + break; + default: + maxl = 1; + break; + } + } else { + cp += 2; maxl = 0; - break; - default: - maxl = 1; - break; } stnam = cp; /* Advance to the end of the name. */ + arg_complete = 1; for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) { if ('\0' == *cp) { - mandoc_msg - (MANDOCERR_BADESCAPE, - r->parse, ln, - (int)(stesc - *bufp), NULL); - continue; + mandoc_msg(MANDOCERR_ESC_BAD, r->parse, + ln, (int)(stesc - *bufp), stesc); + arg_complete = 0; + break; } - if (0 == maxl && ']' == *cp) + if (0 == maxl && *cp == term) { + cp++; break; + } } /* @@ -593,33 +606,49 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int * undefined, resume searching for escapes. */ - if (NULL == res) - res = roff_getstrn(r, stnam, naml); - else - snprintf(ubuf, sizeof(ubuf), "%d", - roff_getregn(r, stnam, naml)); + switch (stesc[1]) { + case '*': + if (arg_complete) + res = roff_getstrn(r, stnam, naml); + break; + case 'B': + npos = 0; + ubuf[0] = arg_complete && + roff_evalnum(stnam, &npos, NULL, 0) && + stnam + npos + 1 == cp ? '1' : '0'; + ubuf[1] = '\0'; + break; + case 'n': + if (arg_complete) + (void)snprintf(ubuf, sizeof(ubuf), "%d", + roff_getregn(r, stnam, naml)); + else + ubuf[0] = '\0'; + break; + case 'w': + /* use even incomplete args */ + (void)snprintf(ubuf, sizeof(ubuf), "%d", + 24 * (int)naml); + break; + } if (NULL == res) { - mandoc_msg - (MANDOCERR_BADESCAPE, r->parse, - ln, (int)(stesc - *bufp), NULL); + mandoc_vmsg(MANDOCERR_STR_UNDEF, + r->parse, ln, (int)(stesc - *bufp), + "%.*s", (int)naml, stnam); 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 + (maxl ? 0 : 1), *szp); - /* Prepare for the next replacement. */ start = nbuf + pos; - stesc = nbuf + (stesc - *bufp) + ressz; + stesc = nbuf + (stesc - *bufp) + strlen(res); free(*bufp); *bufp = nbuf; } @@ -683,7 +712,7 @@ roff_parsetext(char **bufp, size_t *szp, int pos, int } enum rofferr -roff_parseln(struct roff *r, int ln, char **bufp, +roff_parseln(struct roff *r, int ln, char **bufp, size_t *szp, int pos, int *offs) { enum rofft t; @@ -714,8 +743,7 @@ roff_parseln(struct roff *r, int ln, char **bufp, if (r->last && ! ctl) { t = r->last->tok; assert(roffs[t].text); - e = (*roffs[t].text) - (r, t, bufp, szp, ln, pos, pos, offs); + e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs); assert(ROFF_IGN == e || ROFF_CONT == e); if (ROFF_CONT != e) return(e); @@ -737,9 +765,8 @@ roff_parseln(struct roff *r, int ln, char **bufp, if (r->last) { t = r->last->tok; assert(roffs[t].sub); - return((*roffs[t].sub) - (r, t, bufp, szp, - ln, ppos, pos, offs)); + return((*roffs[t].sub)(r, t, bufp, szp, + ln, ppos, pos, offs)); } /* @@ -748,33 +775,31 @@ 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); - return((*roffs[t].proc) - (r, t, bufp, szp, - ln, ppos, pos, offs)); + return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs)); } - void roff_endparse(struct roff *r) { if (r->last) - mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, - r->last->line, r->last->col, NULL); + mandoc_msg(MANDOCERR_BLK_NOEND, r->parse, + r->last->line, r->last->col, + roffs[r->last->tok].name); if (r->eqn) { - mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, - r->eqn->eqn.ln, r->eqn->eqn.pos, NULL); + mandoc_msg(MANDOCERR_BLK_NOEND, r->parse, + r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ"); eqn_end(&r->eqn); } if (r->tbl) { - mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, - r->tbl->line, r->tbl->pos, NULL); + mandoc_msg(MANDOCERR_BLK_NOEND, r->parse, + r->tbl->line, r->tbl->pos, "TS"); tbl_end(&r->tbl); } } @@ -784,33 +809,30 @@ 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); } -/* ARGSUSED */ static enum rofferr roff_cblock(ROFF_ARGS) { @@ -821,31 +843,33 @@ roff_cblock(ROFF_ARGS) */ if (NULL == r->last) { - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, ".."); return(ROFF_IGN); } switch (r->last->tok) { - case (ROFF_am): + case ROFF_am: + /* ROFF_am1 is remapped to ROFF_am in roff_block(). */ /* FALLTHROUGH */ - case (ROFF_ami): + case ROFF_ami: /* FALLTHROUGH */ - case (ROFF_am1): - /* FALLTHROUGH */ - case (ROFF_de): + case ROFF_de: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */ /* FALLTHROUGH */ - case (ROFF_dei): + case ROFF_dei: /* FALLTHROUGH */ - case (ROFF_ig): + case ROFF_ig: break; default: - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, ".."); return(ROFF_IGN); } 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); @@ -853,7 +877,6 @@ roff_cblock(ROFF_ARGS) } - static void roffnode_cleanscope(struct roff *r) { @@ -865,30 +888,32 @@ roffnode_cleanscope(struct roff *r) } } - static void roff_ccond(struct roff *r, int ln, int ppos) { if (NULL == r->last) { - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "\\}"); return; } switch (r->last->tok) { - case (ROFF_el): + case ROFF_el: /* FALLTHROUGH */ - case (ROFF_ie): + case ROFF_ie: /* FALLTHROUGH */ - case (ROFF_if): + case ROFF_if: break; default: - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "\\}"); return; } if (r->last->endspan > -1) { - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "\\}"); return; } @@ -897,41 +922,49 @@ roff_ccond(struct roff *r, int ln, int ppos) return; } - -/* ARGSUSED */ static enum rofferr roff_block(ROFF_ARGS) { - int sv; - size_t sz; - char *name; + const char *name; + char *iname, *cp; + size_t namesz; - name = NULL; + /* Ignore groff compatibility mode for now. */ - if (ROFF_ig != tok) { - if ('\0' == (*bufp)[pos]) { - mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL); - return(ROFF_IGN); - } + if (ROFF_de1 == tok) + tok = ROFF_de; + else if (ROFF_am1 == tok) + tok = ROFF_am; - /* - * Re-write `de1', since we don't really care about - * groff's strange compatibility mode, into `de'. - */ + /* Parse the macro name argument. */ - if (ROFF_de1 == tok) - tok = ROFF_de; - if (ROFF_de == tok) - name = *bufp + pos; - else - mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, - roffs[tok].name); + cp = *bufp + pos; + if (ROFF_ig == tok) { + iname = NULL; + namesz = 0; + } else { + iname = cp; + namesz = roff_getname(r, &cp, ln, ppos); + iname[namesz] = '\0'; + } - while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos])) - pos++; + /* Resolve the macro name argument if it is indirect. */ - while (isspace((unsigned char)(*bufp)[pos])) - (*bufp)[pos++] = '\0'; + if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) { + if (NULL == (name = roff_getstrn(r, iname, namesz))) { + mandoc_vmsg(MANDOCERR_STR_UNDEF, + r->parse, ln, (int)(iname - *bufp), + "%.*s", (int)namesz, iname); + namesz = 0; + } else + namesz = strlen(name); + } else + name = iname; + + if (0 == namesz && ROFF_ig != tok) { + mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse, + ln, ppos, roffs[tok].name); + return(ROFF_IGN); } roffnode_push(r, tok, name, ln, ppos); @@ -942,43 +975,40 @@ roff_block(ROFF_ARGS) * appended from roff_block_text() in multiline mode. */ - if (ROFF_de == tok) - roff_setstr(r, name, "", 0); + if (ROFF_de == tok || ROFF_dei == 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. */ + /* Get the custom end marker. */ - sv = pos; - while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos])) - pos++; + iname = cp; + namesz = roff_getname(r, &cp, ln, ppos); - /* - * 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. - */ + /* Resolve the end marker if it is indirect. */ - assert(pos > sv); - sz = (size_t)(pos - sv); + if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) { + if (NULL == (name = roff_getstrn(r, iname, namesz))) { + mandoc_vmsg(MANDOCERR_STR_UNDEF, + r->parse, ln, (int)(iname - *bufp), + "%.*s", (int)namesz, iname); + namesz = 0; + } else + namesz = strlen(name); + } else + name = iname; - if (1 == sz && '.' == (*bufp)[sv]) - return(ROFF_IGN); + if (namesz) + r->last->end = mandoc_strndup(name, namesz); - r->last->end = mandoc_malloc(sz + 1); + if ('\0' != *cp) + mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse, + ln, pos, ".%s ... %s", roffs[tok].name, cp); - 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); } - -/* ARGSUSED */ static enum rofferr roff_block_sub(ROFF_ARGS) { @@ -999,10 +1029,10 @@ roff_block_sub(ROFF_ARGS) if ((*bufp)[i] != r->last->end[j]) break; - if ('\0' == r->last->end[j] && - ('\0' == (*bufp)[i] || - ' ' == (*bufp)[i] || - '\t' == (*bufp)[i])) { + if ('\0' == r->last->end[j] && + ('\0' == (*bufp)[i] || + ' ' == (*bufp)[i] || + '\t' == (*bufp)[i])) { roffnode_pop(r); roffnode_cleanscope(r); @@ -1010,7 +1040,7 @@ 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); } @@ -1021,37 +1051,28 @@ 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 - * in `de' blocks; elsewhere, simply throw them away. - */ if (ROFF_cblock != t) { - if (ROFF_de == tok) + if (ROFF_ig != tok) roff_setstr(r, r->last->name, *bufp + ppos, 2); return(ROFF_IGN); } assert(roffs[t].proc); - return((*roffs[t].proc)(r, t, bufp, szp, - ln, ppos, pos, offs)); + return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs)); } - -/* ARGSUSED */ static enum rofferr roff_block_text(ROFF_ARGS) { - if (ROFF_de == tok) + if (ROFF_ig != tok) roff_setstr(r, r->last->name, *bufp + pos, 2); return(ROFF_IGN); } - -/* ARGSUSED */ static enum rofferr roff_cond_sub(ROFF_ARGS) { @@ -1061,7 +1082,7 @@ 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 @@ -1072,7 +1093,7 @@ roff_cond_sub(ROFF_ARGS) (rr || ROFFMAC_STRUCT & roffs[t].flags)) { assert(roffs[t].proc); return((*roffs[t].proc)(r, t, bufp, szp, - ln, ppos, pos, offs)); + ln, ppos, pos, offs)); } /* @@ -1096,7 +1117,6 @@ roff_cond_sub(ROFF_ARGS) return(rr ? ROFF_CONT : ROFF_IGN); } -/* ARGSUSED */ static enum rofferr roff_cond_text(ROFF_ARGS) { @@ -1126,8 +1146,11 @@ roff_cond_text(ROFF_ARGS) static int roff_getnum(const char *v, int *pos, int *res) { - int p, n; + int myres, n, p; + if (NULL == res) + res = &myres; + p = *pos; n = v[p] == '-'; if (n) @@ -1204,20 +1227,20 @@ roff_evalcond(const char *v, int *pos) wanttrue = 1; switch (v[*pos]) { - case ('n'): + case 'n': /* FALLTHROUGH */ - case ('o'): + case 'o': (*pos)++; return(wanttrue); - case ('c'): + case 'c': /* FALLTHROUGH */ - case ('d'): + case 'd': /* FALLTHROUGH */ - case ('e'): + case 'e': /* FALLTHROUGH */ - case ('r'): + case 'r': /* FALLTHROUGH */ - case ('t'): + case 't': (*pos)++; return(!wanttrue); default: @@ -1230,7 +1253,6 @@ roff_evalcond(const char *v, int *pos) return(roff_evalstrcond(v, pos) == wanttrue); } -/* ARGSUSED */ static enum rofferr roff_line_ignore(ROFF_ARGS) { @@ -1238,24 +1260,23 @@ roff_line_ignore(ROFF_ARGS) return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr roff_cond(ROFF_ARGS) { roffnode_push(r, tok, NULL, ln, ppos); - /* + /* * An `.el' has no conditional body: it will consume the value * of the current rstack entry set in prior `ie' calls or - * defaults to DENY. + * defaults to DENY. * * If we're not an `el', however, then evaluate the conditional. */ r->last->rule = ROFF_el == tok ? - (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) : - roff_evalcond(*bufp, &pos); + (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) : + roff_evalcond(*bufp, &pos); /* * An if-else will put the NEGATION of the current evaluated @@ -1263,10 +1284,10 @@ roff_cond(ROFF_ARGS) */ if (ROFF_ie == tok) { - if (r->rstackpos == RSTACK_MAX - 1) { - mandoc_msg(MANDOCERR_MEM, - r->parse, ln, ppos, NULL); - return(ROFF_ERR); + if (r->rstackpos + 1 == r->rstacksz) { + r->rstacksz += 16; + r->rstack = mandoc_reallocarray(r->rstack, + r->rstacksz, sizeof(int)); } r->rstack[++r->rstackpos] = !r->last->rule; } @@ -1296,7 +1317,7 @@ roff_cond(ROFF_ARGS) r->last->endspan = -1; pos += 2; goto out; - } + } /* * Anything else following the conditional causes @@ -1305,7 +1326,8 @@ 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; @@ -1314,34 +1336,34 @@ out: return(ROFF_RERUN); } - -/* ARGSUSED */ 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); } @@ -1357,31 +1379,31 @@ roff_getop(const char *v, int *pos, char *res) *res = v[*pos]; switch (*res) { - case ('+'): + case '+': /* FALLTHROUGH */ - case ('-'): + case '-': /* FALLTHROUGH */ - case ('*'): + case '*': /* FALLTHROUGH */ - case ('/'): + case '/': /* FALLTHROUGH */ - case ('%'): + case '%': /* FALLTHROUGH */ - case ('&'): + case '&': /* FALLTHROUGH */ - case (':'): + case ':': break; case '<': switch (v[*pos + 1]) { - case ('='): + case '=': *res = 'l'; (*pos)++; break; - case ('>'): + case '>': *res = '!'; (*pos)++; break; - case ('?'): + case '?': *res = 'i'; (*pos)++; break; @@ -1391,11 +1413,11 @@ roff_getop(const char *v, int *pos, char *res) break; case '>': switch (v[*pos + 1]) { - case ('='): + case '=': *res = 'g'; (*pos)++; break; - case ('?'): + case '?': *res = 'a'; (*pos)++; break; @@ -1430,9 +1452,16 @@ roff_evalpar(const char *v, int *pos, int *res) if ( ! roff_evalnum(v, pos, res, 1)) return(0); - /* If the trailing parenthesis is missing, ignore the error. */ + /* + * Omission of the closing parenthesis + * is an error in validation mode, + * but ignored in evaluation mode. + */ + if (')' == v[*pos]) (*pos)++; + else if (NULL == res) + return(0); return(1); } @@ -1478,51 +1507,54 @@ roff_evalnum(const char *v, int *pos, int *res, int sk while (isspace((unsigned char)v[*pos])) (*pos)++; + if (NULL == res) + continue; + switch (operator) { - case ('+'): + case '+': *res += operand2; break; - case ('-'): + case '-': *res -= operand2; break; - case ('*'): + case '*': *res *= operand2; break; - case ('/'): + case '/': *res /= operand2; break; - case ('%'): + case '%': *res %= operand2; break; - case ('<'): + case '<': *res = *res < operand2; break; - case ('>'): + case '>': *res = *res > operand2; break; - case ('l'): + case 'l': *res = *res <= operand2; break; - case ('g'): + case 'g': *res = *res >= operand2; break; - case ('='): + case '=': *res = *res == operand2; break; - case ('!'): + case '!': *res = *res != operand2; break; - case ('&'): + case '&': *res = *res && operand2; break; - case (':'): + case ':': *res = *res || operand2; break; - case ('i'): + case 'i': if (operand2 < *res) *res = operand2; break; - case ('a'): + case 'a': if (operand2 > *res) *res = operand2; break; @@ -1573,17 +1605,17 @@ roff_getregro(const char *name) { switch (*name) { - case ('A'): /* ASCII approximation mode is always off. */ + case 'A': /* ASCII approximation mode is always off. */ return(0); - case ('g'): /* Groff compatibility mode is always on. */ + case 'g': /* Groff compatibility mode is always on. */ return(1); - case ('H'): /* Fixed horizontal resolution. */ + case 'H': /* Fixed horizontal resolution. */ return (24); - case ('j'): /* Always adjust left margin only. */ + case 'j': /* Always adjust left margin only. */ return(0); - case ('T'): /* Some output device is always defined. */ + case 'T': /* Some output device is always defined. */ return(1); - case ('V'): /* Fixed vertical resolution. */ + case 'V': /* Fixed vertical resolution. */ return (40); default: return (-1); @@ -1645,14 +1677,20 @@ 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++; @@ -1667,11 +1705,14 @@ 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) { @@ -1688,23 +1729,24 @@ roff_rr(ROFF_ARGS) return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr 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); } -/* ARGSUSED */ static enum rofferr roff_it(ROFF_ARGS) { @@ -1717,8 +1759,8 @@ roff_it(ROFF_ARGS) len = strcspn(cp, " \t"); cp[len] = '\0'; if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) { - mandoc_msg(MANDOCERR_NUMERIC, r->parse, - ln, ppos, *bufp + 1); + mandoc_msg(MANDOCERR_IT_NONUM, r->parse, + ln, ppos, *bufp + 1); return(ROFF_IGN); } cp += len + 1; @@ -1729,7 +1771,6 @@ roff_it(ROFF_ARGS) return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr roff_Dd(ROFF_ARGS) { @@ -1742,7 +1783,6 @@ roff_Dd(ROFF_ARGS) return(ROFF_CONT); } -/* ARGSUSED */ static enum rofferr roff_TH(ROFF_ARGS) { @@ -1755,26 +1795,26 @@ roff_TH(ROFF_ARGS) return(ROFF_CONT); } -/* ARGSUSED */ static enum rofferr roff_TE(ROFF_ARGS) { if (NULL == r->tbl) - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "TE"); else tbl_end(&r->tbl); return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr roff_T_(ROFF_ARGS) { if (NULL == r->tbl) - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "T&"); else tbl_restart(ppos, ln, r->tbl); @@ -1791,7 +1831,7 @@ roff_closeeqn(struct roff *r) #endif static void -roff_openeqn(struct roff *r, const char *name, int line, +roff_openeqn(struct roff *r, const char *name, int line, int offs, const char *buf) { struct eqn_node *e; @@ -1813,7 +1853,6 @@ roff_openeqn(struct roff *r, const char *name, int lin } } -/* ARGSUSED */ static enum rofferr roff_EQ(ROFF_ARGS) { @@ -1822,23 +1861,22 @@ roff_EQ(ROFF_ARGS) return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr roff_EN(ROFF_ARGS) { - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN"); return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr roff_TS(ROFF_ARGS) { struct tbl_node *tbl; if (r->tbl) { - mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse, + ln, ppos, "TS breaks TS"); tbl_end(&r->tbl); } @@ -1853,7 +1891,6 @@ roff_TS(ROFF_ARGS) return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr roff_cc(ROFF_ARGS) { @@ -1870,7 +1907,6 @@ roff_cc(ROFF_ARGS) return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr roff_tr(ROFF_ARGS) { @@ -1892,9 +1928,8 @@ roff_tr(ROFF_ARGS) if ('\\' == *first) { esc = mandoc_escape(&p, NULL, NULL); if (ESCAPE_ERROR == esc) { - mandoc_msg - (MANDOCERR_BADESCAPE, r->parse, - ln, (int)(p - *bufp), NULL); + mandoc_msg(MANDOCERR_ESC_BAD, r->parse, + ln, (int)(p - *bufp), first); return(ROFF_IGN); } fsz = (size_t)(p - first); @@ -1904,28 +1939,27 @@ roff_tr(ROFF_ARGS) if ('\\' == *second) { esc = mandoc_escape(&p, NULL, NULL); if (ESCAPE_ERROR == esc) { - mandoc_msg - (MANDOCERR_BADESCAPE, r->parse, - ln, (int)(p - *bufp), NULL); + mandoc_msg(MANDOCERR_ESC_BAD, r->parse, + ln, (int)(p - *bufp), second); return(ROFF_IGN); } ssz = (size_t)(p - second); } else if ('\0' == *second) { - mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, - ln, (int)(p - *bufp), NULL); + mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, + ln, (int)(p - *bufp), NULL); second = " "; p--; } if (fsz > 1) { - roff_setstrn(&r->xmbtab, first, - fsz, second, ssz, 0); + roff_setstrn(&r->xmbtab, first, fsz, + second, ssz, 0); continue; } if (NULL == r->xtab) - r->xtab = mandoc_calloc - (128, sizeof(struct roffstr)); + r->xtab = mandoc_calloc(128, + sizeof(struct roffstr)); free(r->xtab[(int)*first].p); r->xtab[(int)*first].p = mandoc_strndup(second, ssz); @@ -1935,13 +1969,13 @@ roff_tr(ROFF_ARGS) return(ROFF_IGN); } -/* ARGSUSED */ static enum rofferr 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 @@ -1950,9 +1984,9 @@ 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); } @@ -1960,7 +1994,6 @@ roff_so(ROFF_ARGS) return(ROFF_SO); } -/* ARGSUSED */ static enum rofferr roff_userdef(ROFF_ARGS) { @@ -1989,14 +2022,9 @@ 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; @@ -2015,37 +2043,42 @@ 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_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos, + "%.*s", (int)(cp - name + 1), name); + 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); } /* @@ -2061,7 +2094,7 @@ roff_setstr(struct roff *r, const char *name, const ch { roff_setstrn(&r->strtab, name, strlen(name), string, - string ? strlen(string) : 0, append); + string ? strlen(string) : 0, append); } static void @@ -2076,7 +2109,8 @@ 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) { @@ -2142,8 +2176,8 @@ roff_getstrn(const struct roff *r, const char *name, s int i; for (n = r->strtab; n; n = n->next) - if (0 == strncmp(name, n->key.p, len) && - '\0' == n->key.p[(int)len]) + if (0 == strncmp(name, n->key.p, len) && + '\0' == n->key.p[(int)len]) return(n->val.p); for (i = 0; i < PREDEFS_MAX; i++) @@ -2170,14 +2204,14 @@ roff_freestr(struct roffkv *r) const struct tbl_span * roff_span(const struct roff *r) { - + return(r->tbl ? tbl_span(r->tbl) : NULL); } const struct eqn * roff_eqn(const struct roff *r) { - + return(r->last_eqn ? &r->last_eqn->eqn : NULL); } @@ -2236,8 +2270,8 @@ roff_strdup(const struct roff *r, const char *p) * Append the match to the array and move * forward by its keysize. */ - res = mandoc_realloc - (res, ssz + cp->val.sz + 1); + res = mandoc_realloc(res, + ssz + cp->val.sz + 1); memcpy(res + ssz, cp->val.p, cp->val.sz); ssz += cp->val.sz; p += (int)cp->key.sz; @@ -2258,8 +2292,8 @@ roff_strdup(const struct roff *r, const char *p) memcpy(res + ssz, pp, sz); break; } - /* - * We bail out on bad escapes. + /* + * We bail out on bad escapes. * No need to warn: we already did so when * roff_res() was called. */ @@ -2274,7 +2308,7 @@ roff_strdup(const struct roff *r, const char *p) } /* - * Find out whether a line is a macro line or not. + * Find out whether a line is a macro line or not. * If it is, adjust the current position and return one; if it isn't, * return zero and don't change the current position. * If the control character has been set with `.cc', then let that grain