=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.212 retrieving revision 1.219 diff -u -p -r1.212 -r1.219 --- mandoc/roff.c 2014/06/29 22:14:10 1.212 +++ mandoc/roff.c 2014/07/06 19:09:00 1.219 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.212 2014/06/29 22:14:10 schwarze Exp $ */ +/* $Id: roff.c,v 1.219 2014/07/06 19:09:00 schwarze Exp $ */ /* * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons * Copyright (c) 2010-2014 Ingo Schwarze @@ -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); @@ -499,7 +500,7 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int size_t naml; /* actual length of the escape name */ int expand_count; /* to avoid infinite loops */ int npos; /* position in numeric expression */ - int irc; /* return code from roff_evalnum() */ + int arg_complete; /* argument not interrupted by eol */ char term; /* character terminating the escape */ expand_count = 0; @@ -541,8 +542,9 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int 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; } @@ -584,10 +586,12 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int /* 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); + mandoc_msg(MANDOCERR_ESC_BAD, r->parse, + ln, (int)(stesc - *bufp), stesc); + arg_complete = 0; break; } if (0 == maxl && *cp == term) { @@ -603,28 +607,34 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int switch (stesc[1]) { case '*': - res = roff_getstrn(r, stnam, naml); + if (arg_complete) + res = roff_getstrn(r, stnam, naml); break; case 'B': npos = 0; - irc = roff_evalnum(stnam, &npos, NULL, 0); - ubuf[0] = irc && stnam + npos + 1 == cp - ? '1' : '0'; + ubuf[0] = arg_complete && + roff_evalnum(stnam, &npos, NULL, 0) && + stnam + npos + 1 == cp ? '1' : '0'; ubuf[1] = '\0'; break; case 'n': - (void)snprintf(ubuf, sizeof(ubuf), "%d", - roff_getregn(r, stnam, naml)); + 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 = ""; } @@ -764,7 +774,7 @@ 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); @@ -797,29 +807,27 @@ 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); } @@ -857,7 +865,8 @@ 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); @@ -910,15 +919,16 @@ 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); } @@ -929,19 +939,15 @@ 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); /* @@ -950,38 +956,23 @@ 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); } @@ -1016,7 +1007,7 @@ 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); } @@ -1027,7 +1018,7 @@ 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 @@ -1062,7 +1053,7 @@ 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 @@ -1306,7 +1297,8 @@ 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; @@ -1904,9 +1896,8 @@ 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); @@ -1916,9 +1907,8 @@ 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); @@ -2039,10 +2029,12 @@ roff_getname(struct roff *r, char **cpp, int ln, int p } if ('\\' != *cp) continue; + namesz = cp - name; + if ('{' == cp[1] || '}' == cp[1]) + break; cp++; if ('\\' == *cp) continue; - namesz = cp - name - 1; mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL); mandoc_escape((const char **)&cp, NULL, NULL); break;