=================================================================== RCS file: /cvs/mandoc/mdoc_macro.c,v retrieving revision 1.189 retrieving revision 1.205 diff -u -p -r1.189 -r1.205 --- mandoc/mdoc_macro.c 2015/04/18 16:06:40 1.189 +++ mandoc/mdoc_macro.c 2015/10/17 00:21:07 1.205 @@ -1,4 +1,4 @@ -/* $Id: mdoc_macro.c,v 1.189 2015/04/18 16:06:40 schwarze Exp $ */ +/* $Id: mdoc_macro.c,v 1.205 2015/10/17 00:21:07 schwarze Exp $ */ /* * Copyright (c) 2008-2012 Kristaps Dzonsons * Copyright (c) 2010, 2012-2015 Ingo Schwarze @@ -30,6 +30,7 @@ #include "roff.h" #include "mdoc.h" #include "libmandoc.h" +#include "roff_int.h" #include "libmdoc.h" static void blk_full(MACRO_PROT_ARGS); @@ -209,7 +210,7 @@ const struct mdoc_macro * const mdoc_macros = __mdoc_m * are errors. */ void -mdoc_macroend(struct roff_man *mdoc) +mdoc_endparse(struct roff_man *mdoc) { struct roff_node *n; @@ -231,24 +232,28 @@ mdoc_macroend(struct roff_man *mdoc) /* * Look up the macro at *p called by "from", - * or as a line macro if from == MDOC_MAX. + * or as a line macro if from == TOKEN_NONE. */ static int lookup(struct roff_man *mdoc, int from, int line, int ppos, const char *p) { int res; - if (from == MDOC_MAX || mdoc_macros[from].flags & MDOC_PARSED) { + if (mdoc->flags & MDOC_PHRASEQF) { + mdoc->flags &= ~MDOC_PHRASEQF; + return TOKEN_NONE; + } + if (from == TOKEN_NONE || mdoc_macros[from].flags & MDOC_PARSED) { res = mdoc_hash_find(p); - if (res != MDOC_MAX) { + if (res != TOKEN_NONE) { if (mdoc_macros[res].flags & MDOC_CALLABLE) - return(res); + return res; if (res != MDOC_br && res != MDOC_sp && res != MDOC_ll) mandoc_msg(MANDOCERR_MACRO_CALL, mdoc->parse, line, ppos, p); } } - return(MDOC_MAX); + return TOKEN_NONE; } /* @@ -257,12 +262,11 @@ lookup(struct roff_man *mdoc, int from, int line, int static void rew_last(struct roff_man *mdoc, const struct roff_node *to) { - struct roff_node *n, *np; + struct roff_node *np; if (to->flags & MDOC_VALID) return; - mdoc->next = ROFF_NEXT_SIBLING; while (mdoc->last != to) { /* * Save the parent here, because we may delete the @@ -272,11 +276,10 @@ rew_last(struct roff_man *mdoc, const struct roff_node */ np = mdoc->last->parent; mdoc_valid_post(mdoc); - n = mdoc->last; mdoc->last = np; assert(mdoc->last); - mdoc->last->last = n; } + mdoc->next = ROFF_NEXT_SIBLING; mdoc_valid_post(mdoc); } @@ -290,19 +293,22 @@ rew_pending(struct roff_man *mdoc, const struct roff_n for (;;) { rew_last(mdoc, n); - switch (n->type) { - case ROFFT_HEAD: - mdoc_body_alloc(mdoc, n->line, n->pos, n->tok); - return; - case ROFFT_BLOCK: - break; - default: - return; - } + if (mdoc->last == n) { + switch (n->type) { + case ROFFT_HEAD: + roff_body_alloc(mdoc, n->line, n->pos, + n->tok); + return; + case ROFFT_BLOCK: + break; + default: + return; + } + if ( ! (n->flags & MDOC_BROKEN)) + return; + } else + n = mdoc->last; - if ( ! (n->flags & MDOC_BROKEN)) - return; - for (;;) { if ((n = n->parent) == NULL) return; @@ -327,41 +333,40 @@ rew_alt(int tok) { switch (tok) { case MDOC_Ac: - return(MDOC_Ao); + return MDOC_Ao; case MDOC_Bc: - return(MDOC_Bo); + return MDOC_Bo; case MDOC_Brc: - return(MDOC_Bro); + return MDOC_Bro; case MDOC_Dc: - return(MDOC_Do); + return MDOC_Do; case MDOC_Ec: - return(MDOC_Eo); + return MDOC_Eo; case MDOC_Ed: - return(MDOC_Bd); + return MDOC_Bd; case MDOC_Ef: - return(MDOC_Bf); + return MDOC_Bf; case MDOC_Ek: - return(MDOC_Bk); + return MDOC_Bk; case MDOC_El: - return(MDOC_Bl); + return MDOC_Bl; case MDOC_Fc: - return(MDOC_Fo); + return MDOC_Fo; case MDOC_Oc: - return(MDOC_Oo); + return MDOC_Oo; case MDOC_Pc: - return(MDOC_Po); + return MDOC_Po; case MDOC_Qc: - return(MDOC_Qo); + return MDOC_Qo; case MDOC_Re: - return(MDOC_Rs); + return MDOC_Rs; case MDOC_Sc: - return(MDOC_So); + return MDOC_So; case MDOC_Xc: - return(MDOC_Xo); + return MDOC_Xo; default: - return(tok); + return tok; } - /* NOTREACHED */ } static void @@ -412,7 +417,7 @@ find_pending(struct roff_man *mdoc, int tok, int line, } } } - return(irc); + return irc; } /* @@ -431,11 +436,11 @@ dword(struct roff_man *mdoc, int line, int col, const ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) && d == DELIM_NONE && mdoc->last->type == ROFFT_TEXT && mdoc_isdelim(mdoc->last->string) == DELIM_NONE) { - mdoc_word_append(mdoc, p); + roff_word_append(mdoc, p); return; } - mdoc_word_alloc(mdoc, line, col, p); + roff_word_alloc(mdoc, line, col, p); /* * If the word consists of a bare delimiter, @@ -464,7 +469,8 @@ append_delims(struct roff_man *mdoc, int line, int *po for (;;) { la = *pos; - if (mdoc_args(mdoc, line, pos, buf, MDOC_MAX, &p) == ARGS_EOLN) + if (mdoc_args(mdoc, line, pos, buf, TOKEN_NONE, &p) == + ARGS_EOLN) break; dword(mdoc, line, la, p, DELIM_MAX, 1); @@ -497,23 +503,23 @@ macro_or_word(MACRO_PROT_ARGS, int parsed) int ntok; p = buf + ppos; - ntok = MDOC_MAX; + ntok = TOKEN_NONE; if (*p == '"') p++; else if (parsed && ! (mdoc->flags & MDOC_PHRASELIT)) ntok = lookup(mdoc, tok, line, ppos, p); - if (ntok == MDOC_MAX) { - dword(mdoc, line, ppos, p, DELIM_MAX, tok == MDOC_MAX || + if (ntok == TOKEN_NONE) { + dword(mdoc, line, ppos, p, DELIM_MAX, tok == TOKEN_NONE || mdoc_macros[tok].flags & MDOC_JOIN); - return(0); + return 0; } else { if (mdoc_macros[tok].fp == in_line_eoln) rew_elem(mdoc, tok); mdoc_macro(mdoc, ntok, line, ppos, pos, buf); - if (tok == MDOC_MAX) + if (tok == TOKEN_NONE) append_delims(mdoc, line, pos, buf); - return(1); + return 1; } } @@ -639,12 +645,14 @@ blk_exp_close(MACRO_PROT_ARGS) if (body == NULL) { mandoc_msg(MANDOCERR_BLK_NOTOPEN, mdoc->parse, line, ppos, mdoc_macronames[tok]); + if (later != NULL) + later->flags &= ~MDOC_BROKEN; if (maxargs && endbody == NULL) { /* * Stray .Ec without previous .Eo: * Break the output line, keep the arguments. */ - mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL); + roff_elem_alloc(mdoc, line, ppos, MDOC_br); rew_elem(mdoc, MDOC_br); } } else if (endbody == NULL) { @@ -666,6 +674,8 @@ blk_exp_close(MACRO_PROT_ARGS) if (endbody != NULL) n = endbody; + + ntok = TOKEN_NONE; for (j = 0; ; j++) { lastarg = *pos; @@ -676,10 +686,10 @@ blk_exp_close(MACRO_PROT_ARGS) if (ac == ARGS_PUNCT || ac == ARGS_EOLN) break; - ntok = ac == ARGS_QWORD ? MDOC_MAX : + ntok = ac == ARGS_QWORD ? TOKEN_NONE : lookup(mdoc, tok, line, lastarg, p); - if (ntok == MDOC_MAX) { + if (ntok == TOKEN_NONE) { dword(mdoc, line, lastarg, p, DELIM_MAX, MDOC_JOIN & mdoc_macros[tok].flags); continue; @@ -693,7 +703,7 @@ blk_exp_close(MACRO_PROT_ARGS) } if (n != NULL) { - if (n != mdoc->last && n->flags & MDOC_BROKEN) { + if (ntok != TOKEN_NONE && n->flags & MDOC_BROKEN) { target = n; do target = target->parent; @@ -728,15 +738,10 @@ in_line(MACRO_PROT_ARGS) switch (tok) { case MDOC_An: - /* FALLTHROUGH */ case MDOC_Ar: - /* FALLTHROUGH */ case MDOC_Fl: - /* FALLTHROUGH */ case MDOC_Mt: - /* FALLTHROUGH */ case MDOC_Nm: - /* FALLTHROUGH */ case MDOC_Pa: nc = 1; break; @@ -780,7 +785,7 @@ in_line(MACRO_PROT_ARGS) } ntok = (ac == ARGS_QWORD || (tok == MDOC_Fn && !cnt)) ? - MDOC_MAX : lookup(mdoc, tok, line, la, p); + TOKEN_NONE : lookup(mdoc, tok, line, la, p); /* * In this case, we've located a submacro and must @@ -789,7 +794,7 @@ in_line(MACRO_PROT_ARGS) * or raise a warning. */ - if (ntok != MDOC_MAX) { + if (ntok != TOKEN_NONE) { if (scope) rew_elem(mdoc, tok); if (nc && ! cnt) { @@ -940,7 +945,6 @@ blk_full(MACRO_PROT_ARGS) if (mdoc_macros[n->tok].flags & MDOC_EXPLICIT) { switch (tok) { case MDOC_Sh: - /* FALLTHROUGH */ case MDOC_Ss: mandoc_vmsg(MANDOCERR_BLK_BROKEN, mdoc->parse, line, ppos, @@ -988,7 +992,7 @@ blk_full(MACRO_PROT_ARGS) if (tok == MDOC_It && (n == NULL || n->tok != MDOC_Bl)) { mandoc_vmsg(MANDOCERR_IT_STRAY, mdoc->parse, line, ppos, "It %s", buf + *pos); - mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL); + roff_elem_alloc(mdoc, line, ppos, MDOC_br); rew_elem(mdoc, MDOC_br); return; } @@ -1022,34 +1026,47 @@ blk_full(MACRO_PROT_ARGS) */ if (tok == MDOC_Nd) { - head = mdoc_head_alloc(mdoc, line, ppos, tok); + head = roff_head_alloc(mdoc, line, ppos, tok); rew_last(mdoc, head); - body = mdoc_body_alloc(mdoc, line, ppos, tok); + body = roff_body_alloc(mdoc, line, ppos, tok); } if (tok == MDOC_Bk) mdoc->flags |= MDOC_KEEP; - ac = ARGS_PEND; + ac = ARGS_EOLN; for (;;) { + + /* + * If we are right after a tab character, + * do not parse the first word for macros. + */ + + if (mdoc->flags & MDOC_PHRASEQN) { + mdoc->flags &= ~MDOC_PHRASEQN; + mdoc->flags |= MDOC_PHRASEQF; + } + la = *pos; lac = ac; ac = mdoc_args(mdoc, line, pos, buf, tok, &p); if (ac == ARGS_EOLN) { - if (lac != ARGS_PPHRASE && lac != ARGS_PHRASE) + if (lac != ARGS_PHRASE || + ! (mdoc->flags & MDOC_PHRASEQF)) break; + /* - * This is necessary: if the last token on a - * line is a `Ta' or tab, then we'll get - * ARGS_EOLN, so we must be smart enough to - * reopen our scope if the last parse was a - * phrase or partial phrase. + * This line ends in a tab; start the next + * column now, with a leading blank. */ + if (body != NULL) rew_last(mdoc, body); - body = mdoc_body_alloc(mdoc, line, ppos, tok); + body = roff_body_alloc(mdoc, line, ppos, tok); + roff_word_alloc(mdoc, line, ppos, "\\&"); break; } + if (tok == MDOC_Bd || tok == MDOC_Bk) { mandoc_vmsg(MANDOCERR_ARG_EXCESS, mdoc->parse, line, la, "%s ... %s", @@ -1070,9 +1087,7 @@ blk_full(MACRO_PROT_ARGS) */ if (head == NULL && - ac != ARGS_PEND && ac != ARGS_PHRASE && - ac != ARGS_PPHRASE && ac != ARGS_QWORD && mdoc_isdelim(p) == DELIM_OPEN) { dword(mdoc, line, la, p, DELIM_OPEN, 0); @@ -1082,11 +1097,9 @@ blk_full(MACRO_PROT_ARGS) /* Open a head if one hasn't been opened. */ if (head == NULL) - head = mdoc_head_alloc(mdoc, line, ppos, tok); + head = roff_head_alloc(mdoc, line, ppos, tok); - if (ac == ARGS_PHRASE || - ac == ARGS_PEND || - ac == ARGS_PPHRASE) { + if (ac == ARGS_PHRASE) { /* * If we haven't opened a body yet, rewind the @@ -1094,20 +1107,18 @@ blk_full(MACRO_PROT_ARGS) */ rew_last(mdoc, body == NULL ? head : body); - body = mdoc_body_alloc(mdoc, line, ppos, tok); + body = roff_body_alloc(mdoc, line, ppos, tok); - /* - * Process phrases: set whether we're in a - * partial-phrase (this effects line handling) - * then call down into the phrase parser. - */ + /* Process to the tab or to the end of the line. */ - if (ac == ARGS_PPHRASE) - mdoc->flags |= MDOC_PPHRASE; - if (ac == ARGS_PEND && lac == ARGS_PPHRASE) - mdoc->flags |= MDOC_PPHRASE; - parse_rest(mdoc, MDOC_MAX, line, &la, buf); - mdoc->flags &= ~MDOC_PPHRASE; + mdoc->flags |= MDOC_PHRASE; + parse_rest(mdoc, TOKEN_NONE, line, &la, buf); + mdoc->flags &= ~MDOC_PHRASE; + + /* There may have been `Ta' macros. */ + + while (body->next != NULL) + body = body->next; continue; } @@ -1118,7 +1129,7 @@ blk_full(MACRO_PROT_ARGS) if (blk->flags & MDOC_VALID) return; if (head == NULL) - head = mdoc_head_alloc(mdoc, line, ppos, tok); + head = roff_head_alloc(mdoc, line, ppos, tok); if (nl && tok != MDOC_Bd && tok != MDOC_Bl && tok != MDOC_Rs) append_delims(mdoc, line, pos, buf); if (body != NULL) @@ -1129,7 +1140,7 @@ blk_full(MACRO_PROT_ARGS) /* Close out scopes to remain in a consistent state. */ rew_last(mdoc, head); - body = mdoc_body_alloc(mdoc, line, ppos, tok); + body = roff_body_alloc(mdoc, line, ppos, tok); out: if (mdoc->flags & MDOC_FREECOL) { rew_last(mdoc, body); @@ -1160,7 +1171,7 @@ blk_part_imp(MACRO_PROT_ARGS) */ blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL); - rew_last(mdoc, mdoc_head_alloc(mdoc, line, ppos, tok)); + rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok)); /* * Open the body scope "on-demand", that is, after we've @@ -1181,13 +1192,13 @@ blk_part_imp(MACRO_PROT_ARGS) } if (body == NULL) - body = mdoc_body_alloc(mdoc, line, ppos, tok); + body = roff_body_alloc(mdoc, line, ppos, tok); if (macro_or_word(mdoc, tok, line, la, pos, buf, 1)) break; } if (body == NULL) - body = mdoc_body_alloc(mdoc, line, ppos, tok); + body = roff_body_alloc(mdoc, line, ppos, tok); if (find_pending(mdoc, tok, line, ppos, body)) return; @@ -1221,7 +1232,7 @@ blk_part_exp(MACRO_PROT_ARGS) * case of `Eo'); and a body that may be empty. */ - mdoc_block_alloc(mdoc, line, ppos, tok, NULL); + roff_block_alloc(mdoc, line, ppos, tok); head = NULL; for (;;) { la = *pos; @@ -1238,11 +1249,11 @@ blk_part_exp(MACRO_PROT_ARGS) } if (head == NULL) { - head = mdoc_head_alloc(mdoc, line, ppos, tok); + head = roff_head_alloc(mdoc, line, ppos, tok); if (tok == MDOC_Eo) /* Not parsed. */ dword(mdoc, line, la, p, DELIM_MAX, 0); rew_last(mdoc, head); - mdoc_body_alloc(mdoc, line, ppos, tok); + roff_body_alloc(mdoc, line, ppos, tok); if (tok == MDOC_Eo) continue; } @@ -1254,8 +1265,8 @@ blk_part_exp(MACRO_PROT_ARGS) /* Clean-up to leave in a consistent state. */ if (head == NULL) { - rew_last(mdoc, mdoc_head_alloc(mdoc, line, ppos, tok)); - mdoc_body_alloc(mdoc, line, ppos, tok); + rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok)); + roff_body_alloc(mdoc, line, ppos, tok); } if (nl) append_delims(mdoc, line, pos, buf); @@ -1283,16 +1294,12 @@ in_line_argn(MACRO_PROT_ARGS) switch (tok) { case MDOC_Ap: - /* FALLTHROUGH */ case MDOC_Ns: - /* FALLTHROUGH */ case MDOC_Ux: maxargs = 0; break; case MDOC_Bx: - /* FALLTHROUGH */ case MDOC_Es: - /* FALLTHROUGH */ case MDOC_Xr: maxargs = 2; break; @@ -1336,9 +1343,9 @@ in_line_argn(MACRO_PROT_ARGS) } ntok = (ac == ARGS_QWORD || (tok == MDOC_Pf && state == 0)) ? - MDOC_MAX : lookup(mdoc, tok, line, la, p); + TOKEN_NONE : lookup(mdoc, tok, line, la, p); - if (ntok != MDOC_MAX) { + if (ntok != TOKEN_NONE) { if (state >= 0) { rew_elem(mdoc, tok); state = -2; @@ -1390,7 +1397,7 @@ in_line_eoln(MACRO_PROT_ARGS) if (mdoc->next == ROFF_NEXT_SIBLING) n = n->parent; if (n->tok == MDOC_Nm) - rew_last(mdoc, mdoc->last->parent); + rew_last(mdoc, n->parent); } if (buf[*pos] == '\0' && @@ -1420,9 +1427,9 @@ parse_rest(struct roff_man *mdoc, int tok, int line, i for (;;) { la = *pos; if (mdoc_args(mdoc, line, pos, buf, tok, NULL) == ARGS_EOLN) - return(0); + return 0; if (macro_or_word(mdoc, tok, line, la, pos, buf, 1)) - return(1); + return 1; } } @@ -1471,6 +1478,6 @@ phrase_ta(MACRO_PROT_ARGS) /* Advance to the next column. */ rew_last(mdoc, body); - mdoc_body_alloc(mdoc, line, ppos, MDOC_It); - parse_rest(mdoc, MDOC_MAX, line, pos, buf); + roff_body_alloc(mdoc, line, ppos, MDOC_It); + parse_rest(mdoc, TOKEN_NONE, line, pos, buf); }