=================================================================== RCS file: /cvs/mandoc/mdoc_macro.c,v retrieving revision 1.44 retrieving revision 1.49 diff -u -p -r1.44 -r1.49 --- mandoc/mdoc_macro.c 2010/03/29 19:45:42 1.44 +++ mandoc/mdoc_macro.c 2010/03/31 08:04:57 1.49 @@ -1,4 +1,4 @@ -/* $Id: mdoc_macro.c,v 1.44 2010/03/29 19:45:42 kristaps Exp $ */ +/* $Id: mdoc_macro.c,v 1.49 2010/03/31 08:04:57 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -42,13 +42,13 @@ static int blk_exp_close(MACRO_PROT_ARGS); static int blk_part_imp(MACRO_PROT_ARGS); static int phrase(struct mdoc *, int, int, char *); -static int rew_dohalt(int, enum mdoc_type, +static int rew_dohalt(enum mdoct, enum mdoc_type, const struct mdoc_node *); -static int rew_alt(int); -static int rew_dobreak(int, const struct mdoc_node *); -static int rew_elem(struct mdoc *, int); +static enum mdoct rew_alt(enum mdoct); +static int rew_dobreak(enum mdoct, const struct mdoc_node *); +static int rew_elem(struct mdoc *, enum mdoct); static int rew_sub(enum mdoc_type, struct mdoc *, - int, int, int); + enum mdoct, int, int); static int rew_last(struct mdoc *, const struct mdoc_node *); static int append_delims(struct mdoc *, int, int *, char *); @@ -318,8 +318,8 @@ rew_last(struct mdoc *mdoc, const struct mdoc_node *to * Return the opening macro of a closing one, e.g., `Ec' has `Eo' as its * matching pair. */ -static int -rew_alt(int tok) +static enum mdoct +rew_alt(enum mdoct tok) { switch (tok) { case (MDOC_Ac): @@ -369,7 +369,8 @@ rew_alt(int tok) * The scope-closing and so on occurs in the various rew_* routines. */ static int -rew_dohalt(int tok, enum mdoc_type type, const struct mdoc_node *p) +rew_dohalt(enum mdoct tok, enum mdoc_type type, + const struct mdoc_node *p) { if (MDOC_ROOT == p->type) @@ -508,7 +509,7 @@ rew_dohalt(int tok, enum mdoc_type type, const struct * REWIND_NOHALT). */ static int -rew_dobreak(int tok, const struct mdoc_node *p) +rew_dobreak(enum mdoct tok, const struct mdoc_node *p) { assert(MDOC_ROOT != p->type); @@ -537,7 +538,6 @@ rew_dobreak(int tok, const struct mdoc_node *p) return(1); break; case (MDOC_Oc): - /* XXX - experimental! */ if (MDOC_Op == p->tok) return(1); break; @@ -555,7 +555,7 @@ rew_dobreak(int tok, const struct mdoc_node *p) static int -rew_elem(struct mdoc *mdoc, int tok) +rew_elem(struct mdoc *mdoc, enum mdoct tok) { struct mdoc_node *n; @@ -571,7 +571,7 @@ rew_elem(struct mdoc *mdoc, int tok) static int rew_sub(enum mdoc_type t, struct mdoc *m, - int tok, int line, int ppos) + enum mdoct tok, int line, int ppos) { struct mdoc_node *n; int c; @@ -595,7 +595,25 @@ rew_sub(enum mdoc_type t, struct mdoc *m, } assert(n); - return(rew_last(m, n)); + if ( ! rew_last(m, n)) + return(0); + +#ifdef UGLY + /* + * The current block extends an enclosing block beyond a line + * break. Now that the current block ends, close the enclosing + * block, too. + */ + if (NULL != (n = n->pending)) { + assert(MDOC_HEAD == n->type); + if ( ! rew_last(m, n)) + return(0); + if ( ! mdoc_body_alloc(m, n->line, n->pos, n->tok)) + return(0); + } +#endif + + return(1); } @@ -714,6 +732,7 @@ in_line(MACRO_PROT_ARGS) * Whether we allow ignored elements (those without content, * usually because of reserved words) to squeak by. */ + switch (tok) { case (MDOC_An): /* FALLTHROUGH */ @@ -835,8 +854,8 @@ in_line(MACRO_PROT_ARGS) * If no elements have been collected and we're allowed to have * empties (nc), open a scope and close it out. Otherwise, * raise a warning. - * */ + if (nc && 0 == cnt) { if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg)) return(0); @@ -857,26 +876,17 @@ in_line(MACRO_PROT_ARGS) static int blk_full(MACRO_PROT_ARGS) { - int c, lastarg, reopen, dohead; + int c, la; struct mdoc_arg *arg; + struct mdoc_node *head; /* save of head macro */ + struct mdoc_node *body; /* save of body macro */ +#ifdef UGLY + struct mdoc_node *n; +#endif char *p; - /* - * Whether to process a block-head section. If this is - * non-zero, then a head will be opened for all line arguments. - * If not, then the head will always be empty and only a body - * will be opened, which will stay open at the eoln. - */ + /* Close out prior implicit scope. */ - switch (tok) { - case (MDOC_Nd): - dohead = 0; - break; - default: - dohead = 1; - break; - } - if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) { if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos)) return(0); @@ -884,12 +894,21 @@ blk_full(MACRO_PROT_ARGS) return(0); } + /* + * This routine accomodates implicitly- and explicitly-scoped + * macro openings. Implicit ones first close out prior scope + * (seen above). Delay opening the head until necessary to + * allow leading punctuation to print. Special consideration + * for `It -column', which has phrase-part syntax instead of + * regular child nodes. + */ + for (arg = NULL;; ) { - lastarg = *pos; + la = *pos; c = mdoc_argv(m, line, tok, &arg, pos, buf); if (ARGV_WORD == c) { - *pos = lastarg; + *pos = la; break; } @@ -904,70 +923,104 @@ blk_full(MACRO_PROT_ARGS) if ( ! mdoc_block_alloc(m, line, ppos, tok, arg)) return(0); - if ( ! mdoc_head_alloc(m, line, ppos, tok)) - return(0); - if ('\0' == buf[*pos]) { - if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) - return(0); - if ( ! mdoc_body_alloc(m, line, ppos, tok)) - return(0); - return(1); - } + head = body = NULL; - /* Immediately close out head and enter body, if applicable. */ + /* + * The `Nd' macro has all arguments in its body: it's a hybrid + * of block partial-explicit and full-implicit. Stupid. + */ - if (0 == dohead) { + if (MDOC_Nd == tok) { + if ( ! mdoc_head_alloc(m, line, ppos, tok)) + return(0); + head = m->last; if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) return(0); if ( ! mdoc_body_alloc(m, line, ppos, tok)) return(0); + body = m->last; } - for (reopen = 0;; ) { - lastarg = *pos; + for (;;) { + la = *pos; c = mdoc_args(m, line, pos, buf, tok, &p); if (ARGS_ERROR == c) return(0); if (ARGS_EOLN == c) break; - if (ARGS_PHRASE == c) { - assert(dohead); - if (reopen && ! mdoc_head_alloc(m, line, ppos, tok)) + + /* Don't emit leading punct. for phrases. */ + + if (NULL == head && ARGS_PHRASE != c && + 1 == mdoc_isdelim(p)) { + if ( ! mdoc_word_alloc(m, line, la, p)) return(0); - /* - * Phrases are self-contained macro phrases used - * in the columnar output of a macro. They need - * special handling. - */ - if ( ! phrase(m, line, lastarg, buf)) + continue; + } + + /* Always re-open head for phrases. */ + + if (NULL == head || ARGS_PHRASE == c) { + if ( ! mdoc_head_alloc(m, line, ppos, tok)) return(0); + head = m->last; + } + + if (ARGS_PHRASE == c) { + if ( ! phrase(m, line, la, buf)) + return(0); if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) return(0); - - reopen = 1; continue; } - if (MDOC_MAX == (c = lookup(tok, p))) { - if ( ! mdoc_word_alloc(m, line, lastarg, p)) + c = lookup(tok, p); + if (MDOC_MAX != c) { + if ( ! mdoc_macro(m, c, line, la, pos, buf)) return(0); - continue; - } + break; + } + if ( ! mdoc_word_alloc(m, line, la, p)) + return(0); - if ( ! mdoc_macro(m, c, line, lastarg, pos, buf)) + } + + if (NULL == head) { + if ( ! mdoc_head_alloc(m, line, ppos, tok)) return(0); - break; + head = m->last; } if (1 == ppos && ! append_delims(m, line, pos, buf)) return(0); - /* If the body's already open, then just return. */ - if (0 == dohead) + /* If we've already opened our body, exit now. */ + + if (NULL != body) return(1); +#ifdef UGLY + /* + * If there is an open (i.e., unvalidated) sub-block requiring + * explicit close-out, postpone switching the current block from + * head to body until the rew_sub() call closing out that + * sub-block. + */ + for (n = m->last; n && n != head; n = n->parent) { + if (MDOC_BLOCK == n->type && + MDOC_EXPLICIT & mdoc_macros[n->tok].flags && + ! (MDOC_VALID & n->flags)) { + assert( ! (MDOC_ACTED & n->flags)); + n->pending = head; + return(1); + } + } +#endif + + /* Close out scopes to remain in a consistent state. */ + if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) return(0); if ( ! mdoc_body_alloc(m, line, ppos, tok)) @@ -1024,7 +1077,7 @@ blk_part_imp(MACRO_PROT_ARGS) if (ARGS_PUNCT == c) break; - if (1 == mdoc_isdelim(p) && NULL == body) { + if (NULL == body && 1 == mdoc_isdelim(p)) { if ( ! mdoc_word_alloc(m, line, la, p)) return(0); continue; @@ -1058,9 +1111,8 @@ blk_part_imp(MACRO_PROT_ARGS) * If we can't rewind to our body, then our scope has already * been closed by another macro (like `Oc' closing `Op'). This * is ugly behaviour nodding its head to OpenBSD's overwhelming - * crufty use of `Op' breakage. XXX: DEPRECATE. + * crufty use of `Op' breakage. */ - for (n = m->last; n; n = n->parent) if (body == n) break; @@ -1117,7 +1169,7 @@ blk_part_exp(MACRO_PROT_ARGS) /* Flush out leading punctuation. */ - if (1 == mdoc_isdelim(p) && NULL == head) { + if (NULL == head && 1 == mdoc_isdelim(p)) { assert(NULL == body); if ( ! mdoc_word_alloc(m, line, la, p)) return(0); @@ -1197,7 +1249,13 @@ in_line_argn(MACRO_PROT_ARGS) struct mdoc_arg *arg; char *p; - /* Fixed maximum arguments per macro, if applicable. */ + /* + * A line macro that has a fixed number of arguments (maxargs). + * Only open the scope once the first non-leading-punctuation is + * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then + * keep it open until the maximum number of arguments are + * exhausted. + */ switch (tok) { case (MDOC_Ap): @@ -1217,9 +1275,7 @@ in_line_argn(MACRO_PROT_ARGS) break; } - /* Macro argument processing. */ - - for (arg = NULL;; ) { + for (arg = NULL; ; ) { la = *pos; c = mdoc_argv(m, line, tok, &arg, pos, buf); @@ -1237,22 +1293,8 @@ in_line_argn(MACRO_PROT_ARGS) return(0); } - /* Open the element scope. */ - - if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg)) - return(0); - - /* Process element arguments. */ - - for (flushed = j = 0; ; j++) { + for (flushed = j = 0; ; ) { la = *pos; - - if (j == maxargs && ! flushed) { - if ( ! rew_elem(m, tok)) - return(0); - flushed = 1; - } - c = mdoc_args(m, line, pos, buf, tok, &p); if (ARGS_ERROR == c) @@ -1262,12 +1304,28 @@ in_line_argn(MACRO_PROT_ARGS) if (ARGS_EOLN == c) break; + if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) && + 0 == j && 1 == mdoc_isdelim(p)) { + if ( ! mdoc_word_alloc(m, line, la, p)) + return(0); + continue; + } else if (0 == j) + if ( ! mdoc_elem_alloc(m, line, la, tok, arg)) + return(0); + + if (j == maxargs && ! flushed) { + if ( ! rew_elem(m, tok)) + return(0); + flushed = 1; + } + if (MDOC_MAX != (c = lookup(tok, p))) { if ( ! flushed && ! rew_elem(m, tok)) return(0); flushed = 1; if ( ! mdoc_macro(m, c, line, la, pos, buf)) return(0); + j++; break; } @@ -1285,18 +1343,24 @@ in_line_argn(MACRO_PROT_ARGS) * code is no here, it's unlikely to be removed. */ +#ifdef __OpenBSD__ if (MDOC_Xr == tok && j == maxargs) { - if ( ! mdoc_elem_alloc(m, line, ppos, MDOC_Ns, NULL)) + if ( ! mdoc_elem_alloc(m, line, la, MDOC_Ns, NULL)) return(0); if ( ! rew_elem(m, MDOC_Ns)) return(0); } +#endif if ( ! mdoc_word_alloc(m, line, la, p)) return(0); + j++; } - /* Close out and append delimiters. */ + if (0 == j && ! mdoc_elem_alloc(m, line, la, tok, arg)) + return(0); + + /* Close out in a consistent state. */ if ( ! flushed && ! rew_elem(m, tok)) return(0);