=================================================================== RCS file: /cvs/mandoc/Attic/macro.c,v retrieving revision 1.49 retrieving revision 1.68 diff -u -p -r1.49 -r1.68 --- mandoc/Attic/macro.c 2009/01/22 14:56:21 1.49 +++ mandoc/Attic/macro.c 2009/03/12 06:32:17 1.68 @@ -1,4 +1,4 @@ -/* $Id: macro.c,v 1.49 2009/01/22 14:56:21 kristaps Exp $ */ +/* $Id: macro.c,v 1.68 2009/03/12 06:32:17 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -21,62 +21,27 @@ #include #include #include -#ifdef __linux__ -#include -#endif +#include "private.h" + /* * This has scanning/parsing routines, each of which extract a macro and * its arguments and parameters, then know how to progress to the next - * macro. Macros are parsed according as follows: - * - * ELEMENT: TEXT | epsilon - * BLOCK: HEAD PUNCT BODY PUNCT BLOCK_TAIL PUNCT - * BLOCK_TAIL: TAIL | epsilon - * HEAD: ELEMENT | TEXT | BLOCK | epsilon - * BODY: ELEMENT | TEXT | BLOCK | epsilon - * TAIL: TEXT | epsilon - * PUNCT: TEXT (delimiters) | epsilon - * - * These are arranged into a parse tree, an example of which follows: - * - * ROOT - * BLOCK (.Sh) - * HEAD - * TEXT (`NAME') - * BODY - * ELEMENT (.Nm) - * TEXT (`mdocml') - * ELEMENT (.Nd) - * TEXT (`mdoc macro compiler') - * BLOCK (.Op) - * HEAD - * ELEMENT (.Fl) - * TEXT (`v') - * BLOCK (.Op) - * HEAD - * ELEMENT (.Fl) - * TEXT (`v') - * ELEMENT (.Fl) - * TEXT (`W') - * ELEMENT (.Ns) - * ELEMENT (.Ar) - * TEXT (`err...') - * - * These types are always per-line except for block bodies, which may - * span multiple lines. Macros are assigned a parsing routine, which - * corresponds to the type, in the mdoc_macros table. - * - * Note that types are general: there can be several parsing routines - * corresponding to a single type. The macro_text function, for - * example, parses an ELEMENT type (see the function definition for - * details) that may be interrupted by further macros; the - * macro_constant function, on the other hand, parses an ELEMENT type - * spanning a single line. + * macro. */ -#include "private.h" +/* FIXME: .Fl, .Ar, .Cd handling of `|'. */ +static int macro_obsolete(MACRO_PROT_ARGS); +static int macro_constant(MACRO_PROT_ARGS); +static int macro_constant_scoped(MACRO_PROT_ARGS); +static int macro_constant_delimited(MACRO_PROT_ARGS); +static int macro_text(MACRO_PROT_ARGS); +static int macro_scoped(MACRO_PROT_ARGS); +static int macro_scoped_close(MACRO_PROT_ARGS); +static int macro_scoped_line(MACRO_PROT_ARGS); +static int macro_phrase(struct mdoc *, int, int, char *); + #define REWIND_REWIND (1 << 0) #define REWIND_NOHALT (1 << 1) #define REWIND_HALT (1 << 2) @@ -95,15 +60,170 @@ static int append_delims(struct mdoc *, int, int *, static int lookup(struct mdoc *, int, int, int, const char *); static int pwarn(struct mdoc *, int, int, int); static int perr(struct mdoc *, int, int, int); +static int scopewarn(struct mdoc *, enum mdoc_type, int, int, + const struct mdoc_node *); #define WMACPARM (1) #define WOBS (2) #define ENOCTX (1) #define ENOPARMS (2) -#define EARGVLIM (3) +/* Central table of library: who gets parsed how. */ +const struct mdoc_macro __mdoc_macros[MDOC_MAX] = { + { NULL, 0 }, /* \" */ + { macro_constant, MDOC_PROLOGUE }, /* Dd */ + { macro_constant, MDOC_PROLOGUE }, /* Dt */ + { macro_constant, MDOC_PROLOGUE }, /* Os */ + { macro_scoped, 0 }, /* Sh */ + { macro_scoped, 0 }, /* Ss */ + { macro_text, 0 }, /* Pp */ + { macro_scoped_line, MDOC_PARSED }, /* D1 */ + { macro_scoped_line, MDOC_PARSED }, /* Dl */ + { macro_scoped, MDOC_EXPLICIT }, /* Bd */ + { macro_scoped_close, MDOC_EXPLICIT }, /* Ed */ + { macro_scoped, MDOC_EXPLICIT }, /* Bl */ + { macro_scoped_close, MDOC_EXPLICIT }, /* El */ + { macro_scoped, MDOC_PARSED }, /* It */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */ + { macro_text, MDOC_PARSED }, /* An */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */ + { macro_constant, 0 }, /* Cd */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Er */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */ + { macro_constant, 0 }, /* Ex */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */ + { macro_constant, 0 }, /* Fd */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */ + { macro_text, MDOC_PARSED }, /* Ft */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Ic */ + { macro_constant, 0 }, /* In */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Li */ + { macro_constant, 0 }, /* Nd */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Op */ + { macro_obsolete, 0 }, /* Ot */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */ + { macro_constant, 0 }, /* Rv */ + /* XXX - .St supposed to be (but isn't) callable. */ + { macro_constant_delimited, MDOC_PARSED }, /* St */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Va */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */ + { macro_constant, 0 }, /* %A */ + { macro_constant, 0 }, /* %B */ + { macro_constant, 0 }, /* %D */ + { macro_constant, 0 }, /* %I */ + { macro_constant, 0 }, /* %J */ + { macro_constant, 0 }, /* %N */ + { macro_constant, 0 }, /* %O */ + { macro_constant, 0 }, /* %P */ + { macro_constant, 0 }, /* %R */ + { macro_constant, 0 }, /* %T */ + { macro_constant, 0 }, /* %V */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ac */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ao */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Aq */ + { macro_constant_delimited, 0 }, /* At */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Bc */ + { macro_scoped, MDOC_EXPLICIT }, /* Bf */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bo */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Bq */ + { macro_constant_delimited, MDOC_PARSED }, /* Bsx */ + { macro_constant_delimited, MDOC_PARSED }, /* Bx */ + { macro_constant, 0 }, /* Db */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Dc */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Do */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dq */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ec */ + { macro_scoped_close, MDOC_EXPLICIT }, /* Ef */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Em */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */ + { macro_constant_delimited, MDOC_PARSED }, /* Fx */ + { macro_text, MDOC_PARSED }, /* Ms */ + { macro_constant_delimited, MDOC_CALLABLE | MDOC_PARSED }, /* No */ + { macro_constant_delimited, MDOC_CALLABLE | MDOC_PARSED }, /* Ns */ + { macro_constant_delimited, MDOC_PARSED }, /* Nx */ + { macro_constant_delimited, MDOC_PARSED }, /* Ox */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Pc */ + { macro_constant_delimited, MDOC_PARSED }, /* Pf */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Po */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pq */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Qc */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ql */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Qo */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Qq */ + { macro_scoped_close, MDOC_EXPLICIT }, /* Re */ + { macro_scoped, MDOC_EXPLICIT }, /* Rs */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Sc */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* So */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sq */ + { macro_constant, 0 }, /* Sm */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Sx */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Sy */ + { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */ + { macro_constant_delimited, MDOC_PARSED }, /* Ux */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */ + /* XXX - .Fo supposed to be (but isn't) callable. */ + { macro_scoped, MDOC_EXPLICIT }, /* Fo */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Fc */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Oo */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Oc */ + { macro_scoped, MDOC_EXPLICIT }, /* Bk */ + { macro_scoped_close, MDOC_EXPLICIT }, /* Ek */ + { macro_constant, 0 }, /* Bt */ + { macro_constant, 0 }, /* Hf */ + { macro_obsolete, 0 }, /* Fr */ + { macro_constant, 0 }, /* Ud */ + { macro_constant, 0 }, /* Lb */ + { macro_constant_delimited, MDOC_CALLABLE | MDOC_PARSED }, /* Ap */ + { macro_text, 0 }, /* Lp */ + { macro_text, MDOC_PARSED }, /* Lk */ + { macro_text, MDOC_PARSED }, /* Mt */ + { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Brq */ + { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bro */ + { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Brc */ +}; + +const struct mdoc_macro * const mdoc_macros = __mdoc_macros; + + +/* + * This is called at the end of parsing. It must traverse up the tree, + * closing out open [implicit] scopes. Obviously, open explicit scopes + * are errors. + */ +int +macro_end(struct mdoc *mdoc) +{ + struct mdoc_node *n; + + assert(mdoc->first); + assert(mdoc->last); + + /* Scan for open explicit scopes. */ + + n = MDOC_VALID & mdoc->last->flags ? + mdoc->last->parent : mdoc->last; + + for ( ; n; n = n->parent) { + if (MDOC_BLOCK != n->type) + continue; + if ( ! (MDOC_EXPLICIT & mdoc_macros[n->tok].flags)) + continue; + return(mdoc_nerr(mdoc, n, + "macro scope still open on exit")); + } + + return(rewind_last(mdoc, mdoc->first)); +} + + static int perr(struct mdoc *mdoc, int line, int pos, int type) { @@ -112,17 +232,12 @@ perr(struct mdoc *mdoc, int line, int pos, int type) switch (type) { case (ENOCTX): c = mdoc_perr(mdoc, line, pos, - "closing macro has prior context"); + "closing macro has no prior context"); break; case (ENOPARMS): c = mdoc_perr(mdoc, line, pos, "macro doesn't expect parameters"); break; - case (EARGVLIM): - c = mdoc_perr(mdoc, line, pos, - "argument hard-limit %d reached", - MDOC_LINEARG_MAX); - break; default: abort(); /* NOTREACHED */ @@ -153,11 +268,58 @@ pwarn(struct mdoc *mdoc, int line, int pos, int type) static int +scopewarn(struct mdoc *mdoc, enum mdoc_type type, + int line, int pos, const struct mdoc_node *p) +{ + const char *n, *t, *tt; + + n = t = ""; + tt = "block"; + + switch (type) { + case (MDOC_BODY): + tt = "multi-line"; + break; + case (MDOC_HEAD): + tt = "line"; + break; + default: + break; + } + + switch (p->type) { + case (MDOC_BLOCK): + n = mdoc_macronames[p->tok]; + t = "block"; + break; + case (MDOC_BODY): + n = mdoc_macronames[p->tok]; + t = "multi-line"; + break; + case (MDOC_HEAD): + n = mdoc_macronames[p->tok]; + t = "line"; + break; + default: + break; + } + + if ( ! (MDOC_IGN_SCOPE & mdoc->pflags)) + return(mdoc_perr(mdoc, line, pos, + "%s scope breaks %s scope of %s", + tt, t, n)); + return(mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX, + "%s scope breaks %s scope of %s", + tt, t, n)); +} + + +static int lookup(struct mdoc *mdoc, int line, int pos, int from, const char *p) { int res; - res = mdoc_find(mdoc, p); + res = mdoc_tokhash_find(mdoc->htab, p); if (MDOC_PARSED & mdoc_macros[from].flags) return(res); if (MDOC_MAX == res) @@ -199,6 +361,8 @@ rewind_alt(int tok) return(MDOC_Ao); case (MDOC_Bc): return(MDOC_Bo); + case (MDOC_Brc): + return(MDOC_Bro); case (MDOC_Dc): return(MDOC_Do); case (MDOC_Ec): @@ -248,6 +412,8 @@ rewind_dohalt(int tok, enum mdoc_type type, const stru /* FALLTHROUGH */ case (MDOC_Bq): /* FALLTHROUGH */ + case (MDOC_Brq): + /* FALLTHROUGH */ case (MDOC_D1): /* FALLTHROUGH */ case (MDOC_Dl): @@ -302,6 +468,8 @@ rewind_dohalt(int tok, enum mdoc_type type, const stru /* FALLTHROUGH */ case (MDOC_Bo): /* FALLTHROUGH */ + case (MDOC_Bro): + /* FALLTHROUGH */ case (MDOC_Do): /* FALLTHROUGH */ case (MDOC_Eo): @@ -328,6 +496,8 @@ rewind_dohalt(int tok, enum mdoc_type type, const stru /* FALLTHROUGH */ case (MDOC_Bc): /* FALLTHROUGH */ + case (MDOC_Brc): + /* FALLTHROUGH */ case (MDOC_Dc): /* FALLTHROUGH */ case (MDOC_Ec): @@ -437,7 +607,8 @@ rewind_subblock(enum mdoc_type type, struct mdoc *mdoc break; else if (rewind_dobreak(tok, n)) continue; - return(mdoc_perr(mdoc, line, ppos, "scope breaks prior %s", mdoc_node2a(n))); + if ( ! scopewarn(mdoc, type, line, ppos, n)) + return(0); } assert(n); @@ -460,7 +631,8 @@ rewind_expblock(struct mdoc *mdoc, int tok, int line, break; else if (rewind_dobreak(tok, n)) continue; - return(mdoc_perr(mdoc, line, ppos, "scope breaks prior %s", mdoc_node2a(n))); + if ( ! scopewarn(mdoc, MDOC_BLOCK, line, ppos, n)) + return(0); } assert(n); @@ -483,7 +655,8 @@ rewind_impblock(struct mdoc *mdoc, int tok, int line, break; else if (rewind_dobreak(tok, n)) continue; - return(mdoc_perr(mdoc, line, ppos, "scope breaks prior %s", mdoc_node2a(n))); + if ( ! scopewarn(mdoc, MDOC_BLOCK, line, ppos, n)) + return(0); } assert(n); @@ -523,7 +696,7 @@ append_delims(struct mdoc *mdoc, int line, int *pos, c * Close out an explicit scope. This optionally parses a TAIL type with * a set number of TEXT children. */ -int +static int macro_scoped_close(MACRO_PROT_ARGS) { int tt, j, c, lastarg, maxargs, flushed; @@ -545,7 +718,8 @@ macro_scoped_close(MACRO_PROT_ARGS) if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) { if (0 == buf[*pos]) { - if ( ! rewind_subblock(MDOC_BODY, mdoc, tok, line, ppos)) + if ( ! rewind_subblock(MDOC_BODY, mdoc, + tok, line, ppos)) return(0); return(rewind_expblock(mdoc, tok, line, ppos)); } @@ -587,7 +761,8 @@ macro_scoped_close(MACRO_PROT_ARGS) return(0); else if (MDOC_MAX != c) { if ( ! flushed) { - if ( ! rewind_expblock(mdoc, tok, line, ppos)) + if ( ! rewind_expblock(mdoc, tok, + line, ppos)) return(0); flushed = 1; } @@ -634,19 +809,20 @@ macro_scoped_close(MACRO_PROT_ARGS) * TEXT (`;') * TEXT (`;') */ -int +static int macro_text(MACRO_PROT_ARGS) { - int la, lastpunct, c, w, argc; - struct mdoc_arg argv[MDOC_LINEARG_MAX]; + int la, lastpunct, c, w; + struct mdoc_arg *arg; char *p; la = ppos; lastpunct = 0; + arg = NULL; - for (argc = 0; argc < MDOC_LINEARG_MAX; argc++) { + for (;;) { la = *pos; - c = mdoc_argv(mdoc, line, tok, &argv[argc], pos, buf); + c = mdoc_argv(mdoc, line, tok, &arg, pos, buf); if (ARGV_EOLN == c) break; if (ARGV_WORD == c) { @@ -654,22 +830,12 @@ macro_text(MACRO_PROT_ARGS) break; } else if (ARGV_ARG == c) continue; - - mdoc_argv_free(argc, argv); + mdoc_argv_free(arg); return(0); } - if (MDOC_LINEARG_MAX == argc) { - mdoc_argv_free(argc - 1, argv); - return(perr(mdoc, line, ppos, EARGVLIM)); - } - - c = mdoc_elem_alloc(mdoc, line, ppos, tok, argc, argv); - - if (0 == c) { - mdoc_argv_free(argc, argv); + if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg)) return(0); - } mdoc->next = MDOC_NEXT_CHILD; @@ -677,51 +843,46 @@ macro_text(MACRO_PROT_ARGS) for (;;) { la = *pos; w = mdoc_args(mdoc, line, pos, buf, tok, &p); - assert(ARGS_PHRASE != c); + assert(ARGS_PHRASE != w); - if (ARGS_ERROR == w) { - mdoc_argv_free(argc, argv); + if (ARGS_ERROR == w) return(0); - } - if (ARGS_EOLN == w) break; if (ARGS_PUNCT == w) break; + /* Quoted words shouldn't be looked-up. */ + c = ARGS_QWORD == w ? MDOC_MAX : lookup(mdoc, line, la, tok, p); + /* MDOC_MAX (not a macro) or -1 (error). */ + if (MDOC_MAX != c && -1 != c) { - if (0 == lastpunct && ! rewind_elem(mdoc, tok)) { - mdoc_argv_free(argc, argv); + if (0 == lastpunct && ! rewind_elem(mdoc, tok)) return(0); - } - mdoc_argv_free(argc, argv); c = mdoc_macro(mdoc, c, line, la, pos, buf); if (0 == c) return(0); if (ppos > 1) return(1); return(append_delims(mdoc, line, pos, buf)); - } else if (-1 == c) { - mdoc_argv_free(argc, argv); + } else if (-1 == c) return(0); - } + /* Non-quote-enclosed punctuation. */ + if (ARGS_QWORD != w && mdoc_isdelim(p)) { - if (0 == lastpunct && ! rewind_elem(mdoc, tok)) { - mdoc_argv_free(argc, argv); + if (0 == lastpunct && ! rewind_elem(mdoc, tok)) return(0); - } lastpunct = 1; } else if (lastpunct) { - c = mdoc_elem_alloc(mdoc, line, - ppos, tok, argc, argv); - if (0 == c) { - mdoc_argv_free(argc, argv); + c = mdoc_elem_alloc(mdoc, line, ppos, tok, arg); + + if (0 == c) return(0); - } + mdoc->next = MDOC_NEXT_CHILD; lastpunct = 0; } @@ -731,8 +892,6 @@ macro_text(MACRO_PROT_ARGS) mdoc->next = MDOC_NEXT_SIBLING; } - mdoc_argv_free(argc, argv); - if (0 == lastpunct && ! rewind_elem(mdoc, tok)) return(0); if (ppos > 1) @@ -763,13 +922,15 @@ macro_text(MACRO_PROT_ARGS) * TEXT (`Another.') * * Note that the `.It' macro, possibly the most difficult (as it has - * embedded scope, etc.) is handled by this routine. + * embedded scope, etc.) is handled by this routine. It handles + * columnar output, where columns are "phrases" and denote multiple + * block heads. */ -int +static int macro_scoped(MACRO_PROT_ARGS) { - int c, lastarg, argc; - struct mdoc_arg argv[MDOC_LINEARG_MAX]; + int c, lastarg, reopen; + struct mdoc_arg *arg; char *p; assert ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)); @@ -784,10 +945,12 @@ macro_scoped(MACRO_PROT_ARGS) } /* Parse arguments. */ + + arg = NULL; - for (argc = 0; argc < MDOC_LINEARG_MAX; argc++) { + for (;;) { lastarg = *pos; - c = mdoc_argv(mdoc, line, tok, &argv[argc], pos, buf); + c = mdoc_argv(mdoc, line, tok, &arg, pos, buf); if (ARGV_EOLN == c) break; if (ARGV_WORD == c) { @@ -795,20 +958,11 @@ macro_scoped(MACRO_PROT_ARGS) break; } else if (ARGV_ARG == c) continue; - mdoc_argv_free(argc, argv); + mdoc_argv_free(arg); return(0); } - if (MDOC_LINEARG_MAX == argc) { - mdoc_argv_free(argc - 1, argv); - return(perr(mdoc, line, ppos, EARGVLIM)); - } - - c = mdoc_block_alloc(mdoc, line, ppos, - tok, (size_t)argc, argv); - mdoc_argv_free(argc, argv); - - if (0 == c) + if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, arg)) return(0); mdoc->next = MDOC_NEXT_CHILD; @@ -816,7 +970,8 @@ macro_scoped(MACRO_PROT_ARGS) if (0 == buf[*pos]) { if ( ! mdoc_head_alloc(mdoc, line, ppos, tok)) return(0); - if ( ! rewind_subblock(MDOC_HEAD, mdoc, tok, line, ppos)) + if ( ! rewind_subblock(MDOC_HEAD, mdoc, + tok, line, ppos)) return(0); if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) return(0); @@ -826,6 +981,9 @@ macro_scoped(MACRO_PROT_ARGS) if ( ! mdoc_head_alloc(mdoc, line, ppos, tok)) return(0); + + /* Indicate that columnar scope shouldn't be reopened. */ + reopen = 0; mdoc->next = MDOC_NEXT_CHILD; for (;;) { @@ -834,22 +992,27 @@ macro_scoped(MACRO_PROT_ARGS) if (ARGS_ERROR == c) return(0); - if (ARGS_PUNCT == c) - break; if (ARGS_EOLN == c) break; - if (ARGS_PHRASE == c) { + if (reopen && ! mdoc_head_alloc(mdoc, line, ppos, tok)) + return(0); + mdoc->next = MDOC_NEXT_CHILD; + /* - if ( ! mdoc_phrase(mdoc, line, lastarg, buf)) + * Phrases are self-contained macro phrases used + * in the columnar output of a macro. They need + * special handling. + */ + if ( ! macro_phrase(mdoc, line, lastarg, buf)) return(0); - */ + if ( ! rewind_subblock(MDOC_HEAD, mdoc, tok, line, ppos)) + return(0); + + reopen = 1; continue; } - /* FIXME: if .It -column, the lookup must be for a - * sub-line component. BLAH. */ - if (-1 == (c = lookup(mdoc, line, lastarg, tok, p))) return(0); @@ -865,10 +1028,10 @@ macro_scoped(MACRO_PROT_ARGS) break; } - if ( ! rewind_subblock(MDOC_HEAD, mdoc, tok, line, ppos)) - return(0); if (1 == ppos && ! append_delims(mdoc, line, pos, buf)) return(0); + if ( ! rewind_subblock(MDOC_HEAD, mdoc, tok, line, ppos)) + return(0); if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) return(0); @@ -897,13 +1060,13 @@ macro_scoped(MACRO_PROT_ARGS) * TEXT (`;') * TEXT (`;') */ -int +static int macro_scoped_line(MACRO_PROT_ARGS) { int lastarg, c; char *p; - if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, 0, NULL)) + if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, NULL)) return(0); mdoc->next = MDOC_NEXT_CHILD; @@ -969,7 +1132,7 @@ macro_scoped_line(MACRO_PROT_ARGS) * TEXT (`b') * TEXT (';') */ -int +static int macro_constant_scoped(MACRO_PROT_ARGS) { int lastarg, flushed, j, c, maxargs; @@ -987,7 +1150,7 @@ macro_constant_scoped(MACRO_PROT_ARGS) break; } - if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, 0, NULL)) + if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, NULL)) return(0); mdoc->next = MDOC_NEXT_CHILD; @@ -1030,20 +1193,24 @@ macro_constant_scoped(MACRO_PROT_ARGS) return(0); else if (MDOC_MAX != c) { if ( ! flushed) { - if ( ! rewind_subblock(MDOC_HEAD, mdoc, tok, line, ppos)) + if ( ! rewind_subblock(MDOC_HEAD, mdoc, + tok, line, ppos)) return(0); flushed = 1; - if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) + if ( ! mdoc_body_alloc(mdoc, line, + ppos, tok)) return(0); mdoc->next = MDOC_NEXT_CHILD; } - if ( ! mdoc_macro(mdoc, c, line, lastarg, pos, buf)) + if ( ! mdoc_macro(mdoc, c, line, lastarg, + pos, buf)) return(0); break; } if ( ! flushed && mdoc_isdelim(p)) { - if ( ! rewind_subblock(MDOC_HEAD, mdoc, tok, line, ppos)) + if ( ! rewind_subblock(MDOC_HEAD, mdoc, + tok, line, ppos)) return(0); flushed = 1; if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) @@ -1083,26 +1250,30 @@ macro_constant_scoped(MACRO_PROT_ARGS) * ELEMENT (.No) * TEXT (`b') */ -int +static int macro_constant_delimited(MACRO_PROT_ARGS) { - int lastarg, flushed, j, c, maxargs, argc; - struct mdoc_arg argv[MDOC_LINEARG_MAX]; + int lastarg, flushed, j, c, maxargs, + igndelim, ignargs; + struct mdoc_arg *arg; char *p; lastarg = ppos; flushed = 0; + + /* + * Maximum arguments per macro. Some of these have none and + * exit as soon as they're parsed. + */ switch (tok) { + case (MDOC_Ap): + /* FALLTHROUGH */ case (MDOC_No): /* FALLTHROUGH */ case (MDOC_Ns): /* FALLTHROUGH */ - case (MDOC_Pf): - /* FALLTHROUGH */ case (MDOC_Ux): - /* FALLTHROUGH */ - case (MDOC_St): maxargs = 0; break; default: @@ -1110,29 +1281,53 @@ macro_constant_delimited(MACRO_PROT_ARGS) break; } - for (argc = 0; argc < MDOC_LINEARG_MAX; argc++) { - lastarg = *pos; - c = mdoc_argv(mdoc, line, tok, &argv[argc], pos, buf); - if (ARGV_EOLN == c) - break; - if (ARGV_WORD == c) { - *pos = lastarg; - break; - } else if (ARGV_ARG == c) - continue; - mdoc_argv_free(argc, argv); - return(0); + /* + * Whether to ignore delimiter characters. `Pf' accepts its + * first token as a parameter no matter what it looks like (if + * it's text). + */ + + switch (tok) { + case (MDOC_Pf): + igndelim = 1; + break; + default: + igndelim = 0; + break; } - if (MDOC_LINEARG_MAX == argc) { - mdoc_argv_free(argc - 1, argv); - return(perr(mdoc, line, ppos, EARGVLIM)); + /* + * Whether to ignore arguments: `St', for example, handles its + * argument-like parameters as regular parameters. + */ + + switch (tok) { + case (MDOC_St): + ignargs = 1; + break; + default: + ignargs = 0; + break; } - c = mdoc_elem_alloc(mdoc, line, ppos, tok, argc, argv); - mdoc_argv_free(argc, argv); + arg = NULL; - if (0 == c) + if ( ! ignargs) + for (;;) { + lastarg = *pos; + c = mdoc_argv(mdoc, line, tok, &arg, pos, buf); + if (ARGV_EOLN == c) + break; + if (ARGV_WORD == c) { + *pos = lastarg; + break; + } else if (ARGV_ARG == c) + continue; + mdoc_argv_free(arg); + return(0); + } + + if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg)) return(0); mdoc->next = MDOC_NEXT_CHILD; @@ -1167,7 +1362,7 @@ macro_constant_delimited(MACRO_PROT_ARGS) break; } - if ( ! flushed && mdoc_isdelim(p)) { + if ( ! flushed && mdoc_isdelim(p) && ! igndelim) { if ( ! rewind_elem(mdoc, tok)) return(0); flushed = 1; @@ -1191,18 +1386,20 @@ macro_constant_delimited(MACRO_PROT_ARGS) * A constant macro is the simplest classification. It spans an entire * line. */ -int +static int macro_constant(MACRO_PROT_ARGS) { - int c, w, la, argc; - struct mdoc_arg argv[MDOC_LINEARG_MAX]; + int c, w, la; + struct mdoc_arg *arg; char *p; assert( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)); - for (argc = 0; argc < MDOC_LINEARG_MAX; argc++) { + arg = NULL; + + for (;;) { la = *pos; - c = mdoc_argv(mdoc, line, tok, &argv[argc], pos, buf); + c = mdoc_argv(mdoc, line, tok, &arg, pos, buf); if (ARGV_EOLN == c) break; if (ARGV_WORD == c) { @@ -1210,20 +1407,11 @@ macro_constant(MACRO_PROT_ARGS) break; } else if (ARGV_ARG == c) continue; - - mdoc_argv_free(argc, argv); + mdoc_argv_free(arg); return(0); } - if (MDOC_LINEARG_MAX == argc) { - mdoc_argv_free(argc - 1, argv); - return(perr(mdoc, line, ppos, EARGVLIM)); - } - - c = mdoc_elem_alloc(mdoc, line, ppos, tok, argc, argv); - mdoc_argv_free(argc, argv); - - if (0 == c) + if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg)) return(0); mdoc->next = MDOC_NEXT_CHILD; @@ -1258,7 +1446,7 @@ macro_constant(MACRO_PROT_ARGS) /* ARGSUSED */ -int +static int macro_obsolete(MACRO_PROT_ARGS) { @@ -1266,31 +1454,45 @@ macro_obsolete(MACRO_PROT_ARGS) } -/* - * This is called at the end of parsing. It must traverse up the tree, - * closing out open [implicit] scopes. Obviously, open explicit scopes - * are errors. - */ -int -macro_end(struct mdoc *mdoc) +static int +macro_phrase(struct mdoc *mdoc, int line, int ppos, char *buf) { - struct mdoc_node *n; + int i, la, c; - assert(mdoc->first); - assert(mdoc->last); + for (i = ppos; buf[i]; ) { + assert(' ' != buf[i]); - /* Scan for open explicit scopes. */ + la = i; + if ('\"' == buf[i]) { + la = ++i; + while (buf[i] && '\"' != buf[i]) + i++; + if (0 == buf[i]) + return(mdoc_err(mdoc, "unterminated quoted parameter")); + } else + while (buf[i] && ! isspace ((unsigned char)buf[i])) + i++; - n = MDOC_VALID & mdoc->last->flags ? - mdoc->last->parent : mdoc->last; + if (buf[i]) + buf[i++] = 0; - for ( ; n; n = n->parent) { - if (MDOC_BLOCK != n->type) - continue; - if ( ! (MDOC_EXPLICIT & mdoc_macros[n->tok].flags)) - continue; - return(mdoc_nerr(mdoc, n, "macro scope still open on exit")); + while (buf[i] && isspace((unsigned char)buf[i])) + i++; + + if (MDOC_MAX != (c = mdoc_tokhash_find(mdoc->htab, &buf[la]))) { + if ( ! mdoc_macro(mdoc, c, line, la, &i, buf)) + return(0); + + return(1); + } + + if ( ! mdoc_word_alloc(mdoc, line, la, &buf[la])) + return(0); + mdoc->next = MDOC_NEXT_SIBLING; + + while (buf[i] && isspace((unsigned char)buf[i])) + i++; } - return(rewind_last(mdoc, mdoc->first)); + return(1); }