=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.19 retrieving revision 1.24 diff -u -p -r1.19 -r1.24 --- mandoc/roff.c 2008/11/29 16:23:22 1.19 +++ mandoc/roff.c 2008/11/30 21:41:35 1.24 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.19 2008/11/29 16:23:22 kristaps Exp $ */ +/* $Id: roff.c,v 1.24 2008/11/30 21:41:35 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -34,12 +34,10 @@ /* FIXME: warn about empty lists. */ -/* FIXME: ; : } ) (etc.) after text macros? */ +/* FIXME: roff_layout and roff_text have identical-ish lower bodies. */ /* FIXME: NAME section needs specific elements. */ -/* FIXME: don't print Os, just do roffbegin. */ - #define ROFF_MAXARG 32 enum roffd { @@ -70,9 +68,7 @@ struct rofftok { int flags; #define ROFF_PARSED (1 << 0) /* "Parsed". */ #define ROFF_CALLABLE (1 << 1) /* "Callable". */ -#define ROFF_QUOTES (1 << 2) /* Quoted args. */ -#define ROFF_SHALLOW (1 << 3) /* Nesting block. */ -#define ROFF_PUNCT (1 << 4) +#define ROFF_SHALLOW (1 << 2) /* Nesting block. */ }; struct roffarg { @@ -180,9 +176,9 @@ static const struct rofftok tokens[ROFF_MAX] = { {roff_comment, NULL, NULL, NULL, 0, ROFF_COMMENT, 0 }, /* \" */ { roff_Dd, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Dd */ { roff_Dt, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Dt */ - { roff_Os, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Os */ - { roff_layout, NULL, NULL, NULL, ROFF_Sh, ROFF_LAYOUT, ROFF_PARSED }, /* Sh */ - { roff_layout, NULL, NULL, NULL, ROFF_Ss, ROFF_LAYOUT, ROFF_PARSED }, /* Ss */ + { roff_Os, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Os */ + { roff_layout, NULL, NULL, NULL, ROFF_Sh, ROFF_LAYOUT, 0 }, /* Sh */ + { roff_layout, NULL, NULL, NULL, ROFF_Ss, ROFF_LAYOUT, 0 }, /* Ss */ { roff_text, NULL, NULL, NULL, ROFF_Pp, ROFF_TEXT, 0 }, /* Pp */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* D1 */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Dl */ @@ -190,34 +186,34 @@ static const struct rofftok tokens[ROFF_MAX] = { { roff_close, NULL, NULL, NULL, ROFF_Bd, ROFF_LAYOUT, 0 }, /* Ed */ { roff_layout, roffarg_Bl, NULL, roffchild_Bl, 0, ROFF_LAYOUT, 0 }, /* Bl */ { roff_close, NULL, roffparent_El, NULL, ROFF_Bl, ROFF_LAYOUT, 0 }, /* El */ - { roff_layout, NULL, roffparent_It, NULL, ROFF_It, ROFF_LAYOUT, ROFF_SHALLOW }, /* It */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Ad */ - { roff_text, roffarg_An, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* An */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Ar */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Cd */ /* XXX man.4 only */ + { roff_layout, NULL, roffparent_It, NULL, ROFF_It, ROFF_LAYOUT, ROFF_PARSED | ROFF_SHALLOW }, /* It */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ad */ /* FIXME */ + { roff_text, roffarg_An, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* An */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ar */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Cd */ /* XXX man.4 only */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Cm */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Dv */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Er */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Ev */ /* XXX needs arg */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dv */ /* XXX needs arg */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Er */ /* XXX needs arg */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ev */ /* XXX needs arg */ { roff_text, roffarg_Ex, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ex */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Fa */ /* XXX needs arg */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fa */ /* XXX needs arg */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Fd */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Fl */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Fn */ /* XXX needs arg */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* Ft */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Ic */ /* XXX needs arg */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fl */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fn */ /* XXX needs arg */ /* FIXME */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ft */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ic */ /* XXX needs arg */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* In */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Li */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Nd */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Nm */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Op */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Li */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Nd */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Nm */ /* FIXME */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Op */ { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ot */ /* XXX deprecated */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Pa */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pa */ { roff_text, roffarg_Rv, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Rv */ { roff_text, roffarg_St, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* St */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Va */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Vt */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Xr */ /* XXX needs arg */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Va */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Vt */ /* XXX needs arg */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xr */ /* XXX needs arg */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %A */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %B */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %D */ @@ -237,22 +233,22 @@ static const struct rofftok tokens[ROFF_MAX] = { { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Bf */ /* FIXME */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bo */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* Bsx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* Bx */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bsx */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bx */ {roff_special, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Db */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dc */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Do */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dq */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ec */ { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ef */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Em */ /* XXX needs arg */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Em */ /* XXX needs arg */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Eo */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* Fx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* Ms */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Fx */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ms */ { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* No */ { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ns */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* Nx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* Ox */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Nx */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ox */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pc */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Pf */ { roff_text, NULL, NULL, NULL, 0, ROFF_LAYOUT, ROFF_PARSED | ROFF_CALLABLE }, /* Po */ @@ -268,9 +264,9 @@ static const struct rofftok tokens[ROFF_MAX] = { { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sq */ {roff_special, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Sm */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Sy */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_PUNCT }, /* Tn */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_PUNCT }, /* Ux */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sy */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Tn */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ux */ { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xc */ { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xo */ { roff_layout, NULL, NULL, roffchild_Fo, 0, ROFF_LAYOUT, 0 }, /* Fo */ @@ -311,8 +307,11 @@ const char *const toknamesp[ROFF_MAX] = { "Ic", "In", "Li", "Nd", "Nm", "Op", "Ot", "Pa", "Rv", "St", "Va", "Vt", + /* LINTED */ "Xr", "\%A", "\%B", "\%D", + /* LINTED */ "\%I", "\%J", "\%N", "\%O", + /* LINTED */ "\%P", "\%R", "\%T", "\%V", "Ac", "Ao", "Aq", "At", "Bc", "Bf", "Bo", "Bq", @@ -442,7 +441,7 @@ static int textparse(const struct rofftree *tree, char *buf) { - return((*tree->cb.roffdata)(tree->arg, buf)); + return((*tree->cb.roffdata)(tree->arg, 1, buf)); } @@ -675,7 +674,7 @@ static int rofffindtok(const char *buf) { char token[4]; - size_t i; + int i; for (i = 0; *buf && ! isspace(*buf) && i < 3; i++, buf++) token[i] = *buf; @@ -698,6 +697,48 @@ rofffindtok(const char *buf) static int +roffispunct(const char *p) +{ + + if (0 == *p) + return(0); + if (0 != *(p + 1)) + return(0); + + switch (*p) { + case('{'): + /* FALLTHROUGH */ + case('.'): + /* FALLTHROUGH */ + case(','): + /* FALLTHROUGH */ + case(';'): + /* FALLTHROUGH */ + case(':'): + /* FALLTHROUGH */ + case('?'): + /* FALLTHROUGH */ + case('!'): + /* FALLTHROUGH */ + case('('): + /* FALLTHROUGH */ + case(')'): + /* FALLTHROUGH */ + case('['): + /* FALLTHROUGH */ + case(']'): + /* FALLTHROUGH */ + case('}'): + return(1); + default: + break; + } + + return(0); +} + + +static int rofffindcallable(const char *name) { int c; @@ -942,13 +983,16 @@ roff_layout(ROFFCALL_ARGS) return(0); if ( ! (ROFF_PARSED & tokens[tok].flags)) { + i = 0; while (*argv) { - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++)) return(0); + i = 1; } return((*tree->cb.roffout)(tree->arg, tok)); } + i = 0; while (*argv) { if (ROFF_MAX != (c = rofffindcallable(*argv))) { if (NULL == tokens[c].cb) { @@ -963,10 +1007,38 @@ roff_layout(ROFFCALL_ARGS) } assert(tree->arg); - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++)) return(0); + i = 1; } + /* + * If we're the first parser (*argv == tree->cur) then purge out + * any additional punctuation, should there be any remaining at + * the end of line. + */ + + if (ROFF_PARSED & tokens[tok].flags && *argv) { + i = 0; + while (argv[i]) + i++; + + assert(i > 0); + if ( ! roffispunct(argv[--i])) + return((*tree->cb.roffout)(tree->arg, tok)); + + while (i >= 0 && roffispunct(argv[i])) + i--; + + assert(0 != i); + i++; + + /* LINTED */ + while (argv[i]) + if ( ! (*tree->cb.roffdata)(tree->arg, 0, argv[i++])) + return(0); + } + return((*tree->cb.roffout)(tree->arg, tok)); } @@ -975,7 +1047,7 @@ roff_layout(ROFFCALL_ARGS) static int roff_text(ROFFCALL_ARGS) { - int i, c, argcp[ROFF_MAXARG]; + int i, j, first, c, argcp[ROFF_MAXARG]; char *v, *argvp[ROFF_MAXARG]; if (ROFF_PRELUDE & tree->state) { @@ -984,6 +1056,9 @@ roff_text(ROFFCALL_ARGS) return(0); } + /* FIXME: breaks if passed from roff_layout. */ + first = *argv == tree->cur; + i = 0; argv++; @@ -1004,31 +1079,86 @@ roff_text(ROFFCALL_ARGS) return(0); if ( ! (ROFF_PARSED & tokens[tok].flags)) { + i = 0; while (*argv) { - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++)) return(0); + i = 1; } return((*tree->cb.roffout)(tree->arg, tok)); } + i = 0; while (*argv) { - if (ROFF_MAX != (c = rofffindcallable(*argv))) { - if (NULL == tokens[c].cb) { - roff_err(tree, *argv, "unsupported " - "macro `%s'", - toknames[c]); - return(0); + if (ROFF_MAX == (c = rofffindcallable(*argv))) { + /* + * If all that remains is roff punctuation, then + * close out our scope and return. + */ + if (roffispunct(*argv)) { + for (j = 0; argv[j]; j++) + if ( ! roffispunct(argv[j])) + break; + if (NULL == argv[j]) + break; + i = 1; } - if ( ! (*tokens[c].cb)(c, tree, - argv, ROFF_ENTER)) + + if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++)) return(0); - break; + + i = 1; + continue; } - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + + /* + * A sub-command has been found. Execute it and + * discontinue parsing for arguments. + */ + + if (NULL == tokens[c].cb) { + roff_err(tree, *argv, "unsupported macro `%s'", + toknames[c]); return(0); + } + + if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) + return(0); + + break; } - return((*tree->cb.roffout)(tree->arg, tok)); + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); + + /* + * If we're the first parser (*argv == tree->cur) then purge out + * any additional punctuation, should there be any remaining at + * the end of line. + */ + + if (first && *argv) { + i = 0; + while (argv[i]) + i++; + + assert(i > 0); + if ( ! roffispunct(argv[--i])) + return(1); + + while (i >= 0 && roffispunct(argv[i])) + i--; + + assert(0 != i); + i++; + + /* LINTED */ + while (argv[i]) + if ( ! (*tree->cb.roffdata)(tree->arg, 0, argv[i++])) + return(0); + } + + return(1); }