=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.17 retrieving revision 1.26 diff -u -p -r1.17 -r1.26 --- mandoc/roff.c 2008/11/29 14:14:21 1.17 +++ mandoc/roff.c 2008/12/01 09:25:18 1.26 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.17 2008/11/29 14:14:21 kristaps Exp $ */ +/* $Id: roff.c,v 1.26 2008/12/01 09:25:18 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -34,7 +34,7 @@ /* 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. */ @@ -54,7 +54,7 @@ enum rofftype { #define ROFFCALL_ARGS \ int tok, struct rofftree *tree, \ - const char *argv[], enum roffd type + char *argv[], enum roffd type struct rofftree; @@ -68,8 +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_SHALLOW (1 << 2) /* Nesting block. */ }; struct roffarg { @@ -106,12 +105,14 @@ struct rofftree { static int roff_Dd(ROFFCALL_ARGS); static int roff_Dt(ROFFCALL_ARGS); static int roff_Os(ROFFCALL_ARGS); +#ifdef notyet +static int roff_Ns(ROFFCALL_ARGS); +#endif static int roff_layout(ROFFCALL_ARGS); static int roff_text(ROFFCALL_ARGS); static int roff_comment(ROFFCALL_ARGS); static int roff_close(ROFFCALL_ARGS); -static int roff_special(ROFFCALL_ARGS); static struct roffnode *roffnode_new(int, struct rofftree *); static void roffnode_free(struct rofftree *); @@ -129,7 +130,7 @@ static int roffargs(const struct rofftree *, int, char *, char **); static int roffargok(int, int); static int roffnextopt(const struct rofftree *, - int, const char ***, char **); + int, char ***, char **); static int roffparse(struct rofftree *, char *); static int textparse(const struct rofftree *, char *); @@ -177,9 +178,9 @@ 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 */ @@ -187,11 +188,11 @@ 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 }, /* Ad */ + { 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, ROFF_QUOTES }, /* Cd */ /* XXX man.4 only */ + { 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 }, /* Dv */ /* XXX needs arg */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Er */ /* XXX needs arg */ @@ -205,7 +206,7 @@ static const struct rofftok tokens[ROFF_MAX] = { { 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 }, /* Li */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Nd */ + { 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 */ @@ -233,10 +234,10 @@ static const struct rofftok tokens[ROFF_MAX] = { { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bc */ { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Bf */ /* FIXME */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bo */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bq */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bq */ { 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 */ + { NULL, 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 */ @@ -247,11 +248,11 @@ static const struct rofftok tokens[ROFF_MAX] = { { 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 */ + { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ns */ { 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 */ + { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Pf */ { roff_text, NULL, NULL, NULL, 0, ROFF_LAYOUT, ROFF_PARSED | ROFF_CALLABLE }, /* Po */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pq */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qc */ @@ -263,7 +264,7 @@ static const struct rofftok tokens[ROFF_MAX] = { { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sc */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* So */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sq */ - {roff_special, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Sm */ + { NULL, 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 }, /* Sy */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Tn */ @@ -284,6 +285,7 @@ static const int tokenargs[ROFF_ARGMAX] = { 0, ROFF_VALUE, ROFF_VALUE, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, ROFF_VALUE, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, @@ -294,7 +296,6 @@ static const int tokenargs[ROFF_ARGMAX] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, }; const char *const toknamesp[ROFF_MAX] = { @@ -308,8 +309,11 @@ 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", @@ -385,9 +389,6 @@ roff_free(struct rofftree *tree, int flush) goto end; } - if ( ! (*tree->cb.rofftail)(tree->arg)) - goto end; - error = 0; end: @@ -417,11 +418,6 @@ roff_alloc(const struct roffcb *cb, void *args) (void)memcpy(&tree->cb, cb, sizeof(struct roffcb)); - if ( ! (*tree->cb.roffhead)(args)) { - free(tree); - return(NULL); - } - return(tree); } @@ -447,7 +443,7 @@ static int textparse(const struct rofftree *tree, char *buf) { - return((*tree->cb.roffdata)(tree->arg, buf)); + return((*tree->cb.roffdata)(tree->arg, 1, buf)); } @@ -500,7 +496,6 @@ roffargs(const struct rofftree *tree, } -/* XXX */ static int roffscan(int tok, const int *tokv) { @@ -522,8 +517,12 @@ roffparse(struct rofftree *tree, char *buf) int tok, t; struct roffnode *n; char *argv[ROFF_MAXARG]; - const char **argvp; + char **argvp; + if (0 != *buf && 0 != *(buf + 1) && 0 != *(buf + 2)) + if (0 == strncmp(buf, ".\\\"", 3)) + return(1); + if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { roff_err(tree, buf + 1, "bogus line macro"); return(0); @@ -531,13 +530,13 @@ roffparse(struct rofftree *tree, char *buf) roff_err(tree, buf + 1, "unsupported macro `%s'", toknames[tok]); return(0); - } else if (ROFF_COMMENT == tokens[tok].type) - return(1); - + } + + assert(ROFF___ != tok); if ( ! roffargs(tree, tok, buf, argv)) return(0); - argvp = (const char **)argv; + argvp = (char **)argv; /* * Prelude macros break some assumptions, so branch now. @@ -564,9 +563,9 @@ roffparse(struct rofftree *tree, char *buf) } if ( ! roffscan(tok, tokens[tree->last->tok].children)) { - roff_err(tree, *argvp, "`%s' is invalid child `%s'", - toknames[tree->last->tok], - toknames[tok]); + roff_err(tree, *argvp, "`%s' is invalid child of `%s'", + toknames[tok], + toknames[tree->last->tok]); return(0); } @@ -680,7 +679,7 @@ 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; @@ -703,6 +702,48 @@ 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; @@ -761,9 +802,9 @@ roffnode_free(struct rofftree *tree) static int roffnextopt(const struct rofftree *tree, int tok, - const char ***in, char **val) + char ***in, char **val) { - const char *arg, **argv; + char *arg, **argv; int v; *val = NULL; @@ -870,8 +911,8 @@ roff_Os(ROFFCALL_ARGS) { if (ROFF_EXIT == type) { - assert(ROFF_PRELUDE_Os & tree->state); - return(roff_layout(tok, tree, argv, type)); + roffnode_free(tree); + return((*tree->cb.rofftail)(tree->arg)); } else if (ROFF_BODY & tree->state) { assert( ! (ROFF_PRELUDE & tree->state)); assert(ROFF_PRELUDE_Os & tree->state); @@ -893,7 +934,10 @@ roff_Os(ROFFCALL_ARGS) assert(NULL == tree->last); - return(roff_layout(tok, tree, argv, type)); + if (NULL == roffnode_new(tok, tree)) + return(0); + + return((*tree->cb.roffhead)(tree->arg)); } @@ -934,42 +978,77 @@ roff_layout(ROFFCALL_ARGS) if (NULL == roffnode_new(tok, tree)) return(0); + if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, argvp)) + return(0); + + if (NULL == *argv) + return(1); + if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) 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; } - - if ( ! ((*tree->cb.roffout)(tree->arg, tok))) - return(0); - return((*tree->cb.roffblkin)(tree->arg, tok)); + 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]); + if (ROFF_MAX == (c = rofffindcallable(*argv))) { + assert(tree->arg); + if ( ! (*tree->cb.roffdata) + (tree->arg, i, *argv++)) return(0); - } - if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) - return(0); - break; + i = 1; + continue; } - assert(tree->arg); - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + 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; } - if ( ! ((*tree->cb.roffout)(tree->arg, tok))) - return(0); - return((*tree->cb.roffblkin)(tree->arg, tok)); + /* + * 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)) + return((*tree->cb.roffout)(tree->arg, tok)); + + 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)); } @@ -977,7 +1056,7 @@ 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) { @@ -986,6 +1065,9 @@ roff_text(ROFFCALL_ARGS) return(0); } + /* FIXME: breaks if passed from roff_layout. */ + first = *argv == tree->cur; + i = 0; argv++; @@ -1006,31 +1088,88 @@ 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)) + return(1); + + 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); } @@ -1052,13 +1191,52 @@ roff_close(ROFFCALL_ARGS) } +#if notyet /* ARGSUSED */ static int -roff_special(ROFFCALL_ARGS) +roff_Ns(ROFFCALL_ARGS) { + int c; - return((*tree->cb.roffspecial)(tree->arg, tok)); + argv++; + + if (ROFF_MAX != (c = rofffindcallable(*argv))) { + if (NULL == tokens[c].cb) { + roff_err(tree, *argv, "unsupported macro `%s'", + toknames[c]); + return(0); + } + if ( ! (*tree->cb.roffspecial)(tree->arg, tok)) + return(0); + if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) + return(0); + + return(1); + } else if ( ! (*tree->cb.roffdata)(tree->arg, 0, *argv++)) + return(0); + + while (*argv) { + if (ROFF_MAX == (c = rofffindcallable(*argv))) { + assert(tree->arg); + if ( ! (*tree->cb.roffdata) + (tree->arg, 1, *argv++)) + return(0); + continue; + } + 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(1); } +#endif static void