=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.13 retrieving revision 1.20 diff -u -p -r1.13 -r1.20 --- mandoc/roff.c 2008/11/27 17:27:50 1.13 +++ mandoc/roff.c 2008/11/30 12:41:45 1.20 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.13 2008/11/27 17:27:50 kristaps Exp $ */ +/* $Id: roff.c,v 1.20 2008/11/30 12:41:45 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -38,6 +38,8 @@ /* FIXME: NAME section needs specific elements. */ +/* FIXME: don't print Os, just do roffbegin. */ + #define ROFF_MAXARG 32 enum roffd { @@ -54,7 +56,7 @@ enum rofftype { #define ROFFCALL_ARGS \ int tok, struct rofftree *tree, \ - const char *argv[], enum roffd type + char *argv[], enum roffd type struct rofftree; @@ -80,28 +82,27 @@ struct roffarg { struct roffnode { int tok; /* Token id. */ struct roffnode *parent; /* Parent (or NULL). */ - size_t line; /* Parsed at line. */ }; struct rofftree { struct roffnode *last; /* Last parsed node. */ - time_t date; /* `Dd' results. */ char *cur; + + time_t date; /* `Dd' results. */ char os[64]; /* `Os' results. */ char title[64]; /* `Dt' results. */ char section[64]; /* `Dt' results. */ char volume[64]; /* `Dt' results. */ + int state; #define ROFF_PRELUDE (1 << 1) /* In roff prelude. */ - /* FIXME: if we had prev ptrs, this wouldn't be necessary. */ #define ROFF_PRELUDE_Os (1 << 2) /* `Os' is parsed. */ #define ROFF_PRELUDE_Dt (1 << 3) /* `Dt' is parsed. */ #define ROFF_PRELUDE_Dd (1 << 4) /* `Dd' is parsed. */ #define ROFF_BODY (1 << 5) /* In roff body. */ - struct md_mbuf *mbuf; /* Output (or NULL). */ - const struct md_args *args; /* Global args. */ - const struct md_rbuf *rbuf; /* Input. */ - const struct roffcb *cb; + + struct roffcb cb; + void *arg; }; static int roff_Dd(ROFFCALL_ARGS); @@ -115,7 +116,7 @@ static int roff_close(ROFFCALL_ARGS); static int roff_special(ROFFCALL_ARGS); static struct roffnode *roffnode_new(int, struct rofftree *); -static void roffnode_free(int, struct rofftree *); +static void roffnode_free(struct rofftree *); static void roff_warn(const struct rofftree *, const char *, char *, ...); @@ -130,10 +131,9 @@ static int roffargs(const struct rofftree *, int, char *, char **); static int roffargok(int, int); static int roffnextopt(const struct rofftree *, - int, const char ***, char **); -static int roffparse(struct rofftree *, char *, size_t); -static int textparse(const struct rofftree *, - const char *, size_t); + int, char ***, char **); +static int roffparse(struct rofftree *, char *); +static int textparse(const struct rofftree *, char *); static const int roffarg_An[] = { ROFF_Split, ROFF_Nosplit, @@ -175,8 +175,7 @@ static const int roffparent_It[] = { ROFF_Bl, ROFF_It, static const int roffparent_Re[] = { ROFF_Rs, ROFF_MAX }; /* Table of all known tokens. */ -static const struct rofftok tokens[ROFF_MAX] = - { +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 */ @@ -191,7 +190,7 @@ static const struct rofftok tokens[ROFF_MAX] = { 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_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 */ @@ -236,7 +235,7 @@ 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 */ @@ -273,21 +272,21 @@ static const struct rofftok tokens[ROFF_MAX] = { 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, NULL, 0, ROFF_LAYOUT, 0 }, /* Fo */ + { roff_layout, NULL, NULL, roffchild_Fo, 0, ROFF_LAYOUT, 0 }, /* Fo */ { roff_close, NULL, roffparent_Fc, NULL, ROFF_Fo, ROFF_LAYOUT, 0 }, /* Fc */ - { roff_layout, NULL, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Oo */ + { roff_layout, NULL, NULL, roffchild_Oo, 0, ROFF_LAYOUT, 0 }, /* Oo */ { roff_close, NULL, roffparent_Oc, NULL, ROFF_Oo, ROFF_LAYOUT, 0 }, /* Oc */ { roff_layout, roffarg_Bk, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Bk */ { roff_close, NULL, NULL, NULL, ROFF_Bk, ROFF_LAYOUT, 0 }, /* Ek */ }; /* Table of all known token arguments. */ -static const int tokenargs[ROFF_ARGMAX] = - { +static const int tokenargs[ROFF_ARGMAX] = { 0, 0, 0, 0, 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, @@ -298,11 +297,9 @@ 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] = - { +const char *const toknamesp[ROFF_MAX] = { "\\\"", "Dd", "Dt", "Os", "Sh", "Ss", "Pp", "D1", "Dl", "Bd", "Ed", "Bl", @@ -331,8 +328,7 @@ const char *const toknamesp[ROFF_MAX] = "Bk", "Ek", }; -const char *const tokargnamesp[ROFF_ARGMAX] = - { +const char *const tokargnamesp[ROFF_ARGMAX] = { "split", "nosplit", "ragged", "unfilled", "literal", "file", "offset", "bullet", "dash", @@ -355,7 +351,6 @@ const char *const tokargnamesp[ROFF_ARGMAX] = "svid4", "filled", "words", }; - const char *const *toknames = toknamesp; const char *const *tokargnames = tokargnamesp; @@ -363,72 +358,90 @@ const char *const *tokargnames = tokargnamesp; int roff_free(struct rofftree *tree, int flush) { - int error; + int error, t; + struct roffnode *n; - assert(tree->mbuf); + error = 0; + if ( ! flush) - tree->mbuf = NULL; + goto end; - /* LINTED */ - while (tree->last) - if ( ! (*tokens[tree->last->tok].cb) - (tree->last->tok, tree, NULL, ROFF_EXIT)) - /* Disallow flushing. */ - tree->mbuf = NULL; + error = 1; - error = tree->mbuf ? 0 : 1; + if (ROFF_PRELUDE & tree->state) { + roff_warn(tree, NULL, "prelude never finished"); + goto end; + } - if (tree->mbuf && (ROFF_PRELUDE & tree->state)) { - /*roff_warn(tree, "prelude never finished");*/ - error = 1; + for (n = tree->last; n->parent; n = n->parent) { + if (0 != tokens[n->tok].ctx) + continue; + roff_warn(tree, NULL, "closing explicit scope `%s'", + toknames[n->tok]); + goto end; } + while (tree->last) { + t = tree->last->tok; + if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) + goto end; + } + + error = 0; + +end: + + while (tree->last) + roffnode_free(tree); + free(tree); + return(error ? 0 : 1); } struct rofftree * -roff_alloc(const struct md_args *args, struct md_mbuf *out, - const struct md_rbuf *in, const struct roffcb *cb) +roff_alloc(const struct roffcb *cb, void *args) { struct rofftree *tree; + assert(args); + assert(cb); + if (NULL == (tree = calloc(1, sizeof(struct rofftree)))) err(1, "calloc"); tree->state = ROFF_PRELUDE; - tree->args = args; - tree->mbuf = out; - tree->rbuf = in; - tree->cb = cb; + tree->arg = args; + (void)memcpy(&tree->cb, cb, sizeof(struct roffcb)); + return(tree); } int -roff_engine(struct rofftree *tree, char *buf, size_t sz) +roff_engine(struct rofftree *tree, char *buf) { tree->cur = buf; + assert(buf); - if (0 == sz) { + if (0 == *buf) { roff_warn(tree, buf, "blank line"); return(0); } else if ('.' != *buf) - return(textparse(tree, buf, sz)); + return(textparse(tree, buf)); - return(roffparse(tree, buf, sz)); + return(roffparse(tree, buf)); } static int -textparse(const struct rofftree *tree, const char *buf, size_t sz) +textparse(const struct rofftree *tree, char *buf) { - - /* Print text. */ - return(1); + + return((*tree->cb.roffdata)(tree->arg, buf)); } @@ -498,15 +511,13 @@ roffscan(int tok, const int *tokv) static int -roffparse(struct rofftree *tree, char *buf, size_t sz) +roffparse(struct rofftree *tree, char *buf) { int tok, t; struct roffnode *n; char *argv[ROFF_MAXARG]; - const char **argvp; + char **argvp; - assert(sz > 0); - if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { roff_err(tree, buf + 1, "bogus line macro"); return(0); @@ -520,7 +531,7 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) if ( ! roffargs(tree, tok, buf, argv)) return(0); - argvp = (const char **)argv; + argvp = (char **)argv; /* * Prelude macros break some assumptions, so branch now. @@ -540,18 +551,16 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) */ if ( ! roffscan(tree->last->tok, tokens[tok].parents)) { - warnx("%s: invalid parent `%s' for `%s' (line %zu)", - tree->rbuf->name, - toknames[tree->last->tok], - toknames[tok], tree->rbuf->line); + roff_err(tree, *argvp, "`%s' has invalid parent `%s'", + toknames[tok], + toknames[tree->last->tok]); return(0); } if ( ! roffscan(tok, tokens[tree->last->tok].children)) { - warnx("%s: invalid child `%s' for `%s' (line %zu)", - tree->rbuf->name, toknames[tok], - toknames[tree->last->tok], - tree->rbuf->line); + roff_err(tree, *argvp, "`%s' is invalid child of `%s'", + toknames[tok], + toknames[tree->last->tok]); return(0); } @@ -632,6 +641,7 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) assert(tree->last); assert(tok != tokens[tok].ctx && 0 != tokens[tok].ctx); + /* LINTED */ do { t = tree->last->tok; if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) @@ -686,7 +696,51 @@ rofffindtok(const char *buf) } +#if notyet 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); +} +#endif + + +static int rofffindcallable(const char *name) { int c; @@ -706,7 +760,6 @@ roffnode_new(int tokid, struct rofftree *tree) if (NULL == (p = malloc(sizeof(struct roffnode)))) err(1, "malloc"); - p->line = tree->rbuf->line; p->tok = tokid; p->parent = tree->last; tree->last = p; @@ -732,12 +785,11 @@ roffargok(int tokid, int argid) static void -roffnode_free(int tokid, struct rofftree *tree) +roffnode_free(struct rofftree *tree) { struct roffnode *p; assert(tree->last); - assert(tree->last->tok == tokid); p = tree->last; tree->last = tree->last->parent; @@ -747,9 +799,9 @@ roffnode_free(int tokid, 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; @@ -856,8 +908,8 @@ 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); @@ -879,7 +931,10 @@ 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)); } @@ -897,8 +952,8 @@ roff_layout(ROFFCALL_ARGS) } if (ROFF_EXIT == type) { - roffnode_free(tok, tree); - return((*tree->cb->roffblkout)(tree->args, tok)); + roffnode_free(tree); + return((*tree->cb.roffblkout)(tree->arg, tok)); } i = 0; @@ -920,16 +975,21 @@ roff_layout(ROFFCALL_ARGS) if (NULL == roffnode_new(tok, tree)) return(0); - if ( ! (*tree->cb->roffin)(tree->args, tok, argcp, argvp)) + if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, argvp)) return(0); - if ( ! (ROFF_PARSED & tokens[tok].flags)) { + if (NULL == *argv) + return(1); - /* TODO: print all tokens. */ + if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) + return(0); - if ( ! ((*tree->cb->roffout)(tree->args, tok))) - return(0); - return((*tree->cb->roffblkin)(tree->args, tok)); + if ( ! (ROFF_PARSED & tokens[tok].flags)) { + while (*argv) { + if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + return(0); + } + return((*tree->cb.roffout)(tree->arg, tok)); } while (*argv) { @@ -940,20 +1000,17 @@ roff_layout(ROFFCALL_ARGS) toknames[c]); return(0); } - if ( ! (*tokens[c].cb)(c, tree, - argv, ROFF_ENTER)) + if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) return(0); + break; } - /* TODO: print token. */ - - argv++; + assert(tree->arg); + if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + return(0); } - if ( ! ((*tree->cb->roffout)(tree->args, tok))) - return(0); - - return((*tree->cb->roffblkin)(tree->args, tok)); + return((*tree->cb.roffout)(tree->arg, tok)); } @@ -986,35 +1043,42 @@ roff_text(ROFFCALL_ARGS) argcp[i] = ROFF_ARGMAX; argvp[i] = NULL; - if ( ! (*tree->cb->roffin)(tree->args, tok, argcp, argvp)) + if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) return(0); if ( ! (ROFF_PARSED & tokens[tok].flags)) { - - /* TODO: print all tokens. */ - - return((*tree->cb->roffout)(tree->args, tok)); + while (*argv) { + if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + return(0); + } + return((*tree->cb.roffout)(tree->arg, tok)); } 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))) { + if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) return(0); - } - if ( ! (*tokens[c].cb)(c, tree, - argv, ROFF_ENTER)) - return(0); + continue; } - /* TODO: print token. */ + /* + * A sub-command has been found. Execute it and + * discontinue parsing for arguments. + */ - 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; } - return((*tree->cb->roffout)(tree->args, tok)); + return((*tree->cb.roffout)(tree->arg, tok)); } @@ -1041,7 +1105,7 @@ static int roff_special(ROFFCALL_ARGS) { - return((*tree->cb->roffspecial)(tok)); + return((*tree->cb.roffspecial)(tree->arg, tok)); } @@ -1055,8 +1119,8 @@ roff_warn(const struct rofftree *tree, const char *pos (void)vsnprintf(buf, sizeof(buf), fmt, ap); va_end(ap); - (*tree->cb->roffmsg)(tree->args, ROFF_WARN, tree->cur, pos, - tree->rbuf->name, tree->rbuf->line, buf); + (*tree->cb.roffmsg)(tree->arg, + ROFF_WARN, tree->cur, pos, buf); } @@ -1070,6 +1134,6 @@ roff_err(const struct rofftree *tree, const char *pos, (void)vsnprintf(buf, sizeof(buf), fmt, ap); va_end(ap); - (*tree->cb->roffmsg)(tree->args, ROFF_ERROR, tree->cur, pos, - tree->rbuf->name, tree->rbuf->line, buf); + (*tree->cb.roffmsg)(tree->arg, + ROFF_ERROR, tree->cur, pos, buf); }