=================================================================== RCS file: /cvs/mandoc/mdoc_validate.c,v retrieving revision 1.120 retrieving revision 1.132 diff -u -p -r1.120 -r1.132 --- mandoc/mdoc_validate.c 2010/10/11 13:24:33 1.120 +++ mandoc/mdoc_validate.c 2010/11/30 12:59:20 1.132 @@ -1,4 +1,4 @@ -/* $Id: mdoc_validate.c,v 1.120 2010/10/11 13:24:33 kristaps Exp $ */ +/* $Id: mdoc_validate.c,v 1.132 2010/11/30 12:59:20 kristaps Exp $ */ /* * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons * @@ -18,6 +18,10 @@ #include "config.h" #endif +#ifndef OSNAME +#include +#endif + #include #include @@ -26,6 +30,7 @@ #include #include #include +#include #include "mandoc.h" #include "libmdoc.h" @@ -36,6 +41,9 @@ #define PRE_ARGS struct mdoc *mdoc, struct mdoc_node *n #define POST_ARGS struct mdoc *mdoc +#define NUMSIZ 32 +#define DATESIZ 32 + enum check_ineq { CHECK_LT, CHECK_GT, @@ -65,6 +73,9 @@ static int check_argv(struct mdoc *, struct mdoc_node *, struct mdoc_argv *); static int check_args(struct mdoc *, struct mdoc_node *); +static int concat(struct mdoc *, char *, + const struct mdoc_node *, size_t); + static int ebool(POST_ARGS); static int berr_ge1(POST_ARGS); static int bwarn_ge1(POST_ARGS); @@ -84,18 +95,27 @@ static int post_an(POST_ARGS); static int post_at(POST_ARGS); static int post_bf(POST_ARGS); static int post_bl(POST_ARGS); +static int post_bl_block(POST_ARGS); +static int post_bl_block_width(POST_ARGS); +static int post_bl_block_tag(POST_ARGS); static int post_bl_head(POST_ARGS); +static int post_dd(POST_ARGS); static int post_dt(POST_ARGS); +static int post_defaults(POST_ARGS); +static int post_literal(POST_ARGS); +static int post_eoln(POST_ARGS); static int post_it(POST_ARGS); static int post_lb(POST_ARGS); static int post_nm(POST_ARGS); +static int post_os(POST_ARGS); +static int post_prol(POST_ARGS); static int post_root(POST_ARGS); static int post_rs(POST_ARGS); static int post_sh(POST_ARGS); static int post_sh_body(POST_ARGS); static int post_sh_head(POST_ARGS); static int post_st(POST_ARGS); -static int post_eoln(POST_ARGS); +static int post_std(POST_ARGS); static int post_vt(POST_ARGS); static int pre_an(PRE_ARGS); static int pre_bd(PRE_ARGS); @@ -104,40 +124,48 @@ static int pre_dd(PRE_ARGS); static int pre_display(PRE_ARGS); static int pre_dt(PRE_ARGS); static int pre_it(PRE_ARGS); +static int pre_literal(PRE_ARGS); static int pre_os(PRE_ARGS); -static int pre_pp(PRE_ARGS); +static int pre_par(PRE_ARGS); static int pre_rv(PRE_ARGS); static int pre_sh(PRE_ARGS); static int pre_ss(PRE_ARGS); static v_post posts_an[] = { post_an, NULL }; -static v_post posts_at[] = { post_at, NULL }; -static v_post posts_bd_bk[] = { hwarn_eq0, bwarn_ge1, NULL }; +static v_post posts_at[] = { post_at, post_defaults, NULL }; +static v_post posts_bd[] = { post_literal, hwarn_eq0, bwarn_ge1, NULL }; static v_post posts_bf[] = { hwarn_le1, post_bf, NULL }; +static v_post posts_bk[] = { hwarn_eq0, bwarn_ge1, NULL }; static v_post posts_bl[] = { bwarn_ge1, post_bl, NULL }; static v_post posts_bool[] = { eerr_eq1, ebool, NULL }; static v_post posts_eoln[] = { post_eoln, NULL }; -static v_post posts_dt[] = { post_dt, NULL }; +static v_post posts_defaults[] = { post_defaults, NULL }; +static v_post posts_dd[] = { ewarn_ge1, post_dd, post_prol, NULL }; +static v_post posts_dl[] = { post_literal, bwarn_ge1, herr_eq0, NULL }; +static v_post posts_dt[] = { post_dt, post_prol, NULL }; static v_post posts_fo[] = { hwarn_eq1, bwarn_ge1, NULL }; static v_post posts_it[] = { post_it, NULL }; static v_post posts_lb[] = { eerr_eq1, post_lb, NULL }; static v_post posts_nd[] = { berr_ge1, NULL }; static v_post posts_nm[] = { post_nm, NULL }; static v_post posts_notext[] = { ewarn_eq0, NULL }; +static v_post posts_os[] = { post_os, post_prol, NULL }; static v_post posts_rs[] = { berr_ge1, herr_eq0, post_rs, NULL }; static v_post posts_sh[] = { herr_ge1, bwarn_ge1, post_sh, NULL }; static v_post posts_sp[] = { eerr_le1, NULL }; static v_post posts_ss[] = { herr_ge1, NULL }; static v_post posts_st[] = { eerr_eq1, post_st, NULL }; +static v_post posts_std[] = { post_std, NULL }; static v_post posts_text[] = { eerr_ge1, NULL }; static v_post posts_text1[] = { eerr_eq1, NULL }; static v_post posts_vt[] = { post_vt, NULL }; static v_post posts_wline[] = { bwarn_ge1, herr_eq0, NULL }; static v_post posts_wtext[] = { ewarn_ge1, NULL }; static v_pre pres_an[] = { pre_an, NULL }; -static v_pre pres_bd[] = { pre_display, pre_bd, pre_pp, NULL }; -static v_pre pres_bl[] = { pre_bl, pre_pp, NULL }; +static v_pre pres_bd[] = { pre_display, pre_bd, pre_literal, pre_par, NULL }; +static v_pre pres_bl[] = { pre_bl, pre_par, NULL }; static v_pre pres_d1[] = { pre_display, NULL }; +static v_pre pres_dl[] = { pre_literal, pre_display, NULL }; static v_pre pres_dd[] = { pre_dd, NULL }; static v_pre pres_dt[] = { pre_dt, NULL }; static v_pre pres_er[] = { NULL, NULL }; @@ -145,35 +173,35 @@ static v_pre pres_ex[] = { NULL, NULL }; static v_pre pres_fd[] = { NULL, NULL }; static v_pre pres_it[] = { pre_it, NULL }; static v_pre pres_os[] = { pre_os, NULL }; -static v_pre pres_pp[] = { pre_pp, NULL }; +static v_pre pres_pp[] = { pre_par, NULL }; static v_pre pres_rv[] = { pre_rv, NULL }; static v_pre pres_sh[] = { pre_sh, NULL }; static v_pre pres_ss[] = { pre_ss, NULL }; const struct valids mdoc_valids[MDOC_MAX] = { { NULL, NULL }, /* Ap */ - { pres_dd, posts_wtext }, /* Dd */ + { pres_dd, posts_dd }, /* Dd */ { pres_dt, posts_dt }, /* Dt */ - { pres_os, NULL }, /* Os */ + { pres_os, posts_os }, /* Os */ { pres_sh, posts_sh }, /* Sh */ { pres_ss, posts_ss }, /* Ss */ { pres_pp, posts_notext }, /* Pp */ { pres_d1, posts_wline }, /* D1 */ - { pres_d1, posts_wline }, /* Dl */ - { pres_bd, posts_bd_bk }, /* Bd */ + { pres_dl, posts_dl }, /* Dl */ + { pres_bd, posts_bd }, /* Bd */ { NULL, NULL }, /* Ed */ { pres_bl, posts_bl }, /* Bl */ { NULL, NULL }, /* El */ { pres_it, posts_it }, /* It */ { NULL, posts_text }, /* Ad */ { pres_an, posts_an }, /* An */ - { NULL, NULL }, /* Ar */ + { NULL, posts_defaults }, /* Ar */ { NULL, posts_text }, /* Cd */ { NULL, NULL }, /* Cm */ { NULL, NULL }, /* Dv */ { pres_er, posts_text }, /* Er */ { NULL, NULL }, /* Ev */ - { pres_ex, NULL }, /* Ex */ + { pres_ex, posts_std }, /* Ex */ { NULL, NULL }, /* Fa */ { pres_fd, posts_wtext }, /* Fd */ { NULL, NULL }, /* Fl */ @@ -181,13 +209,13 @@ const struct valids mdoc_valids[MDOC_MAX] = { { NULL, posts_wtext }, /* Ft */ { NULL, posts_text }, /* Ic */ { NULL, posts_text1 }, /* In */ - { NULL, NULL }, /* Li */ + { NULL, posts_defaults }, /* Li */ { NULL, posts_nd }, /* Nd */ { NULL, posts_nm }, /* Nm */ { NULL, posts_wline }, /* Op */ { NULL, NULL }, /* Ot */ - { NULL, NULL }, /* Pa */ - { pres_rv, NULL }, /* Rv */ + { NULL, posts_defaults }, /* Pa */ + { pres_rv, posts_std }, /* Rv */ { NULL, posts_st }, /* St */ { NULL, NULL }, /* Va */ { NULL, posts_vt }, /* Vt */ @@ -251,7 +279,7 @@ const struct valids mdoc_valids[MDOC_MAX] = { { NULL, NULL }, /* Fc */ { NULL, NULL }, /* Oo */ { NULL, NULL }, /* Oc */ - { NULL, posts_bd_bk }, /* Bk */ + { NULL, posts_bk }, /* Bk */ { NULL, NULL }, /* Ek */ { NULL, posts_eoln }, /* Bt */ { NULL, NULL }, /* Hf */ @@ -260,7 +288,7 @@ const struct valids mdoc_valids[MDOC_MAX] = { { NULL, posts_lb }, /* Lb */ { NULL, posts_notext }, /* Lp */ { NULL, posts_text }, /* Lk */ - { NULL, posts_text }, /* Mt */ + { NULL, posts_defaults }, /* Mt */ { NULL, posts_wline }, /* Brq */ { NULL, NULL }, /* Bro */ { NULL, NULL }, /* Brc */ @@ -275,7 +303,26 @@ const struct valids mdoc_valids[MDOC_MAX] = { { NULL, NULL }, /* Ta */ }; +#define RSORD_MAX 14 /* Number of `Rs' blocks. */ +static const enum mdoct rsord[RSORD_MAX] = { + MDOC__A, + MDOC__T, + MDOC__B, + MDOC__I, + MDOC__J, + MDOC__R, + MDOC__N, + MDOC__V, + MDOC__P, + MDOC__Q, + MDOC__D, + MDOC__O, + MDOC__C, + MDOC__U +}; + + int mdoc_valid_pre(struct mdoc *mdoc, struct mdoc_node *n) { @@ -561,6 +608,7 @@ pre_display(PRE_ARGS) if (MDOC_BLOCK == node->type) if (MDOC_Bd == node->tok) break; + if (NULL == node) return(1); @@ -893,15 +941,16 @@ pre_it(PRE_ARGS) static int pre_an(PRE_ARGS) { + int i; if (NULL == n->args) return(1); - if (n->args->argc > 1) - if ( ! mdoc_nmsg(mdoc, n, MANDOCERR_ARGCOUNT)) + + for (i = 1; i < (int)n->args->argc; i++) + if ( ! mdoc_pmsg(mdoc, n->args->argv[i].line, + n->args->argv[i].pos, MANDOCERR_IGNARGV)) return(0); - /* FIXME: this should use a different error message. */ - if (MDOC_Split == n->args->argv[0].arg) n->data.An.auth = AUTH_split; else if (MDOC_Nosplit == n->args->argv[0].arg) @@ -912,7 +961,6 @@ pre_an(PRE_ARGS) return(1); } - static int pre_rv(PRE_ARGS) { @@ -920,27 +968,7 @@ pre_rv(PRE_ARGS) return(check_stdarg(mdoc, n)); } - static int -post_dt(POST_ARGS) -{ - const struct mdoc_node *nn; - const char *p; - - if (NULL != (nn = mdoc->last->child)) - for (p = nn->string; *p; p++) { - if (toupper((u_char)*p) == *p) - continue; - if ( ! mdoc_nmsg(mdoc, nn, MANDOCERR_UPPERCASE)) - return(0); - break; - } - - return(1); -} - - -static int pre_dt(PRE_ARGS) { @@ -953,7 +981,6 @@ pre_dt(PRE_ARGS) return(1); } - static int pre_os(PRE_ARGS) { @@ -967,7 +994,6 @@ pre_os(PRE_ARGS) return(1); } - static int pre_dd(PRE_ARGS) { @@ -1055,17 +1081,38 @@ post_bf(POST_ARGS) return(1); } - static int post_lb(POST_ARGS) { + const char *p; + char *buf; + size_t sz; - if (mdoc_a2lib(mdoc->last->child->string)) + assert(mdoc->last->child); + assert(MDOC_TEXT == mdoc->last->child->type); + + p = mdoc_a2lib(mdoc->last->child->string); + + /* If lookup ok, replace with table value. */ + + if (p) { + free(mdoc->last->child->string); + mdoc->last->child->string = mandoc_strdup(p); return(1); - return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADLIB)); -} + } + /* If not, use "library ``xxxx''. */ + sz = strlen(mdoc->last->child->string) + + 2 + strlen("\\(lqlibrary\\(rq"); + buf = mandoc_malloc(sz); + snprintf(buf, sz, "library \\(lq%s\\(rq", + mdoc->last->child->string); + free(mdoc->last->child->string); + mdoc->last->child->string = buf; + return(1); +} + static int post_eoln(POST_ARGS) { @@ -1105,28 +1152,129 @@ post_vt(POST_ARGS) static int post_nm(POST_ARGS) { + char buf[BUFSIZ]; - if (mdoc->last->child) + /* If no child specified, make sure we have the meta name. */ + + if (NULL == mdoc->last->child && NULL == mdoc->meta.name) { + mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_NONAME); return(1); - if (mdoc->meta.name) + } else if (mdoc->meta.name) return(1); - return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_NONAME)); + + /* If no meta name, set it from the child. */ + + if ( ! concat(mdoc, buf, mdoc->last->child, BUFSIZ)) + return(0); + + mdoc->meta.name = mandoc_strdup(buf); + + return(1); } +static int +post_literal(POST_ARGS) +{ + + /* + * The `Dl' (note "el" not "one") and `Bd' macros unset the + * MDOC_LITERAL flag as they leave. Note that `Bd' only sets + * this in literal mode, but it doesn't hurt to just switch it + * off in general since displays can't be nested. + */ + if (MDOC_BODY == mdoc->last->type) + mdoc->last->flags &= ~MDOC_LITERAL; + + return(1); +} + static int +post_defaults(POST_ARGS) +{ + struct mdoc_node *nn; + + /* + * The `Ar' defaults to "file ..." if no value is provided as an + * argument; the `Mt' and `Pa' macros use "~"; the `Li' just + * gets an empty string. + */ + + if (mdoc->last->child) + return(1); + + nn = mdoc->last; + mdoc->next = MDOC_NEXT_CHILD; + + switch (nn->tok) { + case (MDOC_Ar): + if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "file")) + return(0); + if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "...")) + return(0); + break; + case (MDOC_At): + if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "AT&T")) + return(0); + if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "UNIX")) + return(0); + break; + case (MDOC_Li): + if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "")) + return(0); + break; + case (MDOC_Pa): + /* FALLTHROUGH */ + case (MDOC_Mt): + if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "~")) + return(0); + break; + default: + abort(); + /* NOTREACHED */ + } + + mdoc->last = nn; + return(1); +} + +static int post_at(POST_ARGS) { + const char *p, *q; + char *buf; + size_t sz; + /* + * If we have a child, look it up in the standard keys. If a + * key exist, use that instead of the child; if it doesn't, + * prefix "AT&T UNIX " to the existing data. + */ + if (NULL == mdoc->last->child) return(1); + assert(MDOC_TEXT == mdoc->last->child->type); - if (mdoc_a2att(mdoc->last->child->string)) - return(1); - return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADATT)); -} + p = mdoc_a2att(mdoc->last->child->string); + if (p) { + free(mdoc->last->child->string); + mdoc->last->child->string = mandoc_strdup(p); + } else { + mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADATT); + p = "AT&T UNIX "; + q = mdoc->last->child->string; + sz = strlen(p) + strlen(q) + 1; + buf = mandoc_malloc(sz); + strlcpy(buf, p, sz); + strlcat(buf, q, sz); + free(mdoc->last->child->string); + mdoc->last->child->string = buf; + } + return(1); +} + static int post_an(POST_ARGS) { @@ -1231,28 +1379,220 @@ post_it(POST_ARGS) return(1); } - static int -post_bl_head(POST_ARGS) +post_bl_block(POST_ARGS) { struct mdoc_node *n; - assert(mdoc->last->parent); - n = mdoc->last->parent; + /* + * These are fairly complicated, so we've broken them into two + * functions. post_bl_block_tag() is called when a -tag is + * specified, but no -width (it must be guessed). The second + * when a -width is specified (macro indicators must be + * rewritten into real lengths). + */ - if (LIST_column == n->data.Bl->type) { - if (n->data.Bl->ncols && mdoc->last->nchild) { - mdoc_nmsg(mdoc, n, MANDOCERR_COLUMNS); + n = mdoc->last; + + if (LIST_tag == n->data.Bl->type && + NULL == n->data.Bl->width) { + if ( ! post_bl_block_tag(mdoc)) return(0); + } else if (NULL != n->data.Bl->width) { + if ( ! post_bl_block_width(mdoc)) + return(0); + } else + return(1); + + assert(n->data.Bl->width); + return(1); +} + +static int +post_bl_block_width(POST_ARGS) +{ + size_t width; + int i; + enum mdoct tok; + struct mdoc_node *n; + char buf[NUMSIZ]; + + n = mdoc->last; + + /* + * Calculate the real width of a list from the -width string, + * which may contain a macro (with a known default width), a + * literal string, or a scaling width. + * + * If the value to -width is a macro, then we re-write it to be + * the macro's width as set in share/tmac/mdoc/doc-common. + */ + + if (0 == strcmp(n->data.Bl->width, "Ds")) + width = 6; + else if (MDOC_MAX == (tok = mdoc_hash_find(n->data.Bl->width))) + return(1); + else if (0 == (width = mdoc_macro2len(tok))) + return(mdoc_nmsg(mdoc, n, MANDOCERR_BADWIDTH)); + + /* The value already exists: free and reallocate it. */ + + assert(n->args); + + for (i = 0; i < (int)n->args->argc; i++) + if (MDOC_Width == n->args->argv[i].arg) + break; + + assert(i < (int)n->args->argc); + + snprintf(buf, NUMSIZ, "%zun", width); + free(n->args->argv[i].value[0]); + n->args->argv[i].value[0] = mandoc_strdup(buf); + + /* Set our width! */ + n->data.Bl->width = n->args->argv[i].value[0]; + return(1); +} + +static int +post_bl_block_tag(POST_ARGS) +{ + struct mdoc_node *n, *nn; + size_t sz, ssz; + int i; + char buf[NUMSIZ]; + + /* + * Calculate the -width for a `Bl -tag' list if it hasn't been + * provided. Uses the first head macro. NOTE AGAIN: this is + * ONLY if the -width argument has NOT been provided. See + * post_bl_block_width() for converting the -width string. + */ + + sz = 10; + n = mdoc->last; + + for (nn = n->body->child; nn; nn = nn->next) { + if (MDOC_It != nn->tok) + continue; + + assert(MDOC_BLOCK == nn->type); + nn = nn->head->child; + + if (nn == NULL) { + /* No -width for .Bl and first .It is emtpy */ + if ( ! mdoc_nmsg(mdoc, n, MANDOCERR_NOWIDTHARG)) + return(0); + break; } + + if (MDOC_TEXT == nn->type) { + sz = strlen(nn->string) + 1; + break; + } + + if (0 != (ssz = mdoc_macro2len(nn->tok))) + sz = ssz; + else if ( ! mdoc_nmsg(mdoc, n, MANDOCERR_NOWIDTHARG)) + return(0); + + break; + } + + /* Defaults to ten ens. */ + + snprintf(buf, NUMSIZ, "%zun", sz); + + /* + * We have to dynamically add this to the macro's argument list. + * We're guaranteed that a MDOC_Width doesn't already exist. + */ + + assert(n->args); + i = (int)(n->args->argc)++; + + n->args->argv = mandoc_realloc(n->args->argv, + n->args->argc * sizeof(struct mdoc_argv)); + + n->args->argv[i].arg = MDOC_Width; + n->args->argv[i].line = n->line; + n->args->argv[i].pos = n->pos; + n->args->argv[i].sz = 1; + n->args->argv[i].value = mandoc_malloc(sizeof(char *)); + n->args->argv[i].value[0] = mandoc_strdup(buf); + + /* Set our width! */ + n->data.Bl->width = n->args->argv[i].value[0]; + return(1); +} + + +static int +post_bl_head(POST_ARGS) +{ + struct mdoc_node *np, *nn, *nnp; + int i, j; + + if (LIST_column != mdoc->last->data.Bl->type) + /* FIXME: this should be ERROR class... */ + return(hwarn_eq0(mdoc)); + + /* + * Convert old-style lists, where the column width specifiers + * trail as macro parameters, to the new-style ("normal-form") + * lists where they're argument values following -column. + */ + + /* First, disallow both types and allow normal-form. */ + + /* + * TODO: technically, we can accept both and just merge the two + * lists, but I'll leave that for another day. + */ + + if (mdoc->last->data.Bl->ncols && mdoc->last->nchild) { + mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_COLUMNS); + return(0); + } else if (NULL == mdoc->last->child) return(1); + + np = mdoc->last->parent; + assert(np->args); + + for (j = 0; j < (int)np->args->argc; j++) + if (MDOC_Column == np->args->argv[j].arg) + break; + + assert(j < (int)np->args->argc); + assert(0 == np->args->argv[j].sz); + + /* + * Accomodate for new-style groff column syntax. Shuffle the + * child nodes, all of which must be TEXT, as arguments for the + * column field. Then, delete the head children. + */ + + np->args->argv[j].sz = (size_t)mdoc->last->nchild; + np->args->argv[j].value = mandoc_malloc + ((size_t)mdoc->last->nchild * sizeof(char *)); + + mdoc->last->data.Bl->ncols = np->args->argv[j].sz; + mdoc->last->data.Bl->cols = (const char **)np->args->argv[j].value; + + for (i = 0, nn = mdoc->last->child; nn; i++) { + np->args->argv[j].value[i] = nn->string; + nn->string = NULL; + nnp = nn; + nn = nn->next; + mdoc_node_delete(NULL, nnp); } - /* FIXME: should be ERROR class. */ - return(hwarn_eq0(mdoc)); + mdoc->last->nchild = 0; + mdoc->last->child = NULL; + + return(1); } - static int post_bl(POST_ARGS) { @@ -1260,6 +1600,8 @@ post_bl(POST_ARGS) if (MDOC_HEAD == mdoc->last->type) return(post_bl_head(mdoc)); + if (MDOC_BLOCK == mdoc->last->type) + return(post_bl_block(mdoc)); if (MDOC_BODY != mdoc->last->type) return(1); if (NULL == mdoc->last->child) @@ -1325,60 +1667,119 @@ post_root(POST_ARGS) return(0); } - static int post_st(POST_ARGS) { + const char *p; - if (mdoc_a2st(mdoc->last->child->string)) - return(1); - return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADSTANDARD)); -} + assert(MDOC_TEXT == mdoc->last->child->type); + p = mdoc_a2st(mdoc->last->child->string); + if (p == NULL) { + mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADSTANDARD); + mdoc_node_delete(mdoc, mdoc->last); + } else { + free(mdoc->last->child->string); + mdoc->last->child->string = mandoc_strdup(p); + } + + return(1); +} + static int post_rs(POST_ARGS) { - struct mdoc_node *nn; + struct mdoc_node *nn, *next, *prev; + int i, j; if (MDOC_BODY != mdoc->last->type) return(1); - for (nn = mdoc->last->child; nn; nn = nn->next) - switch (nn->tok) { - case(MDOC__U): - /* FALLTHROUGH */ - case(MDOC__Q): - /* FALLTHROUGH */ - case(MDOC__C): - /* FALLTHROUGH */ - case(MDOC__A): - /* FALLTHROUGH */ - case(MDOC__B): - /* FALLTHROUGH */ - case(MDOC__D): - /* FALLTHROUGH */ - case(MDOC__I): - /* FALLTHROUGH */ - case(MDOC__J): - /* FALLTHROUGH */ - case(MDOC__N): - /* FALLTHROUGH */ - case(MDOC__O): - /* FALLTHROUGH */ - case(MDOC__P): - /* FALLTHROUGH */ - case(MDOC__R): - /* FALLTHROUGH */ - case(MDOC__T): - /* FALLTHROUGH */ - case(MDOC__V): - break; - default: - mdoc_nmsg(mdoc, nn, MANDOCERR_SYNTCHILD); - return(0); + /* + * Make sure only certain types of nodes are allowed within the + * the `Rs' body. Delete offending nodes and raise a warning. + * Do this before re-ordering for the sake of clarity. + */ + + next = NULL; + for (nn = mdoc->last->child; nn; nn = next) { + for (i = 0; i < RSORD_MAX; i++) + if (nn->tok == rsord[i]) + break; + + if (i < RSORD_MAX) { + next = nn->next; + continue; } + next = nn->next; + mdoc_nmsg(mdoc, nn, MANDOCERR_CHILD); + mdoc_node_delete(mdoc, nn); + } + + /* + * The full `Rs' block needs special handling to order the + * sub-elements according to `rsord'. Pick through each element + * and correctly order it. This is a insertion sort. + */ + + next = NULL; + for (nn = mdoc->last->child->next; nn; nn = next) { + /* Determine order of `nn'. */ + for (i = 0; i < RSORD_MAX; i++) + if (rsord[i] == nn->tok) + break; + + /* + * Remove `nn' from the chain. This somewhat + * repeats mdoc_node_unlink(), but since we're + * just re-ordering, there's no need for the + * full unlink process. + */ + + if (NULL != (next = nn->next)) + next->prev = nn->prev; + + if (NULL != (prev = nn->prev)) + prev->next = nn->next; + + nn->prev = nn->next = NULL; + + /* + * Scan back until we reach a node that's + * ordered before `nn'. + */ + + for ( ; prev ; prev = prev->prev) { + /* Determine order of `prev'. */ + for (j = 0; j < RSORD_MAX; j++) + if (rsord[j] == prev->tok) + break; + + if (j <= i) + break; + } + + /* + * Set `nn' back into its correct place in front + * of the `prev' node. + */ + + nn->prev = prev; + + if (prev) { + if (prev->next) + prev->next->prev = nn; + nn->next = prev->next; + prev->next = nn; + } else { + mdoc->last->child->prev = nn; + nn->next = mdoc->last->child; + mdoc->last->child = nn; + } + } + return(1); } @@ -1432,77 +1833,75 @@ post_sh_body(POST_ARGS) static int post_sh_head(POST_ARGS) { - char buf[BUFSIZ]; - enum mdoc_sec sec; - const struct mdoc_node *n; + char buf[BUFSIZ]; + enum mdoc_sec sec; /* * Process a new section. Sections are either "named" or - * "custom"; custom sections are user-defined, while named ones - * usually follow a conventional order and may only appear in - * certain manual sections. + * "custom". Custom sections are user-defined, while named ones + * follow a conventional order and may only appear in certain + * manual sections. */ - buf[0] = '\0'; + if ( ! concat(mdoc, buf, mdoc->last->child, BUFSIZ)) + return(0); - /* - * FIXME: yes, these can use a dynamic buffer, but I don't do so - * in the interests of simplicity. - */ + sec = mdoc_str2sec(buf); - for (n = mdoc->last->child; n; n = n->next) { - /* XXX - copied from compact(). */ - assert(MDOC_TEXT == n->type); + /* The NAME should be first. */ - if (strlcat(buf, n->string, BUFSIZ) >= BUFSIZ) { - mdoc_nmsg(mdoc, n, MANDOCERR_MEM); - return(0); - } - if (NULL == n->next) - continue; - if (strlcat(buf, " ", BUFSIZ) >= BUFSIZ) { - mdoc_nmsg(mdoc, n, MANDOCERR_MEM); - return(0); - } - } + if (SEC_NAME != sec && SEC_NONE == mdoc->lastnamed) + mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_NAMESECFIRST); - sec = mdoc_str2sec(buf); + /* The SYNOPSIS gets special attention in other areas. */ - /* - * Check: NAME should always be first, CUSTOM has no roles, - * non-CUSTOM has a conventional order to be followed. - */ + if (SEC_SYNOPSIS == sec) + mdoc->flags |= MDOC_SYNOPSIS; + else + mdoc->flags &= ~MDOC_SYNOPSIS; - if (SEC_NAME != sec && SEC_NONE == mdoc->lastnamed) - if ( ! mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_NAMESECFIRST)) - return(0); + /* Mark our last section. */ + mdoc->lastsec = sec; + + /* We don't care about custom sections after this. */ + if (SEC_CUSTOM == sec) return(1); + /* + * Check whether our non-custom section is being repeated or is + * out of order. + */ + if (sec == mdoc->lastnamed) - if ( ! mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECREP)) - return(0); + mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECREP); if (sec < mdoc->lastnamed) - if ( ! mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECOOO)) - return(0); + mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECOOO); - /* - * Check particular section/manual conventions. LIBRARY can - * only occur in manual section 2, 3, and 9. - */ + /* Mark the last named section. */ + mdoc->lastnamed = sec; + + /* Check particular section/manual conventions. */ + + assert(mdoc->meta.msec); + switch (sec) { + case (SEC_RETURN_VALUES): + /* FALLTHROUGH */ + case (SEC_ERRORS): + /* FALLTHROUGH */ case (SEC_LIBRARY): - assert(mdoc->meta.msec); if (*mdoc->meta.msec == '2') break; if (*mdoc->meta.msec == '3') break; if (*mdoc->meta.msec == '9') break; - return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECMSEC)); + mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECMSEC); + break; default: break; } @@ -1510,15 +1909,17 @@ post_sh_head(POST_ARGS) return(1); } - static int -pre_pp(PRE_ARGS) +pre_par(PRE_ARGS) { if (NULL == mdoc->last) return(1); - /* Don't allow prior `Lp' or `Pp'. */ + /* + * Don't allow prior `Lp' or `Pp' prior to a paragraph-type + * block: `Lp', `Pp', or non-compact `Bd' or `Bl'. + */ if (MDOC_Pp != mdoc->last->tok && MDOC_Lp != mdoc->last->tok) return(1); @@ -1532,3 +1933,299 @@ pre_pp(PRE_ARGS) mdoc_node_delete(mdoc, mdoc->last); return(1); } + +static int +pre_literal(PRE_ARGS) +{ + + if (MDOC_BODY != n->type) + return(1); + + /* + * The `Dl' (note "el" not "one") and `Bd -literal' and `Bd + * -unfilled' macros set MDOC_LITERAL on entrance to the body. + */ + + switch (n->tok) { + case (MDOC_Dl): + mdoc->flags |= MDOC_LITERAL; + break; + case (MDOC_Bd): + assert(n->data.Bd); + if (DISP_literal == n->data.Bd->type) + mdoc->flags |= MDOC_LITERAL; + if (DISP_unfilled == n->data.Bd->type) + mdoc->flags |= MDOC_LITERAL; + break; + default: + abort(); + /* NOTREACHED */ + } + + return(1); +} + +static int +post_dd(POST_ARGS) +{ + char buf[DATESIZ]; + struct mdoc_node *n; + + n = mdoc->last; + + if (NULL == n->child) { + mdoc->meta.date = time(NULL); + return(post_prol(mdoc)); + } + + if ( ! concat(mdoc, buf, n->child, DATESIZ)) + return(0); + + mdoc->meta.date = mandoc_a2time + (MTIME_MDOCDATE | MTIME_CANONICAL, buf); + + if (0 == mdoc->meta.date) { + if ( ! mdoc_nmsg(mdoc, n, MANDOCERR_BADDATE)) + return(0); + mdoc->meta.date = time(NULL); + } + + return(1); +} + +static int +post_dt(POST_ARGS) +{ + struct mdoc_node *nn, *n; + const char *cp; + char *p; + + n = mdoc->last; + + if (mdoc->meta.title) + free(mdoc->meta.title); + if (mdoc->meta.vol) + free(mdoc->meta.vol); + if (mdoc->meta.arch) + free(mdoc->meta.arch); + + mdoc->meta.title = mdoc->meta.vol = mdoc->meta.arch = NULL; + + /* First make all characters uppercase. */ + + if (NULL != (nn = n->child)) + for (p = nn->string; *p; p++) { + if (toupper((u_char)*p) == *p) + continue; + if ( ! mdoc_nmsg(mdoc, nn, MANDOCERR_UPPERCASE)) + return(0); + break; + } + + /* Handles: `.Dt' + * --> title = unknown, volume = local, msec = 0, arch = NULL + */ + + if (NULL == (nn = n->child)) { + /* XXX: make these macro values. */ + /* FIXME: warn about missing values. */ + mdoc->meta.title = mandoc_strdup("UNKNOWN"); + mdoc->meta.vol = mandoc_strdup("LOCAL"); + mdoc->meta.msec = mandoc_strdup("1"); + return(1); + } + + /* Handles: `.Dt TITLE' + * --> title = TITLE, volume = local, msec = 0, arch = NULL + */ + + mdoc->meta.title = mandoc_strdup + ('\0' == nn->string[0] ? "UNKNOWN" : nn->string); + + if (NULL == (nn = nn->next)) { + /* FIXME: warn about missing msec. */ + /* XXX: make this a macro value. */ + mdoc->meta.vol = mandoc_strdup("LOCAL"); + mdoc->meta.msec = mandoc_strdup("1"); + return(1); + } + + /* Handles: `.Dt TITLE SEC' + * --> title = TITLE, volume = SEC is msec ? + * format(msec) : SEC, + * msec = SEC is msec ? atoi(msec) : 0, + * arch = NULL + */ + + cp = mdoc_a2msec(nn->string); + if (cp) { + mdoc->meta.vol = mandoc_strdup(cp); + mdoc->meta.msec = mandoc_strdup(nn->string); + } else if (mdoc_nmsg(mdoc, n, MANDOCERR_BADMSEC)) { + mdoc->meta.vol = mandoc_strdup(nn->string); + mdoc->meta.msec = mandoc_strdup(nn->string); + } else + return(0); + + if (NULL == (nn = nn->next)) + return(1); + + /* Handles: `.Dt TITLE SEC VOL' + * --> title = TITLE, volume = VOL is vol ? + * format(VOL) : + * VOL is arch ? format(arch) : + * VOL + */ + + cp = mdoc_a2vol(nn->string); + if (cp) { + free(mdoc->meta.vol); + mdoc->meta.vol = mandoc_strdup(cp); + } else { + /* FIXME: warn about bad arch. */ + cp = mdoc_a2arch(nn->string); + if (NULL == cp) { + free(mdoc->meta.vol); + mdoc->meta.vol = mandoc_strdup(nn->string); + } else + mdoc->meta.arch = mandoc_strdup(cp); + } + + /* Ignore any subsequent parameters... */ + /* FIXME: warn about subsequent parameters. */ + + return(1); +} + +static int +post_prol(POST_ARGS) +{ + /* + * Remove prologue macros from the document after they're + * processed. The final document uses mdoc_meta for these + * values and discards the originals. + */ + + mdoc_node_delete(mdoc, mdoc->last); + if (mdoc->meta.title && mdoc->meta.date && mdoc->meta.os) + mdoc->flags |= MDOC_PBODY; + + return(1); +} + +static int +post_os(POST_ARGS) +{ + struct mdoc_node *n; + char buf[BUFSIZ]; +#ifndef OSNAME + struct utsname utsname; +#endif + + n = mdoc->last; + + /* + * Set the operating system by way of the `Os' macro. Note that + * if an argument isn't provided and -DOSNAME="\"foo\"" is + * provided during compilation, this value will be used instead + * of filling in "sysname release" from uname(). + */ + + if (mdoc->meta.os) + free(mdoc->meta.os); + + if ( ! concat(mdoc, buf, n->child, BUFSIZ)) + return(0); + + /* XXX: yes, these can all be dynamically-adjusted buffers, but + * it's really not worth the extra hackery. + */ + + if ('\0' == buf[0]) { +#ifdef OSNAME + if (strlcat(buf, OSNAME, BUFSIZ) >= BUFSIZ) { + mdoc_nmsg(mdoc, n, MANDOCERR_MEM); + return(0); + } +#else /*!OSNAME */ + if (-1 == uname(&utsname)) + return(mdoc_nmsg(mdoc, n, MANDOCERR_UTSNAME)); + + if (strlcat(buf, utsname.sysname, BUFSIZ) >= BUFSIZ) { + mdoc_nmsg(mdoc, n, MANDOCERR_MEM); + return(0); + } + if (strlcat(buf, " ", 64) >= BUFSIZ) { + mdoc_nmsg(mdoc, n, MANDOCERR_MEM); + return(0); + } + if (strlcat(buf, utsname.release, BUFSIZ) >= BUFSIZ) { + mdoc_nmsg(mdoc, n, MANDOCERR_MEM); + return(0); + } +#endif /*!OSNAME*/ + } + + mdoc->meta.os = mandoc_strdup(buf); + return(1); +} + +static int +post_std(POST_ARGS) +{ + struct mdoc_node *nn, *n; + + n = mdoc->last; + + /* + * Macros accepting `-std' as an argument have the name of the + * current document (`Nm') filled in as the argument if it's not + * provided. + */ + + if (n->child) + return(1); + if (NULL == mdoc->meta.name) + return(1); + + nn = n; + mdoc->next = MDOC_NEXT_CHILD; + + if ( ! mdoc_word_alloc(mdoc, n->line, n->pos, mdoc->meta.name)) + return(0); + mdoc->last = nn; + return(1); +} + +static int +concat(struct mdoc *m, char *p, const struct mdoc_node *n, size_t sz) +{ + + p[0] = '\0'; + + /* + * Concatenate sibling nodes together. All siblings must be of + * type MDOC_TEXT or an assertion is raised. Concatenation is + * separated by a single whitespace. + */ + + for ( ; n; n = n->next) { + assert(MDOC_TEXT == n->type); + + if (strlcat(p, n->string, sz) >= sz) { + mdoc_nmsg(m, n, MANDOCERR_MEM); + return(0); + } + + if (NULL == n->next) + continue; + + if (strlcat(p, " ", sz) >= sz) { + mdoc_nmsg(m, n, MANDOCERR_MEM); + return(0); + } + } + + return(1); +} +