Correctly catch `Vt' semicolon in mandoc-db.

[mandoc.git] / mdoc_validate.c

diff --git a/mdoc_validate.c b/mdoc_validate.c
index 56b249f01d3c480d082ac5678ff4f173041debf9..a34a221d697a70506a1ad31f193ac37cb09dd3cf 100644 (file)
--- a/mdoc_validate.c
+++ b/mdoc_validate.c
@@ -1,4 +1,4 @@
-/*     $Id: mdoc_validate.c,v 1.156 2011/02/06 21:44:36 kristaps Exp $ */
+/*     $Id: mdoc_validate.c,v 1.169 2011/04/30 10:18:24 kristaps Exp $ */

 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>


@@ -33,6 +33,7 @@
 #include <string.h>
 #include <time.h>
 
 #include <string.h>
 #include <time.h>
 

+#include "mdoc.h"

 #include "mandoc.h"
 #include "libmdoc.h"
 #include "libmandoc.h"
 #include "mandoc.h"
 #include "libmdoc.h"
 #include "libmandoc.h"


@@ -74,6 +75,8 @@ static        void     check_args(struct mdoc *, struct mdoc_node *);
 
 static int      concat(struct mdoc *, char *, 
                        const struct mdoc_node *, size_t);
 
 static int      concat(struct mdoc *, char *, 
                        const struct mdoc_node *, size_t);

+static enum mdoc_sec   a2sec(const char *);
+static size_t          macro2len(enum mdoct);

 
 static int      ebool(POST_ARGS);
 static int      berr_ge1(POST_ARGS);
 
 static int      ebool(POST_ARGS);
 static int      berr_ge1(POST_ARGS);


@@ -140,7 +143,7 @@ static      v_post   posts_bx[] = { post_bx, NULL };
 static v_post   posts_bool[] = { ebool, NULL };
 static v_post   posts_eoln[] = { post_eoln, NULL };
 static v_post   posts_defaults[] = { post_defaults, NULL };
 static v_post   posts_bool[] = { ebool, NULL };
 static v_post   posts_eoln[] = { post_eoln, 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_dd[] = { post_dd, post_prol, NULL };

 static v_post   posts_dl[] = { post_literal, bwarn_ge1, 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_dl[] = { post_literal, bwarn_ge1, NULL };
 static v_post   posts_dt[] = { post_dt, post_prol, NULL };
 static v_post   posts_fo[] = { hwarn_eq1, bwarn_ge1, NULL };


@@ -152,9 +155,9 @@ static      v_post   posts_notext[] = { ewarn_eq0, NULL };
 static v_post   posts_ns[] = { post_ns, NULL };
 static v_post   posts_os[] = { post_os, post_prol, NULL };
 static v_post   posts_rs[] = { post_rs, NULL };
 static v_post   posts_ns[] = { post_ns, NULL };
 static v_post   posts_os[] = { post_os, post_prol, NULL };
 static v_post   posts_rs[] = { post_rs, NULL };

-static v_post   posts_sh[] = { post_ignpar, hwarn_ge1, bwarn_ge1, post_sh, NULL };
+static v_post   posts_sh[] = { post_ignpar, hwarn_ge1, post_sh, NULL };

 static v_post   posts_sp[] = { ewarn_le1, NULL };
 static v_post   posts_sp[] = { ewarn_le1, NULL };

-static v_post   posts_ss[] = { post_ignpar, hwarn_ge1, bwarn_ge1, NULL };
+static v_post   posts_ss[] = { post_ignpar, hwarn_ge1, NULL };

 static v_post   posts_st[] = { post_st, NULL };
 static v_post   posts_std[] = { post_std, NULL };
 static v_post   posts_text[] = { ewarn_ge1, NULL };
 static v_post   posts_st[] = { post_st, NULL };
 static v_post   posts_std[] = { post_std, NULL };
 static v_post   posts_text[] = { ewarn_ge1, NULL };


@@ -177,7 +180,7 @@ static      v_pre    pres_sh[] = { pre_sh, NULL };
 static v_pre    pres_ss[] = { pre_ss, NULL };
 static v_pre    pres_std[] = { pre_std, NULL };
 
 static v_pre    pres_ss[] = { pre_ss, NULL };
 static v_pre    pres_std[] = { pre_std, NULL };
 

-const  struct valids mdoc_valids[MDOC_MAX] = {
+static const struct valids mdoc_valids[MDOC_MAX] = {

        { NULL, NULL },                         /* Ap */
        { pres_dd, posts_dd },                  /* Dd */
        { pres_dt, posts_dt },                  /* Dt */
        { NULL, NULL },                         /* Ap */
        { pres_dd, posts_dd },                  /* Dd */
        { pres_dt, posts_dt },                  /* Dt */


@@ -221,7 +224,7 @@ const       struct valids mdoc_valids[MDOC_MAX] = {
        { NULL, posts_text },                   /* Xr */ 
        { NULL, posts_text },                   /* %A */
        { NULL, posts_text },                   /* %B */ /* FIXME: can be used outside Rs/Re. */
        { NULL, posts_text },                   /* Xr */ 
        { NULL, posts_text },                   /* %A */
        { NULL, posts_text },                   /* %B */ /* FIXME: can be used outside Rs/Re. */

-       { NULL, posts_text },                   /* %D */ /* FIXME: check date with mandoc_a2time(). */
+       { NULL, posts_text },                   /* %D */

        { NULL, posts_text },                   /* %I */
        { NULL, posts_text },                   /* %J */
        { NULL, posts_text },                   /* %N */
        { NULL, posts_text },                   /* %I */
        { NULL, posts_text },                   /* %J */
        { NULL, posts_text },                   /* %N */


@@ -321,6 +324,30 @@ static     const enum mdoct rsord[RSORD_MAX] = {
        MDOC__U
 };
 
        MDOC__U
 };
 

+static const char * const secnames[SEC__MAX] = {
+       NULL,
+       "NAME",
+       "LIBRARY",
+       "SYNOPSIS",
+       "DESCRIPTION",
+       "IMPLEMENTATION NOTES",
+       "RETURN VALUES",
+       "ENVIRONMENT",
+       "FILES",
+       "EXIT STATUS",
+       "EXAMPLES",
+       "DIAGNOSTICS",
+       "COMPATIBILITY",
+       "ERRORS",
+       "SEE ALSO",
+       "STANDARDS",
+       "HISTORY",
+       "AUTHORS",
+       "CAVEATS",
+       "BUGS",
+       "SECURITY CONSIDERATIONS",
+       NULL
+};

 
 int
 mdoc_valid_pre(struct mdoc *mdoc, struct mdoc_node *n)
 
 int
 mdoc_valid_pre(struct mdoc *mdoc, struct mdoc_node *n)


@@ -369,6 +396,8 @@ mdoc_valid_post(struct mdoc *mdoc)
        switch (mdoc->last->type) {
        case (MDOC_TEXT):
                /* FALLTHROUGH */
        switch (mdoc->last->type) {
        case (MDOC_TEXT):
                /* FALLTHROUGH */

+       case (MDOC_EQN):
+               /* FALLTHROUGH */

        case (MDOC_TBL):
                return(1);
        case (MDOC_ROOT):
        case (MDOC_TBL):
                return(1);
        case (MDOC_ROOT):


@@ -418,10 +447,10 @@ check_count(struct mdoc *m, enum mdoc_type type,
        }
 
        t = lvl == CHECK_WARN ? MANDOCERR_ARGCWARN : MANDOCERR_ARGCOUNT;
        }
 
        t = lvl == CHECK_WARN ? MANDOCERR_ARGCWARN : MANDOCERR_ARGCOUNT;

-
-       return(mdoc_vmsg(m, t, m->last->line, m->last->pos,
+       mandoc_vmsg(t, m->parse, m->last->line, m->last->pos,

                        "want %s%d children (have %d)",
                        "want %s%d children (have %d)",

-                       p, val, m->last->nchild));
+                       p, val, m->last->nchild);
+       return(1);

 }
 
 static int
 }
 
 static int


@@ -516,31 +545,39 @@ check_argv(struct mdoc *m, struct mdoc_node *n, struct mdoc_argv *v)
 static void
 check_text(struct mdoc *m, int ln, int pos, char *p)
 {
 static void
 check_text(struct mdoc *m, int ln, int pos, char *p)
 {

-       int              c;
+       char            *cpp, *pp;

        size_t           sz;
 
        size_t           sz;
 

-       for ( ; *p; p++, pos++) {
+       while ('\0' != *p) {

                sz = strcspn(p, "\t\\");
                sz = strcspn(p, "\t\\");

-               p += (int)sz;
-
-               if ('\0' == *p)
-                       break;

 
 

+               p += (int)sz;

                pos += (int)sz;
 
                if ('\t' == *p) {
                        if ( ! (MDOC_LITERAL & m->flags))
                                mdoc_pmsg(m, ln, pos, MANDOCERR_BADTAB);
                pos += (int)sz;
 
                if ('\t' == *p) {
                        if ( ! (MDOC_LITERAL & m->flags))
                                mdoc_pmsg(m, ln, pos, MANDOCERR_BADTAB);

+                       p++;
+                       pos++;

                        continue;
                        continue;

-               }
+               } else if ('\0' == *p)
+                       break;

 
 

-               if (0 == (c = mandoc_special(p))) {
+               pos++;
+               pp = ++p;
+
+               if (ESCAPE_ERROR == mandoc_escape
+                               ((const char **)&pp, NULL, NULL)) {

                        mdoc_pmsg(m, ln, pos, MANDOCERR_BADESCAPE);
                        mdoc_pmsg(m, ln, pos, MANDOCERR_BADESCAPE);

-                       continue;
+                       break;

                }
 
                }
 

-               p += c - 1;
-               pos += c - 1;
+               cpp = p;
+               while (NULL != (cpp = memchr(cpp, ASCII_HYPH, pp - cpp)))
+                       *cpp = '-';
+
+               pos += pp - p;
+               p = pp;

        }
 }
 
        }
 }
 


@@ -553,10 +590,9 @@ check_parent(PRE_ARGS, enum mdoct tok, enum mdoc_type t)
                        (t == n->parent->type))
                return(1);
 
                        (t == n->parent->type))
                return(1);
 

-       mdoc_vmsg(mdoc, MANDOCERR_SYNTCHILD,
-                               n->line, n->pos, "want parent %s",
-                               MDOC_ROOT == t ? "<root>" : 
-                                       mdoc_macronames[tok]);
+       mandoc_vmsg(MANDOCERR_SYNTCHILD, mdoc->parse, n->line, 
+                       n->pos, "want parent %s", MDOC_ROOT == t ? 
+                       "<root>" : mdoc_macronames[tok]);

        return(0);
 }
 
        return(0);
 }
 


@@ -917,7 +953,7 @@ static int
 pre_dt(PRE_ARGS)
 {
 
 pre_dt(PRE_ARGS)
 {
 

-       if (0 == mdoc->meta.date || mdoc->meta.os)
+       if (NULL == mdoc->meta.date || mdoc->meta.os)

                mdoc_nmsg(mdoc, n, MANDOCERR_PROLOGOOO);
 
        if (mdoc->meta.title)
                mdoc_nmsg(mdoc, n, MANDOCERR_PROLOGOOO);
 
        if (mdoc->meta.title)


@@ -930,7 +966,7 @@ static int
 pre_os(PRE_ARGS)
 {
 
 pre_os(PRE_ARGS)
 {
 

-       if (NULL == mdoc->meta.title || 0 == mdoc->meta.date)
+       if (NULL == mdoc->meta.title || NULL == mdoc->meta.date)

                mdoc_nmsg(mdoc, n, MANDOCERR_PROLOGOOO);
 
        if (mdoc->meta.os)
                mdoc_nmsg(mdoc, n, MANDOCERR_PROLOGOOO);
 
        if (mdoc->meta.os)


@@ -1238,7 +1274,7 @@ post_an(POST_ARGS)
 static int
 post_it(POST_ARGS)
 {
 static int
 post_it(POST_ARGS)
 {

-       int               i, cols, rc;
+       int               i, cols;

        enum mdoc_list    lt;
        struct mdoc_node *n, *c;
        enum mandocerr    er;
        enum mdoc_list    lt;
        struct mdoc_node *n, *c;
        enum mandocerr    er;


@@ -1304,10 +1340,10 @@ post_it(POST_ARGS)
                else
                        er = MANDOCERR_SYNTARGCOUNT;
 
                else
                        er = MANDOCERR_SYNTARGCOUNT;
 

-               rc = mdoc_vmsg(mdoc, er, 
-                               mdoc->last->line, mdoc->last->pos, 
+               mandoc_vmsg(er, mdoc->parse, mdoc->last->line, 
+                               mdoc->last->pos, 

                                "columns == %d (have %d)", cols, i);
                                "columns == %d (have %d)", cols, i);

-               return(rc);
+               return(MANDOCERR_ARGCOUNT == er);

        default:
                break;
        }
        default:
                break;
        }


@@ -1368,7 +1404,7 @@ post_bl_block_width(POST_ARGS)
                width = 6;
        else if (MDOC_MAX == (tok = mdoc_hash_find(n->norm->Bl.width)))
                return(1);
                width = 6;
        else if (MDOC_MAX == (tok = mdoc_hash_find(n->norm->Bl.width)))
                return(1);

-       else if (0 == (width = mdoc_macro2len(tok)))  {
+       else if (0 == (width = macro2len(tok)))  {

                mdoc_nmsg(mdoc, n, MANDOCERR_BADWIDTH);
                return(1);
        }
                mdoc_nmsg(mdoc, n, MANDOCERR_BADWIDTH);
                return(1);
        }


@@ -1425,7 +1461,7 @@ post_bl_block_tag(POST_ARGS)
                        break;
                }
 
                        break;
                }
 

-               if (0 != (ssz = mdoc_macro2len(nn->tok)))
+               if (0 != (ssz = macro2len(nn->tok)))

                        sz = ssz;
 
                break;
                        sz = ssz;
 
                break;


@@ -1499,7 +1535,7 @@ post_bl_head(POST_ARGS)
        assert(0 == np->args->argv[j].sz);
 
        /*
        assert(0 == np->args->argv[j].sz);
 
        /*

-        * Accomodate for new-style groff column syntax.  Shuffle the
+        * Accommodate 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.
         */
         * child nodes, all of which must be TEXT, as arguments for the
         * column field.  Then, delete the head children.
         */


@@ -1819,7 +1855,7 @@ post_sh_head(POST_ARGS)
        if ( ! concat(mdoc, buf, mdoc->last->child, BUFSIZ))
                return(0);
 
        if ( ! concat(mdoc, buf, mdoc->last->child, BUFSIZ))
                return(0);
 

-       sec = mdoc_str2sec(buf);
+       sec = a2sec(buf);

 
        /* The NAME should be first. */
 
 
        /* The NAME should be first. */
 


@@ -1969,23 +2005,21 @@ post_dd(POST_ARGS)
        char              buf[DATESIZE];
        struct mdoc_node *n;
 
        char              buf[DATESIZE];
        struct mdoc_node *n;
 

-       n = mdoc->last;
+       if (mdoc->meta.date)
+               free(mdoc->meta.date);

 
 

-       if (NULL == n->child) {
-               mdoc->meta.date = time(NULL);
+       n = mdoc->last;
+       if (NULL == n->child || '\0' == n->child->string[0]) {
+               mdoc->meta.date = mandoc_normdate
+                       (mdoc->parse, NULL, n->line, n->pos);

                return(1);
        }
 
        if ( ! concat(mdoc, buf, n->child, DATESIZE))
                return(0);
 
                return(1);
        }
 
        if ( ! concat(mdoc, buf, n->child, DATESIZE))
                return(0);
 

-       mdoc->meta.date = mandoc_a2time
-               (MTIME_MDOCDATE | MTIME_CANONICAL, buf);
-
-       if (0 == mdoc->meta.date) {
-               mdoc_nmsg(mdoc, n, MANDOCERR_BADDATE);
-               mdoc->meta.date = time(NULL);
-       }
+       mdoc->meta.date = mandoc_normdate
+               (mdoc->parse, buf, n->line, n->pos);

 
        return(1);
 }
 
        return(1);
 }


@@ -2127,7 +2161,8 @@ post_bx(POST_ARGS)
 
        n = mdoc->last->child;
        if (n && NULL != (n = n->next))
 
        n = mdoc->last->child;
        if (n && NULL != (n = n->next))

-               *n->string = toupper((unsigned char)*n->string);
+               *n->string = (char)toupper
+                       ((unsigned char)*n->string);

 
        return(1);
 }
 
        return(1);
 }


@@ -2167,7 +2202,7 @@ post_os(POST_ARGS)
                        return(0);
                }
 #else /*!OSNAME */
                        return(0);
                }
 #else /*!OSNAME */

-               if (uname(&utsname)) {
+               if (-1 == uname(&utsname)) {

                        mdoc_nmsg(mdoc, n, MANDOCERR_UNAME);
                         mdoc->meta.os = mandoc_strdup("UNKNOWN");
                         return(post_prol(mdoc));
                        mdoc_nmsg(mdoc, n, MANDOCERR_UNAME);
                         mdoc->meta.os = mandoc_strdup("UNKNOWN");
                         return(post_prol(mdoc));


@@ -2254,3 +2289,107 @@ concat(struct mdoc *m, char *p, const struct mdoc_node *n, size_t sz)
        return(1);
 }
 
        return(1);
 }
 

+static enum mdoc_sec 
+a2sec(const char *p)
+{
+       int              i;
+
+       for (i = 0; i < (int)SEC__MAX; i++) 
+               if (secnames[i] && 0 == strcmp(p, secnames[i]))
+                       return((enum mdoc_sec)i);
+
+       return(SEC_CUSTOM);
+}
+
+static size_t
+macro2len(enum mdoct macro)
+{
+
+       switch (macro) {
+       case(MDOC_Ad):
+               return(12);
+       case(MDOC_Ao):
+               return(12);
+       case(MDOC_An):
+               return(12);
+       case(MDOC_Aq):
+               return(12);
+       case(MDOC_Ar):
+               return(12);
+       case(MDOC_Bo):
+               return(12);
+       case(MDOC_Bq):
+               return(12);
+       case(MDOC_Cd):
+               return(12);
+       case(MDOC_Cm):
+               return(10);
+       case(MDOC_Do):
+               return(10);
+       case(MDOC_Dq):
+               return(12);
+       case(MDOC_Dv):
+               return(12);
+       case(MDOC_Eo):
+               return(12);
+       case(MDOC_Em):
+               return(10);
+       case(MDOC_Er):
+               return(17);
+       case(MDOC_Ev):
+               return(15);
+       case(MDOC_Fa):
+               return(12);
+       case(MDOC_Fl):
+               return(10);
+       case(MDOC_Fo):
+               return(16);
+       case(MDOC_Fn):
+               return(16);
+       case(MDOC_Ic):
+               return(10);
+       case(MDOC_Li):
+               return(16);
+       case(MDOC_Ms):
+               return(6);
+       case(MDOC_Nm):
+               return(10);
+       case(MDOC_No):
+               return(12);
+       case(MDOC_Oo):
+               return(10);
+       case(MDOC_Op):
+               return(14);
+       case(MDOC_Pa):
+               return(32);
+       case(MDOC_Pf):
+               return(12);
+       case(MDOC_Po):
+               return(12);
+       case(MDOC_Pq):
+               return(12);
+       case(MDOC_Ql):
+               return(16);
+       case(MDOC_Qo):
+               return(12);
+       case(MDOC_So):
+               return(12);
+       case(MDOC_Sq):
+               return(12);
+       case(MDOC_Sy):
+               return(6);
+       case(MDOC_Sx):
+               return(16);
+       case(MDOC_Tn):
+               return(10);
+       case(MDOC_Va):
+               return(12);
+       case(MDOC_Vt):
+               return(12);
+       case(MDOC_Xr):
+               return(10);
+       default:
+               break;
+       };
+       return(0);
+}