Get ready for a version.

[mandoc.git] / mdoc_validate.c

diff --git a/mdoc_validate.c b/mdoc_validate.c
index a34a221d697a70506a1ad31f193ac37cb09dd3cf..060ccdadec38c27390e8c7a94d9f4e616a7cd6fe 100644 (file)
--- a/mdoc_validate.c
+++ b/mdoc_validate.c
@@ -1,4 +1,4 @@
-/*     $Id: mdoc_validate.c,v 1.169 2011/04/30 10:18:24 kristaps Exp $ */
+/*     $Id: mdoc_validate.c,v 1.182 2012/03/23 05:50:25 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>


@@ -72,9 +72,7 @@ static        void     check_text(struct mdoc *, int, int, char *);
 static void     check_argv(struct mdoc *, 
                        struct mdoc_node *, struct mdoc_argv *);
 static void     check_args(struct mdoc *, struct mdoc_node *);
 static void     check_argv(struct mdoc *, 
                        struct mdoc_node *, struct mdoc_argv *);
 static void     check_args(struct mdoc *, struct mdoc_node *);

-
-static int      concat(struct mdoc *, char *, 
-                       const struct mdoc_node *, size_t);
+static int      concat(char *, const struct mdoc_node *, size_t);

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


@@ -545,40 +543,13 @@ 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)
 {

-       char            *cpp, *pp;
-       size_t           sz;
-
-       while ('\0' != *p) {
-               sz = strcspn(p, "\t\\");
-
-               p += (int)sz;
-               pos += (int)sz;
-
-               if ('\t' == *p) {
-                       if ( ! (MDOC_LITERAL & m->flags))
-                               mdoc_pmsg(m, ln, pos, MANDOCERR_BADTAB);
-                       p++;
-                       pos++;
-                       continue;
-               } else if ('\0' == *p)
-                       break;
+       char            *cp;

 
 

-               pos++;
-               pp = ++p;
-
-               if (ESCAPE_ERROR == mandoc_escape
-                               ((const char **)&pp, NULL, NULL)) {
-                       mdoc_pmsg(m, ln, pos, MANDOCERR_BADESCAPE);
-                       break;
-               }
-
-               cpp = p;
-               while (NULL != (cpp = memchr(cpp, ASCII_HYPH, pp - cpp)))
-                       *cpp = '-';
+       if (MDOC_LITERAL & m->flags)
+               return;

 
 

-               pos += pp - p;
-               p = pp;
-       }
+       for (cp = p; NULL != (p = strchr(p, '\t')); p++)
+               mdoc_pmsg(m, ln, pos + (int)(p - cp), MANDOCERR_BADTAB);

 }
 
 static int
 }
 
 static int


@@ -690,8 +661,13 @@ pre_bl(PRE_ARGS)
                        comp = 1;
                        break;
                case (MDOC_Width):
                        comp = 1;
                        break;
                case (MDOC_Width):

-                       dup = (NULL != n->norm->Bl.width);
-                       width = n->args->argv[i].value[0];
+                       /* NB: this can be empty! */
+                       if (n->args->argv[i].sz) {
+                               width = n->args->argv[i].value[0];
+                               dup = (NULL != n->norm->Bl.width);
+                               break;
+                       }
+                       mdoc_nmsg(mdoc, n, MANDOCERR_IGNARGV);

                        break;
                case (MDOC_Offset):
                        /* NB: this can be empty! */
                        break;
                case (MDOC_Offset):
                        /* NB: this can be empty! */


@@ -731,7 +707,7 @@ pre_bl(PRE_ARGS)
                        if (LIST_column == lt) {
                                n->norm->Bl.ncols = 
                                        n->args->argv[i].sz;
                        if (LIST_column == lt) {
                                n->norm->Bl.ncols = 
                                        n->args->argv[i].sz;

-                               n->norm->Bl.cols = (const char **)
+                               n->norm->Bl.cols = (void *)

                                        n->args->argv[i].value;
                        }
                }
                                        n->args->argv[i].value;
                        }
                }


@@ -899,7 +875,7 @@ pre_sh(PRE_ARGS)
        if (MDOC_BLOCK != n->type)
                return(1);
 
        if (MDOC_BLOCK != n->type)
                return(1);
 

-       mdoc->regs->regs[(int)REG_nS].set = 0;
+       roff_regunset(mdoc->roff, REG_nS);

        return(check_parent(mdoc, n, MDOC_MAX, MDOC_ROOT));
 }
 
        return(check_parent(mdoc, n, MDOC_MAX, MDOC_ROOT));
 }
 


@@ -1133,6 +1109,7 @@ static int
 post_nm(POST_ARGS)
 {
        char             buf[BUFSIZ];
 post_nm(POST_ARGS)
 {
        char             buf[BUFSIZ];

+       int              c;

 
        /* If no child specified, make sure we have the meta name. */
 
 
        /* If no child specified, make sure we have the meta name. */
 


@@ -1144,11 +1121,14 @@ post_nm(POST_ARGS)
 
        /* If no meta name, set it from the child. */
 
 
        /* If no meta name, set it from the child. */
 

-       if ( ! concat(mdoc, buf, mdoc->last->child, BUFSIZ))
+       buf[0] = '\0';
+       if (-1 == (c = concat(buf, mdoc->last->child, BUFSIZ))) {
+               mdoc_nmsg(mdoc, mdoc->last->child, MANDOCERR_MEM);

                return(0);
                return(0);

+       }

 
 

+       assert(c);

        mdoc->meta.name = mandoc_strdup(buf);
        mdoc->meta.name = mandoc_strdup(buf);

-

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


@@ -1419,7 +1399,7 @@ post_bl_block_width(POST_ARGS)
 
        assert(i < (int)n->args->argc);
 
 
        assert(i < (int)n->args->argc);
 

-       snprintf(buf, NUMSIZ, "%zun", width);
+       snprintf(buf, NUMSIZ, "%un", (unsigned int)width);

        free(n->args->argv[i].value[0]);
        n->args->argv[i].value[0] = mandoc_strdup(buf);
 
        free(n->args->argv[i].value[0]);
        n->args->argv[i].value[0] = mandoc_strdup(buf);
 


@@ -1469,7 +1449,7 @@ post_bl_block_tag(POST_ARGS)
 
        /* Defaults to ten ens. */
 
 
        /* Defaults to ten ens. */
 

-       snprintf(buf, NUMSIZ, "%zun", sz);
+       snprintf(buf, NUMSIZ, "%un", (unsigned int)sz);

 
        /*
         * We have to dynamically add this to the macro's argument list.
 
        /*
         * We have to dynamically add this to the macro's argument list.


@@ -1545,7 +1525,7 @@ post_bl_head(POST_ARGS)
                ((size_t)mdoc->last->nchild * sizeof(char *));
 
        mdoc->last->norm->Bl.ncols = np->args->argv[j].sz;
                ((size_t)mdoc->last->nchild * sizeof(char *));
 
        mdoc->last->norm->Bl.ncols = np->args->argv[j].sz;

-       mdoc->last->norm->Bl.cols = (const char **)np->args->argv[j].value;
+       mdoc->last->norm->Bl.cols = (void *)np->args->argv[j].value;

 
        for (i = 0, nn = mdoc->last->child; nn; i++) {
                np->args->argv[j].value[i] = nn->string;
 
        for (i = 0, nn = mdoc->last->child; nn; i++) {
                np->args->argv[j].value[i] = nn->string;


@@ -1718,6 +1698,14 @@ post_rs(POST_ARGS)
                mdoc_node_delete(mdoc, nn);
        }
 
                mdoc_node_delete(mdoc, nn);
        }
 

+       /*
+        * Nothing to sort if only invalid nodes were found
+        * inside the `Rs' body.
+        */
+
+       if (NULL == mdoc->last->child)
+               return(1);
+

        /*
         * The full `Rs' block needs special handling to order the
         * sub-elements according to `rsord'.  Pick through each element
        /*
         * The full `Rs' block needs special handling to order the
         * sub-elements according to `rsord'.  Pick through each element


@@ -1843,7 +1831,9 @@ static int
 post_sh_head(POST_ARGS)
 {
        char             buf[BUFSIZ];
 post_sh_head(POST_ARGS)
 {
        char             buf[BUFSIZ];

+       struct mdoc_node *n;

        enum mdoc_sec    sec;
        enum mdoc_sec    sec;

+       int              c;

 
        /*
         * Process a new section.  Sections are either "named" or
 
        /*
         * Process a new section.  Sections are either "named" or


@@ -1852,10 +1842,13 @@ post_sh_head(POST_ARGS)
         * manual sections.
         */
 
         * manual sections.
         */
 

-       if ( ! concat(mdoc, buf, mdoc->last->child, BUFSIZ))
+       sec = SEC_CUSTOM;
+       buf[0] = '\0';
+       if (-1 == (c = concat(buf, mdoc->last->child, BUFSIZ))) {
+               mdoc_nmsg(mdoc, mdoc->last->child, MANDOCERR_MEM);

                return(0);
                return(0);

-
-       sec = a2sec(buf);
+       } else if (1 == c)
+               sec = a2sec(buf);

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


@@ -1873,6 +1866,20 @@ post_sh_head(POST_ARGS)
 
        mdoc->lastsec = sec;
 
 
        mdoc->lastsec = sec;
 

+       /*
+        * Set the section attribute for the current HEAD, for its
+        * parent BLOCK, and for the HEAD children; the latter can
+        * only be TEXT nodes, so no recursion is needed.
+        * For other blocks and elements, including .Sh BODY, this is
+        * done when allocating the node data structures, but for .Sh
+        * BLOCK and HEAD, the section is still unknown at that time.
+        */
+
+       mdoc->last->parent->sec = sec;
+       mdoc->last->sec = sec;
+       for (n = mdoc->last->child; n; n = n->next)
+               n->sec = sec;
+

        /* We don't care about custom sections after this. */
 
        if (SEC_CUSTOM == sec)
        /* We don't care about custom sections after this. */
 
        if (SEC_CUSTOM == sec)


@@ -2004,6 +2011,7 @@ post_dd(POST_ARGS)
 {
        char              buf[DATESIZE];
        struct mdoc_node *n;
 {
        char              buf[DATESIZE];
        struct mdoc_node *n;

+       int               c;

 
        if (mdoc->meta.date)
                free(mdoc->meta.date);
 
        if (mdoc->meta.date)
                free(mdoc->meta.date);


@@ -2015,9 +2023,13 @@ post_dd(POST_ARGS)
                return(1);
        }
 
                return(1);
        }
 

-       if ( ! concat(mdoc, buf, n->child, DATESIZE))
+       buf[0] = '\0';
+       if (-1 == (c = concat(buf, n->child, DATESIZE))) {
+               mdoc_nmsg(mdoc, n->child, MANDOCERR_MEM);

                return(0);
                return(0);

+       }

 
 

+       assert(c);

        mdoc->meta.date = mandoc_normdate
                (mdoc->parse, buf, n->line, n->pos);
 
        mdoc->meta.date = mandoc_normdate
                (mdoc->parse, buf, n->line, n->pos);
 


@@ -2046,7 +2058,7 @@ post_dt(POST_ARGS)
 
        if (NULL != (nn = n->child))
                for (p = nn->string; *p; p++) {
 
        if (NULL != (nn = n->child))
                for (p = nn->string; *p; p++) {

-                       if (toupper((u_char)*p) == *p)
+                       if (toupper((unsigned char)*p) == *p)

                                continue;
 
                        /* 
                                continue;
 
                        /* 


@@ -2092,7 +2104,7 @@ post_dt(POST_ARGS)
         *       arch = NULL
         */
 
         *       arch = NULL
         */
 

-       cp = mdoc_a2msec(nn->string);
+       cp = mandoc_a2msec(nn->string);

        if (cp) {
                mdoc->meta.vol = mandoc_strdup(cp);
                mdoc->meta.msec = mandoc_strdup(nn->string);
        if (cp) {
                mdoc->meta.vol = mandoc_strdup(cp);
                mdoc->meta.msec = mandoc_strdup(nn->string);


@@ -2172,6 +2184,7 @@ post_os(POST_ARGS)
 {
        struct mdoc_node *n;
        char              buf[BUFSIZ];
 {
        struct mdoc_node *n;
        char              buf[BUFSIZ];

+       int               c;

 #ifndef OSNAME
        struct utsname    utsname;
 #endif
 #ifndef OSNAME
        struct utsname    utsname;
 #endif


@@ -2188,8 +2201,13 @@ post_os(POST_ARGS)
        if (mdoc->meta.os)
                free(mdoc->meta.os);
 
        if (mdoc->meta.os)
                free(mdoc->meta.os);
 

-       if ( ! concat(mdoc, buf, n->child, BUFSIZ))
+       buf[0] = '\0';
+       if (-1 == (c = concat(buf, n->child, BUFSIZ))) {
+               mdoc_nmsg(mdoc, n->child, MANDOCERR_MEM);

                return(0);
                return(0);

+       }
+
+       assert(c);

 
        /* XXX: yes, these can all be dynamically-adjusted buffers, but
         * it's really not worth the extra hackery.
 
        /* XXX: yes, these can all be dynamically-adjusted buffers, but
         * it's really not worth the extra hackery.


@@ -2256,34 +2274,24 @@ post_std(POST_ARGS)
        return(1);
 }
 
        return(1);
 }
 

+/*
+ * Concatenate a node, stopping at the first non-text.
+ * Concatenation is separated by a single whitespace.  
+ * Returns -1 on fatal (string overrun) error, 0 if child nodes were
+ * encountered, 1 otherwise.
+ */

 static int
 static int

-concat(struct mdoc *m, char *p, const struct mdoc_node *n, size_t sz)
+concat(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.  Returns 0 on fatal (string
-        * overrun) error.
-        */
-
-       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);
+       for ( ; NULL != n; n = n->next) {
+               if (MDOC_TEXT != n->type) 

                        return(0);
                        return(0);

-               }
+               if ('\0' != p[0] && strlcat(p, " ", sz) >= sz)
+                       return(-1);
+               if (strlcat(p, n->string, sz) >= sz)
+                       return(-1);
+               concat(p, n->child, sz);

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