Protect against malicious manual pages containing .ll requests with

[mandoc.git] / man_macro.c

diff --git a/man_macro.c b/man_macro.c
index 73b9e6c44ac1272762d40e373ea03fd552265e13..aa8b200196a6eb36e24aecaae246b7634188b20d 100644 (file)
--- a/man_macro.c
+++ b/man_macro.c
@@ -1,7 +1,7 @@
-/*     $Id: man_macro.c,v 1.104 2015/04/03 23:19:15 schwarze Exp $ */
+/*     $Id: man_macro.c,v 1.123 2017/06/25 11:45:37 schwarze Exp $ */

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

- * Copyright (c) 2012, 2013, 2014, 2015 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2012-2015, 2017 Ingo Schwarze <schwarze@openbsd.org>

  * Copyright (c) 2013 Franco Fichtner <franco@lastsummer.de>
  *
  * Permission to use, copy, modify, and distribute this software for any
  * Copyright (c) 2013 Franco Fichtner <franco@lastsummer.de>
  *
  * Permission to use, copy, modify, and distribute this software for any


@@ -29,18 +29,18 @@
 #include "roff.h"
 #include "man.h"
 #include "libmandoc.h"
 #include "roff.h"
 #include "man.h"
 #include "libmandoc.h"

+#include "roff_int.h"

 #include "libman.h"
 
 static void             blk_close(MACRO_PROT_ARGS);
 static void             blk_exp(MACRO_PROT_ARGS);
 static void             blk_imp(MACRO_PROT_ARGS);
 static void             in_line_eoln(MACRO_PROT_ARGS);
 #include "libman.h"
 
 static void             blk_close(MACRO_PROT_ARGS);
 static void             blk_exp(MACRO_PROT_ARGS);
 static void             blk_imp(MACRO_PROT_ARGS);
 static void             in_line_eoln(MACRO_PROT_ARGS);

-static int              man_args(struct man *, int,
+static int              man_args(struct roff_man *, int,

                                int *, char *, char **);
                                int *, char *, char **);

-static void             rew_scope(struct man *, int);
+static void             rew_scope(struct roff_man *, enum roff_tok);

 
 

-const  struct man_macro __man_macros[MAN_MAX] = {
-       { in_line_eoln, MAN_NSCOPED }, /* br */
+const  struct man_macro __man_macros[MAN_MAX - MAN_TH] = {

        { in_line_eoln, MAN_BSCOPE }, /* TH */
        { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SH */
        { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SS */
        { in_line_eoln, MAN_BSCOPE }, /* TH */
        { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SH */
        { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SS */


@@ -61,30 +61,28 @@ const       struct man_macro __man_macros[MAN_MAX] = {
        { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* I */
        { in_line_eoln, 0 }, /* IR */
        { in_line_eoln, 0 }, /* RI */
        { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* I */
        { in_line_eoln, 0 }, /* IR */
        { in_line_eoln, 0 }, /* RI */

-       { in_line_eoln, MAN_NSCOPED }, /* sp */
-       { in_line_eoln, MAN_BSCOPE }, /* nf */
-       { in_line_eoln, MAN_BSCOPE }, /* fi */
+       { in_line_eoln, MAN_NSCOPED }, /* nf */
+       { in_line_eoln, MAN_NSCOPED }, /* fi */

        { blk_close, MAN_BSCOPE }, /* RE */
        { blk_exp, MAN_BSCOPE }, /* RS */
        { in_line_eoln, 0 }, /* DT */
        { in_line_eoln, 0 }, /* UC */
        { in_line_eoln, MAN_NSCOPED }, /* PD */
        { in_line_eoln, 0 }, /* AT */
        { blk_close, MAN_BSCOPE }, /* RE */
        { blk_exp, MAN_BSCOPE }, /* RS */
        { in_line_eoln, 0 }, /* DT */
        { in_line_eoln, 0 }, /* UC */
        { in_line_eoln, MAN_NSCOPED }, /* PD */
        { in_line_eoln, 0 }, /* AT */

-       { in_line_eoln, 0 }, /* in */
-       { in_line_eoln, 0 }, /* ft */
+       { in_line_eoln, MAN_NSCOPED }, /* in */

        { in_line_eoln, 0 }, /* OP */
        { in_line_eoln, MAN_BSCOPE }, /* EX */
        { in_line_eoln, MAN_BSCOPE }, /* EE */
        { blk_exp, MAN_BSCOPE }, /* UR */
        { blk_close, MAN_BSCOPE }, /* UE */
        { in_line_eoln, 0 }, /* OP */
        { in_line_eoln, MAN_BSCOPE }, /* EX */
        { in_line_eoln, MAN_BSCOPE }, /* EE */
        { blk_exp, MAN_BSCOPE }, /* UR */
        { blk_close, MAN_BSCOPE }, /* UE */

-       { in_line_eoln, 0 }, /* ll */
+       { blk_exp, MAN_BSCOPE }, /* MT */
+       { blk_close, MAN_BSCOPE }, /* ME */

 };
 };

-
-const  struct man_macro * const man_macros = __man_macros;
+const  struct man_macro *const man_macros = __man_macros - MAN_TH;

 
 
 void
 
 
 void

-man_unscope(struct man *man, const struct roff_node *to)
+man_unscope(struct roff_man *man, const struct roff_node *to)

 {
        struct roff_node *n;
 
 {
        struct roff_node *n;
 


@@ -94,13 +92,12 @@ man_unscope(struct man *man, const struct roff_node *to)
 
                /* Reached the end of the document? */
 
 
                /* Reached the end of the document? */
 

-               if (to == NULL && ! (n->flags & MAN_VALID)) {
+               if (to == NULL && ! (n->flags & NODE_VALID)) {

                        if (man->flags & (MAN_BLINE | MAN_ELINE) &&
                            man_macros[n->tok].flags & MAN_SCOPED) {
                                mandoc_vmsg(MANDOCERR_BLK_LINE,
                                    man->parse, n->line, n->pos,
                        if (man->flags & (MAN_BLINE | MAN_ELINE) &&
                            man_macros[n->tok].flags & MAN_SCOPED) {
                                mandoc_vmsg(MANDOCERR_BLK_LINE,
                                    man->parse, n->line, n->pos,

-                                   "EOF breaks %s",
-                                   man_macronames[n->tok]);
+                                   "EOF breaks %s", roff_name[n->tok]);

                                if (man->flags & MAN_ELINE)
                                        man->flags &= ~MAN_ELINE;
                                else {
                                if (man->flags & MAN_ELINE)
                                        man->flags &= ~MAN_ELINE;
                                else {


@@ -110,14 +107,14 @@ man_unscope(struct man *man, const struct roff_node *to)
                                }
                                man->last = n;
                                n = n->parent;
                                }
                                man->last = n;
                                n = n->parent;

-                               man_node_delete(man, man->last);
+                               roff_node_delete(man, man->last);

                                continue;
                        }
                        if (n->type == ROFFT_BLOCK &&
                            man_macros[n->tok].fp == blk_exp)
                                mandoc_msg(MANDOCERR_BLK_NOEND,
                                    man->parse, n->line, n->pos,
                                continue;
                        }
                        if (n->type == ROFFT_BLOCK &&
                            man_macros[n->tok].fp == blk_exp)
                                mandoc_msg(MANDOCERR_BLK_NOEND,
                                    man->parse, n->line, n->pos,

-                                   man_macronames[n->tok]);
+                                   roff_name[n->tok]);

                }
 
                /*
                }
 
                /*


@@ -129,7 +126,7 @@ man_unscope(struct man *man, const struct roff_node *to)
 
                man->last = n;
                n = n->parent;
 
                man->last = n;
                n = n->parent;

-               man_valid_post(man);
+               man->last->flags |= NODE_VALID;

        }
 
        /*
        }
 
        /*


@@ -140,7 +137,7 @@ man_unscope(struct man *man, const struct roff_node *to)
         */
 
        man->next = (man->last == to) ?
         */
 
        man->next = (man->last == to) ?

-           MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
+           ROFF_NEXT_CHILD : ROFF_NEXT_SIBLING;

 }
 
 /*
 }
 
 /*


@@ -149,21 +146,21 @@ man_unscope(struct man *man, const struct roff_node *to)
  * scopes.  When a scope is closed, it must be validated and actioned.
  */
 static void
  * scopes.  When a scope is closed, it must be validated and actioned.
  */
 static void

-rew_scope(struct man *man, int tok)
+rew_scope(struct roff_man *man, enum roff_tok tok)

 {
        struct roff_node *n;
 
        /* Preserve empty paragraphs before RS. */
 
        n = man->last;
 {
        struct roff_node *n;
 
        /* Preserve empty paragraphs before RS. */
 
        n = man->last;

-       if (tok == MAN_RS && n->nchild == 0 &&
+       if (tok == MAN_RS && n->child == NULL &&

            (n->tok == MAN_P || n->tok == MAN_PP || n->tok == MAN_LP))
                return;
 
        for (;;) {
                if (n->type == ROFFT_ROOT)
                        return;
            (n->tok == MAN_P || n->tok == MAN_PP || n->tok == MAN_LP))
                return;
 
        for (;;) {
                if (n->type == ROFFT_ROOT)
                        return;

-               if (n->flags & MAN_VALID) {
+               if (n->flags & NODE_VALID) {

                        n = n->parent;
                        continue;
                }
                        n = n->parent;
                        continue;
                }


@@ -192,7 +189,7 @@ rew_scope(struct man *man, int tok)
 void
 blk_close(MACRO_PROT_ARGS)
 {
 void
 blk_close(MACRO_PROT_ARGS)
 {

-       int                      ntok;
+       enum roff_tok            ntok;

        const struct roff_node  *nn;
        char                    *p;
        int                      nrew, target;
        const struct roff_node  *nn;
        char                    *p;
        int                      nrew, target;


@@ -222,9 +219,11 @@ blk_close(MACRO_PROT_ARGS)
        case MAN_UE:
                ntok = MAN_UR;
                break;
        case MAN_UE:
                ntok = MAN_UR;
                break;

+       case MAN_ME:
+               ntok = MAN_MT;
+               break;

        default:
                abort();
        default:
                abort();

-               /* NOTREACHED */

        }
 
        for (nn = man->last->parent; nn; nn = nn->parent)
        }
 
        for (nn = man->last->parent; nn; nn = nn->parent)


@@ -233,7 +232,7 @@ blk_close(MACRO_PROT_ARGS)
 
        if (nn == NULL) {
                mandoc_msg(MANDOCERR_BLK_NOTOPEN, man->parse,
 
        if (nn == NULL) {
                mandoc_msg(MANDOCERR_BLK_NOTOPEN, man->parse,

-                   line, ppos, man_macronames[tok]);
+                   line, ppos, roff_name[tok]);

                rew_scope(man, MAN_PP);
        } else {
                line = man->last->line;
                rew_scope(man, MAN_PP);
        } else {
                line = man->last->line;


@@ -241,6 +240,10 @@ blk_close(MACRO_PROT_ARGS)
                ntok = man->last->tok;
                man_unscope(man, nn);
 
                ntok = man->last->tok;
                man_unscope(man, nn);
 

+               if (tok == MAN_RE && nn->head->aux > 0)
+                       roff_setreg(man->roff, "an-margin",
+                           nn->head->aux, '-');
+

                /* Move a trailing paragraph behind the block. */
 
                if (ntok == MAN_LP || ntok == MAN_PP || ntok == MAN_P) {
                /* Move a trailing paragraph behind the block. */
 
                if (ntok == MAN_LP || ntok == MAN_PP || ntok == MAN_P) {


@@ -258,21 +261,28 @@ blk_exp(MACRO_PROT_ARGS)
        int              la;
 
        rew_scope(man, tok);
        int              la;
 
        rew_scope(man, tok);

-       man_block_alloc(man, line, ppos, tok);
-       man_head_alloc(man, line, ppos, tok);
-       head = man->last;
+       roff_block_alloc(man, line, ppos, tok);
+       head = roff_head_alloc(man, line, ppos, tok);

 
        la = *pos;
 
        la = *pos;

-       if (man_args(man, line, pos, buf, &p))
-               man_word_alloc(man, line, la, p);
+       if (man_args(man, line, pos, buf, &p)) {
+               roff_word_alloc(man, line, la, p);
+               if (tok == MAN_RS) {
+                       if (roff_getreg(man->roff, "an-margin") == 0)
+                               roff_setreg(man->roff, "an-margin",
+                                   7 * 24, '=');
+                       if ((head->aux = strtod(p, NULL) * 24.0) > 0)
+                               roff_setreg(man->roff, "an-margin",
+                                   head->aux, '+');
+               }
+       }

 
        if (buf[*pos] != '\0')
 
        if (buf[*pos] != '\0')

-               mandoc_vmsg(MANDOCERR_ARG_EXCESS,
-                   man->parse, line, *pos, "%s ... %s",
-                   man_macronames[tok], buf + *pos);
+               mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse, line,
+                   *pos, "%s ... %s", roff_name[tok], buf + *pos);

 
        man_unscope(man, head);
 
        man_unscope(man, head);

-       man_body_alloc(man, line, ppos, tok);
+       roff_body_alloc(man, line, ppos, tok);

 }
 
 /*
 }
 
 /*


@@ -289,9 +299,10 @@ blk_imp(MACRO_PROT_ARGS)
        struct roff_node *n;
 
        rew_scope(man, tok);
        struct roff_node *n;
 
        rew_scope(man, tok);

-       man_block_alloc(man, line, ppos, tok);
-       man_head_alloc(man, line, ppos, tok);
-       n = man->last;
+       n = roff_block_alloc(man, line, ppos, tok);
+       if (n->tok == MAN_SH || n->tok == MAN_SS)
+               man->flags &= ~MAN_LITERAL;
+       n = roff_head_alloc(man, line, ppos, tok);

 
        /* Add line arguments. */
 
 
        /* Add line arguments. */
 


@@ -299,7 +310,7 @@ blk_imp(MACRO_PROT_ARGS)
                la = *pos;
                if ( ! man_args(man, line, pos, buf, &p))
                        break;
                la = *pos;
                if ( ! man_args(man, line, pos, buf, &p))
                        break;

-               man_word_alloc(man, line, la, p);
+               roff_word_alloc(man, line, la, p);

        }
 
        /*
        }
 
        /*


@@ -317,7 +328,7 @@ blk_imp(MACRO_PROT_ARGS)
        /* Close out the head and open the body. */
 
        man_unscope(man, n);
        /* Close out the head and open the body. */
 
        man_unscope(man, n);

-       man_body_alloc(man, line, ppos, tok);
+       roff_body_alloc(man, line, ppos, tok);

 }
 
 void
 }
 
 void


@@ -327,22 +338,20 @@ in_line_eoln(MACRO_PROT_ARGS)
        char            *p;
        struct roff_node *n;
 
        char            *p;
        struct roff_node *n;
 

-       man_elem_alloc(man, line, ppos, tok);
+       roff_elem_alloc(man, line, ppos, tok);

        n = man->last;
 
        for (;;) {
        n = man->last;
 
        for (;;) {

-               if (buf[*pos] != '\0' && (tok == MAN_br ||
-                   tok == MAN_fi || tok == MAN_nf)) {
+               if (buf[*pos] != '\0' && (tok == MAN_fi || tok == MAN_nf)) {

                        mandoc_vmsg(MANDOCERR_ARG_SKIP,
                            man->parse, line, *pos, "%s %s",
                        mandoc_vmsg(MANDOCERR_ARG_SKIP,
                            man->parse, line, *pos, "%s %s",

-                           man_macronames[tok], buf + *pos);
+                           roff_name[tok], buf + *pos);

                        break;
                }
                        break;
                }

-               if (buf[*pos] != '\0' && man->last != n &&
-                   (tok == MAN_PD || tok == MAN_ft || tok == MAN_sp)) {
+               if (buf[*pos] != '\0' && man->last != n && tok == MAN_PD) {

                        mandoc_vmsg(MANDOCERR_ARG_EXCESS,
                            man->parse, line, *pos, "%s ... %s",
                        mandoc_vmsg(MANDOCERR_ARG_EXCESS,
                            man->parse, line, *pos, "%s ... %s",

-                           man_macronames[tok], buf + *pos);
+                           roff_name[tok], buf + *pos);

                        break;
                }
                la = *pos;
                        break;
                }
                la = *pos;


@@ -350,19 +359,19 @@ in_line_eoln(MACRO_PROT_ARGS)
                        break;
                if (man_macros[tok].flags & MAN_JOIN &&
                    man->last->type == ROFFT_TEXT)
                        break;
                if (man_macros[tok].flags & MAN_JOIN &&
                    man->last->type == ROFFT_TEXT)

-                       man_word_append(man, p);
+                       roff_word_append(man, p);

                else
                else

-                       man_word_alloc(man, line, la, p);
+                       roff_word_alloc(man, line, la, p);

        }
 
        /*
        }
 
        /*

-        * Append MAN_EOS in case the last snipped argument
+        * Append NODE_EOS in case the last snipped argument

         * ends with a dot, e.g. `.IR syslog (3).'
         */
 
        if (n != man->last &&
            mandoc_eos(man->last->string, strlen(man->last->string)))
         * ends with a dot, e.g. `.IR syslog (3).'
         */
 
        if (n != man->last &&
            mandoc_eos(man->last->string, strlen(man->last->string)))

-               man->last->flags |= MAN_EOS;
+               man->last->flags |= NODE_EOS;

 
        /*
         * If no arguments are specified and this is MAN_SCOPED (i.e.,
 
        /*
         * If no arguments are specified and this is MAN_SCOPED (i.e.,


@@ -377,42 +386,27 @@ in_line_eoln(MACRO_PROT_ARGS)
        }
 
        assert(man->last->type != ROFFT_ROOT);
        }
 
        assert(man->last->type != ROFFT_ROOT);

-       man->next = MAN_NEXT_SIBLING;
+       man->next = ROFF_NEXT_SIBLING;

 
 

-       /*
-        * Rewind our element scope.  Note that when TH is pruned, we'll
-        * be back at the root, so make sure that we don't clobber as
-        * its sibling.
-        */
+       /* Rewind our element scope. */

 
        for ( ; man->last; man->last = man->last->parent) {
 
        for ( ; man->last; man->last = man->last->parent) {

+               man_state(man, man->last);

                if (man->last == n)
                        break;
                if (man->last == n)
                        break;

-               if (man->last->type == ROFFT_ROOT)
-                       break;
-               man_valid_post(man);

        }
        }

-
-       assert(man->last);
-
-       /*
-        * Same here regarding whether we're back at the root.
-        */
-
-       if (man->last->type != ROFFT_ROOT)
-               man_valid_post(man);

 }
 
 }
 

-

 void
 void

-man_macroend(struct man *man)
+man_endparse(struct roff_man *man)

 {
 
        man_unscope(man, man->first);
 {
 
        man_unscope(man, man->first);

+       man->flags &= ~MAN_LITERAL;

 }
 
 static int
 }
 
 static int

-man_args(struct man *man, int line, int *pos, char *buf, char **v)
+man_args(struct roff_man *man, int line, int *pos, char *buf, char **v)

 {
        char     *start;
 
 {
        char     *start;
 


@@ -421,8 +415,8 @@ man_args(struct man *man, int line, int *pos, char *buf, char **v)
        assert(' ' != *start);
 
        if ('\0' == *start)
        assert(' ' != *start);
 
        if ('\0' == *start)

-               return(0);
+               return 0;

 
        *v = mandoc_getarg(man->parse, v, line, pos);
 
        *v = mandoc_getarg(man->parse, v, line, pos);

-       return(1);
+       return 1;

 }
 }