Initial revision

1 files changed, 1804 insertions, 0 deletions

diff --git a/pwmanager/libcrypt/crypt/sexp.c b/pwmanager/libcrypt/crypt/sexp.c
new file mode 100644
index 0000000..c19dc33
--- a/dev/null
+++ b/pwmanager/libcrypt/crypt/sexp.c
@@ -0,0 +1,1804 @@
+/* sexp.c  -  S-Expression handling
+ * Copyright (C) 1999, 2000, 2001, 2002, 2003,
+ *               2004 Free Software Foundation, Inc.
+ *
+ * This file is part of Libgcrypt.
+ *
+ * Libgcrypt is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser general Public License as
+ * published by the Free Software Foundation; either version 2.1 of
+ * the License, or (at your option) any later version.
+ *
+ * Libgcrypt is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
+ */
+
+
+#include <config.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdarg.h>
+#include <ctype.h>
+#include <assert.h>
+
+#define GCRYPT_NO_MPI_MACROS 1
+#include "g10lib.h"
+#include "memory.h"
+
+typedef struct gcry_sexp *NODE;
+typedef unsigned short DATALEN;
+
+struct gcry_sexp
+{
+  byte d[1];
+};
+
+#define ST_STOP  0
+#define ST_DATA  1  /* datalen follows */
+#define ST_HINT  2  /* datalen follows */
+#define ST_OPEN  3
+#define ST_CLOSE 4
+
+/* the atoi macros assume that the buffer has only valid digits */
+#define atoi_1(p)   (*(p) - '0' )
+#define xtoi_1(p)   (*(p) <= '9'? (*(p)- '0'): \
+                     *(p) <= 'F'? (*(p)-'A'+10):(*(p)-'a'+10))
+#define xtoi_2(p)   ((xtoi_1(p) * 16) + xtoi_1((p)+1))
+
+#define TOKEN_SPECIALS  "-./_:*+="
+
+static gcry_error_t
+sexp_sscan (gcry_sexp_t *retsexp, size_t *erroff,
+	    const char *buffer, size_t length, int argflag,
+	    va_list arg_ptr, void **arg_list);
+
+#if 0
+static void
+dump_mpi( gcry_mpi_t a )
+{
+    char buffer[1000];
+    size_t n = 1000;
+
+    if( !a )
+	fputs("[no MPI]", stderr );
+    else if( gcry_mpi_print( GCRYMPI_FMT_HEX, buffer, &n, a ) )
+	fputs("[MPI too large to print]", stderr );
+    else
+	fputs( buffer, stderr );
+}
+#endif
+
+static void
+dump_string (const byte *p, size_t n, int delim )
+{
+  for (; n; n--, p++ )
+    {
+      if ((*p & 0x80) || iscntrl( *p ) || *p == delim ) 
+        {
+          if( *p == '\n' )
+            log_printf ("\\n");
+          else if( *p == '\r' )
+            log_printf ("\\r");
+          else if( *p == '\f' )
+            log_printf ("\\f");
+          else if( *p == '\v' )
+            log_printf ("\\v");
+	    else if( *p == '\b' )
+              log_printf ("\\b");
+          else if( !*p )       
+            log_printf ("\\0");
+          else
+            log_printf ("\\x%02x", *p );
+	}
+      else
+        log_printf ("%c", *p);
+    }
+}
+
+
+void
+gcry_sexp_dump (const gcry_sexp_t a)
+{
+  const byte *p;
+  int indent = 0;
+  int type;
+
+  if (!a)
+    {
+      log_printf ( "[nil]\n");
+      return;
+    }
+
+  p = a->d;
+  while ( (type = *p) != ST_STOP )
+    {
+      p++;
+      switch ( type )
+        {
+        case ST_OPEN:
+          log_printf ("%*s[open]\n", 2*indent, "");
+          indent++;
+          break;
+        case ST_CLOSE:
+          if( indent )
+            indent--;
+          log_printf ("%*s[close]\n", 2*indent, "");
+          break;
+        case ST_DATA: {
+          DATALEN n;
+          memcpy ( &n, p, sizeof n );
+          p += sizeof n;
+          log_printf ("%*s[data=\"", 2*indent, "" );
+          dump_string (p, n, '\"' );
+          log_printf ("\"]\n");
+          p += n;
+        }
+        break;
+        default:
+          log_printf ("%*s[unknown tag %d]\n", 2*indent, "", type);
+          break;
+	}
+    }
+}
+
+/****************
+ * Pass list through except when it is an empty list - in that case
+ * return NULL and release the passed list.
+ */
+static gcry_sexp_t
+normalize ( gcry_sexp_t list )
+{
+    char *p;
+    if ( !list )
+	return NULL;
+    p = list->d;
+    if ( *p == ST_STOP ) {
+	/* this is "" */
+	gcry_sexp_release ( list );
+	return NULL;
+    }
+    if( *p == ST_OPEN && p[1] == ST_CLOSE ) {
+	/* this is "()" */
+	gcry_sexp_release ( list );
+	return NULL;
+    }
+
+    return list;
+}
+
+/* Create a new S-expression object by reading LENGTH bytes from
+   BUFFER, assuming it is canonilized encoded or autodetected encoding
+   when AUTODETECT is set to 1.  With FREEFNC not NULL, ownership of
+   the buffer is transferred to tyhe newle created object.  FREEFNC
+   should be the freefnc used to release BUFFER; there is no guarantee
+   at which point this function is called; most likey you want to use
+   free() or gcry_free(). 
+ 
+   Passing LENGTH and AUTODETECT as 0 is allowed to indicate that
+   BUFFER points to a valid canonical encoded S-expression.  A LENGTH
+   of 0 and AUTODETECT 1 indicates that buffer points to a
+   null-terminated string.
+  
+   This function returns 0 and and the pointer to the new object in
+   RETSEXP or an error code in which case RETSEXP is set to NULL.  */
+gcry_error_t
+gcry_sexp_create (gcry_sexp_t *retsexp, void *buffer, size_t length,
+                  int autodetect, void (*freefnc)(void*) )
+{
+  gcry_error_t errcode;
+  gcry_sexp_t se;
+  volatile va_list dummy_arg_ptr;
+
+  if (!retsexp)
+    return gcry_error (GPG_ERR_INV_ARG);
+  *retsexp = NULL;
+  if (autodetect < 0 || autodetect > 1 || !buffer)
+    return gcry_error (GPG_ERR_INV_ARG);
+
+  if (!length && !autodetect)
+    { /* What a brave caller to assume that there is really a canonical
+         encoded S-expression in buffer */
+      length = gcry_sexp_canon_len (buffer, 0, NULL, &errcode);
+      if (!length)
+        return errcode;
+    }
+  else if (!length && autodetect)
+    { /* buffer is a string */
+      length = strlen ((char *)buffer);
+    }
+
+  errcode = sexp_sscan (&se, NULL, buffer, length, 0, dummy_arg_ptr, NULL);
+  if (errcode)
+    return errcode;
+
+  *retsexp = se;
+  if (freefnc)
+    {
+      /* For now we release the buffer immediately.  As soon as we
+         have changed the internal represenation of S-expression to
+         the canoncial format - which has the advantage of faster
+         parsing - we will use this function as a closure in our
+         GCRYSEXP object and use the BUFFER directly */
+      freefnc (buffer);
+    }
+  return gcry_error (GPG_ERR_NO_ERROR);
+}
+
+/* Same as gcry_sexp_create but don't transfer ownership */
+gcry_error_t
+gcry_sexp_new (gcry_sexp_t *retsexp, const void *buffer, size_t length,
+               int autodetect)
+{
+  return gcry_sexp_create (retsexp, (void *)buffer, length, autodetect, NULL);
+}
+
+
+/****************
+ * Release resource of the given SEXP object.
+ */
+void
+gcry_sexp_release( gcry_sexp_t sexp )
+{
+  if (sexp)
+    {
+      if (gcry_is_secure (sexp))
+        {
+          /* Extra paranoid wiping. */
+          const byte *p = sexp->d;
+          int type;
+
+          while ( (type = *p) != ST_STOP )
+            {
+              p++;
+              switch ( type )
+                {
+                case ST_OPEN:
+                  break;
+                case ST_CLOSE:
+                  break;
+                case ST_DATA: 
+                  {
+                    DATALEN n;
+                    memcpy ( &n, p, sizeof n );
+                    p += sizeof n;
+                    p += n;
+                  }
+                  break;
+                default:
+                  break;
+                }
+            }
+          wipememory (sexp->d, p - sexp->d);
+        }
+      gcry_free ( sexp );
+    }
+}
+
+
+/****************
+ * Make a pair from lists a and b, don't use a or b later on.
+ * Special behaviour:  If one is a single element list we put the
+ * element straight into the new pair.
+ */
+gcry_sexp_t
+gcry_sexp_cons( const gcry_sexp_t a, const gcry_sexp_t b )
+{
+  /* NYI: Implementation should be quite easy with our new data
+     representation */
+  BUG ();
+  return NULL;
+}
+
+
+/****************
+ * Make a list from all items in the array the end of the array is marked
+ * with a NULL.
+ */
+gcry_sexp_t
+gcry_sexp_alist( const gcry_sexp_t *array )
+{
+  /* NYI: Implementation should be quite easy with our new data
+     representation. */
+  BUG ();
+  return NULL;
+}
+
+/****************
+ * Make a list from all items, the end of list is indicated by a NULL
+ */
+gcry_sexp_t
+gcry_sexp_vlist( const gcry_sexp_t a, ... )
+{
+  /* NYI: Implementation should be quite easy with our new data
+     representation. */
+  BUG ();
+  return NULL;
+}
+
+
+/****************
+ * Append n to the list a
+ * Returns: a new ist (which maybe a)
+ */
+gcry_sexp_t
+gcry_sexp_append( const gcry_sexp_t a, const gcry_sexp_t n )
+{
+  /* NYI: Implementation should be quite easy with our new data
+     representation. */
+  BUG ();
+  return NULL;
+}
+
+gcry_sexp_t
+gcry_sexp_prepend( const gcry_sexp_t a, const gcry_sexp_t n )
+{
+  /* NYI: Implementation should be quite easy with our new data
+     representation. */
+  BUG ();
+  return NULL;
+}
+
+
+
+/****************
+ * Locate token in a list. The token must be the car of a sublist.
+ * Returns: A new list with this sublist or NULL if not found.
+ */
+gcry_sexp_t
+gcry_sexp_find_token( const gcry_sexp_t list, const char *tok, size_t toklen )
+{
+    const byte *p;
+    DATALEN n;
+
+    if ( !list )
+	return NULL;
+
+    if( !toklen )
+	toklen = strlen(tok);
+    p = list->d;
+    while ( *p != ST_STOP ) {
+	if ( *p == ST_OPEN && p[1] == ST_DATA ) {
+	    const byte *head = p;
+
+	    p += 2;
+	    memcpy ( &n, p, sizeof n ); p += sizeof n;
+	    if ( n == toklen && !memcmp( p, tok, toklen ) ) { /* found it */
+		gcry_sexp_t newlist;
+		byte *d;
+		int level = 1;
+
+		/* look for the end of the list */
+		for ( p += n; level; p++ ) {
+		    if ( *p == ST_DATA ) {
+			memcpy ( &n, ++p, sizeof n );
+			p += sizeof n + n;
+			p--; /* compensate for later increment */
+		    }
+		    else if ( *p == ST_OPEN ) {
+			level++;
+		    }
+		    else if ( *p == ST_CLOSE ) {
+			level--;
+		    }
+		    else if ( *p == ST_STOP ) {
+			BUG ();
+		    }
+		} while ( level );
+		n = p - head;
+
+		newlist = gcry_xmalloc ( sizeof *newlist + n );
+		d = newlist->d;
+		memcpy ( d, head, n ); d += n;
+		*d++ = ST_STOP;
+		return normalize ( newlist );
+	    }
+	    p += n;
+	}
+	else if ( *p == ST_DATA ) {
+	    memcpy ( &n, ++p, sizeof n ); p += sizeof n;
+	    p += n;
+	}
+	else
+	    p++;
+    }
+    return NULL;
+}
+
+/****************
+ * Return the length of the given list
+ */
+int
+gcry_sexp_length( const gcry_sexp_t list )
+{
+    const byte *p;
+    DATALEN n;
+    int type;
+    int length = 0;
+    int level = 0;
+
+    if ( !list )
+	return 0;
+
+    p = list->d;
+    while ( (type=*p) != ST_STOP ) {
+	p++;
+	if ( type == ST_DATA ) {
+	    memcpy ( &n, p, sizeof n );
+	    p += sizeof n + n;
+	    if ( level == 1 )
+		length++;
+	}
+	else if ( type == ST_OPEN ) {
+	    if ( level == 1 )
+		length++;
+	    level++;
+	}
+	else if ( type == ST_CLOSE ) {
+	    level--;
+	}
+    }
+    return length;
+}
+
+
+
+/****************
+ * Extract the CAR of the given list
+ */
+gcry_sexp_t
+gcry_sexp_nth( const gcry_sexp_t list, int number )
+{
+    const byte *p;
+    DATALEN n;
+    gcry_sexp_t newlist;
+    byte *d;
+    int level = 0;
+
+    if ( !list || list->d[0] != ST_OPEN )
+	return NULL;
+    p = list->d;
+
+    while ( number > 0 ) {
+	p++;
+	if ( *p == ST_DATA ) {
+	    memcpy ( &n, ++p, sizeof n );
+	    p += sizeof n + n;
+	    p--;
+	    if ( !level )
+		number--;
+	}
+	else if ( *p == ST_OPEN ) {
+	    level++;
+	}
+	else if ( *p == ST_CLOSE ) {
+	    level--;
+	    if ( !level )
+		number--;
+	}
+	else if ( *p == ST_STOP ) {
+	    return NULL;
+	}
+    }
+    p++;
+
+    if ( *p == ST_DATA ) {
+	memcpy ( &n, p, sizeof n ); p += sizeof n;
+	newlist = gcry_xmalloc ( sizeof *newlist + n + 1 );
+	d = newlist->d;
+	memcpy ( d, p, n ); d += n;
+	*d++ = ST_STOP;
+    }
+    else if ( *p == ST_OPEN ) {
+	const byte *head = p;
+
+	level = 1;
+	do {
+	    p++;
+	    if ( *p == ST_DATA ) {
+		memcpy ( &n, ++p, sizeof n );
+		p += sizeof n + n;
+		p--;
+	    }
+	    else if ( *p == ST_OPEN ) {
+		level++;
+	    }
+	    else if ( *p == ST_CLOSE ) {
+		level--;
+	    }
+	    else if ( *p == ST_STOP ) {
+		BUG ();
+	    }
+	} while ( level );
+	n = p + 1 - head;
+
+	newlist = gcry_xmalloc ( sizeof *newlist + n );
+	d = newlist->d;
+	memcpy ( d, head, n ); d += n;
+	*d++ = ST_STOP;
+    }
+    else
+	newlist = NULL;
+
+    return normalize (newlist);
+}
+
+gcry_sexp_t
+gcry_sexp_car( const gcry_sexp_t list )
+{
+    return gcry_sexp_nth ( list, 0 );
+}
+
+/****************
+ * Get data from the car.  The returned value is valid as long as the list
+ * is not modified.
+ */
+const char *
+gcry_sexp_nth_data( const gcry_sexp_t list, int number, size_t *datalen )
+{
+    const byte *p;
+    DATALEN n;
+    int level = 0;
+
+    *datalen = 0;
+    if ( !list ) {
+	return NULL;
+    }
+    p = list->d;
+    if ( *p == ST_OPEN )
+	p++;	     /* yep, a list */
+    else if (number )
+	return NULL; /* not a list but an n > 0 element requested */
+
+    /* skip n elements */
+    while ( number > 0 ) {
+	if ( *p == ST_DATA ) {
+	    memcpy ( &n, ++p, sizeof n );
+	    p += sizeof n + n;
+	    p--;
+	    if ( !level )
+		number--;
+	}
+	else if ( *p == ST_OPEN ) {
+	    level++;
+	}
+	else if ( *p == ST_CLOSE ) {
+	    level--;
+	    if ( !level )
+		number--;
+	}
+	else if ( *p == ST_STOP ) {
+	    return NULL;
+	}
+	p++;
+    }
+
+
+    if ( *p == ST_DATA ) {
+	memcpy ( &n, ++p, sizeof n );
+	*datalen = n;
+	return p + sizeof n;
+    }
+
+    return NULL;
+}
+
+/****************
+ * Get a MPI from the car
+ */
+gcry_mpi_t
+gcry_sexp_nth_mpi( gcry_sexp_t list, int number, int mpifmt )
+{
+    const byte *p;
+    DATALEN n;
+    int level = 0;
+
+    if ( !list )
+	return NULL;
+    if ( !mpifmt )
+	mpifmt = GCRYMPI_FMT_STD;
+
+    p = list->d;
+    if ( *p == ST_OPEN )
+	p++;	     /* yep, a list */
+    else if (number )
+	return NULL; /* not a list but an n > 0 element requested */
+
+    /* skip n elements */
+    while ( number > 0 ) {
+	if ( *p == ST_DATA ) {
+	    memcpy ( &n, ++p, sizeof n );
+	    p += sizeof n + n;
+	    p--;
+	    if ( !level )
+		number--;
+	}
+	else if ( *p == ST_OPEN ) {
+	    level++;
+	}
+	else if ( *p == ST_CLOSE ) {
+	    level--;
+	    if ( !level )
+		number--;
+	}
+	else if ( *p == ST_STOP ) {
+	    return NULL;
+	}
+	p++;
+    }
+
+    if ( *p == ST_DATA ) {
+	gcry_mpi_t a;
+	size_t nbytes;
+
+	memcpy ( &n, ++p, sizeof n );
+	p += sizeof n;
+	nbytes = n;
+	if( !gcry_mpi_scan( &a, mpifmt, p, n, &nbytes ) )
+	    return a;
+    }
+
+    return NULL;
+}
+
+
+/****************
+ * Get the CDR
+ */
+gcry_sexp_t
+gcry_sexp_cdr( const gcry_sexp_t list )
+{
+    const byte *p;
+    const byte *head;
+    DATALEN n;
+    gcry_sexp_t newlist;
+    byte *d;
+    int level = 0;
+    int skip = 1;
+
+    if ( !list || list->d[0] != ST_OPEN )
+	return NULL;
+    p = list->d;
+
+    while ( skip > 0 ) {
+	p++;
+	if ( *p == ST_DATA ) {
+	    memcpy ( &n, ++p, sizeof n );
+	    p += sizeof n + n;
+	    p--;
+	    if ( !level )
+		skip--;
+	}
+	else if ( *p == ST_OPEN ) {
+	    level++;
+	}
+	else if ( *p == ST_CLOSE ) {
+	    level--;
+	    if ( !level )
+		skip--;
+	}
+	else if ( *p == ST_STOP ) {
+	    return NULL;
+	}
+    }
+    p++;
+
+    head = p;
+    level = 0;
+    do {
+	if ( *p == ST_DATA ) {
+	    memcpy ( &n, ++p, sizeof n );
+	    p += sizeof n + n;
+	    p--;
+	}
+	else if ( *p == ST_OPEN ) {
+	    level++;
+	}
+	else if ( *p == ST_CLOSE ) {
+	    level--;
+	}
+	else if ( *p == ST_STOP ) {
+	    return NULL;
+	}
+	p++;
+    } while ( level );
+    n = p - head;
+
+    newlist = gcry_xmalloc ( sizeof *newlist + n + 2 );
+    d = newlist->d;
+    *d++ = ST_OPEN;
+    memcpy ( d, head, n ); d += n;
+    *d++ = ST_CLOSE;
+    *d++ = ST_STOP;
+
+    return normalize (newlist);
+}
+
+gcry_sexp_t
+gcry_sexp_cadr ( const gcry_sexp_t list )
+{
+    gcry_sexp_t a, b;
+
+    a = gcry_sexp_cdr ( list );
+    b = gcry_sexp_car ( a );
+    gcry_sexp_release ( a );
+    return b;
+}
+
+
+
+static int
+hextobyte( const byte *s )
+{
+    int c=0;
+
+    if( *s >= '0' && *s <= '9' )
+	c = 16 * (*s - '0');
+    else if( *s >= 'A' && *s <= 'F' )
+	c = 16 * (10 + *s - 'A');
+    else if( *s >= 'a' && *s <= 'f' ) {
+	c = 16 * (10 + *s - 'a');
+    }
+    s++;
+    if( *s >= '0' && *s <= '9' )
+	c += *s - '0';
+    else if( *s >= 'A' && *s <= 'F' )
+	c += 10 + *s - 'A';
+    else if( *s >= 'a' && *s <= 'f' ) {
+	c += 10 + *s - 'a';
+    }
+    return c;
+}
+
+struct make_space_ctx {
+    gcry_sexp_t sexp;
+    size_t allocated;
+    byte *pos;
+};
+
+static void
+make_space ( struct make_space_ctx *c, size_t n )
+{
+    size_t used = c->pos - c->sexp->d;
+
+    if ( used + n + sizeof(DATALEN) + 1 >= c->allocated ) {
+	gcry_sexp_t newsexp;
+	byte *newhead;
+
+	c->allocated += 2*(n+sizeof(DATALEN)+1);
+	newsexp = gcry_xrealloc ( c->sexp, sizeof *newsexp + c->allocated - 1 );
+	newhead = newsexp->d;
+	c->pos = newhead + used;
+	c->sexp = newsexp;
+    }
+}
+
+
+/* Unquote STRING of LENGTH and store it into BUF.  The surrounding
+   quotes are must already be removed from STRING.  We assume that the
+   quoted string is syntacillay correct.  */
+static size_t
+unquote_string (const unsigned char *string, size_t length, unsigned char *buf)
+{
+  int esc = 0;
+  const unsigned char *s = string;
+  unsigned char *d = buf;
+  size_t n = length;
+
+  for (; n; n--, s++)
+    {
+      if (esc)
+        {
+          switch (*s)
+            {
+            case 'b':  *d++ = '\b'; break;
+            case 't':  *d++ = '\t'; break;
+            case 'v':  *d++ = '\v'; break;
+            case 'n':  *d++ = '\n'; break;
+            case 'f':  *d++ = '\f'; break;
+            case 'r':  *d++ = '\r'; break;
+            case '"':  *d++ = '\"'; break;
+            case '\'': *d++ = '\''; break;
+            case '\\': *d++ = '\\'; break;
+
+            case '\r':  /* ignore CR[,LF] */
+              if (n>1 && s[1] == '\n')
+                {
+                  s++; n--;
+                }
+              esc = 0;
+              break;
+              
+            case '\n':  /* ignore LF[,CR] */
+              if (n>1 && s[1] == '\r')
+                {
+                  s++; n--;
+                }
+              break;
+
+            case 'x': /* hex value */
+              if (n>2 && hexdigitp (s+1) && hexdigitp (s+2))
+                {
+                  s++; n--;
+                  *d++ = xtoi_2 (s);
+                  s++; n--;
+                }
+              break;
+
+            default:
+              if (n>2 && octdigitp (s) && octdigitp (s+1) && octdigitp (s+2))
+                {
+                  *d++ = (atoi_1 (s)*64) + (atoi_1 (s+1)*8) + atoi_1 (s+2);
+                  s += 2;
+                  n -= 2;
+                }
+              break;
+	    }
+          esc = 0;
+        }
+      else if( *s == '\\' )
+        esc = 1;
+      else
+        *d++ = *s;
+    } 
+
+  return d - buf;
+}
+
+/****************
+ * Scan the provided buffer and return the S expression in our internal
+ * format.  Returns a newly allocated expression.  If erroff is not NULL and
+ * a parsing error has occured, the offset into buffer will be returned.
+ * If ARGFLAG is true, the function supports some printf like
+ * expressions.
+ *  These are:
+ *	%m - MPI
+ *	%s - string (no autoswitch to secure allocation)
+ *	%d - integer stored as string (no autoswitch to secure allocation)
+ *      %b - memory buffer; this takes _two_ arguments: an integer with the 
+ *           length of the buffer and a pointer to the buffer.
+ *  all other format elements are currently not defined and return an error.
+ *  this includes the "%%" sequence becauce the percent sign is not an
+ *  allowed character.
+ * FIXME: We should find a way to store the secure-MPIs not in the string
+ * but as reference to somewhere - this can help us to save huge amounts
+ * of secure memory.  The problem is, that if only one element is secure, all
+ * other elements are automagicaly copied to secure memory too, so the most
+ * common operation gcry_sexp_cdr_mpi() will always return a secure MPI
+ * regardless whether it is needed or not.
+ */
+static gcry_error_t
+sexp_sscan (gcry_sexp_t *retsexp, size_t *erroff,
+	    const char *buffer, size_t length, int argflag,
+	    va_list arg_ptr, void **arg_list)
+{
+  gcry_err_code_t err = GPG_ERR_NO_ERROR;
+  static const char tokenchars[] =
+    "abcdefghijklmnopqrstuvwxyz"
+    "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+    "0123456789-./_:*+=";
+  const char *p;
+  size_t n;
+  const char *digptr = NULL;
+  const char *quoted = NULL;
+  const char *tokenp = NULL;
+  const char *hexfmt = NULL;
+  const char *base64 = NULL;
+  const char *disphint = NULL;
+  const char *percent = NULL;
+  int hexcount = 0;
+  int quoted_esc = 0;
+  int datalen = 0;
+  size_t dummy_erroff;
+  struct make_space_ctx c;
+  int arg_counter = 0;
+  int level = 0;
+
+  /* FIXME: invent better error codes (?).  */
+
+  if (! erroff)
+    erroff = &dummy_erroff;
+
+  /* Depending on wether ARG_LIST is non-zero or not, this macro gives
+     us the next argument, either from the variable argument list as
+     specified by ARG_PTR or from the argument array ARG_LIST.  */
+#define ARG_NEXT(storage, type)                          \
+  do                                                     \
+    {                                                    \
+      if (! arg_list)                                    \
+	storage = va_arg (arg_ptr, type);                \
+      else                                               \
+	storage = *((type *) (arg_list[arg_counter++])); \
+    }                                                    \
+  while (0)
+
+#define MAKE_SPACE(n)  do { make_space ( &c, (n) ); } while (0)
+#define STORE_LEN(p,n) do {						   \
+			    DATALEN ashort = (n);			   \
+			    memcpy ( (p), &ashort, sizeof(ashort) );	   \
+			    (p) += sizeof (ashort);			   \
+			} while (0)
+
+  /* We assume that the internal representation takes less memory
+   * than the provided one.  However, we add space for one extra datalen
+   * so that the code which does the ST_CLOSE can use MAKE_SPACE */
+  c.allocated = length + sizeof(DATALEN);
+  if (buffer && length && gcry_is_secure (buffer))
+    c.sexp = gcry_xmalloc_secure (sizeof *c.sexp + c.allocated - 1);
+  else
+    c.sexp = gcry_xmalloc (sizeof *c.sexp + c.allocated - 1);
+  c.pos = c.sexp->d;
+
+  for (p = buffer, n = length; n; p++, n--)
+    {
+      if (tokenp && (! hexfmt))
+	{
+	  if (strchr (tokenchars, *p))
+	    continue;
+	  else
+	    {
+	      datalen = p - tokenp;
+	      MAKE_SPACE (datalen);
+	      *c.pos++ = ST_DATA;
+	      STORE_LEN (c.pos, datalen);
+	      memcpy (c.pos, tokenp, datalen);
+	      c.pos += datalen;
+	      tokenp = NULL;
+	    }
+	}
+
+      if (quoted)
+	{
+	  if (quoted_esc)
+	    {
+	      switch (*p)
+		{
+		case 'b': case 't': case 'v': case 'n': case 'f':
+		case 'r': case '"': case '\'': case '\\':
+		  quoted_esc = 0;
+		  break;
+
+		case '0': case '1': case '2': case '3': case '4':
+		case '5': case '6': case '7':
+		  if (! ((n > 2)
+			 && (p[1] >= '0') && (p[1] <= '7')
+			 && (p[2] >= '0') && (p[2] <= '7')))
+		    {
+		      *erroff = p - buffer;
+		      /* Invalid octal value.  */
+		      err = GPG_ERR_SEXP_BAD_QUOTATION;
+		      //return gcry_error (GPG_ERR_SEXP_BAD_QUOTATION);
+		    }
+		  p += 2;
+		  n -= 2;
+		  quoted_esc = 0;
+		  break;
+		  
+		case 'x':
+		  if (! ((n > 2) && isxdigit(p[1]) && isxdigit(p[2])))
+		    {
+		      *erroff = p - buffer;
+		      /* Invalid hex value.  */
+		      err = GPG_ERR_SEXP_BAD_QUOTATION;
+		      //return gcry_error (GPG_ERR_SEXP_BAD_QUOTATION);
+		    }
+		  p += 2;
+		  n -= 2;
+		  quoted_esc = 0;
+		  break;
+
+		case '\r':
+		  /* ignore CR[,LF] */
+		  if (n && (p[1] == '\n'))
+		    {
+		      p++;
+		      n--;
+		    }
+		  quoted_esc = 0;
+		  break;
+
+		case '\n':
+		  /* ignore LF[,CR] */
+		  if (n && (p[1] == '\r'))
+		    {
+		      p++;
+		      n--;
+		    }
+		  quoted_esc = 0;
+		  break;
+
+		default:
+		  *erroff = p - buffer;
+		  /* Invalid quoted string escape.  */
+		  err = GPG_ERR_SEXP_BAD_QUOTATION;
+		}
+	    }
+	  else if (*p == '\\')
+	    quoted_esc = 1;
+	  else if (*p == '\"')
+	    {
+	      /* Keep it easy - we know that the unquoted string will
+		 never be larger. */
+	      char *save;
+	      size_t len;
+	      
+	      quoted++; /* Skip leading quote.  */
+	      MAKE_SPACE (p - quoted);
+	      *c.pos++ = ST_DATA;
+	      save = c.pos;
+	      STORE_LEN (c.pos, 0); /* Will be fixed up later.  */
+	      len = unquote_string (quoted, p - quoted, c.pos);
+	      c.pos += len;
+	      STORE_LEN (save, len);
+	      quoted = NULL;
+	    }
+	}
+      else if (hexfmt)
+	{
+	  if (isxdigit (*p))
+	    hexcount++;
+	  else if (*p == '#')
+	    {
+	      if ((hexcount & 1))
+		{
+		  *erroff = p - buffer;
+		  err = GPG_ERR_SEXP_ODD_HEX_NUMBERS;
+		}
+
+	      datalen = hexcount / 2;
+	      MAKE_SPACE (datalen);
+	      *c.pos++ = ST_DATA;
+	      STORE_LEN (c.pos, datalen);
+	      for (hexfmt++; hexfmt < p; hexfmt++)
+		{
+		  if (isspace (*hexfmt))
+		    continue;
+		  *c.pos++ = hextobyte (hexfmt);
+		  hexfmt++;
+		}
+	      hexfmt = NULL;
+	    }
+	  else if (! isspace (*p))
+	    {
+	      *erroff = p - buffer;
+	      err = GPG_ERR_SEXP_BAD_HEX_CHAR;
+	    }
+	}
+      else if (base64)
+	{
+	  if (*p == '|')
+	    base64 = NULL;
+	}
+      else if (digptr)
+	{
+	  if (isdigit (*p))
+	    ;
+	  else if (*p == ':')
+	    {
+	      datalen = atoi (digptr); /* FIXME: check for overflow.  */
+	      digptr = NULL;
+	      if (datalen > n - 1)
+		{
+		  *erroff = p - buffer;
+		  /* Buffer too short.  */
+		  err = GPG_ERR_SEXP_STRING_TOO_LONG;
+		}
+	      /* Make a new list entry.  */
+	      MAKE_SPACE (datalen);
+	      *c.pos++ = ST_DATA;
+	      STORE_LEN (c.pos, datalen);
+	      memcpy (c.pos, p + 1, datalen);
+	      c.pos += datalen;
+	      n -= datalen;
+	      p += datalen;
+	    }
+	  else if (*p == '\"')
+	    {
+	      digptr = NULL; /* We ignore the optional length.  */
+	      quoted = p;
+	      quoted_esc = 0;
+	    }
+	  else if (*p == '#')
+	    {
+	      digptr = NULL; /* We ignore the optional length.  */
+	      hexfmt = p;
+	      hexcount = 0;
+	    }
+	  else if (*p == '|')
+	    {
+	      digptr = NULL; /* We ignore the optional length.  */
+	      base64 = p;
+	    }
+	  else
+	    {
+	      *erroff = p - buffer;
+	      err = GPG_ERR_SEXP_INV_LEN_SPEC;
+	    }
+	}
+      else if (percent)
+	{
+	  if (*p == 'm')
+	    {
+	      /* Insert an MPI.  */
+	      gcry_mpi_t m;
+	      size_t nm = 0;
+
+	      ARG_NEXT (m, gcry_mpi_t);
+	      
+	      if (gcry_mpi_print (GCRYMPI_FMT_STD, NULL, 0, &nm, m))
+		BUG ();
+
+	      MAKE_SPACE (nm);
+	      if ((! gcry_is_secure (c.sexp->d))
+		  &&  gcry_mpi_get_flag ( m, GCRYMPI_FLAG_SECURE))
+		{
+		  /* We have to switch to secure allocation.  */
+		  gcry_sexp_t newsexp;
+		  byte *newhead;
+
+		  newsexp = gcry_xmalloc_secure (sizeof *newsexp
+						 + c.allocated - 1);
+		  newhead = newsexp->d;
+		  memcpy (newhead, c.sexp->d, (c.pos - c.sexp->d));
+		  c.pos = newhead + (c.pos - c.sexp->d);
+		  gcry_free (c.sexp);
+		  c.sexp = newsexp;
+		}
+
+	      *c.pos++ = ST_DATA;
+	      STORE_LEN (c.pos, nm);
+	      if (gcry_mpi_print (GCRYMPI_FMT_STD, c.pos, nm, &nm, m))
+		BUG ();
+	      c.pos += nm;
+	    }
+	  else if (*p == 's')
+	    {
+	      /* Insert an string.  */
+	      const char *astr;
+	      size_t alen;
+
+	      ARG_NEXT (astr, const char *);
+	      alen = strlen (astr);
+	      
+	      MAKE_SPACE (alen);
+	      *c.pos++ = ST_DATA;
+	      STORE_LEN (c.pos, alen);
+	      memcpy (c.pos, astr, alen);
+	      c.pos += alen;
+	    }
+	  else if (*p == 'b')
+	    {
+	      /* Insert a memory buffer.  */
+	      const char *astr;
+	      int alen;
+
+	      ARG_NEXT (alen, int);
+	      ARG_NEXT (astr, const char *);
+	      
+	      MAKE_SPACE (alen);
+	      if (alen
+                  && !gcry_is_secure (c.sexp->d)
+		  && gcry_is_secure (astr))
+              {
+		  /* We have to switch to secure allocation.  */
+		  gcry_sexp_t newsexp;
+		  byte *newhead;
+
+		  newsexp = gcry_xmalloc_secure (sizeof *newsexp
+ 						 + c.allocated - 1);
+		  newhead = newsexp->d;
+		  memcpy (newhead, c.sexp->d, (c.pos - c.sexp->d));
+		  c.pos = newhead + (c.pos - c.sexp->d);
+		  gcry_free (c.sexp);
+		  c.sexp = newsexp;
+		}
+
+	      *c.pos++ = ST_DATA;
+	      STORE_LEN (c.pos, alen);
+	      memcpy (c.pos, astr, alen);
+	      c.pos += alen;
+	    }
+	  else if (*p == 'd')
+	    {
+	      /* Insert an integer as string.  */
+	      int aint;
+	      size_t alen;
+	      char buf[20];
+	      
+	      ARG_NEXT (aint, int);
+	      sprintf (buf, "%d", aint);
+	      alen = strlen (buf);
+	      MAKE_SPACE (alen);
+	      *c.pos++ = ST_DATA;
+	      STORE_LEN (c.pos, alen);
+	      memcpy (c.pos, buf, alen);
+	      c.pos += alen;
+	    }
+	  else
+	    {
+	      *erroff = p - buffer;
+	      /* Invalid format specifier.  */
+	      err = GPG_ERR_SEXP_INV_LEN_SPEC;
+	    }
+	  percent = NULL;
+	}
+      else if (*p == '(')
+	{
+	  if (disphint)
+	    {
+	      *erroff = p - buffer;
+	      /* Open display hint.  */
+	      err = GPG_ERR_SEXP_UNMATCHED_DH;
+	    }
+	  MAKE_SPACE (0);
+	  *c.pos++ = ST_OPEN;
+	  level++;
+	}
+      else if (*p == ')')
+	{
+	  /* Walk up.  */
+	  if (disphint)
+	    {
+	      *erroff = p - buffer;
+	      /* Open display hint.  */
+	      err = GPG_ERR_SEXP_UNMATCHED_DH;
+	    }
+	  MAKE_SPACE (0);
+	  *c.pos++ = ST_CLOSE;
+	  level--;
+	}
+      else if (*p == '\"')
+	{
+	  quoted = p;
+	  quoted_esc = 0;
+	}
+      else if (*p == '#')
+	{
+	  hexfmt = p;
+	  hexcount = 0;
+	}
+      else if (*p == '|')
+	base64 = p;
+      else if (*p == '[')
+	{
+	  if (disphint)
+	    {
+	      *erroff = p - buffer;
+	      /* Open display hint.  */
+	      err = GPG_ERR_SEXP_NESTED_DH;
+	    }
+	  disphint = p;
+	}
+      else if (*p == ']')
+	{
+	  if (! disphint)
+	    {
+	      *erroff = p - buffer;
+	      /* Open display hint.  */
+	      err = GPG_ERR_SEXP_UNMATCHED_DH;
+	    }
+	  disphint = NULL;
+	}
+      else if (isdigit (*p))
+	{
+	  if (*p == '0')
+	    {
+	      /* A length may not begin with zero.  */
+	      *erroff = p - buffer;
+	      err = GPG_ERR_SEXP_ZERO_PREFIX;
+	    }
+	  digptr = p;
+	}
+      else if (strchr (tokenchars, *p))
+	tokenp = p;
+      else if (isspace (*p))
+	;
+      else if (*p == '{')
+	{
+	  /* fixme: handle rescanning: we can do this by saving our
+	     current state and start over at p+1 -- Hmmm. At this
+	     point here we are in a well defined state, so we don't
+	     need to save it.  Great.  */
+	  *erroff = p - buffer;
+	  err = GPG_ERR_SEXP_UNEXPECTED_PUNC;
+	}
+      else if (strchr ("&\\", *p))
+	{
+	  /* Reserved punctuation.  */
+	  *erroff = p - buffer;
+	  err = GPG_ERR_SEXP_UNEXPECTED_PUNC;
+	}
+      else if (argflag && (*p == '%'))
+	percent = p;
+      else
+	{
+	  /* Bad or unavailable.  */
+	  *erroff = p - buffer;
+	  err = GPG_ERR_SEXP_BAD_CHARACTER;
+	}
+    }
+  MAKE_SPACE (0);
+  *c.pos++ = ST_STOP;
+
+  if (level)
+    err = GPG_ERR_SEXP_UNMATCHED_PAREN;
+
+  if (err)
+    {
+      /* Error -> deallocate.  */
+      if (c.sexp)
+        {
+          /* Extra paranoid wipe on error. */
+          if (gcry_is_secure (c.sexp))
+            wipememory (c.sexp, sizeof (struct gcry_sexp) + c.allocated - 1);
+          gcry_free (c.sexp);
+        }
+      /* This might be expected by existing code...  */
+      *retsexp = NULL;
+    }
+  else
+    *retsexp = normalize (c.sexp);
+
+  return gcry_error (err);
+#undef MAKE_SPACE
+#undef STORE_LEN
+}
+
+gcry_error_t
+gcry_sexp_build (gcry_sexp_t *retsexp, size_t *erroff, const char *format, ...)
+{
+  gcry_error_t rc;
+  va_list arg_ptr;
+  
+  va_start (arg_ptr, format);
+  rc = sexp_sscan (retsexp, erroff, format, strlen(format), 1,
+		   arg_ptr, NULL);
+  va_end (arg_ptr);
+  
+  return rc;
+}
+
+/* Like gcry_sexp_build, but uses an array instead of variable
+   function arguments.  */
+gcry_error_t
+gcry_sexp_build_array (gcry_sexp_t *retsexp, size_t *erroff,
+		       const char *format, void **arg_list)
+{
+  /* We don't need the va_list because it is controlled by the
+     following flag, however we have to pass it but can't initialize
+     it as there is no portable way to do so.  volatile is needed to
+     suppress the compiler warning */
+  volatile va_list dummy_arg_ptr;
+  
+  gcry_error_t rc;
+
+  rc = sexp_sscan (retsexp, erroff, format, strlen(format), 1,
+		   dummy_arg_ptr, arg_list);
+
+  return rc;
+}
+
+gcry_error_t
+gcry_sexp_sscan (gcry_sexp_t *retsexp, size_t *erroff,
+		 const char *buffer, size_t length)
+{
+  /* We don't need the va_list because it is controlled by the
+     following flag, however we have to pass it but can't initialize
+     it as there is no portable way to do so.  volatile is needed to
+     suppress the compiler warning */
+  volatile va_list dummy_arg_ptr;
+
+  return sexp_sscan (retsexp, erroff, buffer, length, 0,
+		     dummy_arg_ptr, NULL);
+}
+
+
+/* Figure out a suitable encoding for BUFFER of LENGTH.
+   Returns: 0 = Binary
+            1 = String possible
+            2 = Token possible
+*/
+static int
+suitable_encoding (const unsigned char *buffer, size_t length)
+{
+  const unsigned char *s;
+  int maybe_token = 1;
+
+  if (!length)
+    return 1;
+  
+  for (s=buffer; length; s++, length--)
+    {
+      if ( (*s < 0x20 || (*s >= 0x7f && *s <= 0xa0))
+           && !strchr ("\b\t\v\n\f\r\"\'\\", *s))
+        return 0; /*binary*/
+      if ( maybe_token
+           && !alphap (s) && !digitp (s)  && !strchr (TOKEN_SPECIALS, *s))
+        maybe_token = 0;
+    }
+  s = buffer;
+  if ( maybe_token && !digitp (s) )
+    return 2;
+  return 1;
+}
+
+
+static int
+convert_to_hex (const unsigned char *src, size_t len, unsigned char *dest)
+{
+  int i;
+
+  if (dest)
+    {
+      *dest++ = '#';
+      for (i=0; i < len; i++, dest += 2 )
+        sprintf (dest, "%02X", src[i]);
+      *dest++ = '#';
+    }
+  return len*2+2;
+}
+
+static int
+convert_to_string (const unsigned char *s, size_t len, unsigned char *dest)
+{
+  if (dest)
+    {
+      unsigned char *p = dest;
+      *p++ = '\"';
+      for (; len; len--, s++ )
+        {
+          switch (*s)
+            {
+            case '\b': *p++ = '\\'; *p++ = 'b';  break;
+            case '\t': *p++ = '\\'; *p++ = 't';  break;
+            case '\v': *p++ = '\\'; *p++ = 'v';  break;
+            case '\n': *p++ = '\\'; *p++ = 'n';  break;
+            case '\f': *p++ = '\\'; *p++ = 'f';  break;
+            case '\r': *p++ = '\\'; *p++ = 'r';  break;
+            case '\"': *p++ = '\\'; *p++ = '\"';  break;
+            case '\'': *p++ = '\\'; *p++ = '\'';  break;
+            case '\\': *p++ = '\\'; *p++ = '\\';  break;
+            default: 
+              if ( (*s < 0x20 || (*s >= 0x7f && *s <= 0xa0)))
+                {
+                  sprintf (p, "\\x%02x", *s); 
+                  p += 4;
+                }
+              else
+                *p++ = *s;
+            }
+        }
+      *p++ = '\"';
+      return p - dest;
+    }
+  else
+    {
+      int count = 2;
+      for (; len; len--, s++ )
+        {
+          switch (*s)
+            {
+            case '\b': 
+            case '\t': 
+            case '\v': 
+            case '\n': 
+            case '\f': 
+            case '\r': 
+            case '\"':
+            case '\'':
+            case '\\': count += 2; break;
+            default: 
+              if ( (*s < 0x20 || (*s >= 0x7f && *s <= 0xa0)))
+                count += 4;
+              else
+                count++;
+            }
+        }
+      return count;
+    }
+}
+
+
+
+static int
+convert_to_token (const unsigned char *src, size_t len, unsigned char *dest)
+{
+  if (dest)
+    memcpy (dest, src, len);
+  return len;
+}
+
+
+/****************
+ * Print SEXP to buffer using the MODE.  Returns the length of the
+ * SEXP in buffer or 0 if the buffer is too short (We have at least an
+ * empty list consisting of 2 bytes).  If a buffer of NULL is provided,
+ * the required length is returned.
+ */
+size_t
+gcry_sexp_sprint( const gcry_sexp_t list, int mode,
+					char *buffer, size_t maxlength )
+{
+  static byte empty[3] = { ST_OPEN, ST_CLOSE, ST_STOP };
+  const byte *s;
+  char *d;
+  DATALEN n;
+  char numbuf[20];
+  size_t len = 0;
+  int i, indent = 0;
+  
+  s = list? list->d : empty;
+  d = buffer;
+  while ( *s != ST_STOP )
+    {
+      switch ( *s )
+        {
+        case ST_OPEN:
+          s++;
+          if ( mode != GCRYSEXP_FMT_CANON )
+            {
+              if (indent)
+                len++; 
+              len += indent;
+            }
+          len++;
+          if ( buffer ) 
+            {
+              if ( len >= maxlength )
+                return 0;
+              if ( mode != GCRYSEXP_FMT_CANON )
+                {
+                  if (indent)
+                    *d++ = '\n'; 
+                  for (i=0; i < indent; i++)
+                    *d++ = ' ';
+                }
+              *d++ = '(';
+	    }
+          indent++;
+          break;
+        case ST_CLOSE:
+          s++;
+          len++;
+          if ( buffer ) 
+            {
+              if ( len >= maxlength )
+                return 0;
+              *d++ = ')';
+	    }
+          indent--;
+          if (*s != ST_OPEN && *s != ST_STOP && mode != GCRYSEXP_FMT_CANON)
+            {
+              len++;
+              len += indent;
+              if (buffer)
+                {
+                  if (len >= maxlength)
+                    return 0;
+                  *d++ = '\n';
+                  for (i=0; i < indent; i++)
+                    *d++ = ' ';
+                }
+            }
+          break;
+        case ST_DATA:
+          s++;
+          memcpy ( &n, s, sizeof n ); s += sizeof n;
+          if (mode == GCRYSEXP_FMT_ADVANCED)
+            {
+              int type;
+              size_t nn;
+
+              switch ( (type=suitable_encoding (s, n)))
+                {
+                case 1: nn = convert_to_string (s, n, NULL); break;
+                case 2: nn = convert_to_token (s, n, NULL); break;
+                default: nn = convert_to_hex (s, n, NULL); break;
+                }
+              len += nn;
+              if (buffer)
+                {
+                  if (len >= maxlength)
+                    return 0;
+                  switch (type)
+                    {
+                    case 1: convert_to_string (s, n, d); break;
+                    case 2: convert_to_token (s, n, d); break;
+                    default: convert_to_hex (s, n, d); break;
+                    }
+                  d += nn;
+                }
+              if (s[n] != ST_CLOSE)
+                {
+                  len++;
+                  if (buffer)
+                    {
+                      if (len >= maxlength)
+                        return 0;
+                      *d++ = ' ';
+                    }
+                }
+            }
+          else
+            {
+              sprintf (numbuf, "%u:", (unsigned int)n );
+              len += strlen (numbuf) + n;
+              if ( buffer ) 
+                {
+                  if ( len >= maxlength )
+		    return 0;
+                  d = stpcpy ( d, numbuf );
+                  memcpy ( d, s, n ); d += n;
+                }
+            }
+          s += n;
+          break;
+        default:
+          BUG ();
+	}
+    }
+  if ( mode != GCRYSEXP_FMT_CANON )
+    {
+      len++;
+      if (buffer)
+        {
+          if ( len >= maxlength )
+            return 0;
+          *d++ = '\n'; 
+        }
+    }
+  if (buffer) 
+    {
+      if ( len >= maxlength )
+        return 0;
+      *d++ = 0; /* for convenience we make a C string */
+    }
+  else
+    len++; /* we need one byte more for this */
+
+  return len;
+}
+
+
+/* Scan a cannocial encoded buffer with implicit length values and
+   return the actual length this S-expression uses.  For a valid S-Exp
+   it should never return 0.  If LENGTH is not zero, the maximum
+   length to scan is given - this can be used for syntax checks of
+   data passed from outside. errorcode and erroff may both be passed as
+   NULL.  */
+size_t
+gcry_sexp_canon_len (const unsigned char *buffer, size_t length, 
+                     size_t *erroff, gcry_error_t *errcode)
+{
+  const unsigned char *p;
+  const char *disphint=NULL;
+  unsigned int datalen = 0;
+  size_t dummy_erroff;
+  gcry_error_t dummy_errcode;
+  size_t count = 0;
+  int level = 0;
+
+  if (!erroff)
+    erroff = &dummy_erroff;
+  if (!errcode)
+    errcode = &dummy_errcode;
+
+  *errcode = gcry_error (GPG_ERR_NO_ERROR);
+  *erroff = 0;
+  if (!buffer)
+    return 0;
+  if (*buffer != '(')
+    {
+      *errcode = gcry_error (GPG_ERR_SEXP_NOT_CANONICAL);
+      return 0;
+    }
+
+  for (p=buffer; ; p++, count++ )
+    {
+      if (length && count >= length)
+        {
+          *erroff = count;
+          *errcode = gcry_error (GPG_ERR_SEXP_STRING_TOO_LONG);
+          return 0;
+        }
+      
+      if (datalen)
+        {
+          if (*p == ':')
+            {
+              if (length && (count+datalen) >= length)
+                {
+                  *erroff = count;
+                  *errcode = gcry_error (GPG_ERR_SEXP_STRING_TOO_LONG);
+                  return 0;
+                }
+              count += datalen;
+              p += datalen;
+              datalen = 0;
+	    }
+          else if (digitp(p))
+            datalen = datalen*10 + atoi_1(p);
+          else 
+            {
+              *erroff = count;
+              *errcode = gcry_error (GPG_ERR_SEXP_INV_LEN_SPEC);
+              return 0;
+	    }
+	}
+      else if (*p == '(')
+        {
+          if (disphint)
+            {
+              *erroff = count;
+              *errcode = gcry_error (GPG_ERR_SEXP_UNMATCHED_DH);
+              return 0;
+	    }
+          level++;
+	}
+      else if (*p == ')')
+        { /* walk up */
+          if (!level)
+            {
+              *erroff = count;
+              *errcode = gcry_error (GPG_ERR_SEXP_UNMATCHED_PAREN);
+              return 0;
+	    }
+          if (disphint)
+            {
+              *erroff = count;
+              *errcode = gcry_error (GPG_ERR_SEXP_UNMATCHED_DH);
+              return 0;
+	    }
+          if (!--level)
+            return ++count; /* ready */
+	}
+      else if (*p == '[')
+        {
+          if (disphint) 
+            {
+              *erroff = count;
+              *errcode = gcry_error (GPG_ERR_SEXP_NESTED_DH);
+              return 0;
+            }
+          disphint = p;
+	}
+      else if (*p == ']')
+        {
+          if( !disphint ) 
+            {
+              *erroff = count;
+              *errcode = gcry_error (GPG_ERR_SEXP_UNMATCHED_DH);
+              return 0;
+	    }
+          disphint = NULL;
+	}
+      else if (digitp (p) )
+        {
+          if (*p == '0')
+            { 
+              *erroff = count;
+              *errcode = gcry_error (GPG_ERR_SEXP_ZERO_PREFIX);
+              return 0;
+	    }
+          datalen = atoi_1 (p);
+	}
+      else if (*p == '&' || *p == '\\')
+        {
+          *erroff = count;
+          *errcode = gcry_error (GPG_ERR_SEXP_UNEXPECTED_PUNC);
+          return 0;
+	}
+      else
+        { 
+          *erroff = count;
+          *errcode = gcry_error (GPG_ERR_SEXP_BAD_CHARACTER);
+          return 0;
+	}
+    }
+}