freetype2/src/psaux/psobjs.c

/****************************************************************************
 *
 * psobjs.c
 *
 *   Auxiliary functions for PostScript fonts (body).
 *
 * Copyright (C) 1996-2022 by
 * David Turner, Robert Wilhelm, and Werner Lemberg.
 *
 * This file is part of the FreeType project, and may only be used,
 * modified, and distributed under the terms of the FreeType project
 * license, LICENSE.TXT.  By continuing to use, modify, or distribute
 * this file you indicate that you have read the license and
 * understand and accept it fully.
 *
 */


#include <freetype/internal/psaux.h>
#include <freetype/internal/ftdebug.h>
#include <freetype/internal/ftcalc.h>
#include <freetype/ftdriver.h>

#include "psobjs.h"
#include "psconv.h"

#include "psauxerr.h"
#include "psauxmod.h"


  /**************************************************************************
   *
   * The macro FT_COMPONENT is used in trace mode.  It is an implicit
   * parameter of the FT_TRACE() and FT_ERROR() macros, used to print/log
   * messages during execution.
   */
#undef  FT_COMPONENT
#define FT_COMPONENT  psobjs


  /*************************************************************************/
  /*************************************************************************/
  /*****                                                               *****/
  /*****                             PS_TABLE                          *****/
  /*****                                                               *****/
  /*************************************************************************/
  /*************************************************************************/

  /**************************************************************************
   *
   * @Function:
   *   ps_table_new
   *
   * @Description:
   *   Initializes a PS_Table.
   *
   * @InOut:
   *   table ::
   *     The address of the target table.
   *
   * @Input:
   *   count ::
   *     The table size = the maximum number of elements.
   *
   *   memory ::
   *     The memory object to use for all subsequent
   *     reallocations.
   *
   * @Return:
   *   FreeType error code.  0 means success.
   */
  FT_LOCAL_DEF( FT_Error )
  ps_table_new( PS_Table   table,
                FT_Int     count,
                FT_Memory  memory )
  {
    FT_Error  error;


    table->memory = memory;
    if ( FT_NEW_ARRAY( table->elements, count ) ||
         FT_NEW_ARRAY( table->lengths,  count ) )
      goto Exit;

    table->max_elems = count;
    table->init      = 0xDEADBEEFUL;
    table->num_elems = 0;
    table->block     = NULL;
    table->capacity  = 0;
    table->cursor    = 0;

    *(PS_Table_FuncsRec*)&table->funcs = ps_table_funcs;

  Exit:
    if ( error )
      FT_FREE( table->elements );

    return error;
  }


  static FT_Error
  ps_table_realloc( PS_Table   table,
                    FT_Offset  new_size )
  {
    FT_Memory  memory   = table->memory;
    FT_Byte*   old_base = table->block;
    FT_Error   error;


    /* (re)allocate the base block */
    if ( FT_REALLOC( table->block, table->capacity, new_size ) )
      return error;

    /* rebase offsets if necessary */
    if ( old_base && table->block != old_base )
    {
      FT_Byte**   offset = table->elements;
      FT_Byte**   limit  = offset + table->max_elems;


      for ( ; offset < limit; offset++ )
      {
        if ( *offset )
          *offset = table->block + ( *offset - old_base );
      }
    }

    table->capacity = new_size;

    return FT_Err_Ok;
  }


  /**************************************************************************
   *
   * @Function:
   *   ps_table_add
   *
   * @Description:
   *   Adds an object to a PS_Table, possibly growing its memory block.
   *
   * @InOut:
   *   table ::
   *     The target table.
   *
   * @Input:
   *   idx ::
   *     The index of the object in the table.
   *
   *   object ::
   *     The address of the object to copy in memory.
   *
   *   length ::
   *     The length in bytes of the source object.
   *
   * @Return:
   *   FreeType error code.  0 means success.  An error is returned if a
   *   reallocation fails.
   */
  FT_LOCAL_DEF( FT_Error )
  ps_table_add( PS_Table     table,
                FT_Int       idx,
                const void*  object,
                FT_UInt      length )
  {
    if ( idx < 0 || idx >= table->max_elems )
    {
      FT_ERROR(( "ps_table_add: invalid index\n" ));
      return FT_THROW( Invalid_Argument );
    }

    /* grow the base block if needed */
    if ( table->cursor + length > table->capacity )
    {
      FT_Error    error;
      FT_Offset   new_size = table->capacity;
      FT_PtrDist  in_offset;


      in_offset = (FT_Byte*)object - table->block;
      if ( in_offset < 0 || (FT_Offset)in_offset >= table->capacity )
        in_offset = -1;

      while ( new_size < table->cursor + length )
      {
        /* increase size by 25% and round up to the nearest multiple
           of 1024 */
        new_size += ( new_size >> 2 ) + 1;
        new_size  = FT_PAD_CEIL( new_size, 1024 );
      }

      error = ps_table_realloc( table, new_size );
      if ( error )
        return error;

      if ( in_offset >= 0 )
        object = table->block + in_offset;
    }

    /* add the object to the base block and adjust offset */
    table->elements[idx] = FT_OFFSET( table->block, table->cursor );
    table->lengths [idx] = length;
    FT_MEM_COPY( table->block + table->cursor, object, length );

    table->cursor += length;
    return FT_Err_Ok;
  }


  /**************************************************************************
   *
   * @Function:
   *   ps_table_done
   *
   * @Description:
   *   Finalizes a PS_TableRec (i.e., reallocate it to its current
   *   cursor).
   *
   * @InOut:
   *   table ::
   *     The target table.
   */
  FT_LOCAL_DEF( void )
  ps_table_done( PS_Table  table )
  {
    /* no problem if shrinking fails */
    ps_table_realloc( table, table->cursor );
  }


  FT_LOCAL_DEF( void )
  ps_table_release( PS_Table  table )
  {
    FT_Memory  memory = table->memory;


    if ( (FT_ULong)table->init == 0xDEADBEEFUL )
    {
      FT_FREE( table->block );
      FT_FREE( table->elements );
      FT_FREE( table->lengths );
      table->init = 0;
    }
  }


  /*************************************************************************/
  /*************************************************************************/
  /*****                                                               *****/
  /*****                            T1 PARSER                          *****/
  /*****                                                               *****/
  /*************************************************************************/
  /*************************************************************************/


  /* first character must be already part of the comment */

  static void
  skip_comment( FT_Byte*  *acur,
                FT_Byte*   limit )
  {
    FT_Byte*  cur = *acur;


    while ( cur < limit )
    {
      if ( IS_PS_NEWLINE( *cur ) )
        break;
      cur++;
    }

    *acur = cur;
  }


  static void
  skip_spaces( FT_Byte*  *acur,
               FT_Byte*   limit )
  {
    FT_Byte*  cur = *acur;


    while ( cur < limit )
    {
      if ( !IS_PS_SPACE( *cur ) )
      {
        if ( *cur == '%' )
          /* According to the PLRM, a comment is equal to a space. */
          skip_comment( &cur, limit );
        else
          break;
      }
      cur++;
    }

    *acur = cur;
  }


#define IS_OCTAL_DIGIT( c ) ( '0' <= (c) && (c) <= '7' )


  /* first character must be `(';                               */
  /* *acur is positioned at the character after the closing `)' */

  static FT_Error
  skip_literal_string( FT_Byte*  *acur,
                       FT_Byte*   limit )
  {
    FT_Byte*      cur   = *acur;
    FT_Int        embed = 0;
    FT_Error      error = FT_ERR( Invalid_File_Format );
    unsigned int  i;


    while ( cur < limit )
    {
      FT_Byte  c = *cur;


      cur++;

      if ( c == '\\' )
      {
        /* Red Book 3rd ed., section `Literal Text Strings', p. 29:     */
        /* A backslash can introduce three different types              */
        /* of escape sequences:                                         */
        /*   - a special escaped char like \r, \n, etc.                 */
        /*   - a one-, two-, or three-digit octal number                */
        /*   - none of the above in which case the backslash is ignored */

        if ( cur == limit )
          /* error (or to be ignored?) */
          break;

        switch ( *cur )
        {
          /* skip `special' escape */
        case 'n':
        case 'r':
        case 't':
        case 'b':
        case 'f':
        case '\\':
        case '(':
        case ')':
          cur++;
          break;

        default:
          /* skip octal escape or ignore backslash */
          for ( i = 0; i < 3 && cur < limit; i++ )
          {
            if ( !IS_OCTAL_DIGIT( *cur ) )
              break;

            cur++;
          }
        }
      }
      else if ( c == '(' )
        embed++;
      else if ( c == ')' )
      {
        embed--;
        if ( embed == 0 )
        {
          error = FT_Err_Ok;
          break;
        }
      }
    }

    *acur = cur;

    return error;
  }


  /* first character must be `<' */

  static FT_Error
  skip_string( FT_Byte*  *acur,
               FT_Byte*   limit )
  {
    FT_Byte*  cur = *acur;
    FT_Error  err =  FT_Err_Ok;


    while ( ++cur < limit )
    {
      /* All whitespace characters are ignored. */
      skip_spaces( &cur, limit );
      if ( cur >= limit )
        break;

      if ( !IS_PS_XDIGIT( *cur ) )
        break;
    }

    if ( cur < limit && *cur != '>' )
    {
      FT_ERROR(( "skip_string: missing closing delimiter `>'\n" ));
      err = FT_THROW( Invalid_File_Format );
    }
    else
      cur++;

    *acur = cur;
    return err;
  }


  /* first character must be the opening brace that */
  /* starts the procedure                           */

  /* NB: [ and ] need not match:                    */
  /* `/foo {[} def' is a valid PostScript fragment, */
  /* even within a Type1 font                       */

  static FT_Error
  skip_procedure( FT_Byte*  *acur,
                  FT_Byte*   limit )
  {
    FT_Byte*  cur;
    FT_Int    embed = 0;
    FT_Error  error = FT_Err_Ok;


    FT_ASSERT( **acur == '{' );

    for ( cur = *acur; cur < limit && error == FT_Err_Ok; cur++ )
    {
      switch ( *cur )
      {
      case '{':
        embed++;
        break;

      case '}':
        embed--;
        if ( embed == 0 )
        {
          cur++;
          goto end;
        }
        break;

      case '(':
        error = skip_literal_string( &cur, limit );
        break;

      case '<':
        error = skip_string( &cur, limit );
        break;

      case '%':
        skip_comment( &cur, limit );
        break;
      }
    }

  end:
    if ( embed != 0 )
      error = FT_THROW( Invalid_File_Format );

    *acur = cur;

    return error;
  }


  /************************************************************************
   *
   * All exported parsing routines handle leading whitespace and stop at
   * the first character which isn't part of the just handled token.
   *
   */


  FT_LOCAL_DEF( void )
  ps_parser_skip_PS_token( PS_Parser  parser )
  {
    /* Note: PostScript allows any non-delimiting, non-whitespace        */
    /*       character in a name (PS Ref Manual, 3rd ed, p31).           */
    /*       PostScript delimiters are (, ), <, >, [, ], {, }, /, and %. */

    FT_Byte*  cur   = parser->cursor;
    FT_Byte*  limit = parser->limit;
    FT_Error  error = FT_Err_Ok;


    skip_spaces( &cur, limit );             /* this also skips comments */
    if ( cur >= limit )
      goto Exit;

    /* self-delimiting, single-character tokens */
    if ( *cur == '[' || *cur == ']' )
    {
      cur++;
      goto Exit;
    }

    /* skip balanced expressions (procedures and strings) */

    if ( *cur == '{' )                              /* {...} */
    {
      error = skip_procedure( &cur, limit );
      goto Exit;
    }

    if ( *cur == '(' )                              /* (...) */
    {
      error = skip_literal_string( &cur, limit );
      goto Exit;
    }

    if ( *cur == '<' )                              /* <...> */
    {
      if ( cur + 1 < limit && *(cur + 1) == '<' )   /* << */
      {
        cur++;
        cur++;
      }
      else
        error = skip_string( &cur, limit );

      goto Exit;
    }

    if ( *cur == '>' )
    {
      cur++;
      if ( cur >= limit || *cur != '>' )             /* >> */
      {
        FT_ERROR(( "ps_parser_skip_PS_token:"
                   " unexpected closing delimiter `>'\n" ));
        error = FT_THROW( Invalid_File_Format );
        goto Exit;
      }
      cur++;
      goto Exit;
    }

    if ( *cur == '/' )
      cur++;

    /* anything else */
    while ( cur < limit )
    {
      /* *cur might be invalid (e.g., ')' or '}'), but this   */
      /* is handled by the test `cur == parser->cursor' below */
      if ( IS_PS_DELIM( *cur ) )
        break;

      cur++;
    }

  Exit:
    if ( cur < limit && cur == parser->cursor )
    {
      FT_ERROR(( "ps_parser_skip_PS_token:"
                 " current token is `%c' which is self-delimiting\n",
                 *cur ));
      FT_ERROR(( "                        "
                 " but invalid at this point\n" ));

      error = FT_THROW( Invalid_File_Format );
    }

    if ( cur > limit )
      cur = limit;

    parser->error  = error;
    parser->cursor = cur;
  }


  FT_LOCAL_DEF( void )
  ps_parser_skip_spaces( PS_Parser  parser )
  {
    skip_spaces( &parser->cursor, parser->limit );
  }


  /* `token' here means either something between balanced delimiters */
  /* or the next token; the delimiters are not removed.              */

  FT_LOCAL_DEF( void )
  ps_parser_to_token( PS_Parser  parser,
                      T1_Token   token )
  {
    FT_Byte*  cur;
    FT_Byte*  limit;
    FT_Int    embed;


    token->type  = T1_TOKEN_TYPE_NONE;
    token->start = NULL;
    token->limit = NULL;

    /* first of all, skip leading whitespace */
    ps_parser_skip_spaces( parser );

    cur   = parser->cursor;
    limit = parser->limit;

    if ( cur >= limit )
      return;

    switch ( *cur )
    {
      /************* check for literal string *****************/
    case '(':
      token->type  = T1_TOKEN_TYPE_STRING;
      token->start = cur;

      if ( skip_literal_string( &cur, limit ) == FT_Err_Ok )
        token->limit = cur;
      break;

      /************* check for programs/array *****************/
    case '{':
      token->type  = T1_TOKEN_TYPE_ARRAY;
      token->start = cur;

      if ( skip_procedure( &cur, limit ) == FT_Err_Ok )
        token->limit = cur;
      break;

      /************* check for table/array ********************/
      /* XXX: in theory we should also look for "<<"          */
      /*      since this is semantically equivalent to "[";   */
      /*      in practice it doesn't matter (?)               */
    case '[':
      token->type  = T1_TOKEN_TYPE_ARRAY;
      embed        = 1;
      token->start = cur++;

      /* we need this to catch `[ ]' */
      parser->cursor = cur;
      ps_parser_skip_spaces( parser );
      cur = parser->cursor;

      while ( cur < limit && !parser->error )
      {
        /* XXX: this is wrong because it does not      */
        /*      skip comments, procedures, and strings */
        if ( *cur == '[' )
          embed++;
        else if ( *cur == ']' )
        {
          embed--;
          if ( embed <= 0 )
          {
            token->limit = ++cur;
            break;
          }
        }

        parser->cursor = cur;
        ps_parser_skip_PS_token( parser );
        /* we need this to catch `[XXX ]' */
        ps_parser_skip_spaces  ( parser );
        cur = parser->cursor;
      }
      break;

      /* ************ otherwise, it is any token **************/
    default:
      token->start = cur;
      token->type  = ( *cur == '/' ) ? T1_TOKEN_TYPE_KEY : T1_TOKEN_TYPE_ANY;
      ps_parser_skip_PS_token( parser );
      cur = parser->cursor;
      if ( !parser->error )
        token->limit = cur;
    }

    if ( !token->limit )
    {
      token->start = NULL;
      token->type  = T1_TOKEN_TYPE_NONE;
    }

    parser->cursor = cur;
  }


  /* NB: `tokens' can be NULL if we only want to count */
  /* the number of array elements                      */

  FT_LOCAL_DEF( void )
  ps_parser_to_token_array( PS_Parser  parser,
                            T1_Token   tokens,
                            FT_UInt    max_tokens,
                            FT_Int*    pnum_tokens )
  {
    T1_TokenRec  master;


    *pnum_tokens = -1;

    /* this also handles leading whitespace */
    ps_parser_to_token( parser, &master );

    if ( master.type == T1_TOKEN_TYPE_ARRAY )
    {
      FT_Byte*  old_cursor = parser->cursor;
      FT_Byte*  old_limit  = parser->limit;
      T1_Token  cur        = tokens;
      T1_Token  limit      = cur + max_tokens;


      /* don't include outermost delimiters */
      parser->cursor = master.start + 1;
      parser->limit  = master.limit - 1;

      while ( parser->cursor < parser->limit )
      {
        T1_TokenRec  token;


        ps_parser_to_token( parser, &token );
        if ( !token.type )
          break;

        if ( tokens && cur < limit )
          *cur = token;

        cur++;
      }

      *pnum_tokens = (FT_Int)( cur - tokens );

      parser->cursor = old_cursor;
      parser->limit  = old_limit;
    }
  }


  /* first character must be a delimiter or a part of a number */
  /* NB: `coords' can be NULL if we just want to skip the      */
  /*     array; in this case we ignore `max_coords'            */

  static FT_Int
  ps_tocoordarray( FT_Byte*  *acur,
                   FT_Byte*   limit,
                   FT_Int     max_coords,
                   FT_Short*  coords )
  {
    FT_Byte*  cur   = *acur;
    FT_Int    count = 0;
    FT_Byte   c, ender;


    if ( cur >= limit )
      goto Exit;

    /* check for the beginning of an array; otherwise, only one number */
    /* will be read                                                    */
    c     = *cur;
    ender = 0;

    if ( c == '[' )
      ender = ']';
    else if ( c == '{' )
      ender = '}';

    if ( ender )
      cur++;

    /* now, read the coordinates */
    while ( cur < limit )
    {
      FT_Short  dummy;
      FT_Byte*  old_cur;


      /* skip whitespace in front of data */
      skip_spaces( &cur, limit );
      if ( cur >= limit )
        goto Exit;

      if ( *cur == ender )
      {
        cur++;
        break;
      }

      old_cur = cur;

      if ( coords && count >= max_coords )
        break;

      /* call PS_Conv_ToFixed() even if coords == NULL */
      /* to properly parse number at `cur'             */
      *( coords ? &coords[count] : &dummy ) =
        (FT_Short)( PS_Conv_ToFixed( &cur, limit, 0 ) >> 16 );

      if ( old_cur == cur )
      {
        count = -1;
        goto Exit;
      }
      else
        count++;

      if ( !ender )
        break;
    }

  Exit:
    *acur = cur;
    return count;
  }


  /* first character must be a delimiter or a part of a number */
  /* NB: `values' can be NULL if we just want to skip the      */
  /*     array; in this case we ignore `max_values'            */
  /*                                                           */
  /* return number of successfully parsed values               */

  static FT_Int
  ps_tofixedarray( FT_Byte*  *acur,
                   FT_Byte*   limit,
                   FT_Int     max_values,
                   FT_Fixed*  values,
                   FT_Int     power_ten )
  {
    FT_Byte*  cur   = *acur;
    FT_Int    count = 0;
    FT_Byte   c, ender;


    if ( cur >= limit )
      goto Exit;

    /* Check for the beginning of an array.  Otherwise, only one number */
    /* will be read.                                                    */
    c     = *cur;
    ender = 0;

    if ( c == '[' )
      ender = ']';
    else if ( c == '{' )
      ender = '}';

    if ( ender )
      cur++;

    /* now, read the values */
    while ( cur < limit )
    {
      FT_Fixed  dummy;
      FT_Byte*  old_cur;


      /* skip whitespace in front of data */
      skip_spaces( &cur, limit );
      if ( cur >= limit )
        goto Exit;

      if ( *cur == ender )
      {
        cur++;
        break;
      }

      old_cur = cur;

      if ( values && count >= max_values )
        break;

      /* call PS_Conv_ToFixed() even if coords == NULL */
      /* to properly parse number at `cur'             */
      *( values ? &values[count] : &dummy ) =
        PS_Conv_ToFixed( &cur, limit, power_ten );

      if ( old_cur == cur )
      {
        count = -1;
        goto Exit;
      }
      else
        count++;

      if ( !ender )
        break;
    }

  Exit:
    *acur = cur;
    return count;
  }


#if 0

  static FT_String*
  ps_tostring( FT_Byte**  cursor,
               FT_Byte*   limit,
               FT_Memory  memory )
  {
    FT_Byte*    cur = *cursor;
    FT_UInt     len = 0;
    FT_Int      count;
    FT_String*  result;
    FT_Error    error;


    /* XXX: some stupid fonts have a `Notice' or `Copyright' string     */
    /*      that simply doesn't begin with an opening parenthesis, even */
    /*      though they have a closing one!  E.g. "amuncial.pfb"        */
    /*                                                                  */
    /*      We must deal with these ill-fated cases there.  Note that   */
    /*      these fonts didn't work with the old Type 1 driver as the   */
    /*      notice/copyright was not recognized as a valid string token */
    /*      and made the old token parser commit errors.                */

    while ( cur < limit && ( *cur == ' ' || *cur == '\t' ) )
      cur++;
    if ( cur + 1 >= limit )
      return 0;

    if ( *cur == '(' )
      cur++;  /* skip the opening parenthesis, if there is one */

    *cursor = cur;
    count   = 0;

    /* then, count its length */
    for ( ; cur < limit; cur++ )
    {
      if ( *cur == '(' )
        count++;

      else if ( *cur == ')' )
      {
        count--;
        if ( count < 0 )
          break;
      }
    }

    len = (FT_UInt)( cur - *cursor );
    if ( cur >= limit || FT_QALLOC( result, len + 1 ) )
      return 0;

    /* now copy the string */
    FT_MEM_COPY( result, *cursor, len );
    result[len] = '\0';
    *cursor = cur;
    return result;
  }

#endif /* 0 */


  static int
  ps_tobool( FT_Byte*  *acur,
             FT_Byte*   limit )
  {
    FT_Byte*  cur    = *acur;
    FT_Bool   result = 0;


    /* return 1 if we find `true', 0 otherwise */
    if ( cur + 3 < limit &&
         cur[0] == 't'   &&
         cur[1] == 'r'   &&
         cur[2] == 'u'   &&
         cur[3] == 'e'   )
    {
      result = 1;
      cur   += 5;
    }
    else if ( cur + 4 < limit &&
              cur[0] == 'f'   &&
              cur[1] == 'a'   &&
              cur[2] == 'l'   &&
              cur[3] == 's'   &&
              cur[4] == 'e'   )
    {
      result = 0;
      cur   += 6;
    }

    *acur = cur;
    return result;
  }


  /* load a simple field (i.e. non-table) into the current list of objects */

  FT_LOCAL_DEF( FT_Error )
  ps_parser_load_field( PS_Parser       parser,
                        const T1_Field  field,
                        void**          objects,
                        FT_UInt         max_objects,
                        FT_ULong*       pflags )
  {
    T1_TokenRec   token;
    FT_Byte*      cur;
    FT_Byte*      limit;
    FT_UInt       count;
    FT_UInt       idx;
    FT_Error      error;
    T1_FieldType  type;


    /* this also skips leading whitespace */
    ps_parser_to_token( parser, &token );
    if ( !token.type )
      goto Fail;

    count = 1;
    idx   = 0;
    cur   = token.start;
    limit = token.limit;

    type = field->type;

    /* we must detect arrays in /FontBBox */
    if ( type == T1_FIELD_TYPE_BBOX )
    {
      T1_TokenRec  token2;
      FT_Byte*     old_cur   = parser->cursor;
      FT_Byte*     old_limit = parser->limit;


      /* don't include delimiters */
      parser->cursor = token.start + 1;
      parser->limit  = token.limit - 1;

      ps_parser_to_token( parser, &token2 );
      parser->cursor = old_cur;
      parser->limit  = old_limit;

      if ( token2.type == T1_TOKEN_TYPE_ARRAY )
      {
        type = T1_FIELD_TYPE_MM_BBOX;
        goto FieldArray;
      }
    }
    else if ( token.type == T1_TOKEN_TYPE_ARRAY )
    {
      count = max_objects;

    FieldArray:
      /* if this is an array and we have no blend, an error occurs */
      if ( max_objects == 0 )
        goto Fail;

      idx = 1;

      /* don't include delimiters */
      cur++;
      limit--;
    }

    for ( ; count > 0; count--, idx++ )
    {
      FT_Byte*    q      = (FT_Byte*)objects[idx] + field->offset;
      FT_Long     val;


      skip_spaces( &cur, limit );

      switch ( type )
      {
      case T1_FIELD_TYPE_BOOL:
        val = ps_tobool( &cur, limit );
        FT_TRACE4(( " %s", val ? "true" : "false" ));
        goto Store_Integer;

      case T1_FIELD_TYPE_FIXED:
        val = PS_Conv_ToFixed( &cur, limit, 0 );
        FT_TRACE4(( " %f", (double)val / 65536 ));
        goto Store_Integer;

      case T1_FIELD_TYPE_FIXED_1000:
        val = PS_Conv_ToFixed( &cur, limit, 3 );
        FT_TRACE4(( " %f", (double)val / 65536 / 1000 ));
        goto Store_Integer;

      case T1_FIELD_TYPE_INTEGER:
        val = PS_Conv_ToInt( &cur, limit );
        FT_TRACE4(( " %ld", val ));
        /* fall through */

      Store_Integer:
        switch ( field->size )
        {
        case (8 / FT_CHAR_BIT):
          *(FT_Byte*)q = (FT_Byte)val;
          break;

        case (16 / FT_CHAR_BIT):
          *(FT_UShort*)q = (FT_UShort)val;
          break;

        case (32 / FT_CHAR_BIT):
          *(FT_UInt32*)q = (FT_UInt32)val;
          break;

        default:                /* for 64-bit systems */
          *(FT_Long*)q = val;
        }
        break;

      case T1_FIELD_TYPE_STRING:
      case T1_FIELD_TYPE_KEY:
        {
          FT_Memory   memory = parser->memory;
          FT_UInt     len    = (FT_UInt)( limit - cur );
          FT_String*  string;


          if ( cur >= limit )
            break;

          /* we allow both a string or a name   */
          /* for cases like /FontName (foo) def */
          if ( token.type == T1_TOKEN_TYPE_KEY )
          {
            /* don't include leading `/' */
            len--;
            cur++;
          }
          else if ( token.type == T1_TOKEN_TYPE_STRING )
          {
            /* don't include delimiting parentheses    */
            /* XXX we don't handle <<...>> here        */
            /* XXX should we convert octal escapes?    */
            /*     if so, what encoding should we use? */
            cur++;
            len -= 2;
          }
          else
          {
            FT_ERROR(( "ps_parser_load_field:"
                       " expected a name or string\n" ));
            FT_ERROR(( "                     "
                       " but found token of type %d instead\n",
                       token.type ));
            error = FT_THROW( Invalid_File_Format );
            goto Exit;
          }

          /* for this to work (FT_String**)q must have been */
          /* initialized to NULL                            */
          if ( *(FT_String**)q )
          {
            FT_TRACE0(( "ps_parser_load_field: overwriting field %s\n",
                        field->ident ));
            FT_FREE( *(FT_String**)q );
          }

          if ( FT_QALLOC( string, len + 1 ) )
            goto Exit;

          FT_MEM_COPY( string, cur, len );
          string[len] = 0;

#ifdef FT_DEBUG_LEVEL_TRACE
          if ( token.type == T1_TOKEN_TYPE_STRING )
            FT_TRACE4(( " (%s)", string ));
          else
            FT_TRACE4(( " /%s", string ));
#endif

          *(FT_String**)q = string;
        }
        break;

      case T1_FIELD_TYPE_BBOX:
        {
          FT_Fixed  temp[4];
          FT_BBox*  bbox = (FT_BBox*)q;
          FT_Int    result;


          result = ps_tofixedarray( &cur, limit, 4, temp, 0 );

          if ( result < 4 )
          {
            FT_ERROR(( "ps_parser_load_field:"
                       " expected four integers in bounding box\n" ));
            error = FT_THROW( Invalid_File_Format );
            goto Exit;
          }

          bbox->xMin = FT_RoundFix( temp[0] );
          bbox->yMin = FT_RoundFix( temp[1] );
          bbox->xMax = FT_RoundFix( temp[2] );
          bbox->yMax = FT_RoundFix( temp[3] );

          FT_TRACE4(( " [%ld %ld %ld %ld]",
                      bbox->xMin / 65536,
                      bbox->yMin / 65536,
                      bbox->xMax / 65536,
                      bbox->yMax / 65536 ));
        }
        break;

      case T1_FIELD_TYPE_MM_BBOX:
        {
          FT_Memory  memory = parser->memory;
          FT_Fixed*  temp   = NULL;
          FT_Int     result;
          FT_UInt    i;


          if ( FT_QNEW_ARRAY( temp, max_objects * 4 ) )
            goto Exit;

          for ( i = 0; i < 4; i++ )
          {
            result = ps_tofixedarray( &cur, limit, (FT_Int)max_objects,
                                      temp + i * max_objects, 0 );
            if ( result < 0 || (FT_UInt)result < max_objects )
            {
              FT_ERROR(( "ps_parser_load_field:"
                         " expected %d integer%s in the %s subarray\n",
                         max_objects, max_objects > 1 ? "s" : "",
                         i == 0 ? "first"
                                : ( i == 1 ? "second"
                                           : ( i == 2 ? "third"
                                                      : "fourth" ) ) ));
              FT_ERROR(( "                     "
                         " of /FontBBox in the /Blend dictionary\n" ));
              error = FT_THROW( Invalid_File_Format );

              FT_FREE( temp );
              goto Exit;
            }

            skip_spaces( &cur, limit );
          }

          FT_TRACE4(( " [" ));
          for ( i = 0; i < max_objects; i++ )
          {
            FT_BBox*  bbox = (FT_BBox*)objects[i];


            bbox->xMin = FT_RoundFix( temp[i                  ] );
            bbox->yMin = FT_RoundFix( temp[i +     max_objects] );
            bbox->xMax = FT_RoundFix( temp[i + 2 * max_objects] );
            bbox->yMax = FT_RoundFix( temp[i + 3 * max_objects] );

            FT_TRACE4(( " [%ld %ld %ld %ld]",
                        bbox->xMin / 65536,
                        bbox->yMin / 65536,
                        bbox->xMax / 65536,
                        bbox->yMax / 65536 ));
          }
          FT_TRACE4(( "]" ));

          FT_FREE( temp );
        }
        break;

      default:
        /* an error occurred */
        goto Fail;
      }
    }

#if 0  /* obsolete -- keep for reference */
    if ( pflags )
      *pflags |= 1L << field->flag_bit;
#else
    FT_UNUSED( pflags );
#endif

    error = FT_Err_Ok;

  Exit:
    return error;

  Fail:
    error = FT_THROW( Invalid_File_Format );
    goto Exit;
  }


#define T1_MAX_TABLE_ELEMENTS  32


  FT_LOCAL_DEF( FT_Error )
  ps_parser_load_field_table( PS_Parser       parser,
                              const T1_Field  field,
                              void**          objects,
                              FT_UInt         max_objects,
                              FT_ULong*       pflags )
  {
    T1_TokenRec  elements[T1_MAX_TABLE_ELEMENTS];
    T1_Token     token;
    FT_Int       num_elements;
    FT_Error     error = FT_Err_Ok;
    FT_Byte*     old_cursor;
    FT_Byte*     old_limit;
    T1_FieldRec  fieldrec = *(T1_Field)field;


    fieldrec.type = T1_FIELD_TYPE_INTEGER;
    if ( field->type == T1_FIELD_TYPE_FIXED_ARRAY ||
         field->type == T1_FIELD_TYPE_BBOX        )
      fieldrec.type = T1_FIELD_TYPE_FIXED;

    ps_parser_to_token_array( parser, elements,
                              T1_MAX_TABLE_ELEMENTS, &num_elements );
    if ( num_elements < 0 )
    {
      error = FT_ERR( Ignore );
      goto Exit;
    }
    if ( (FT_UInt)num_elements > field->array_max )
      num_elements = (FT_Int)field->array_max;

    old_cursor = parser->cursor;
    old_limit  = parser->limit;

    /* we store the elements count if necessary;           */
    /* we further assume that `count_offset' can't be zero */
    if ( field->type != T1_FIELD_TYPE_BBOX && field->count_offset != 0 )
      *(FT_Byte*)( (FT_Byte*)objects[0] + field->count_offset ) =
        (FT_Byte)num_elements;

    FT_TRACE4(( " [" ));

    /* we now load each element, adjusting the field.offset on each one */
    token = elements;
    for ( ; num_elements > 0; num_elements--, token++ )
    {
      parser->cursor = token->start;
      parser->limit  = token->limit;

      error = ps_parser_load_field( parser,
                                    &fieldrec,
                                    objects,
                                    max_objects,
                                    0 );
      if ( error )
        break;

      fieldrec.offset += fieldrec.size;
    }

    FT_TRACE4(( "]" ));

#if 0  /* obsolete -- keep for reference */
    if ( pflags )
      *pflags |= 1L << field->flag_bit;
#else
    FT_UNUSED( pflags );
#endif

    parser->cursor = old_cursor;
    parser->limit  = old_limit;

  Exit:
    return error;
  }


  FT_LOCAL_DEF( FT_Long )
  ps_parser_to_int( PS_Parser  parser )
  {
    ps_parser_skip_spaces( parser );
    return PS_Conv_ToInt( &parser->cursor, parser->limit );
  }


  /* first character must be `<' if `delimiters' is non-zero */

  FT_LOCAL_DEF( FT_Error )
  ps_parser_to_bytes( PS_Parser  parser,
                      FT_Byte*   bytes,
                      FT_Offset  max_bytes,
                      FT_ULong*  pnum_bytes,
                      FT_Bool    delimiters )
  {
    FT_Error  error = FT_Err_Ok;
    FT_Byte*  cur;


    ps_parser_skip_spaces( parser );
    cur = parser->cursor;

    if ( cur >= parser->limit )
      goto Exit;

    if ( delimiters )
    {
      if ( *cur != '<' )
      {
        FT_ERROR(( "ps_parser_to_bytes: Missing starting delimiter `<'\n" ));
        error = FT_THROW( Invalid_File_Format );
        goto Exit;
      }

      cur++;
    }

    *pnum_bytes = PS_Conv_ASCIIHexDecode( &cur,
                                          parser->limit,
                                          bytes,
                                          max_bytes );

    parser->cursor = cur;

    if ( delimiters )
    {
      if ( cur < parser->limit && *cur != '>' )
      {
        FT_ERROR(( "ps_parser_to_bytes: Missing closing delimiter `>'\n" ));
        error = FT_THROW( Invalid_File_Format );
        goto Exit;
      }

      parser->cursor++;
    }

  Exit:
    return error;
  }


  FT_LOCAL_DEF( FT_Fixed )
  ps_parser_to_fixed( PS_Parser  parser,
                      FT_Int     power_ten )
  {
    ps_parser_skip_spaces( parser );
    return PS_Conv_ToFixed( &parser->cursor, parser->limit, power_ten );
  }


  FT_LOCAL_DEF( FT_Int )
  ps_parser_to_coord_array( PS_Parser  parser,
                            FT_Int     max_coords,
                            FT_Short*  coords )
  {
    ps_parser_skip_spaces( parser );
    return ps_tocoordarray( &parser->cursor, parser->limit,
                            max_coords, coords );
  }


  FT_LOCAL_DEF( FT_Int )
  ps_parser_to_fixed_array( PS_Parser  parser,
                            FT_Int     max_values,
                            FT_Fixed*  values,
                            FT_Int     power_ten )
  {
    ps_parser_skip_spaces( parser );
    return ps_tofixedarray( &parser->cursor, parser->limit,
                            max_values, values, power_ten );
  }


#if 0

  FT_LOCAL_DEF( FT_String* )
  T1_ToString( PS_Parser  parser )
  {
    return ps_tostring( &parser->cursor, parser->limit, parser->memory );
  }


  FT_LOCAL_DEF( FT_Bool )
  T1_ToBool( PS_Parser  parser )
  {
    return ps_tobool( &parser->cursor, parser->limit );
  }

#endif /* 0 */


  FT_LOCAL_DEF( void )
  ps_parser_init( PS_Parser  parser,
                  FT_Byte*   base,
                  FT_Byte*   limit,
                  FT_Memory  memory )
  {
    parser->error  = FT_Err_Ok;
    parser->base   = base;
    parser->limit  = limit;
    parser->cursor = base;
    parser->memory = memory;
    parser->funcs  = ps_parser_funcs;
  }


  FT_LOCAL_DEF( void )
  ps_parser_done( PS_Parser  parser )
  {
    FT_UNUSED( parser );
  }


  /*************************************************************************/
  /*************************************************************************/
  /*****                                                               *****/
  /*****                            T1 BUILDER                         *****/
  /*****                                                               *****/
  /*************************************************************************/
  /*************************************************************************/

  /**************************************************************************
   *
   * @Function:
   *   t1_builder_init
   *
   * @Description:
   *   Initializes a given glyph builder.
   *
   * @InOut:
   *   builder ::
   *     A pointer to the glyph builder to initialize.
   *
   * @Input:
   *   face ::
   *     The current face object.
   *
   *   size ::
   *     The current size object.
   *
   *   glyph ::
   *     The current glyph object.
   *
   *   hinting ::
   *     Whether hinting should be applied.
   */
  FT_LOCAL_DEF( void )
  t1_builder_init( T1_Builder    builder,
                   FT_Face       face,
                   FT_Size       size,
                   FT_GlyphSlot  glyph,
                   FT_Bool       hinting )
  {
    builder->parse_state = T1_Parse_Start;
    builder->load_points = 1;

    builder->face   = face;
    builder->glyph  = glyph;
    builder->memory = face->memory;

    if ( glyph )
    {
      FT_GlyphLoader  loader = glyph->internal->loader;


      builder->loader  = loader;
      builder->base    = &loader->base.outline;
      builder->current = &loader->current.outline;
      FT_GlyphLoader_Rewind( loader );

      builder->hints_globals = size->internal->module_data;
      builder->hints_funcs   = NULL;

      if ( hinting )
        builder->hints_funcs = glyph->internal->glyph_hints;
    }

    builder->pos_x = 0;
    builder->pos_y = 0;

    builder->left_bearing.x = 0;
    builder->left_bearing.y = 0;
    builder->advance.x      = 0;
    builder->advance.y      = 0;

    builder->funcs = t1_builder_funcs;
  }


  /**************************************************************************
   *
   * @Function:
   *   t1_builder_done
   *
   * @Description:
   *   Finalizes a given glyph builder.  Its contents can still be used
   *   after the call, but the function saves important information
   *   within the corresponding glyph slot.
   *
   * @Input:
   *   builder ::
   *     A pointer to the glyph builder to finalize.
   */
  FT_LOCAL_DEF( void )
  t1_builder_done( T1_Builder  builder )
  {
    FT_GlyphSlot  glyph = builder->glyph;


    if ( glyph )
      glyph->outline = *builder->base;
  }


  /* check that there is enough space for `count' more points */
  FT_LOCAL_DEF( FT_Error )
  t1_builder_check_points( T1_Builder  builder,
                           FT_Int      count )
  {
    return FT_GLYPHLOADER_CHECK_POINTS( builder->loader, count, 0 );
  }


  /* add a new point, do not check space */
  FT_LOCAL_DEF( void )
  t1_builder_add_point( T1_Builder  builder,
                        FT_Pos      x,
                        FT_Pos      y,
                        FT_Byte     flag )
  {
    FT_Outline*  outline = builder->current;


    if ( builder->load_points )
    {
      FT_Vector*  point   = outline->points + outline->n_points;
      FT_Byte*    control = (FT_Byte*)outline->tags + outline->n_points;


      point->x = FIXED_TO_INT( x );
      point->y = FIXED_TO_INT( y );
      *control = (FT_Byte)( flag ? FT_CURVE_TAG_ON : FT_CURVE_TAG_CUBIC );
    }
    outline->n_points++;
  }


  /* check space for a new on-curve point, then add it */
  FT_LOCAL_DEF( FT_Error )
  t1_builder_add_point1( T1_Builder  builder,
                         FT_Pos      x,
                         FT_Pos      y )
  {
    FT_Error  error;


    error = t1_builder_check_points( builder, 1 );
    if ( !error )
      t1_builder_add_point( builder, x, y, 1 );

    return error;
  }


  /* check space for a new contour, then add it */
  FT_LOCAL_DEF( FT_Error )
  t1_builder_add_contour( T1_Builder  builder )
  {
    FT_Outline*  outline = builder->current;
    FT_Error     error;


    /* this might happen in invalid fonts */
    if ( !outline )
    {
      FT_ERROR(( "t1_builder_add_contour: no outline to add points to\n" ));
      return FT_THROW( Invalid_File_Format );
    }

    if ( !builder->load_points )
    {
      outline->n_contours++;
      return FT_Err_Ok;
    }

    error = FT_GLYPHLOADER_CHECK_POINTS( builder->loader, 0, 1 );
    if ( !error )
    {
      if ( outline->n_contours > 0 )
        outline->contours[outline->n_contours - 1] =
          (short)( outline->n_points - 1 );

      outline->n_contours++;
    }

    return error;
  }


  /* if a path was begun, add its first on-curve point */
  FT_LOCAL_DEF( FT_Error )
  t1_builder_start_point( T1_Builder  builder,
                          FT_Pos      x,
                          FT_Pos      y )
  {
    FT_Error  error = FT_ERR( Invalid_File_Format );


    /* test whether we are building a new contour */

    if ( builder->parse_state == T1_Parse_Have_Path )
      error = FT_Err_Ok;
    else
    {
      builder->parse_state = T1_Parse_Have_Path;
      error = t1_builder_add_contour( builder );
      if ( !error )
        error = t1_builder_add_point1( builder, x, y );
    }

    return error;
  }


  /* close the current contour */
  FT_LOCAL_DEF( void )
  t1_builder_close_contour( T1_Builder  builder )
  {
    FT_Outline*  outline = builder->current;
    FT_Int       first;


    if ( !outline )
      return;

    first = outline->n_contours <= 1
            ? 0 : outline->contours[outline->n_contours - 2] + 1;

    /* in malformed fonts it can happen that a contour was started */
    /* but no points were added                                    */
    if ( outline->n_contours && first == outline->n_points )
    {
      outline->n_contours--;
      return;
    }

    /* We must not include the last point in the path if it */
    /* is located on the first point.                       */
    if ( outline->n_points > 1 )
    {
      FT_Vector*  p1      = outline->points + first;
      FT_Vector*  p2      = outline->points + outline->n_points - 1;
      FT_Byte*    control = (FT_Byte*)outline->tags + outline->n_points - 1;


      /* `delete' last point only if it coincides with the first */
      /* point and it is not a control point (which can happen). */
      if ( p1->x == p2->x && p1->y == p2->y )
        if ( *control == FT_CURVE_TAG_ON )
          outline->n_points--;
    }

    if ( outline->n_contours > 0 )
    {
      /* Don't add contours only consisting of one point, i.e.,  */
      /* check whether the first and the last point is the same. */
      if ( first == outline->n_points - 1 )
      {
        outline->n_contours--;
        outline->n_points--;
      }
      else
        outline->contours[outline->n_contours - 1] =
          (short)( outline->n_points - 1 );
    }
  }


  /*************************************************************************/
  /*************************************************************************/
  /*****                                                               *****/
  /*****                           CFF BUILDER                         *****/
  /*****                                                               *****/
  /*************************************************************************/
  /*************************************************************************/


  /**************************************************************************
   *
   * @Function:
   *   cff_builder_init
   *
   * @Description:
   *   Initializes a given glyph builder.
   *
   * @InOut:
   *   builder ::
   *     A pointer to the glyph builder to initialize.
   *
   * @Input:
   *   face ::
   *     The current face object.
   *
   *   size ::
   *     The current size object.
   *
   *   glyph ::
   *     The current glyph object.
   *
   *   hinting ::
   *     Whether hinting is active.
   */
  FT_LOCAL_DEF( void )
  cff_builder_init( CFF_Builder*   builder,
                    TT_Face        face,
                    CFF_Size       size,
                    CFF_GlyphSlot  glyph,
                    FT_Bool        hinting )
  {
    builder->path_begun  = 0;
    builder->load_points = 1;

    builder->face   = face;
    builder->glyph  = glyph;
    builder->memory = face->root.memory;

    if ( glyph )
    {
      FT_GlyphLoader  loader = glyph->root.internal->loader;


      builder->loader  = loader;
      builder->base    = &loader->base.outline;
      builder->current = &loader->current.outline;
      FT_GlyphLoader_Rewind( loader );

      builder->hints_globals = NULL;
      builder->hints_funcs   = NULL;

      if ( hinting && size )
      {
        FT_Size       ftsize   = FT_SIZE( size );
        CFF_Internal  internal = (CFF_Internal)ftsize->internal->module_data;

        if ( internal )
        {
          builder->hints_globals = (void *)internal->topfont;
          builder->hints_funcs   = glyph->root.internal->glyph_hints;
        }
      }
    }

    builder->pos_x = 0;
    builder->pos_y = 0;

    builder->left_bearing.x = 0;
    builder->left_bearing.y = 0;
    builder->advance.x      = 0;
    builder->advance.y      = 0;

    builder->funcs = cff_builder_funcs;
  }


  /**************************************************************************
   *
   * @Function:
   *   cff_builder_done
   *
   * @Description:
   *   Finalizes a given glyph builder.  Its contents can still be used
   *   after the call, but the function saves important information
   *   within the corresponding glyph slot.
   *
   * @Input:
   *   builder ::
   *     A pointer to the glyph builder to finalize.
   */
  FT_LOCAL_DEF( void )
  cff_builder_done( CFF_Builder*  builder )
  {
    CFF_GlyphSlot  glyph = builder->glyph;


    if ( glyph )
      glyph->root.outline = *builder->base;
  }


  /* check that there is enough space for `count' more points */
  FT_LOCAL_DEF( FT_Error )
  cff_check_points( CFF_Builder*  builder,
                    FT_Int        count )
  {
    return FT_GLYPHLOADER_CHECK_POINTS( builder->loader, count, 0 );
  }


  /* add a new point, do not check space */
  FT_LOCAL_DEF( void )
  cff_builder_add_point( CFF_Builder*  builder,
                         FT_Pos        x,
                         FT_Pos        y,
                         FT_Byte       flag )
  {
    FT_Outline*  outline = builder->current;


    if ( builder->load_points )
    {
      FT_Vector*  point   = outline->points + outline->n_points;
      FT_Byte*    control = (FT_Byte*)outline->tags + outline->n_points;

#ifdef CFF_CONFIG_OPTION_OLD_ENGINE
      PS_Driver  driver   = (PS_Driver)FT_FACE_DRIVER( builder->face );


      if ( driver->hinting_engine == FT_HINTING_FREETYPE )
      {
        point->x = x >> 16;
        point->y = y >> 16;
      }
      else
#endif
      {
        /* cf2_decoder_parse_charstrings uses 16.16 coordinates */
        point->x = x >> 10;
        point->y = y >> 10;
      }
      *control = (FT_Byte)( flag ? FT_CURVE_TAG_ON : FT_CURVE_TAG_CUBIC );
    }

    outline->n_points++;
  }


  /* check space for a new on-curve point, then add it */
  FT_LOCAL_DEF( FT_Error )
  cff_builder_add_point1( CFF_Builder*  builder,
                          FT_Pos        x,
                          FT_Pos        y )
  {
    FT_Error  error;


    error = cff_check_points( builder, 1 );
    if ( !error )
      cff_builder_add_point( builder, x, y, 1 );

    return error;
  }


  /* check space for a new contour, then add it */
  FT_LOCAL_DEF( FT_Error )
  cff_builder_add_contour( CFF_Builder*  builder )
  {
    FT_Outline*  outline = builder->current;
    FT_Error     error;


    if ( !builder->load_points )
    {
      outline->n_contours++;
      return FT_Err_Ok;
    }

    error = FT_GLYPHLOADER_CHECK_POINTS( builder->loader, 0, 1 );
    if ( !error )
    {
      if ( outline->n_contours > 0 )
        outline->contours[outline->n_contours - 1] =
          (short)( outline->n_points - 1 );

      outline->n_contours++;
    }

    return error;
  }


  /* if a path was begun, add its first on-curve point */
  FT_LOCAL_DEF( FT_Error )
  cff_builder_start_point( CFF_Builder*  builder,
                           FT_Pos        x,
                           FT_Pos        y )
  {
    FT_Error  error = FT_Err_Ok;


    /* test whether we are building a new contour */
    if ( !builder->path_begun )
    {
      builder->path_begun = 1;
      error = cff_builder_add_contour( builder );
      if ( !error )
        error = cff_builder_add_point1( builder, x, y );
    }

    return error;
  }


  /* close the current contour */
  FT_LOCAL_DEF( void )
  cff_builder_close_contour( CFF_Builder*  builder )
  {
    FT_Outline*  outline = builder->current;
    FT_Int       first;


    if ( !outline )
      return;

    first = outline->n_contours <= 1
            ? 0 : outline->contours[outline->n_contours - 2] + 1;

    /* in malformed fonts it can happen that a contour was started */
    /* but no points were added                                    */
    if ( outline->n_contours && first == outline->n_points )
    {
      outline->n_contours--;
      return;
    }

    /* We must not include the last point in the path if it */
    /* is located on the first point.                       */
    if ( outline->n_points > 1 )
    {
      FT_Vector*  p1      = outline->points + first;
      FT_Vector*  p2      = outline->points + outline->n_points - 1;
      FT_Byte*    control = (FT_Byte*)outline->tags + outline->n_points - 1;


      /* `delete' last point only if it coincides with the first    */
      /* point and if it is not a control point (which can happen). */
      if ( p1->x == p2->x && p1->y == p2->y )
        if ( *control == FT_CURVE_TAG_ON )
          outline->n_points--;
    }

    if ( outline->n_contours > 0 )
    {
      /* Don't add contours only consisting of one point, i.e., */
      /* check whether begin point and last point are the same. */
      if ( first == outline->n_points - 1 )
      {
        outline->n_contours--;
        outline->n_points--;
      }
      else
        outline->contours[outline->n_contours - 1] =
          (short)( outline->n_points - 1 );
    }
  }


  /*************************************************************************/
  /*************************************************************************/
  /*****                                                               *****/
  /*****                            PS BUILDER                         *****/
  /*****                                                               *****/
  /*************************************************************************/
  /*************************************************************************/

  /**************************************************************************
   *
   * @Function:
   *   ps_builder_init
   *
   * @Description:
   *   Initializes a given glyph builder.
   *
   * @InOut:
   *   builder ::
   *     A pointer to the glyph builder to initialize.
   *
   * @Input:
   *   face ::
   *     The current face object.
   *
   *   size ::
   *     The current size object.
   *
   *   glyph ::
   *     The current glyph object.
   *
   *   hinting ::
   *     Whether hinting should be applied.
   */
  FT_LOCAL_DEF( void )
  ps_builder_init( PS_Builder*  ps_builder,
                   void*        builder,
                   FT_Bool      is_t1 )
  {
    FT_ZERO( ps_builder );

    if ( is_t1 )
    {
      T1_Builder  t1builder = (T1_Builder)builder;


      ps_builder->memory  = t1builder->memory;
      ps_builder->face    = (FT_Face)t1builder->face;
      ps_builder->glyph   = (CFF_GlyphSlot)t1builder->glyph;
      ps_builder->loader  = t1builder->loader;
      ps_builder->base    = t1builder->base;
      ps_builder->current = t1builder->current;

      ps_builder->pos_x = &t1builder->pos_x;
      ps_builder->pos_y = &t1builder->pos_y;

      ps_builder->left_bearing = &t1builder->left_bearing;
      ps_builder->advance      = &t1builder->advance;

      ps_builder->bbox        = &t1builder->bbox;
      ps_builder->path_begun  = 0;
      ps_builder->load_points = t1builder->load_points;
      ps_builder->no_recurse  = t1builder->no_recurse;

      ps_builder->metrics_only = t1builder->metrics_only;
    }
    else
    {
      CFF_Builder*  cffbuilder = (CFF_Builder*)builder;


      ps_builder->memory  = cffbuilder->memory;
      ps_builder->face    = (FT_Face)cffbuilder->face;
      ps_builder->glyph   = cffbuilder->glyph;
      ps_builder->loader  = cffbuilder->loader;
      ps_builder->base    = cffbuilder->base;
      ps_builder->current = cffbuilder->current;

      ps_builder->pos_x = &cffbuilder->pos_x;
      ps_builder->pos_y = &cffbuilder->pos_y;

      ps_builder->left_bearing = &cffbuilder->left_bearing;
      ps_builder->advance      = &cffbuilder->advance;

      ps_builder->bbox        = &cffbuilder->bbox;
      ps_builder->path_begun  = cffbuilder->path_begun;
      ps_builder->load_points = cffbuilder->load_points;
      ps_builder->no_recurse  = cffbuilder->no_recurse;

      ps_builder->metrics_only = cffbuilder->metrics_only;
    }

    ps_builder->is_t1 = is_t1;
    ps_builder->funcs = ps_builder_funcs;
  }


  /**************************************************************************
   *
   * @Function:
   *   ps_builder_done
   *
   * @Description:
   *   Finalizes a given glyph builder.  Its contents can still be used
   *   after the call, but the function saves important information
   *   within the corresponding glyph slot.
   *
   * @Input:
   *   builder ::
   *     A pointer to the glyph builder to finalize.
   */
  FT_LOCAL_DEF( void )
  ps_builder_done( PS_Builder*  builder )
  {
    CFF_GlyphSlot  glyph = builder->glyph;


    if ( glyph )
      glyph->root.outline = *builder->base;
  }


  /* check that there is enough space for `count' more points */
  FT_LOCAL_DEF( FT_Error )
  ps_builder_check_points( PS_Builder*  builder,
                           FT_Int       count )
  {
    return FT_GLYPHLOADER_CHECK_POINTS( builder->loader, count, 0 );
  }


  /* add a new point, do not check space */
  FT_LOCAL_DEF( void )
  ps_builder_add_point( PS_Builder*  builder,
                        FT_Pos       x,
                        FT_Pos       y,
                        FT_Byte      flag )
  {
    FT_Outline*  outline = builder->current;


    if ( builder->load_points )
    {
      FT_Vector*  point   = outline->points + outline->n_points;
      FT_Byte*    control = (FT_Byte*)outline->tags + outline->n_points;

#ifdef CFF_CONFIG_OPTION_OLD_ENGINE
      PS_Driver  driver   = (PS_Driver)FT_FACE_DRIVER( builder->face );


      if ( !builder->is_t1 &&
           driver->hinting_engine == FT_HINTING_FREETYPE )
      {
        point->x = x >> 16;
        point->y = y >> 16;
      }
      else
#endif
#ifdef T1_CONFIG_OPTION_OLD_ENGINE
#ifndef CFF_CONFIG_OPTION_OLD_ENGINE
      PS_Driver  driver   = (PS_Driver)FT_FACE_DRIVER( builder->face );
#endif
      if ( builder->is_t1 &&
           driver->hinting_engine == FT_HINTING_FREETYPE )
      {
        point->x = FIXED_TO_INT( x );
        point->y = FIXED_TO_INT( y );
      }
      else
#endif
      {
        /* cf2_decoder_parse_charstrings uses 16.16 coordinates */
        point->x = x >> 10;
        point->y = y >> 10;
      }
      *control = (FT_Byte)( flag ? FT_CURVE_TAG_ON : FT_CURVE_TAG_CUBIC );
    }
    outline->n_points++;
  }


  /* check space for a new on-curve point, then add it */
  FT_LOCAL_DEF( FT_Error )
  ps_builder_add_point1( PS_Builder*  builder,
                         FT_Pos       x,
                         FT_Pos       y )
  {
    FT_Error  error;


    error = ps_builder_check_points( builder, 1 );
    if ( !error )
      ps_builder_add_point( builder, x, y, 1 );

    return error;
  }


  /* check space for a new contour, then add it */
  FT_LOCAL_DEF( FT_Error )
  ps_builder_add_contour( PS_Builder*  builder )
  {
    FT_Outline*  outline = builder->current;
    FT_Error     error;


    /* this might happen in invalid fonts */
    if ( !outline )
    {
      FT_ERROR(( "ps_builder_add_contour: no outline to add points to\n" ));
      return FT_THROW( Invalid_File_Format );
    }

    if ( !builder->load_points )
    {
      outline->n_contours++;
      return FT_Err_Ok;
    }

    error = FT_GLYPHLOADER_CHECK_POINTS( builder->loader, 0, 1 );
    if ( !error )
    {
      if ( outline->n_contours > 0 )
        outline->contours[outline->n_contours - 1] =
          (short)( outline->n_points - 1 );

      outline->n_contours++;
    }

    return error;
  }


  /* if a path was begun, add its first on-curve point */
  FT_LOCAL_DEF( FT_Error )
  ps_builder_start_point( PS_Builder*  builder,
                          FT_Pos       x,
                          FT_Pos       y )
  {
    FT_Error  error = FT_Err_Ok;


    /* test whether we are building a new contour */
    if ( !builder->path_begun )
    {
      builder->path_begun = 1;
      error = ps_builder_add_contour( builder );
      if ( !error )
        error = ps_builder_add_point1( builder, x, y );
    }

    return error;
  }


  /* close the current contour */
  FT_LOCAL_DEF( void )
  ps_builder_close_contour( PS_Builder*  builder )
  {
    FT_Outline*  outline = builder->current;
    FT_Int       first;


    if ( !outline )
      return;

    first = outline->n_contours <= 1
            ? 0 : outline->contours[outline->n_contours - 2] + 1;

    /* in malformed fonts it can happen that a contour was started */
    /* but no points were added                                    */
    if ( outline->n_contours && first == outline->n_points )
    {
      outline->n_contours--;
      return;
    }

    /* We must not include the last point in the path if it */
    /* is located on the first point.                       */
    if ( outline->n_points > 1 )
    {
      FT_Vector*  p1      = outline->points + first;
      FT_Vector*  p2      = outline->points + outline->n_points - 1;
      FT_Byte*    control = (FT_Byte*)outline->tags + outline->n_points - 1;


      /* `delete' last point only if it coincides with the first */
      /* point and it is not a control point (which can happen). */
      if ( p1->x == p2->x && p1->y == p2->y )
        if ( *control == FT_CURVE_TAG_ON )
          outline->n_points--;
    }

    if ( outline->n_contours > 0 )
    {
      /* Don't add contours only consisting of one point, i.e.,  */
      /* check whether the first and the last point is the same. */
      if ( first == outline->n_points - 1 )
      {
        outline->n_contours--;
        outline->n_points--;
      }
      else
        outline->contours[outline->n_contours - 1] =
          (short)( outline->n_points - 1 );
    }
  }


  /*************************************************************************/
  /*************************************************************************/
  /*****                                                               *****/
  /*****                            OTHER                              *****/
  /*****                                                               *****/
  /*************************************************************************/
  /*************************************************************************/


  /**************************************************************************
   *
   * @Function:
   *   ps_decoder_init
   *
   * @Description:
   *   Creates a wrapper decoder for use in the combined
   *   Type 1 / CFF interpreter.
   *
   * @InOut:
   *   ps_decoder ::
   *     A pointer to the decoder to initialize.
   *
   * @Input:
   *   decoder ::
   *     A pointer to the original decoder.
   *
   *   is_t1 ::
   *     Flag indicating Type 1 or CFF
   */
  FT_LOCAL_DEF( void )
  ps_decoder_init( PS_Decoder*  ps_decoder,
                   void*        decoder,
                   FT_Bool      is_t1 )
  {
    FT_ZERO( ps_decoder );

    if ( is_t1 )
    {
      T1_Decoder  t1_decoder = (T1_Decoder)decoder;


      ps_builder_init( &ps_decoder->builder,
                       &t1_decoder->builder,
                       is_t1 );

      ps_decoder->cf2_instance = &t1_decoder->cf2_instance;
      ps_decoder->psnames      = t1_decoder->psnames;

      ps_decoder->num_glyphs  = t1_decoder->num_glyphs;
      ps_decoder->glyph_names = t1_decoder->glyph_names;
      ps_decoder->hint_mode   = t1_decoder->hint_mode;
      ps_decoder->blend       = t1_decoder->blend;

      ps_decoder->num_locals  = (FT_UInt)t1_decoder->num_subrs;
      ps_decoder->locals      = t1_decoder->subrs;
      ps_decoder->locals_len  = t1_decoder->subrs_len;
      ps_decoder->locals_hash = t1_decoder->subrs_hash;

      ps_decoder->buildchar     = t1_decoder->buildchar;
      ps_decoder->len_buildchar = t1_decoder->len_buildchar;

      ps_decoder->lenIV = t1_decoder->lenIV;
    }
    else
    {
      CFF_Decoder*  cff_decoder = (CFF_Decoder*)decoder;


      ps_builder_init( &ps_decoder->builder,
                       &cff_decoder->builder,
                       is_t1 );

      ps_decoder->cff             = cff_decoder->cff;
      ps_decoder->cf2_instance    = &cff_decoder->cff->cf2_instance;
      ps_decoder->current_subfont = cff_decoder->current_subfont;

      ps_decoder->num_globals  = cff_decoder->num_globals;
      ps_decoder->globals      = cff_decoder->globals;
      ps_decoder->globals_bias = cff_decoder->globals_bias;
      ps_decoder->num_locals   = cff_decoder->num_locals;
      ps_decoder->locals       = cff_decoder->locals;
      ps_decoder->locals_bias  = cff_decoder->locals_bias;

      ps_decoder->glyph_width   = &cff_decoder->glyph_width;
      ps_decoder->width_only    = cff_decoder->width_only;

      ps_decoder->hint_mode = cff_decoder->hint_mode;

      ps_decoder->get_glyph_callback  = cff_decoder->get_glyph_callback;
      ps_decoder->free_glyph_callback = cff_decoder->free_glyph_callback;
    }
  }


  /* Synthesize a SubFont object for Type 1 fonts, for use in the  */
  /* new interpreter to access Private dict data.                  */
  FT_LOCAL_DEF( void )
  t1_make_subfont( FT_Face      face,
                   PS_Private   priv,
                   CFF_SubFont  subfont )
  {
    CFF_Private  cpriv = &subfont->private_dict;
    FT_UInt      n, count;


    FT_ZERO( subfont );
    FT_ZERO( cpriv );

    count = cpriv->num_blue_values = priv->num_blue_values;
    for ( n = 0; n < count; n++ )
      cpriv->blue_values[n] = (FT_Pos)priv->blue_values[n];

    count = cpriv->num_other_blues = priv->num_other_blues;
    for ( n = 0; n < count; n++ )
      cpriv->other_blues[n] = (FT_Pos)priv->other_blues[n];

    count = cpriv->num_family_blues = priv->num_family_blues;
    for ( n = 0; n < count; n++ )
      cpriv->family_blues[n] = (FT_Pos)priv->family_blues[n];

    count = cpriv->num_family_other_blues = priv->num_family_other_blues;
    for ( n = 0; n < count; n++ )
      cpriv->family_other_blues[n] = (FT_Pos)priv->family_other_blues[n];

    cpriv->blue_scale = priv->blue_scale;
    cpriv->blue_shift = (FT_Pos)priv->blue_shift;
    cpriv->blue_fuzz  = (FT_Pos)priv->blue_fuzz;

    cpriv->standard_width  = (FT_Pos)priv->standard_width[0];
    cpriv->standard_height = (FT_Pos)priv->standard_height[0];

    count = cpriv->num_snap_widths = priv->num_snap_widths;
    for ( n = 0; n < count; n++ )
      cpriv->snap_widths[n] = (FT_Pos)priv->snap_widths[n];

    count = cpriv->num_snap_heights = priv->num_snap_heights;
    for ( n = 0; n < count; n++ )
      cpriv->snap_heights[n] = (FT_Pos)priv->snap_heights[n];

    cpriv->force_bold       = priv->force_bold;
    cpriv->lenIV            = priv->lenIV;
    cpriv->language_group   = priv->language_group;
    cpriv->expansion_factor = priv->expansion_factor;

    cpriv->subfont = subfont;


    /* Initialize the random number generator. */
    if ( face->internal->random_seed != -1 )
    {
      /* If we have a face-specific seed, use it.    */
      /* If non-zero, update it to a positive value. */
      subfont->random = (FT_UInt32)face->internal->random_seed;
      if ( face->internal->random_seed )
      {
        do
        {
          face->internal->random_seed = (FT_Int32)cff_random(
            (FT_UInt32)face->internal->random_seed );

        } while ( face->internal->random_seed < 0 );
      }
    }
    if ( !subfont->random )
    {
      FT_UInt32  seed;


      /* compute random seed from some memory addresses */
      seed = (FT_UInt32)( (FT_Offset)(char*)&seed    ^
                          (FT_Offset)(char*)&face    ^
                          (FT_Offset)(char*)&subfont );
      seed = seed ^ ( seed >> 10 ) ^ ( seed >> 20 );
      if ( seed == 0 )
        seed = 0x7384;

      subfont->random = seed;
    }
  }


  FT_LOCAL_DEF( void )
  t1_decrypt( FT_Byte*   buffer,
              FT_Offset  length,
              FT_UShort  seed )
  {
    PS_Conv_EexecDecode( &buffer,
                         FT_OFFSET( buffer, length ),
                         buffer,
                         length,
                         &seed );
  }


  FT_LOCAL_DEF( FT_UInt32 )
  cff_random( FT_UInt32  r )
  {
    /* a 32bit version of the `xorshift' algorithm */
    r ^= r << 13;
    r ^= r >> 17;
    r ^= r << 5;

    return r;
  }


/* END */