update for HEAD-2003050101
[reactos.git] / lib / freetype / src / type42 / t42parse.c
diff --git a/lib/freetype/src/type42/t42parse.c b/lib/freetype/src/type42/t42parse.c
new file mode 100644 (file)
index 0000000..8bb2e7b
--- /dev/null
@@ -0,0 +1,994 @@
+/***************************************************************************/
+/*                                                                         */
+/*  t42parse.c                                                             */
+/*                                                                         */
+/*    Type 42 font parser (body).                                          */
+/*                                                                         */
+/*  Copyright 2002 by Roberto Alameda.                                     */
+/*                                                                         */
+/*  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 "t42parse.h"
+#include "t42error.h"
+#include FT_INTERNAL_DEBUG_H
+#include FT_INTERNAL_STREAM_H
+#include FT_LIST_H
+#include FT_INTERNAL_POSTSCRIPT_AUX_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  trace_t42
+
+
+  static void
+  t42_parse_font_name( T42_Face    face,
+                       T42_Loader  loader );
+                       
+  static void
+  t42_parse_font_bbox( T42_Face    face,
+                       T42_Loader  loader );
+                       
+  static void
+  t42_parse_font_matrix( T42_Face    face,
+                         T42_Loader  loader );
+  static void
+  t42_parse_encoding( T42_Face    face,
+                      T42_Loader  loader );
+                      
+  static void
+  t42_parse_charstrings( T42_Face    face,
+                         T42_Loader  loader );
+                         
+  static void
+  t42_parse_sfnts( T42_Face    face,
+                   T42_Loader  loader );
+
+
+  static const
+  T1_FieldRec  t42_keywords[] = {
+
+#undef  FT_STRUCTURE
+#define FT_STRUCTURE  T1_FontInfo
+#undef  T1CODE
+#define T1CODE        T1_FIELD_LOCATION_FONT_INFO
+
+    T1_FIELD_STRING   ( "version",            version )
+    T1_FIELD_STRING   ( "Notice",             notice )
+    T1_FIELD_STRING   ( "FullName",           full_name )
+    T1_FIELD_STRING   ( "FamilyName",         family_name )
+    T1_FIELD_STRING   ( "Weight",             weight )
+    T1_FIELD_NUM      ( "ItalicAngle",        italic_angle )
+    T1_FIELD_TYPE_BOOL( "isFixedPitch",       is_fixed_pitch )
+    T1_FIELD_NUM      ( "UnderlinePosition",  underline_position )
+    T1_FIELD_NUM      ( "UnderlineThickness", underline_thickness )
+
+#undef  FT_STRUCTURE
+#define FT_STRUCTURE  T1_FontRec
+#undef  T1CODE
+#define T1CODE        T1_FIELD_LOCATION_FONT_DICT
+
+    T1_FIELD_NUM( "PaintType",   paint_type )
+    T1_FIELD_NUM( "FontType",    font_type )
+    T1_FIELD_NUM( "StrokeWidth", stroke_width )
+
+    T1_FIELD_CALLBACK( "FontName",    t42_parse_font_name )
+    T1_FIELD_CALLBACK( "FontBBox",    t42_parse_font_bbox )
+    T1_FIELD_CALLBACK( "FontMatrix",  t42_parse_font_matrix )
+    T1_FIELD_CALLBACK( "Encoding",    t42_parse_encoding )
+    T1_FIELD_CALLBACK( "CharStrings", t42_parse_charstrings )
+    T1_FIELD_CALLBACK( "sfnts",       t42_parse_sfnts )
+
+    { 0, T1_FIELD_LOCATION_CID_INFO, T1_FIELD_TYPE_NONE, 0, 0, 0, 0, 0 }
+  };
+
+
+#define T1_Add_Table( p, i, o, l )  (p)->funcs.add( (p), i, o, l )
+#define T1_Done_Table( p )          \
+          do                        \
+          {                         \
+            if ( (p)->funcs.done )  \
+              (p)->funcs.done( p ); \
+          } while ( 0 )
+#define T1_Release_Table( p )          \
+          do                           \
+          {                            \
+            if ( (p)->funcs.release )  \
+              (p)->funcs.release( p ); \
+          } while ( 0 )
+
+#define T1_Skip_Spaces( p )  (p)->root.funcs.skip_spaces( &(p)->root )
+#define T1_Skip_Alpha( p )   (p)->root.funcs.skip_alpha ( &(p)->root )
+
+#define T1_ToInt( p )       (p)->root.funcs.to_int( &(p)->root )
+#define T1_ToFixed( p, t )  (p)->root.funcs.to_fixed( &(p)->root, t )
+
+#define T1_ToCoordArray( p, m, c )                           \
+          (p)->root.funcs.to_coord_array( &(p)->root, m, c )
+#define T1_ToFixedArray( p, m, f, t )                           \
+          (p)->root.funcs.to_fixed_array( &(p)->root, m, f, t )
+#define T1_ToToken( p, t )                          \
+          (p)->root.funcs.to_token( &(p)->root, t )
+#define T1_ToTokenArray( p, t, m, c )                           \
+          (p)->root.funcs.to_token_array( &(p)->root, t, m, c )
+
+#define T1_Load_Field( p, f, o, m, pf )                         \
+          (p)->root.funcs.load_field( &(p)->root, f, o, m, pf )
+#define T1_Load_Field_Table( p, f, o, m, pf )                         \
+          (p)->root.funcs.load_field_table( &(p)->root, f, o, m, pf )
+
+
+  /********************* Parsing Functions ******************/
+
+  FT_LOCAL_DEF( FT_Error )
+  t42_parser_init( T42_Parser     parser,
+                   FT_Stream      stream,
+                   FT_Memory      memory,
+                   PSAux_Service  psaux )
+  {
+    FT_Error  error = T42_Err_Ok;
+    FT_Long   size;
+
+
+    psaux->ps_parser_funcs->init( &parser->root, 0, 0, memory );
+
+    parser->stream    = stream;
+    parser->base_len  = 0;
+    parser->base_dict = 0;
+    parser->in_memory = 0;
+
+    /*******************************************************************/
+    /*                                                                 */
+    /* Here a short summary of what is going on:                       */
+    /*                                                                 */
+    /*   When creating a new Type 42 parser, we try to locate and load */
+    /*   the base dictionary, loading the whole font into memory.      */
+    /*                                                                 */
+    /*   When `loading' the base dictionary, we only setup pointers in */
+    /*   the case of a memory-based stream.  Otherwise, we allocate    */
+    /*   and load the base dictionary in it.                           */
+    /*                                                                 */
+    /*   parser->in_memory is set if we have a memory stream.          */
+    /*                                                                 */
+
+    if ( FT_STREAM_SEEK( 0L ) )
+      goto Exit;
+
+    size = stream->size;
+
+    /* now, try to load `size' bytes of the `base' dictionary we */
+    /* found previously                                          */
+
+    /* if it is a memory-based resource, set up pointers */
+    if ( !stream->read )
+    {
+      parser->base_dict = (FT_Byte*)stream->base + stream->pos;
+      parser->base_len  = size;
+      parser->in_memory = 1;
+
+      /* check that the `size' field is valid */
+      if ( FT_STREAM_SKIP( size ) )
+        goto Exit;
+    }
+    else
+    {
+      /* read segment in memory */
+      if ( FT_ALLOC( parser->base_dict, size )       ||
+           FT_STREAM_READ( parser->base_dict, size ) )
+        goto Exit;
+
+      parser->base_len = size;
+    }
+
+    /* Now check font format; we must see `%!PS-TrueTypeFont' */
+    if (size <= 17                                    ||
+        ( ft_strncmp( (const char*)parser->base_dict,
+                      "%!PS-TrueTypeFont", 17) )      )
+      error = T42_Err_Unknown_File_Format;
+    else
+    {
+      parser->root.base   = parser->base_dict;
+      parser->root.cursor = parser->base_dict;
+      parser->root.limit  = parser->root.cursor + parser->base_len;
+    }
+
+  Exit:
+    if ( error && !parser->in_memory )
+      FT_FREE( parser->base_dict );
+
+    return error;
+  }
+
+
+  FT_LOCAL_DEF( void )
+  t42_parser_done( T42_Parser  parser )
+  {
+    FT_Memory  memory = parser->root.memory;
+
+
+    /* free the base dictionary only when we have a disk stream */
+    if ( !parser->in_memory )
+      FT_FREE( parser->base_dict );
+
+    parser->root.funcs.done( &parser->root );
+  }
+
+
+  static int
+  t42_is_alpha( FT_Byte  c )
+  {
+    /* Note: we must accept "+" as a valid character, as it is used in */
+    /*       embedded type1 fonts in PDF documents.                    */
+    /*                                                                 */
+    return ( ft_isalnum( c ) ||
+             c == '.'        ||
+             c == '_'        ||
+             c == '-'        ||
+             c == '+'        );
+  }
+
+
+  static int
+  t42_is_space( FT_Byte  c )
+  {
+    return ( c == ' ' || c == '\t' || c == '\r' || c == '\n' );
+  }
+
+
+  static void
+  t42_parse_font_name( T42_Face    face,
+                       T42_Loader  loader )
+  {
+    T42_Parser  parser = &loader->parser;
+    FT_Error    error;
+    FT_Memory   memory = parser->root.memory;
+    FT_Int      len;
+    FT_Byte*    cur;
+    FT_Byte*    cur2;
+    FT_Byte*    limit;
+
+
+    T1_Skip_Spaces( parser );
+
+    cur   = parser->root.cursor;
+    limit = parser->root.limit;
+
+    if ( cur >= limit - 1              ||
+         ( *cur != '/' && *cur != '(') )
+      return;
+
+    cur++;
+    cur2 = cur;
+    while ( cur2 < limit && t42_is_alpha( *cur2 ) )
+      cur2++;
+
+    len = (FT_Int)( cur2 - cur );
+    if ( len > 0 )
+    {
+      if ( FT_ALLOC( face->type1.font_name, len + 1 ) )
+      {
+        parser->root.error = error;
+        return;
+      }
+
+      FT_MEM_COPY( face->type1.font_name, cur, len );
+      face->type1.font_name[len] = '\0';
+    }
+    parser->root.cursor = cur2;
+  }
+
+
+  static void
+  t42_parse_font_bbox( T42_Face   face,
+                       T42_Loader  loader )
+  {
+    T42_Parser  parser = &loader->parser;
+    FT_BBox*    bbox   = &face->type1.font_bbox;
+
+    bbox->xMin = T1_ToInt( parser );
+    bbox->yMin = T1_ToInt( parser );
+    bbox->xMax = T1_ToInt( parser );
+    bbox->yMax = T1_ToInt( parser );
+  }
+
+
+  static void
+  t42_parse_font_matrix( T42_Face    face,
+                         T42_Loader  loader )
+  {
+    T42_Parser  parser = &loader->parser;
+    FT_Matrix*  matrix = &face->type1.font_matrix;
+    FT_Vector*  offset = &face->type1.font_offset;
+    FT_Face     root   = (FT_Face)&face->root;
+    FT_Fixed    temp[6];
+    FT_Fixed    temp_scale;
+
+
+    (void)T1_ToFixedArray( parser, 6, temp, 3 );
+
+    temp_scale = ABS( temp[3] );
+
+    /* Set Units per EM based on FontMatrix values.  We set the value to */
+    /* 1000 / temp_scale, because temp_scale was already multiplied by   */
+    /* 1000 (in t1_tofixed, from psobjs.c).                              */
+
+    root->units_per_EM = (FT_UShort)( FT_DivFix( 1000 * 0x10000L,
+                                                 temp_scale ) >> 16 );
+
+    /* we need to scale the values by 1.0/temp_scale */
+    if ( temp_scale != 0x10000L ) {
+      temp[0] = FT_DivFix( temp[0], temp_scale );
+      temp[1] = FT_DivFix( temp[1], temp_scale );
+      temp[2] = FT_DivFix( temp[2], temp_scale );
+      temp[4] = FT_DivFix( temp[4], temp_scale );
+      temp[5] = FT_DivFix( temp[5], temp_scale );
+      temp[3] = 0x10000L;
+    }
+
+    matrix->xx = temp[0];
+    matrix->yx = temp[1];
+    matrix->xy = temp[2];
+    matrix->yy = temp[3];
+
+    /* note that the offsets must be expressed in integer font units */
+    offset->x  = temp[4] >> 16;
+    offset->y  = temp[5] >> 16;
+  }
+
+
+  static void
+  t42_parse_encoding( T42_Face    face,
+                      T42_Loader  loader )
+  {
+    T42_Parser     parser = &loader->parser;
+    FT_Byte*       cur    = parser->root.cursor;
+    FT_Byte*       limit  = parser->root.limit;
+
+    PSAux_Service  psaux  = (PSAux_Service)face->psaux;
+
+
+    /* skip whitespace */
+    while ( t42_is_space( *cur ) )
+    {
+      cur++;
+      if ( cur >= limit )
+      {
+        FT_ERROR(( "t42_parse_encoding: out of bounds!\n" ));
+        parser->root.error = T42_Err_Invalid_File_Format;
+        return;
+      }
+    }
+
+    /* if we have a number, then the encoding is an array, */
+    /* and we must load it now                             */
+    if ( (FT_Byte)( *cur - '0' ) < 10 )
+    {
+      T1_Encoding  encode     = &face->type1.encoding;
+      FT_Int       count, n;
+      PS_Table     char_table = &loader->encoding_table;
+      FT_Memory    memory     = parser->root.memory;
+      FT_Error     error;
+
+
+      /* read the number of entries in the encoding, should be 256 */
+      count = T1_ToInt( parser );
+      if ( parser->root.error )
+        return;
+
+      /* we use a T1_Table to store our charnames */
+      loader->num_chars = encode->num_chars = count;
+      if ( FT_NEW_ARRAY( encode->char_index, count ) ||
+           FT_NEW_ARRAY( encode->char_name,  count ) ||
+           FT_SET_ERROR( psaux->ps_table_funcs->init(
+                           char_table, count, memory ) ) )
+      {
+        parser->root.error = error;
+        return;
+      }
+
+      /* We need to `zero' out encoding_table.elements */
+      for ( n = 0; n < count; n++ )
+      {
+        char*  notdef = (char *)".notdef";
+
+
+        T1_Add_Table( char_table, n, notdef, 8 );
+      }
+
+      /* Now, we will need to read a record of the form         */
+      /* ... charcode /charname ... for each entry in our table */
+      /*                                                        */
+      /* We simply look for a number followed by an immediate   */
+      /* name.  Note that this ignores correctly the sequence   */
+      /* that is often seen in type1 fonts:                     */
+      /*                                                        */
+      /*   0 1 255 { 1 index exch /.notdef put } for dup        */
+      /*                                                        */
+      /* used to clean the encoding array before anything else. */
+      /*                                                        */
+      /* We stop when we encounter a `def'.                     */
+
+      cur   = parser->root.cursor;
+      limit = parser->root.limit;
+      n     = 0;
+
+      for ( ; cur < limit; )
+      {
+        FT_Byte  c;
+
+
+        c = *cur;
+
+        /* we stop when we encounter a `def' */
+        if ( c == 'd' && cur + 3 < limit )
+        {
+          if ( cur[1] == 'e'       &&
+               cur[2] == 'f'       &&
+               t42_is_space( cur[-1] ) &&
+               t42_is_space( cur[3] )  )
+          {
+            FT_TRACE6(( "encoding end\n" ));
+            break;
+          }
+        }
+
+        /* otherwise, we must find a number before anything else */
+        if ( (FT_Byte)( c - '0' ) < 10 )
+        {
+          FT_Int  charcode;
+
+
+          parser->root.cursor = cur;
+          charcode = T1_ToInt( parser );
+          cur      = parser->root.cursor;
+
+          /* skip whitespace */
+          while ( cur < limit && t42_is_space( *cur ) )
+            cur++;
+
+          if ( cur < limit && *cur == '/' )
+          {
+            /* bingo, we have an immediate name -- it must be a */
+            /* character name                                   */
+            FT_Byte*  cur2 = cur + 1;
+            FT_Int    len;
+
+
+            while ( cur2 < limit && t42_is_alpha( *cur2 ) )
+              cur2++;
+
+            len = (FT_Int)( cur2 - cur - 1 );
+
+            parser->root.error = T1_Add_Table( char_table, charcode,
+                                               cur + 1, len + 1 );
+            char_table->elements[charcode][len] = '\0';
+            if ( parser->root.error )
+              return;
+
+            cur = cur2;
+          }
+        }
+        else
+          cur++;
+      }
+
+      face->type1.encoding_type  = T1_ENCODING_TYPE_ARRAY;
+      parser->root.cursor        = cur;
+    }
+    /* Otherwise, we should have either `StandardEncoding', */
+    /* `ExpertEncoding', or `ISOLatin1Encoding'             */
+    else
+    {
+      if ( cur + 17 < limit                                            &&
+           ft_strncmp( (const char*)cur, "StandardEncoding", 16 ) == 0 )
+        face->type1.encoding_type = T1_ENCODING_TYPE_STANDARD;
+
+      else if ( cur + 15 < limit                                          &&
+                ft_strncmp( (const char*)cur, "ExpertEncoding", 14 ) == 0 )
+        face->type1.encoding_type = T1_ENCODING_TYPE_EXPERT;
+
+      else if ( cur + 18 < limit                                             &&
+                ft_strncmp( (const char*)cur, "ISOLatin1Encoding", 17 ) == 0 )
+        face->type1.encoding_type = T1_ENCODING_TYPE_ISOLATIN1;
+
+      else {
+        FT_ERROR(( "t42_parse_encoding: invalid token!\n" ));
+        parser->root.error = T42_Err_Invalid_File_Format;
+      }
+    }
+  }
+
+
+  static FT_UInt
+  t42_hexval( FT_Byte  v )
+  {
+    FT_UInt  d;
+    
+    d = (FT_UInt)( v - 'A' );
+    if ( d < 6 )
+    {
+      d += 10;
+      goto Exit;
+    }
+      
+    d = (FT_UInt)( v - 'a' );
+    if ( d < 6 )
+    {
+      d += 10;
+      goto Exit;
+    }
+      
+    d = (FT_UInt)( v - '0' );
+    if ( d < 10 )
+      goto Exit;
+      
+    d = 0;
+  Exit:         
+    return d;
+  }
+
+
+  static void
+  t42_parse_sfnts( T42_Face    face,
+                   T42_Loader  loader )
+  {
+    T42_Parser  parser = &loader->parser;
+    FT_Memory   memory = parser->root.memory;
+    FT_Byte*    cur    = parser->root.cursor;
+    FT_Byte*    limit  = parser->root.limit;
+    FT_Error    error;
+    FT_Int      num_tables = 0, status;
+    FT_ULong    count, ttf_size = 0, string_size = 0;
+    FT_Bool     in_string  = 0;
+    FT_Byte     v = 0;
+
+
+    /* The format is `/sfnts [ <...> <...> ... ] def' */
+
+    while ( t42_is_space( *cur ) )
+      cur++;
+
+    if (*cur++ == '[')
+    {
+      status = 0;
+      count = 0;
+    }
+    else
+    {
+      FT_ERROR(( "t42_parse_sfnts: can't find begin of sfnts vector!\n" ));
+      error = T42_Err_Invalid_File_Format;
+      goto Fail;
+    }
+
+    while ( cur < limit - 2 )
+    {
+      while ( t42_is_space( *cur ) )
+        cur++;
+
+      switch ( *cur )
+      {
+      case ']':
+        parser->root.cursor = cur++;
+        return;
+
+      case '<':
+        in_string   = 1;
+        string_size = 0;
+        cur++;
+        continue;
+
+      case '>':
+        if ( !in_string )
+        {
+          FT_ERROR(( "t42_parse_sfnts: found unpaired `>'!\n" ));
+          error = T42_Err_Invalid_File_Format;
+          goto Fail;
+        }
+
+        /* A string can have, as a last byte,         */
+        /* a zero byte for padding.  If so, ignore it */
+        if ( ( v == 0 ) && ( string_size % 2 == 1 ) )
+          count--;
+        in_string = 0;
+        cur++;
+        continue;
+
+      case '%':
+        if ( !in_string )
+        {
+          /* Comment found; skip till end of line */
+          while ( *cur != '\n' )
+            cur++;
+          continue;
+        }
+        else
+        {
+          FT_ERROR(( "t42_parse_sfnts: found `%' in string!\n" ));
+          error = T42_Err_Invalid_File_Format;
+          goto Fail;
+        }
+
+      default:
+        if ( !ft_xdigit( *cur ) || !ft_xdigit( *(cur + 1) ) )
+        {
+          FT_ERROR(( "t42_parse_sfnts: found non-hex characters in string" ));
+          error = T42_Err_Invalid_File_Format;
+          goto Fail;
+        }
+
+        v = (FT_Byte)( 16 * t42_hexval( cur[0] ) + t42_hexval( cur[1] ) );
+        cur += 2;
+        string_size++;
+      }
+
+      switch ( status )
+      {
+      case 0: /* The '[' was read, so load offset table, 12 bytes */
+        if ( count < 12 )
+        {
+          face->ttf_data[count++] = v;
+          continue;
+        }
+        else
+        {
+          num_tables = 16 * face->ttf_data[4] + face->ttf_data[5];
+          status     = 1;
+          ttf_size   = 12 + 16 * num_tables;
+
+          if ( FT_REALLOC( face->ttf_data, 12, ttf_size ) )
+            goto Fail;
+        }
+        /* No break, fall-through */
+
+      case 1: /* The offset table is read; read now the table directory */
+        if ( count < ttf_size )
+        {
+          face->ttf_data[count++] = v;
+          continue;
+        }
+        else
+        {
+          int      i;
+          FT_ULong len;
+
+
+          for ( i = 0; i < num_tables; i++ )
+          {
+            FT_Byte*  p = face->ttf_data + 12 + 16*i + 12;
+            
+            len = FT_PEEK_ULONG( p );
+
+            /* Pad to a 4-byte boundary length */
+            ttf_size += ( len + 3 ) & ~3;
+          }
+
+          status         = 2;
+          face->ttf_size = ttf_size;
+
+          if ( FT_REALLOC( face->ttf_data, 12 + 16 * num_tables,
+                           ttf_size + 1 ) )
+            goto Fail;
+        }
+        /* No break, fall-through */
+
+      case 2: /* We are reading normal tables; just swallow them */
+        face->ttf_data[count++] = v;
+
+      }
+    }
+
+    /* If control reaches this point, the format was not valid */
+    error = T42_Err_Invalid_File_Format;
+
+  Fail:
+    parser->root.error = error;
+  }
+
+
+  static void
+  t42_parse_charstrings( T42_Face    face,
+                         T42_Loader  loader )
+  {
+    T42_Parser     parser     = &loader->parser;
+    PS_Table       code_table = &loader->charstrings;
+    PS_Table       name_table = &loader->glyph_names;
+    FT_Memory      memory     = parser->root.memory;
+    FT_Error       error;
+
+    PSAux_Service  psaux      = (PSAux_Service)face->psaux;
+
+    FT_Byte*       cur;
+    FT_Byte*       limit      = parser->root.limit;
+    FT_Int         n;
+
+
+    loader->num_glyphs = T1_ToInt( parser );
+    if ( parser->root.error )
+      return;
+
+    /* initialize tables */
+
+    error = psaux->ps_table_funcs->init( code_table,
+                                         loader->num_glyphs,
+                                         memory );
+    if ( error )
+      goto Fail;
+
+    error = psaux->ps_table_funcs->init( name_table,
+                                         loader->num_glyphs,
+                                         memory );
+    if ( error )
+      goto Fail;
+
+    n = 0;
+
+    for (;;)
+    {
+      /* the format is simple:                    */
+      /*   `/glyphname' + index + def             */
+      /*                                          */
+      /* note that we stop when we find an `end'  */
+      /*                                          */
+      T1_Skip_Spaces( parser );
+
+      cur = parser->root.cursor;
+      if ( cur >= limit )
+        break;
+
+      /* we stop when we find an `end' keyword */
+      if ( *cur   == 'e'   &&
+           cur + 3 < limit &&
+           cur[1] == 'n'   &&
+           cur[2] == 'd'   )
+        break;
+
+      if ( *cur != '/' )
+        T1_Skip_Alpha( parser );
+      else
+      {
+        FT_Byte*  cur2 = cur + 1;
+        FT_Int    len;
+
+
+        while ( cur2 < limit && t42_is_alpha( *cur2 ) )
+          cur2++;
+        len = (FT_Int)( cur2 - cur - 1 );
+
+        error = T1_Add_Table( name_table, n, cur + 1, len + 1 );
+        if ( error )
+          goto Fail;
+
+        /* add a trailing zero to the name table */
+        name_table->elements[n][len] = '\0';
+
+        parser->root.cursor = cur2;
+        T1_Skip_Spaces( parser );
+
+        cur2 = cur = parser->root.cursor;
+        if ( cur >= limit )
+          break;
+
+        while ( cur2 < limit && t42_is_alpha( *cur2 ) )
+          cur2++;
+        len = (FT_Int)( cur2 - cur );
+
+        error = T1_Add_Table( code_table, n, cur, len + 1 );
+        if ( error )
+          goto Fail;
+
+        code_table->elements[n][len] = '\0';
+
+        n++;
+        if ( n >= loader->num_glyphs )
+          break;
+      }
+    }
+
+    /* Index 0 must be a .notdef element */
+    if ( ft_strcmp( (char *)name_table->elements[0], ".notdef" ) )
+    {
+      FT_ERROR(( "t42_parse_charstrings: Index 0 is not `.notdef'!\n" ));
+      error = T42_Err_Invalid_File_Format;
+      goto Fail;
+    }
+
+    loader->num_glyphs = n;
+    return;
+
+  Fail:
+    parser->root.error = error;
+  }
+
+
+  static FT_Error
+  t42_load_keyword( T42_Face    face,
+                    T42_Loader  loader,
+                    T1_Field    field )
+  {
+    FT_Error  error;
+    void*     dummy_object;
+    void**    objects;
+    FT_UInt   max_objects = 0;
+
+
+    /* if the keyword has a dedicated callback, call it */
+    if ( field->type == T1_FIELD_TYPE_CALLBACK ) {
+      field->reader( (FT_Face)face, loader );
+      error = loader->parser.root.error;
+      goto Exit;
+    }
+
+    /* now, the keyword is either a simple field, or a table of fields; */
+    /* we are now going to take care of it                              */
+    switch ( field->location )
+    {
+    case T1_FIELD_LOCATION_FONT_INFO:
+      dummy_object = &face->type1.font_info;
+      objects      = &dummy_object;
+      break;
+
+    default:
+      dummy_object = &face->type1;
+      objects      = &dummy_object;
+    }
+
+    if ( field->type == T1_FIELD_TYPE_INTEGER_ARRAY ||
+         field->type == T1_FIELD_TYPE_FIXED_ARRAY   )
+      error = T1_Load_Field_Table( &loader->parser, field,
+                                   objects, max_objects, 0 );
+    else
+      error = T1_Load_Field( &loader->parser, field,
+                             objects, max_objects, 0 );
+
+   Exit:
+    return error;
+  }
+
+
+  FT_LOCAL_DEF( FT_Error )
+  t42_parse_dict( T42_Face    face,
+                  T42_Loader  loader,
+                  FT_Byte*    base,
+                  FT_Long     size )
+  {
+    T42_Parser  parser = &loader->parser;
+    FT_Byte*    cur    = base;
+    FT_Byte*    limit  = cur + size;
+    FT_UInt     n_keywords = sizeof ( t42_keywords ) / 
+                             sizeof ( t42_keywords[0] );
+
+
+    parser->root.cursor = base;
+    parser->root.limit  = base + size;
+    parser->root.error  = 0;
+
+    for ( ; cur < limit; cur++ )
+    {
+      /* look for `FontDirectory', which causes problems on some fonts */
+      if ( *cur == 'F' && cur + 25 < limit                    &&
+           ft_strncmp( (char*)cur, "FontDirectory", 13 ) == 0 )
+      {
+        FT_Byte*  cur2;
+
+
+        /* skip the `FontDirectory' keyword */
+        cur += 13;
+        cur2 = cur;
+
+        /* lookup the `known' keyword */
+        while ( cur < limit && *cur != 'k'           &&
+                ft_strncmp( (char*)cur, "known", 5 ) )
+          cur++;
+
+        if ( cur < limit )
+        {
+          T1_TokenRec  token;
+
+
+          /* skip the `known' keyword and the token following it */
+          cur += 5;
+          loader->parser.root.cursor = cur;
+          T1_ToToken( &loader->parser, &token );
+
+          /* if the last token was an array, skip it! */
+          if ( token.type == T1_TOKEN_TYPE_ARRAY )
+            cur2 = parser->root.cursor;
+        }
+        cur = cur2;
+      }
+      /* look for immediates */
+      else if ( *cur == '/' && cur + 2 < limit )
+      {
+        FT_Byte*  cur2;
+        FT_UInt    i, len;
+
+
+        cur++;
+        cur2 = cur;
+        while ( cur2 < limit && t42_is_alpha( *cur2 ) )
+          cur2++;
+
+        len  = (FT_UInt)( cur2 - cur );
+        if ( len > 0 && len < 22 ) /* XXX What shall it this 22? */
+        {
+          /* now, compare the immediate name to the keyword table */
+
+          /* Loop through all known keywords */
+          for ( i = 0; i < n_keywords; i++ )
+          {
+            T1_Field  keyword = (T1_Field)&t42_keywords[i];
+            FT_Byte   *name   = (FT_Byte*)keyword->ident;
+
+
+            if ( !name )
+              continue;
+
+            if ( ( len == ft_strlen( (const char *)name ) ) &&
+                 ( ft_memcmp( cur, name, len ) == 0 )       )
+            {
+              /* we found it -- run the parsing callback! */
+              parser->root.cursor = cur2;
+              T1_Skip_Spaces( parser );
+              parser->root.error = t42_load_keyword(face,
+                                                    loader,
+                                                    keyword );
+              if ( parser->root.error )
+                return parser->root.error;
+              cur = parser->root.cursor;
+              break;
+            }
+          }
+        }
+      }
+    }
+    return parser->root.error;
+  }
+
+
+  FT_LOCAL_DEF( void )
+  t42_loader_init( T42_Loader  loader,
+                   T42_Face    face )
+  {
+    FT_UNUSED( face );
+
+    FT_MEM_ZERO( loader, sizeof ( *loader ) );
+    loader->num_glyphs = 0;
+    loader->num_chars  = 0;
+
+    /* initialize the tables -- simply set their `init' field to 0 */
+    loader->encoding_table.init = 0;
+    loader->charstrings.init    = 0;
+    loader->glyph_names.init    = 0;
+  }
+
+
+  FT_LOCAL_DEF( void )
+  t42_loader_done( T42_Loader  loader )
+  {
+    T42_Parser  parser = &loader->parser;
+
+
+    /* finalize tables */
+    T1_Release_Table( &loader->encoding_table );
+    T1_Release_Table( &loader->charstrings );
+    T1_Release_Table( &loader->glyph_names );
+
+    /* finalize parser */
+    t42_parser_done( parser );
+  }
+
+
+/* END */