/* -*- Mode: java; indent-tabs-mode: nil; c-basic-offset: 2 -*- * * The contents of this file are subject to the Mozilla Public * License Version 1.1 (the "License"); you may not use this file * except in compliance with the License. You may obtain a copy of * the License at http://www.mozilla.org/MPL/ * * Software distributed under the License is distributed on an "AS * IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or * implied. See the License for the specific language governing * rights and limitations under the License. * * The Original Code is the Grendel mail/news client. * * The Initial Developer of the Original Code is Netscape Communications * Corporation. Portions created by Netscape are * Copyright (C) 1997 Netscape Communications Corporation. All * Rights Reserved. * * Contributor(s): * * Created: Jamie Zawinski , 28 Aug 1997. * * Contributors: Edwin Woudt */ package grendel.mime.encoder; import calypso.util.ByteBuf; /** Implements a plaintext -> Base64 encoder. */ public final class MimeBase64Decoder extends MimeEncoder { private byte token[] = new byte[4]; // input buffer private byte bytes[] = new byte[3]; // output buffer private int token_length = 0; // input buffer length static private final byte NUL = 127; // must be out of range 0-64 static private final byte EOF = 126; // must be out of range 0-64 static private final byte map[] = { NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 000-007 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 010-017 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 020-027 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 030-037 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 040-047 !"#$%&' NUL, NUL, NUL, 62, NUL, NUL, NUL, 63, // 050-057 ()*+,-./ 52, 53, 54, 55, 56, 57, 58, 59, // 060-067 01234567 60, 61, NUL, NUL, NUL, EOF, NUL, NUL, // 070-077 89:;<=>? NUL, 0, 1, 2, 3, 4, 5, 6, // 100-107 @ABCDEFG 7, 8, 9, 10, 11, 12, 13, 14, // 110-117 HIJKLMNO 15, 16, 17, 18, 19, 20, 21, 22, // 120-127 PQRSTUVW 23, 24, 25, NUL, NUL, NUL, NUL, NUL, // 130-137 XYZ[\]^_ NUL, 26, 27, 28, 29, 30, 31, 32, // 140-147 `abcdefg 33, 34, 35, 36, 37, 38, 39, 40, // 150-157 hijklmno 41, 42, 43, 44, 45, 46, 47, 48, // 160-167 pqrstuvw 49, 50, 51, NUL, NUL, NUL, NUL, NUL, // 170-177 xyz{|}~ NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 200-207 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 210-217 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 220-227 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 230-237 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 240-247 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 250-257 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 260-267 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 270-277 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 300-307 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 310-317 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 320-327 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 330-337 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 340-347 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 350-357 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 360-367 NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, // 370-377 }; // Fast routine that assumes full 4-char tokens with no '=' in them. // private final void decode_token(ByteBuf out) { int num = ((map[token[0]] << 18) | (map[token[1]] << 12) | (map[token[2]] << 6) | (map[token[3]])); bytes[0] = (byte) (0xFF & (num >> 16)); bytes[1] = (byte) (0xFF & (num >> 8)); bytes[2] = (byte) (0xFF & num); out.append(bytes); } // Hairier routine that deals with the final token, which can have fewer // than four characters, and that might be padded with '='. // private final void decode_final_token(ByteBuf out) { byte b0 = map[token[0]]; byte b1 = map[token[1]]; byte b2 = map[token[2]]; byte b3 = map[token[3]]; int eq_count = 0; if (b0 == EOF) { b0 = 0; eq_count++; } if (b1 == EOF) { b1 = 0; eq_count++; } if (b2 == EOF) { b2 = 0; eq_count++; } if (b3 == EOF) { b3 = 0; eq_count++; } int num = ((b0 << 18) | (b1 << 12) | (b2 << 6) | (b3)); // eq_count will be 0, 1, or 2. // No "=" padding means 4 bytes mapped to 3, the normal case, // not handled in this routine. // "xxx=" means 3 bytes mapped to 2. // "xx==" means 2 bytes mapped to 1. // "x===" can't happen, because "x" would then be encoding // only 6 bits, not 8, the minimum possible. out.append((byte) (num >> 16)); // byte 1, count = 0 or 1 or 2 if (eq_count <= 1) { out.append((byte) ((num >> 8) & 0xFF)); // byte 2, count = 0 or 1 if (eq_count == 0) { out.append((byte) (num & 0xFF)); // byte 3, count = 0 } } } /** Given a sequence of input bytes using the Base64 encoding, produces a sequence of unencoded output bytes. Note that some (small) amount of buffering may be necessary, if the input byte stream didn't fall on an appropriate boundary. If there are bytes in `out' already, the new bytes are appended, so the caller should do `out.setLength(0)' first if that's desired. */ public final void translate(ByteBuf in, ByteBuf out) { if (token == null) // already saw eof marker? return; byte inb[] = in.toBytes(); int in_length = in.length(); for (int i = 0; i < in_length; i++) { byte b = inb[i]; byte t = map[b]; if (t != NUL) token[token_length++] = b; if (t == EOF) { eof(out); return; } else if (token_length == 4) { decode_token(out); token_length = 0; } } } /** Tell the Base64 decoder that no more input data will be forthcoming. This may result in output, as a result of flushing the internal buffer. This object must not be used again after calling eof(). If there are bytes in `out' already, the new bytes are appended, so the caller should do `out.setLength(0)' first if that's desired. */ public final void eof(ByteBuf out) { if (token != null && token_length != 0) { while (token_length < 4) token[token_length++] = (byte)'='; decode_final_token(out); } token_length = 0; token = null; bytes = null; } }