BASE64DecoderStream


1   /*
2    * The contents of this file are subject to the terms 
3    * of the Common Development and Distribution License 
4    * (the "License").  You may not use this file except 
5    * in compliance with the License.
6    * 
7    * You can obtain a copy of the license at 
8    * glassfish/bootstrap/legal/CDDLv1.0.txt or 
9    * https://glassfish.dev.java.net/public/CDDLv1.0.html. 
10   * See the License for the specific language governing 
11   * permissions and limitations under the License.
12   * 
13   * When distributing Covered Code, include this CDDL 
14   * HEADER in each file and include the License file at 
15   * glassfish/bootstrap/legal/CDDLv1.0.txt.  If applicable, 
16   * add the following below this CDDL HEADER, with the 
17   * fields enclosed by brackets "[]" replaced with your 
18   * own identifying information: Portions Copyright [yyyy] 
19   * [name of copyright owner]
20   */
21  
22  /*
23   * @(#)BASE64DecoderStream.java 1.14 05/10/18
24   *
25   * Copyright 1997-2005 Sun Microsystems, Inc. All Rights Reserved.
26   */
27  
28  package com.sun.mail.util;
29  
30  import java.io.*;
31  
32  /**
33   * This class implements a BASE64 Decoder. It is implemented as
34   * a FilterInputStream, so one can just wrap this class around
35   * any input stream and read bytes from this filter. The decoding
36   * is done as the bytes are read out.
37   * 
38   * @author John Mani
39   * @author Bill Shannon
40   */
41  
42  public class BASE64DecoderStream extends FilterInputStream {
43      // buffer of decoded bytes for single byte reads
44      private byte[] buffer = new byte[3];
45      private int bufsize = 0;    // size of the cache
46      private int index = 0;  // index into the cache
47  
48      // buffer for almost 8K of typical 76 chars + CRLF lines,
49      // used by getByte method.  this buffer contains encoded bytes.
50      private byte[] input_buffer = new byte[78*105];
51      private int input_pos = 0;
52      private int input_len = 0;;
53  
54      private boolean ignoreErrors = false;
55  
56      /** 
57       * Create a BASE64 decoder that decodes the specified input stream.
58       * The System property <code>mail.mime.base64.ignoreerrors</code>
59       * controls whether errors in the encoded data cause an exception
60       * or are ignored.  The default is false (errors cause exception).
61       *
62       * @param in    the input stream
63       */
64      public BASE64DecoderStream(InputStream in) {
65      super(in);
66      try {
67          String   s = System.getProperty("mail.mime.base64.ignoreerrors");
68          // default to false
69          ignoreErrors = s != null && !s.equalsIgnoreCase("false");
70      } catch (SecurityException   sex) {
71          // ignore it
72      }
73      }
74  
75      /** 
76       * Create a BASE64 decoder that decodes the specified input stream.
77       *
78       * @param in    the input stream
79       * @param ignoreErrors  ignore errors in encoded data?
80       */
81      public BASE64DecoderStream(InputStream in, boolean ignoreErrors) {
82      super(in);
83      this.ignoreErrors = ignoreErrors;
84      }
85  
86      /**
87       * Read the next decoded byte from this input stream. The byte
88       * is returned as an <code>int</code> in the range <code>0</code> 
89       * to <code>255</code>. If no byte is available because the end of 
90       * the stream has been reached, the value <code>-1</code> is returned.
91       * This method blocks until input data is available, the end of the 
92       * stream is detected, or an exception is thrown.
93       *
94       * @return     next byte of data, or <code>-1</code> if the end of the
95       *             stream is reached.
96       * @exception  IOException  if an I/O error occurs.
97       * @see        java.io.FilterInputStream#in
98       */
99      public int read() throws IOException {
100     if (index >= bufsize) {
101         bufsize = decode(buffer, 0, buffer.length);
102         if (bufsize <= 0) // buffer is empty
103         return -1;
104         index = 0; // reset index into buffer
105     }
106     return buffer[index++] & 0xff; // Zero off the MSB
107     }
108 
109     /**
110      * Reads up to <code>len</code> decoded bytes of data from this input stream
111      * into an array of bytes. This method blocks until some input is
112      * available.
113      * <p>
114      *
115      * @param      buf   the buffer into which the data is read.
116      * @param      off   the start offset of the data.
117      * @param      len   the maximum number of bytes read.
118      * @return     the total number of bytes read into the buffer, or
119      *             <code>-1</code> if there is no more data because the end of
120      *             the stream has been reached.
121      * @exception  IOException  if an I/O error occurs.
122      */
123     public int read(byte[] buf, int off, int len) throws IOException {
124     // empty out single byte read buffer
125     int off0 = off;
126     while (index < bufsize && len > 0) {
127         buf[off++] = buffer[index++];
128         len--;
129     }
130     if (index >= bufsize)
131         bufsize = index = 0;
132 
133     int bsize = (len / 3) * 3;  // round down to multiple of 3 bytes
134     if (bsize > 0) {
135         int size = decode(buf, off, bsize);
136         off += size;
137         len -= size;
138 
139         if (size != bsize) {    // hit EOF?
140         if (off == off0)    // haven't returned any data
141             return -1;
142         else            // returned some data before hitting EOF
143             return off - off0;
144         }
145     }
146 
147     // finish up with a partial read if necessary
148     for (; len > 0; len--) {
149         int c = read();
150         if (c == -1)    // EOF
151         break;
152         buf[off++] = (byte)c;
153     }
154 
155     if (off == off0)    // haven't returned any data
156         return -1;
157     else            // returned some data before hitting EOF
158         return off - off0;
159     }
160 
161     /**
162      * Tests if this input stream supports marks. Currently this class
163      * does not support marks
164      */
165     public boolean markSupported() {
166     return false; // Maybe later ..
167     }
168 
169     /**
170      * Returns the number of bytes that can be read from this input
171      * stream without blocking. However, this figure is only
172      * a close approximation in case the original encoded stream
173      * contains embedded CRLFs; since the CRLFs are discarded, not decoded
174      */ 
175     public int available() throws IOException {
176      // This is only an estimate, since in.available()
177      // might include CRLFs too ..
178      return ((in.available() * 3)/4 + (bufsize-index));
179     }
180 
181     /**
182      * This character array provides the character to value map
183      * based on RFC1521.
184      */  
185     private final static char pem_array[] = {
186     'A','B','C','D','E','F','G','H', // 0
187     'I','J','K','L','M','N','O','P', // 1
188     'Q','R','S','T','U','V','W','X', // 2
189     'Y','Z','a','b','c','d','e','f', // 3
190     'g','h','i','j','k','l','m','n', // 4
191     'o','p','q','r','s','t','u','v', // 5
192     'w','x','y','z','0','1','2','3', // 6
193     '4','5','6','7','8','9','+','/'  // 7
194     };
195 
196     private final static byte pem_convert_array[] = new byte[256];
197 
198     static {
199     for (int i = 0; i < 255; i++)
200         pem_convert_array[i] = -1;
201     for (int i = 0; i < pem_array.length; i++)
202         pem_convert_array[pem_array[i]] = (byte)i;
203     }
204 
205     /**
206      * The decoder algorithm.  Most of the complexity here is dealing
207      * with error cases.  Returns the number of bytes decoded, which
208      * may be zero.  Decoding is done by filling an int with 4 6-bit
209      * values by shifting them in from the bottom and then extracting
210      * 3 8-bit bytes from the int by shifting them out from the bottom.
211      *
212      * @param   outbuf  the buffer into which to put the decoded bytes
213      * @param   pos position in the buffer to start filling
214      * @param   len the number of bytes to fill
215      * @return      the number of bytes filled, always a multiple
216      *          of three, and may be zero
217      * @exception   IOException if the data is incorrectly formatted
218      */
219     private int decode(byte[] outbuf, int pos, int len) throws IOException {
220     int pos0 = pos;
221     while (len >= 3) {
222         /*
223          * We need 4 valid base64 characters before we start decoding.
224          * We skip anything that's not a valid base64 character (usually
225          * just CRLF).
226          */
227         int got = 0;
228         int val = 0;
229         while (got < 4) {
230         int i = getByte();
231         if (i == -1 || i == -2) {
232             boolean atEOF;
233             if (i == -1) {
234             if (got == 0)
235                 return pos - pos0;
236             if (!ignoreErrors)
237                 throw new IOException("Error in encoded stream: " +
238                 "needed 4 valid base64 characters " +
239                 "but only got " + got + " before EOF" +
240                 recentChars());
241             atEOF = true;   // don't read any more
242             } else {    // i == -2
243             // found a padding character, we're at EOF
244             // XXX - should do something to make EOF "sticky"
245             if (got < 2 && !ignoreErrors)
246                 throw new IOException("Error in encoded stream: " +
247                 "needed at least 2 valid base64 characters," +
248                 " but only got " + got +
249                 " before padding character (=)" +
250                 recentChars());
251 
252             // didn't get any characters before padding character?
253             if (got == 0)
254                 return pos - pos0;
255             atEOF = false;  // need to keep reading
256             }
257 
258             // pad partial result with zeroes
259 
260             // how many bytes will we produce on output?
261             // (got always < 4, so size always < 3)
262             int size = got - 1;
263             if (size == 0)
264             size = 1;
265 
266             // handle the one padding character we've seen
267             got++;
268             val <<= 6;
269 
270             while (got < 4) {
271             if (!atEOF) {
272                 // consume the rest of the padding characters,
273                 // filling with zeroes
274                 i = getByte();
275                 if (i == -1) {
276                 if (!ignoreErrors)
277                     throw new IOException(
278                     "Error in encoded stream: " +
279                     "hit EOF while looking for " +
280                     "padding characters (=)" +
281                     recentChars());
282                 } else if (i != -2) {
283                 if (!ignoreErrors)
284                     throw new IOException(
285                     "Error in encoded stream: " +
286                     "found valid base64 character after " +
287                     "a padding character (=)" +
288                     recentChars());
289                 }
290             }
291             val <<= 6;
292             got++;
293             }
294 
295             // now pull out however many valid bytes we got
296             val >>= 8;      // always skip first one
297             if (size == 2)
298             outbuf[pos + 1] = (byte)(val & 0xff);
299             val >>= 8;
300             outbuf[pos] = (byte)(val & 0xff);
301             len -= size;
302             pos += size;
303             return pos - pos0;
304         } else {
305             // got a valid byte
306             val <<= 6;
307             got++;
308             val |= i;
309         }
310         }
311 
312         // read 4 valid characters, now extract 3 bytes
313         outbuf[pos + 2] = (byte)(val & 0xff);
314         val >>= 8;
315         outbuf[pos + 1] = (byte)(val & 0xff);
316         val >>= 8;
317         outbuf[pos] = (byte)(val & 0xff);
318         len -= 3;
319         pos += 3;
320     }
321     return pos - pos0;
322     }
323 
324     /**
325      * Read the next valid byte from the input stream.
326      * Buffer lots of data from underlying stream in input_buffer,
327      * for efficiency.
328      *
329      * @return  the next byte, -1 on EOF, or -2 if next byte is '='
330      *      (padding at end of encoded data)
331      */
332     private int getByte() throws IOException {
333     int c;
334     do {
335         if (input_pos >= input_len) {
336         try {
337             input_len = in.read(input_buffer);
338         } catch (EOFException ex) {
339             return -1;
340         }
341         if (input_len <= 0)
342             return -1;
343         input_pos = 0;
344         }
345         // get the next byte in the buffer
346         c = input_buffer[input_pos++] & 0xff;
347         // is it a padding byte?
348         if (c == '=')
349         return -2;
350         // no, convert it
351         c = pem_convert_array[c];
352         // loop until we get a legitimate byte
353     } while (c == -1);
354     return c;
355     }
356 
357     /**
358      * Return the most recent characters, for use in an error message.
359      */
360     private String   recentChars() {
361     // reach into the input buffer and extract up to 10
362     // recent characters, to help in debugging.
363     String   errstr = "";
364     int nc = input_pos > 10 ? 10 : input_pos;
365     if (nc > 0) {
366         errstr += ", the " + nc +
367                 " most recent characters were: \"";
368         for (int k = input_pos - nc; k < input_pos; k++) {
369         char c = (char)(input_buffer[k] & 0xff);
370         switch (c) {
371         case '\r':  errstr += "\\r"; break;
372         case '\n':  errstr += "\\n"; break;
373         case '\t':  errstr += "\\t"; break;
374         default:
375             if (c >= ' ' && c < 0177)
376             errstr += c;
377             else
378             errstr += ("\\" + (int)c);
379         }
380         }
381         errstr += "\"";
382     }
383     return errstr;
384     }
385 
386     /**
387      * Base64 decode a byte array.  No line breaks are allowed.
388      * This method is suitable for short strings, such as those
389      * in the IMAP AUTHENTICATE protocol, but not to decode the
390      * entire content of a MIME part.
391      *
392      * NOTE: inbuf may only contain valid base64 characters.
393      *       Whitespace is not ignored.
394      */
395     public static byte[] decode(byte[] inbuf) {
396     int size = (inbuf.length / 4) * 3;
397     if (size == 0)
398         return inbuf;
399 
400     if (inbuf[inbuf.length - 1] == '=') {
401         size--;
402         if (inbuf[inbuf.length - 2] == '=')
403         size--;
404     }
405     byte[] outbuf = new byte[size];
406 
407     int inpos = 0, outpos = 0;
408     size = inbuf.length;
409     while (size > 0) {
410         int val;
411         int osize = 3;
412         val = pem_convert_array[inbuf[inpos++] & 0xff];
413         val <<= 6;
414         val |= pem_convert_array[inbuf[inpos++] & 0xff];
415         val <<= 6;
416         if (inbuf[inpos] != '=') // End of this BASE64 encoding
417         val |= pem_convert_array[inbuf[inpos++] & 0xff];
418         else
419         osize--;
420         val <<= 6;
421         if (inbuf[inpos] != '=') // End of this BASE64 encoding
422         val |= pem_convert_array[inbuf[inpos++] & 0xff];
423         else
424         osize--;
425         if (osize > 2)
426         outbuf[outpos + 2] = (byte)(val & 0xff);
427         val >>= 8;
428         if (osize > 1)
429         outbuf[outpos + 1] = (byte)(val & 0xff);
430         val >>= 8;
431         outbuf[outpos] = (byte)(val & 0xff);
432         outpos += osize;
433         size -= 4;
434     }
435     return outbuf;
436     }
437 
438     /*** begin TEST program ***
439     public static void main(String argv[]) throws Exception {
440         FileInputStream infile = new FileInputStream(argv[0]);
441     BASE64DecoderStream decoder = new BASE64DecoderStream(infile);
442     int c;
443 
444     while ((c = decoder.read()) != -1)
445         System.out.print((char)c);
446     System.out.flush();
447     }
448     *** end TEST program ***/
449 }
450
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags