|
/Users/lyon/j4p/src/futils/StreamSniffer.java
|
1 package futils;
2
3 // main class futils.StreamSniffer
4
5 import utils.ByteUtil;
6
7 import java.io.BufferedInputStream;
8 import java.io.IOException;
9 import java.io.*;
10 import java.util.Hashtable;
11
12 public class StreamSniffer {
13 private BufferedInputStream bis;
14 private byte header[] = new byte[6];
15 private int numberActuallyRead = 0;
16
17 public StreamSniffer(InputStream is) {
18 bis = new BufferedInputStream(is);
19 init();
20 sniff();
21 }
22
23 public static void main(String args[]) {
24 FileInputStream fis =
25 futils.Futil.getFileInputStream(
26 "select a new data file");
27 StreamSniffer ss =
28 new StreamSniffer(fis);
29
30 System.out.println("fee fi fo fum I smell:" + ss);
31 futils.Futil.close(fis);
32 }
33
34 public BufferedInputStream getStream() {
35 return bis;
36 }
37
38 // Sniff and cover your tracks!
39 private void sniff() {
40 if (!bis.markSupported()) {
41 System.out.println(
42 "StreamSniffer needs" +
43 " a markable stream");
44 return;
45 }
46 bis.mark(header.length);
47
48 try {
49 numberActuallyRead
50 = bis.read(header);
51 bis.reset();
52 } catch (IOException e) {
53 System.out.println(e);
54 numberActuallyRead = -1;
55 }
56
57 }
58
59 public void printHeader() {
60 System.out.println("In hex...");
61 for (int i = 0; i < header.length; i++)
62 ByteUtil.printToHex(header[i]);
63 System.out.println("\nin base 8...");
64 for (int i = 0; i < header.length; i++)
65 ByteUtil.printToOctal(header[i]);
66 System.out.println("\n in ASCII");
67 System.out.println(new String(header));
68 System.out.println("if (match("
69 + ByteUtil.toString(header[0]) + ","
70 + ByteUtil.toString(header[1]) + ","
71 + ByteUtil.toString(header[2]) + ","
72 + ByteUtil.toString(header[3]) + "))");
73
74 }
75
76 public String getMimeType() {
77 return getStringForId(classifyStream());
78 }
79
80 public String toString() {
81 int id = classifyStream();
82 String s = getStringForId(id);
83 if (id == TYPENOTFOUND)
84 printHeader();
85 return s;
86
87 }
88
89 public boolean match(
90 char c0, char c1) {
91 byte b[] = header;
92 return
93 (b[0] == c0) &&
94 (b[1] == c1);
95 }
96
97 public boolean match(
98 char c0, char c1,
99 char c2, char c3) {
100 byte b[] = header;
101 return
102 (b[0] == (c0 & 0xFF)) &&
103 (b[1] == (c1 & 0xFF)) &&
104 (b[2] == (c2 & 0xFF)) &&
105 (b[3] == (c3 & 0xFF));
106 }
107
108 public boolean match(
109 int c0, int c1,
110 int c2, int c3) {
111 byte b[] = header;
112 return
113 ((b[0] & 0xFF) == (c0 & 0xFF)) &&
114 ((b[1] & 0xFF) == (c1 & 0xFF)) &&
115 ((b[2] & 0xFF) == (c2 & 0xFF)) &&
116 ((b[3] & 0xFF) == (c3 & 0xFF));
117 }
118
119 public boolean match(
120 int c0, int c1) {
121 byte b[] = header;
122 return
123 ((b[0] & 0xFF) == (c0 & 0xFF)) &&
124 ((b[1] & 0xFF) == (c1 & 0xFF));
125 }
126
127
128 public int classifyStream() {
129 byte b[] = header;
130 String s = new String(b);
131 short leshort =
132 (short) (
133 (b[0] << 8) | (b[1]));
134 int belong =
135 (b[0] << 24) |
136 (b[1] << 16) |
137 (b[2] << 8) |
138 (b[3]);
139
140
141 if (s.startsWith("begin"))
142 return UUENCODED;
143 if (s.startsWith("xbtoa"))
144 return BTOAD;
145 if (s.startsWith("P1"))
146 return PBM;
147 if (s.startsWith("P2"))
148 return PGM;
149 if (s.startsWith("P3"))
150 return PPM;
151 if (s.startsWith("P4"))
152 return PBM_RAWBITS;
153 if (s.startsWith("P5"))
154 return PGM_RAWBITS;
155 if (s.startsWith("P6"))
156 return PPM_RAWBITS;
157 if (s.startsWith("yz"))
158 return MGR_BITMAP;
159 if (s.startsWith("ILBM"))
160 return IFF_ILBM;
161 if ((b[0] == 131) &&
162 (b[1] == 246) &&
163 (b[2] == 152) &&
164 (b[3] == 225))
165 return SUNRASTER;
166 if ((b[0] == 1) &&
167 (b[1] == 332))
168 return SGI_IMAGE;
169 if ((b[0] == 361) &&
170 (b[1] == 0) &&
171 (b[2] == 100) &&
172 (b[3] == 273))
173 return CMU_WINDOW_MANAGER_BITMAP;
174 if ((b[0] == 131) &&
175 (b[1] == 246) &&
176 (b[2] == 152) &&
177 (b[3] == 225))
178 return SUN;
179 if ((b[0] == 115) &&
180 (b[1] == 115))
181 return TIFF_BIG_ENDIAN;
182 if ((b[0] == 111) &&
183 (b[1] == 111))
184 return TIFF_LITTLE_ENDIAN;
185 if (match(73, 73, 42, 0))
186 return TIFF_LITTLE_ENDIAN;
187 if (s.startsWith("GIF87a"))
188 return GIF87a;
189 if (s.startsWith("GIF89a"))
190 return GIF89a;
191 if ((b[0] == 131) &&
192 (b[1] == 246) &&
193 (b[2] == 152) &&
194 (b[3] == 225))
195 return SUNRASTER;
196 if (leshort == 0xAF11)
197 return FLI;
198 if ((b[0] == 0) &&
199 (b[1] == 0) &&
200 (b[2] == 1) &&
201 (b[3] == 263))
202 return MPEG;
203 if (s.startsWith(".snd"))
204 return SUN_NEXT_AUDIO;
205 if (s.startsWith("MThd"))
206 return STANDARD_MIDI;
207 if (s.startsWith("RIFF"))
208 return MICROSOFT_RIFF;
209 if (s.startsWith("BZ"))
210 return BZIP;
211 if (s.startsWith("FORM"))
212 return IFF_DATA;
213 if (s.startsWith("IIN1"))
214 return NIFF_IMAGE;
215 if (s.startsWith("BM"))
216 return PC_BITMAP;
217 if (s.startsWith("%PDF-"))
218 return PDF_DOCUMENT;
219 if (s.startsWith("%!"))
220 return POSTSCRIPT_DOCUMENT;
221 if (s.startsWith("MOVI"))
222 return SILICON_GRAPHICS_MOVIE;
223 if (s.startsWith("moov"))
224 return APPLE_QUICKTIME_MOVIE;
225 if (s.startsWith("mdat"))
226 return APPLE_QUICKTIME_MOVIE;
227 if (match('P', 'K'))
228 return ZIP_ARCHIVE;
229 if (match(03, 0235))
230 return UNIX_COMPRESS;
231 if (match(037, 0213))
232 return GZIP;
233 if (match(037, 036))
234 return HUFFMAN;
235 if (match((char) 0x89, 'P', 'N', 'G'))
236 return PNG_IMAGE;
237 if (match(0xFF, 0xD8, 0xFF, 0xE0))
238 return JPEG;
239 if (match(0377, 0330, 0377, 0356))
240 return JPG;
241 if (match('8', 'B', 'P', 'S'))
242 return PSHOP8;
243 if (match(172, 237, 0, 5))
244 return ZIP;
245 if (match(60, 104, 116, 109))
246 return HTML;
247 if (match(60, 63, 120, 109))
248 return XML;
249 if (match(137, 80, 78, 71))
250 return PNG_IMAGE;
251 return TYPENOTFOUND;
252 }
253
254 private void init() {
255 add(TYPENOTFOUND, "TYPENOTFOUND");
256 add(UUENCODED, "UUENCODED");
257 add(BTOAD, "BTOAD");
258 add(PBM, "PBM");
259 add(PGM, "PGM");
260 add(PPM, "PPM");
261 add(PBM_RAWBITS, "PBM_RAWBITS");
262 add(PGM_RAWBITS, "PGM_RAWBITS");
263 add(PPM_RAWBITS, "PPM_RAWBITS");
264 add(MGR_BITMAP, "MGR_BITMAP");
265 add(GIF87a, "image/gif");//gif 87a
266 add(GIF89a, "GIF89a");
267 add(IFF_ILBM, "IFF_ILBM");
268 add(SUNRASTER, "SUNRASTER");
269 add(SGI_IMAGE, "SGI_IMAGE");
270 add(CMU_WINDOW_MANAGER_BITMAP, "CMU_WINDOW_MANAGER_BITMAP");
271 add(SUN, "SUN");
272 add(TIFF_BIG_ENDIAN, "TIFF_BIG_ENDIAN");
273 add(TIFF_LITTLE_ENDIAN, "TIFF_LITTLE_ENDIAN");
274 add(FLI, "FLI");
275 add(MPEG, "MPEG");
276 add(SUN_NEXT_AUDIO, "sound/au");
277 add(STANDARD_MIDI, "STANDARD_MIDI");
278 add(MICROSOFT_RIFF, "MICROSOFT_RIFF");
279 add(BZIP, "BZIP");
280 add(IFF_DATA, "IFF_DATA");
281 add(NIFF_IMAGE, "NIFF_IMAGE");
282 add(PC_BITMAP, "PC_BITMAP");
283 add(PDF_DOCUMENT, "PDF_DOCUMENT");
284 add(POSTSCRIPT_DOCUMENT, "POSTSCRIPT_DOCUMENT");
285 add(SILICON_GRAPHICS_MOVIE, "SILICON_GRAPHICS_MOVIE");
286 add(APPLE_QUICKTIME_MOVIE, "APPLE_QUICKTIME_MOVIE");
287 add(ZIP_ARCHIVE, "ZIP_ARCHIVE");
288 add(UNIX_COMPRESS, "UNIX_COMPRESS");
289 add(GZIP, "GZIP");
290 add(HUFFMAN, "HUFFMAN");
291 add(PNG_IMAGE, "PNG_IMAGE");
292 add(JPEG, "JPEG");
293 add(JPG, "JPG");
294 add(PSHOP8, "Photo Shop...8 bits per pel");
295 add(ZIP, "Zip file...Wavelet encoded image sequence?");
296 add(HTML, "text/html");
297 add(XML, "xml document");
298 }
299
300 protected void add(int i, String s) {
301 h.put(new Integer(i), s);
302 }
303
304 public String getStringForId(int id) {
305 return (String) h.get(new Integer(id));
306 }
307
308
309 public static final int TYPENOTFOUND = 0;
310 public static final int UUENCODED = 1;
311 public static final int BTOAD = 2;
312 public static final int PBM = 3;
313 public static final int PGM = 4;
314 public static final int PPM = 5;
315 public static final int PBM_RAWBITS = 6;
316 public static final int PGM_RAWBITS = 7;
317 public static final int PPM_RAWBITS = 8;
318 public static final int MGR_BITMAP = 9;
319 public static final int GIF87a = 10;
320 public static final int GIF89a = 11;
321 public static final int IFF_ILBM = 12;
322 public static final int SUNRASTER = 13;
323 public static final int SGI_IMAGE = 14;
324 public static final int CMU_WINDOW_MANAGER_BITMAP = 15;
325 public static final int SUN = 16;
326 public static final int TIFF_BIG_ENDIAN = 17;
327 public static final int TIFF_LITTLE_ENDIAN = 18;
328 public static final int FLI = 19;
329 public static final int MPEG = 20;
330 public static final int SUN_NEXT_AUDIO = 21;
331 public static final int STANDARD_MIDI = 22;
332 public static final int MICROSOFT_RIFF = 23;
333 public static final int BZIP = 24;
334 public static final int IFF_DATA = 25;
335 public static final int NIFF_IMAGE = 26;
336 public static final int PC_BITMAP = 27;
337 public static final int PDF_DOCUMENT = 28;
338 public static final int POSTSCRIPT_DOCUMENT = 29;
339 public static final int SILICON_GRAPHICS_MOVIE = 30;
340 public static final int APPLE_QUICKTIME_MOVIE = 31;
341 public static final int ZIP_ARCHIVE = 32;
342 public static final int UNIX_COMPRESS = 33;
343 public static final int GZIP = 34;
344 public static final int HUFFMAN = 35;
345 public static final int PNG_IMAGE = 38;
346 public static final int JPEG = 39;
347 public static final int JPG = 40;
348 public static final int PSHOP8 = 41;
349 public static final int ZIP = 42;
350 public static final int HTML = 43;
351 public static final int XML = 44;
352
353 protected Hashtable h = new Hashtable();
354 }
355
356
357