added "dont compress" option with the default setting of
[rsync/rsync.git] / token.c
1 /* 
2    Copyright (C) Andrew Tridgell 1996
3    Copyright (C) Paul Mackerras 1996
4    
5    This program is free software; you can redistribute it and/or modify
6    it under the terms of the GNU General Public License as published by
7    the Free Software Foundation; either version 2 of the License, or
8    (at your option) any later version.
9    
10    This program is distributed in the hope that it will be useful,
11    but WITHOUT ANY WARRANTY; without even the implied warranty of
12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
13    GNU General Public License for more details.
14    
15    You should have received a copy of the GNU General Public License
16    along with this program; if not, write to the Free Software
17    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
18 */
19
20 #include "rsync.h"
21 #include "zlib/zlib.h"
22
23 extern int do_compression;
24 static int compression_level = Z_DEFAULT_COMPRESSION;
25
26 /* determine the compression level based on a wildcard filename list */
27 void set_compression(char *fname)
28 {
29         extern int module_id;
30         char *dont;
31         char *tok;
32
33         if (!do_compression) return;
34
35         compression_level = Z_DEFAULT_COMPRESSION;
36         dont = lp_dont_compress(module_id);
37
38         if (!dont || !*dont) return;
39
40         dont = strdup(dont);
41         fname = strdup(fname);
42         if (!dont || !fname) return;
43
44         strlower(dont);
45         strlower(fname);
46
47         for (tok=strtok(dont," ");tok;tok=strtok(NULL," ")) {
48                 if (fnmatch(tok, fname, 0) == 0) {
49                         compression_level = 0;
50                         break;
51                 }
52         }
53         free(dont);
54         free(fname);
55 }
56
57 /* non-compressing recv token */
58 static int simple_recv_token(int f,char **data)
59 {
60         static int residue;
61         static char *buf;
62         int n;
63
64         if (!buf) {
65                 buf = (char *)malloc(CHUNK_SIZE);
66                 if (!buf) out_of_memory("simple_recv_token");
67         }
68
69         if (residue == 0) {
70                 int i = read_int(f);
71                 if (i <= 0) return i;
72                 residue = i;
73         }
74
75         *data = buf;
76         n = MIN(CHUNK_SIZE,residue);
77         residue -= n;
78         read_buf(f,buf,n);
79         return n;
80 }
81
82
83 /* non-compressing send token */
84 static void simple_send_token(int f,int token,
85                               struct map_struct *buf,OFF_T offset,int n)
86 {
87         if (n > 0) {
88                 int l = 0;
89                 while (l < n) {
90                         int n1 = MIN(CHUNK_SIZE,n-l);
91                         write_int(f,n1);
92                         write_buf(f,map_ptr(buf,offset+l,n1),n1);
93                         l += n1;
94                 }
95         }
96         /* a -2 token means to send data only and no token */
97         if (token != -2) {
98                 write_int(f,-(token+1));
99         }
100 }
101
102
103 /* Flag bytes in compressed stream are encoded as follows: */
104 #define END_FLAG        0       /* that's all folks */
105 #define TOKEN_LONG      0x20    /* followed by 32-bit token number */
106 #define TOKENRUN_LONG   0x21    /* ditto with 16-bit run count */
107 #define DEFLATED_DATA   0x40    /* + 6-bit high len, then low len byte */
108 #define TOKEN_REL       0x80    /* + 6-bit relative token number */
109 #define TOKENRUN_REL    0xc0    /* ditto with 16-bit run count */
110
111 #define MAX_DATA_COUNT  16383   /* fit 14 bit count into 2 bytes with flags */
112
113 /* For coding runs of tokens */
114 static int last_token = -1;
115 static int run_start;
116 static int last_run_end;
117
118 /* Deflation state */
119 static z_stream tx_strm;
120
121 /* Output buffer */
122 static char *obuf;
123
124 /* Send a deflated token */
125 static void
126 send_deflated_token(int f, int token,
127                     struct map_struct *buf, OFF_T offset, int nb, int toklen)
128 {
129         int n, r;
130         static int init_done, flush_pending;
131
132         if (last_token == -1) {
133                 /* initialization */
134                 if (!init_done) {
135                         tx_strm.next_in = NULL;
136                         tx_strm.zalloc = NULL;
137                         tx_strm.zfree = NULL;
138                         if (deflateInit2(&tx_strm, compression_level,
139                                          Z_DEFLATED, -15, 8,
140                                          Z_DEFAULT_STRATEGY) != Z_OK) {
141                                 rprintf(FERROR, "compression init failed\n");
142                                 exit_cleanup(RERR_STREAMIO);
143                         }
144                         if ((obuf = malloc(MAX_DATA_COUNT+2)) == NULL)
145                                 out_of_memory("send_deflated_token");
146                         init_done = 1;
147                 } else
148                         deflateReset(&tx_strm);
149                 last_run_end = 0;
150                 run_start = token;
151                 flush_pending = 0;
152
153         } else if (last_token == -2) {
154                 run_start = token;
155
156         } else if (nb != 0 || token != last_token + 1
157                    || token >= run_start + 65536) {
158                 /* output previous run */
159                 r = run_start - last_run_end;
160                 n = last_token - run_start;
161                 if (r >= 0 && r <= 63) {
162                         write_byte(f, (n==0? TOKEN_REL: TOKENRUN_REL) + r);
163                 } else {
164                         write_byte(f, (n==0? TOKEN_LONG: TOKENRUN_LONG));
165                         write_int(f, run_start);
166                 }
167                 if (n != 0) {
168                         write_byte(f, n);
169                         write_byte(f, n >> 8);
170                 }
171                 last_run_end = last_token;
172                 run_start = token;
173         }
174
175         last_token = token;
176
177         if (nb != 0 || flush_pending) {
178                 /* deflate the data starting at offset */
179                 int flush = Z_NO_FLUSH;
180                 tx_strm.avail_in = 0;
181                 tx_strm.avail_out = 0;
182                 do {
183                         if (tx_strm.avail_in == 0 && nb != 0) {
184                                 /* give it some more input */
185                                 n = MIN(nb, CHUNK_SIZE);
186                                 tx_strm.next_in = (Bytef *)
187                                         map_ptr(buf, offset, n);
188                                 tx_strm.avail_in = n;
189                                 nb -= n;
190                                 offset += n;
191                         }
192                         if (tx_strm.avail_out == 0) {
193                                 tx_strm.next_out = (Bytef *)(obuf + 2);
194                                 tx_strm.avail_out = MAX_DATA_COUNT;
195                                 if (flush != Z_NO_FLUSH) {
196                                         /*
197                                          * We left the last 4 bytes in the
198                                          * buffer, in case they are the
199                                          * last 4.  Move them to the front.
200                                          */
201                                         memcpy(tx_strm.next_out,
202                                                obuf+MAX_DATA_COUNT-2, 4);
203                                         tx_strm.next_out += 4;
204                                         tx_strm.avail_out -= 4;
205                                 }
206                         }
207                         if (nb == 0 && token != -2)
208                                 flush = Z_SYNC_FLUSH;
209                         r = deflate(&tx_strm, flush);
210                         if (r != Z_OK) {
211                                 rprintf(FERROR, "deflate returned %d\n", r);
212                                 exit_cleanup(RERR_STREAMIO);
213                         }
214                         if (nb == 0 || tx_strm.avail_out == 0) {
215                                 n = MAX_DATA_COUNT - tx_strm.avail_out;
216                                 if (flush != Z_NO_FLUSH) {
217                                         /*
218                                          * We have to trim off the last 4
219                                          * bytes of output when flushing
220                                          * (they are just 0, 0, ff, ff).
221                                          */
222                                         n -= 4;
223                                 }
224                                 if (n > 0) {
225                                         obuf[0] = DEFLATED_DATA + (n >> 8);
226                                         obuf[1] = n;
227                                         write_buf(f, obuf, n+2);
228                                 }
229                         }
230                 } while (nb != 0 || tx_strm.avail_out == 0);
231                 flush_pending = token == -2;
232         }
233
234         if (token == -1) {
235                 /* end of file - clean up */
236                 write_byte(f, END_FLAG);
237
238         } else if (token != -2) {
239                 /* add the data in the current block to the compressor's
240                    history and hash table */
241                 tx_strm.next_in = (Bytef *) map_ptr(buf, offset, toklen);
242                 tx_strm.avail_in = toklen;
243                 tx_strm.next_out = (Bytef *) obuf;
244                 tx_strm.avail_out = MAX_DATA_COUNT;
245                 r = deflate(&tx_strm, Z_INSERT_ONLY);
246                 if (r != Z_OK || tx_strm.avail_in != 0) {
247                         rprintf(FERROR, "deflate on token returned %d (%d bytes left)\n",
248                                 r, tx_strm.avail_in);
249                         exit_cleanup(RERR_STREAMIO);
250                 }
251         }
252 }
253
254
255 /* tells us what the receiver is in the middle of doing */
256 static enum { r_init, r_idle, r_running, r_inflating, r_inflated } recv_state;
257
258 /* for inflating stuff */
259 static z_stream rx_strm;
260 static char *cbuf;
261 static char *dbuf;
262
263 /* for decoding runs of tokens */
264 static int rx_token;
265 static int rx_run;
266
267 /* Receive a deflated token and inflate it */
268 static int
269 recv_deflated_token(int f, char **data)
270 {
271         int n, r, flag;
272         static int init_done;
273         static int saved_flag;
274
275         for (;;) {
276                 switch (recv_state) {
277                 case r_init:
278                         if (!init_done) {
279                                 rx_strm.next_out = NULL;
280                                 rx_strm.zalloc = NULL;
281                                 rx_strm.zfree = NULL;
282                                 if (inflateInit2(&rx_strm, -15) != Z_OK) {
283                                         rprintf(FERROR, "inflate init failed\n");
284                                         exit_cleanup(RERR_STREAMIO);
285                                 }
286                                 if ((cbuf = malloc(MAX_DATA_COUNT)) == NULL
287                                     || (dbuf = malloc(CHUNK_SIZE)) == NULL)
288                                         out_of_memory("recv_deflated_token");
289                                 init_done = 1;
290                         } else {
291                                 inflateReset(&rx_strm);
292                         }
293                         recv_state = r_idle;
294                         rx_token = 0;
295                         break;
296
297                 case r_idle:
298                 case r_inflated:
299                         if (saved_flag) {
300                                 flag = saved_flag & 0xff;
301                                 saved_flag = 0;
302                         } else
303                                 flag = read_byte(f);
304                         if ((flag & 0xC0) == DEFLATED_DATA) {
305                                 n = ((flag & 0x3f) << 8) + read_byte(f);
306                                 read_buf(f, cbuf, n);
307                                 rx_strm.next_in = (Bytef *)cbuf;
308                                 rx_strm.avail_in = n;
309                                 recv_state = r_inflating;
310                                 break;
311                         }
312                         if (recv_state == r_inflated) {
313                                 /* check previous inflated stuff ended correctly */
314                                 rx_strm.avail_in = 0;
315                                 rx_strm.next_out = (Bytef *)dbuf;
316                                 rx_strm.avail_out = CHUNK_SIZE;
317                                 r = inflate(&rx_strm, Z_SYNC_FLUSH);
318                                 n = CHUNK_SIZE - rx_strm.avail_out;
319                                 /*
320                                  * Z_BUF_ERROR just means no progress was
321                                  * made, i.e. the decompressor didn't have
322                                  * any pending output for us.
323                                  */
324                                 if (r != Z_OK && r != Z_BUF_ERROR) {
325                                         rprintf(FERROR, "inflate flush returned %d (%d bytes)\n",
326                                                 r, n);
327                                         exit_cleanup(RERR_STREAMIO);
328                                 }
329                                 if (n != 0 && r != Z_BUF_ERROR) {
330                                         /* have to return some more data and
331                                            save the flag for later. */
332                                         saved_flag = flag + 0x10000;
333                                         *data = dbuf;
334                                         return n;
335                                 }
336                                 /*
337                                  * At this point the decompressor should
338                                  * be expecting to see the 0, 0, ff, ff bytes.
339                                  */
340                                 if (!inflateSyncPoint(&rx_strm)) {
341                                         rprintf(FERROR, "decompressor lost sync!\n");
342                                         exit_cleanup(RERR_STREAMIO);
343                                 }
344                                 rx_strm.avail_in = 4;
345                                 rx_strm.next_in = (Bytef *)cbuf;
346                                 cbuf[0] = cbuf[1] = 0;
347                                 cbuf[2] = cbuf[3] = 0xff;
348                                 inflate(&rx_strm, Z_SYNC_FLUSH);
349                                 recv_state = r_idle;
350                         }
351                         if (flag == END_FLAG) {
352                                 /* that's all folks */
353                                 recv_state = r_init;
354                                 return 0;
355                         }
356
357                         /* here we have a token of some kind */
358                         if (flag & TOKEN_REL) {
359                                 rx_token += flag & 0x3f;
360                                 flag >>= 6;
361                         } else
362                                 rx_token = read_int(f);
363                         if (flag & 1) {
364                                 rx_run = read_byte(f);
365                                 rx_run += read_byte(f) << 8;
366                                 recv_state = r_running;
367                         }
368                         return -1 - rx_token;
369
370                 case r_inflating:
371                         rx_strm.next_out = (Bytef *)dbuf;
372                         rx_strm.avail_out = CHUNK_SIZE;
373                         r = inflate(&rx_strm, Z_NO_FLUSH);
374                         n = CHUNK_SIZE - rx_strm.avail_out;
375                         if (r != Z_OK) {
376                                 rprintf(FERROR, "inflate returned %d (%d bytes)\n", r, n);
377                                 exit_cleanup(RERR_STREAMIO);
378                         }
379                         if (rx_strm.avail_in == 0)
380                                 recv_state = r_inflated;
381                         if (n != 0) {
382                                 *data = dbuf;
383                                 return n;
384                         }
385                         break;
386
387                 case r_running:
388                         ++rx_token;
389                         if (--rx_run == 0)
390                                 recv_state = r_idle;
391                         return -1 - rx_token;
392                 }
393         }
394 }
395
396 /*
397  * put the data corresponding to a token that we've just returned
398  * from recv_deflated_token into the decompressor's history buffer.
399  */
400 static void see_deflate_token(char *buf, int len)
401 {
402         int r, blklen;
403         unsigned char hdr[5];
404
405         rx_strm.avail_in = 0;
406         blklen = 0;
407         hdr[0] = 0;
408         do {
409                 if (rx_strm.avail_in == 0 && len != 0) {
410                         if (blklen == 0) {
411                                 /* Give it a fake stored-block header. */
412                                 rx_strm.next_in = (Bytef *)hdr;
413                                 rx_strm.avail_in = 5;
414                                 blklen = len;
415                                 if (blklen > 0xffff)
416                                         blklen = 0xffff;
417                                 hdr[1] = blklen;
418                                 hdr[2] = blklen >> 8;
419                                 hdr[3] = ~hdr[1];
420                                 hdr[4] = ~hdr[2];
421                         } else {
422                                 rx_strm.next_in = (Bytef *)buf;
423                                 rx_strm.avail_in = blklen;
424                                 len -= blklen;
425                                 blklen = 0;
426                         }
427                 }
428                 rx_strm.next_out = (Bytef *)dbuf;
429                 rx_strm.avail_out = CHUNK_SIZE;
430                 r = inflate(&rx_strm, Z_SYNC_FLUSH);
431                 if (r != Z_OK) {
432                         rprintf(FERROR, "inflate (token) returned %d\n", r);
433                         exit_cleanup(RERR_STREAMIO);
434                 }
435         } while (len || rx_strm.avail_out == 0);
436 }
437
438 /*
439  * transmit a verbatim buffer of length n followed by a token 
440  * If token == -1 then we have reached EOF 
441  * If n == 0 then don't send a buffer
442  */
443 void send_token(int f,int token,struct map_struct *buf,OFF_T offset,
444                 int n,int toklen)
445 {
446         if (!do_compression) {
447                 simple_send_token(f,token,buf,offset,n);
448         } else {
449                 send_deflated_token(f, token, buf, offset, n, toklen);
450         }
451 }
452
453
454 /*
455  * receive a token or buffer from the other end. If the reurn value is >0 then
456  * it is a data buffer of that length, and *data will point at the data.
457  * if the return value is -i then it represents token i-1
458  * if the return value is 0 then the end has been reached
459  */
460 int recv_token(int f,char **data)
461 {
462         int tok;
463
464         if (!do_compression) {
465                 tok = simple_recv_token(f,data);
466         } else {
467                 tok = recv_deflated_token(f, data);
468         }
469         return tok;
470 }
471
472 /*
473  * look at the data corresponding to a token, if necessary
474  */
475 void see_token(char *data, int toklen)
476 {
477         if (do_compression)
478                 see_deflate_token(data, toklen);
479 }