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 extern int module_id;
  25
  26 static int compression_level = Z_DEFAULT_COMPRESSION;
  27
  28 /* determine the compression level based on a wildcard filename list */
  29 void set_compression(char *fname)
  30 {
  31         char *dont;
  32         char *tok;
  33
  34         if (!do_compression)
  35                 return;
  36
  37         compression_level = Z_DEFAULT_COMPRESSION;
  38         dont = lp_dont_compress(module_id);
  39
  40         if (!dont || !*dont)
  41                 return;
  42
  43         if (dont[0] == '*' && !dont[1]) {
  44                 /* an optimization to skip the rest of this routine */
  45                 compression_level = 0;
  46                 return;
  47         }
  48
  49         dont = strdup(dont);
  50         fname = strdup(fname);
  51         if (!dont || !fname)
  52                 return;
  53
  54         strlower(dont);
  55         strlower(fname);
  56
  57         for (tok = strtok(dont, " "); tok; tok = strtok(NULL, " ")) {
  58                 if (wildmatch(tok, fname)) {
  59                         compression_level = 0;
  60                         break;
  61                 }
  62         }
  63         free(dont);
  64         free(fname);
  65 }
  66
  67 /* non-compressing recv token */
  68 static int simple_recv_token(int f,char **data)
  69 {
  70         static int residue;
  71         static char *buf;
  72         int n;
  73
  74         if (!buf) {
  75                 buf = new_array(char, CHUNK_SIZE);
  76                 if (!buf)
  77                         out_of_memory("simple_recv_token");
  78         }
  79
  80         if (residue == 0) {
  81                 int i = read_int(f);
  82                 if (i <= 0)
  83                         return i;
  84                 residue = i;
  85         }
  86
  87         *data = buf;
  88         n = MIN(CHUNK_SIZE,residue);
  89         residue -= n;
  90         read_buf(f,buf,n);
  91         return n;
  92 }
  93
  94
  95 /* non-compressing send token */
  96 static void simple_send_token(int f,int token,
  97                               struct map_struct *buf,OFF_T offset,int n)
  98 {
  99         if (n > 0) {
 100                 int l = 0;
 101                 while (l < n) {
 102                         int n1 = MIN(CHUNK_SIZE,n-l);
 103                         write_int(f,n1);
 104                         write_buf(f,map_ptr(buf,offset+l,n1),n1);
 105                         l += n1;
 106                 }
 107         }
 108         /* a -2 token means to send data only and no token */
 109         if (token != -2) {
 110                 write_int(f,-(token+1));
 111         }
 112 }
 113
 114
 115 /* Flag bytes in compressed stream are encoded as follows: */
 116 #define END_FLAG        0       /* that's all folks */
 117 #define TOKEN_LONG      0x20    /* followed by 32-bit token number */
 118 #define TOKENRUN_LONG   0x21    /* ditto with 16-bit run count */
 119 #define DEFLATED_DATA   0x40    /* + 6-bit high len, then low len byte */
 120 #define TOKEN_REL       0x80    /* + 6-bit relative token number */
 121 #define TOKENRUN_REL    0xc0    /* ditto with 16-bit run count */
 122
 123 #define MAX_DATA_COUNT  16383   /* fit 14 bit count into 2 bytes with flags */
 124
 125 /* zlib.h says that if we want to be able to compress something in a single
 126  * call, avail_out must be at least 0.1% larger than avail_in plus 12 bytes.
 127  * We'll add in 0.1%+16, just to be safe (and we'll avoid floating point,
 128  * to ensure that this is a compile-time value). */
 129 #define AVAIL_OUT_SIZE(avail_in_size) ((avail_in_size)*1001/1000+16)
 130
 131 /* For coding runs of tokens */
 132 static int last_token = -1;
 133 static int run_start;
 134 static int last_run_end;
 135
 136 /* Deflation state */
 137 static z_stream tx_strm;
 138
 139 /* Output buffer */
 140 static char *obuf;
 141
 142 /* We want obuf to be able to hold both MAX_DATA_COUNT+2 bytes as well as
 143  * AVAIL_OUT_SIZE(CHUNK_SIZE) bytes, so make sure that it's large enough. */
 144 #if MAX_DATA_COUNT+2 > AVAIL_OUT_SIZE(CHUNK_SIZE)
 145 #define OBUF_SIZE       (MAX_DATA_COUNT+2)
 146 #else
 147 #define OBUF_SIZE       AVAIL_OUT_SIZE(CHUNK_SIZE)
 148 #endif
 149
 150 /* Send a deflated token */
 151 static void
 152 send_deflated_token(int f, int token,
 153                     struct map_struct *buf, OFF_T offset, int nb, int toklen)
 154 {
 155         int n, r;
 156         static int init_done, flush_pending;
 157
 158         if (last_token == -1) {
 159                 /* initialization */
 160                 if (!init_done) {
 161                         tx_strm.next_in = NULL;
 162                         tx_strm.zalloc = NULL;
 163                         tx_strm.zfree = NULL;
 164                         if (deflateInit2(&tx_strm, compression_level,
 165                                          Z_DEFLATED, -15, 8,
 166                                          Z_DEFAULT_STRATEGY) != Z_OK) {
 167                                 rprintf(FERROR, "compression init failed\n");
 168                                 exit_cleanup(RERR_STREAMIO);
 169                         }
 170                         if ((obuf = new_array(char, OBUF_SIZE)) == NULL)
 171                                 out_of_memory("send_deflated_token");
 172                         init_done = 1;
 173                 } else
 174                         deflateReset(&tx_strm);
 175                 last_run_end = 0;
 176                 run_start = token;
 177                 flush_pending = 0;
 178
 179         } else if (last_token == -2) {
 180                 run_start = token;
 181
 182         } else if (nb != 0 || token != last_token + 1
 183                    || token >= run_start + 65536) {
 184                 /* output previous run */
 185                 r = run_start - last_run_end;
 186                 n = last_token - run_start;
 187                 if (r >= 0 && r <= 63) {
 188                         write_byte(f, (n==0? TOKEN_REL: TOKENRUN_REL) + r);
 189                 } else {
 190                         write_byte(f, (n==0? TOKEN_LONG: TOKENRUN_LONG));
 191                         write_int(f, run_start);
 192                 }
 193                 if (n != 0) {
 194                         write_byte(f, n);
 195                         write_byte(f, n >> 8);
 196                 }
 197                 last_run_end = last_token;
 198                 run_start = token;
 199         }
 200
 201         last_token = token;
 202
 203         if (nb != 0 || flush_pending) {
 204                 /* deflate the data starting at offset */
 205                 int flush = Z_NO_FLUSH;
 206                 tx_strm.avail_in = 0;
 207                 tx_strm.avail_out = 0;
 208                 do {
 209                         if (tx_strm.avail_in == 0 && nb != 0) {
 210                                 /* give it some more input */
 211                                 n = MIN(nb, CHUNK_SIZE);
 212                                 tx_strm.next_in = (Bytef *)
 213                                         map_ptr(buf, offset, n);
 214                                 tx_strm.avail_in = n;
 215                                 nb -= n;
 216                                 offset += n;
 217                         }
 218                         if (tx_strm.avail_out == 0) {
 219                                 tx_strm.next_out = (Bytef *)(obuf + 2);
 220                                 tx_strm.avail_out = MAX_DATA_COUNT;
 221                                 if (flush != Z_NO_FLUSH) {
 222                                         /*
 223                                          * We left the last 4 bytes in the
 224                                          * buffer, in case they are the
 225                                          * last 4.  Move them to the front.
 226                                          */
 227                                         memcpy(tx_strm.next_out,
 228                                                obuf+MAX_DATA_COUNT-2, 4);
 229                                         tx_strm.next_out += 4;
 230                                         tx_strm.avail_out -= 4;
 231                                 }
 232                         }
 233                         if (nb == 0 && token != -2)
 234                                 flush = Z_SYNC_FLUSH;
 235                         r = deflate(&tx_strm, flush);
 236                         if (r != Z_OK) {
 237                                 rprintf(FERROR, "deflate returned %d\n", r);
 238                                 exit_cleanup(RERR_STREAMIO);
 239                         }
 240                         if (nb == 0 || tx_strm.avail_out == 0) {
 241                                 n = MAX_DATA_COUNT - tx_strm.avail_out;
 242                                 if (flush != Z_NO_FLUSH) {
 243                                         /*
 244                                          * We have to trim off the last 4
 245                                          * bytes of output when flushing
 246                                          * (they are just 0, 0, ff, ff).
 247                                          */
 248                                         n -= 4;
 249                                 }
 250                                 if (n > 0) {
 251                                         obuf[0] = DEFLATED_DATA + (n >> 8);
 252                                         obuf[1] = n;
 253                                         write_buf(f, obuf, n+2);
 254                                 }
 255                         }
 256                 } while (nb != 0 || tx_strm.avail_out == 0);
 257                 flush_pending = token == -2;
 258         }
 259
 260         if (token == -1) {
 261                 /* end of file - clean up */
 262                 write_byte(f, END_FLAG);
 263         } else if (token != -2) {
 264                 /* add the data in the current block to the compressor's
 265                    history and hash table */
 266                 tx_strm.next_in = (Bytef *) map_ptr(buf, offset, toklen);
 267                 tx_strm.avail_in = toklen;
 268                 tx_strm.next_out = (Bytef *) obuf;
 269                 tx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 270                 r = deflate(&tx_strm, Z_INSERT_ONLY);
 271                 if (r != Z_OK || tx_strm.avail_in != 0) {
 272                         rprintf(FERROR, "deflate on token returned %d (%d bytes left)\n",
 273                                 r, tx_strm.avail_in);
 274                         exit_cleanup(RERR_STREAMIO);
 275                 }
 276         }
 277 }
 278
 279
 280 /* tells us what the receiver is in the middle of doing */
 281 static enum { r_init, r_idle, r_running, r_inflating, r_inflated } recv_state;
 282
 283 /* for inflating stuff */
 284 static z_stream rx_strm;
 285 static char *cbuf;
 286 static char *dbuf;
 287
 288 /* for decoding runs of tokens */
 289 static int rx_token;
 290 static int rx_run;
 291
 292 /* Receive a deflated token and inflate it */
 293 static int
 294 recv_deflated_token(int f, char **data)
 295 {
 296         int n, r, flag;
 297         static int init_done;
 298         static int saved_flag;
 299
 300         for (;;) {
 301                 switch (recv_state) {
 302                 case r_init:
 303                         if (!init_done) {
 304                                 rx_strm.next_out = NULL;
 305                                 rx_strm.zalloc = NULL;
 306                                 rx_strm.zfree = NULL;
 307                                 if (inflateInit2(&rx_strm, -15) != Z_OK) {
 308                                         rprintf(FERROR, "inflate init failed\n");
 309                                         exit_cleanup(RERR_STREAMIO);
 310                                 }
 311                                 if (!(cbuf = new_array(char, MAX_DATA_COUNT))
 312                                     || !(dbuf = new_array(char, AVAIL_OUT_SIZE(CHUNK_SIZE))))
 313                                         out_of_memory("recv_deflated_token");
 314                                 init_done = 1;
 315                         } else {
 316                                 inflateReset(&rx_strm);
 317                         }
 318                         recv_state = r_idle;
 319                         rx_token = 0;
 320                         break;
 321
 322                 case r_idle:
 323                 case r_inflated:
 324                         if (saved_flag) {
 325                                 flag = saved_flag & 0xff;
 326                                 saved_flag = 0;
 327                         } else
 328                                 flag = read_byte(f);
 329                         if ((flag & 0xC0) == DEFLATED_DATA) {
 330                                 n = ((flag & 0x3f) << 8) + read_byte(f);
 331                                 read_buf(f, cbuf, n);
 332                                 rx_strm.next_in = (Bytef *)cbuf;
 333                                 rx_strm.avail_in = n;
 334                                 recv_state = r_inflating;
 335                                 break;
 336                         }
 337                         if (recv_state == r_inflated) {
 338                                 /* check previous inflated stuff ended correctly */
 339                                 rx_strm.avail_in = 0;
 340                                 rx_strm.next_out = (Bytef *)dbuf;
 341                                 rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 342                                 r = inflate(&rx_strm, Z_SYNC_FLUSH);
 343                                 n = AVAIL_OUT_SIZE(CHUNK_SIZE) - rx_strm.avail_out;
 344                                 /*
 345                                  * Z_BUF_ERROR just means no progress was
 346                                  * made, i.e. the decompressor didn't have
 347                                  * any pending output for us.
 348                                  */
 349                                 if (r != Z_OK && r != Z_BUF_ERROR) {
 350                                         rprintf(FERROR, "inflate flush returned %d (%d bytes)\n",
 351                                                 r, n);
 352                                         exit_cleanup(RERR_STREAMIO);
 353                                 }
 354                                 if (n != 0 && r != Z_BUF_ERROR) {
 355                                         /* have to return some more data and
 356                                            save the flag for later. */
 357                                         saved_flag = flag + 0x10000;
 358                                         *data = dbuf;
 359                                         return n;
 360                                 }
 361                                 /*
 362                                  * At this point the decompressor should
 363                                  * be expecting to see the 0, 0, ff, ff bytes.
 364                                  */
 365                                 if (!inflateSyncPoint(&rx_strm)) {
 366                                         rprintf(FERROR, "decompressor lost sync!\n");
 367                                         exit_cleanup(RERR_STREAMIO);
 368                                 }
 369                                 rx_strm.avail_in = 4;
 370                                 rx_strm.next_in = (Bytef *)cbuf;
 371                                 cbuf[0] = cbuf[1] = 0;
 372                                 cbuf[2] = cbuf[3] = 0xff;
 373                                 inflate(&rx_strm, Z_SYNC_FLUSH);
 374                                 recv_state = r_idle;
 375                         }
 376                         if (flag == END_FLAG) {
 377                                 /* that's all folks */
 378                                 recv_state = r_init;
 379                                 return 0;
 380                         }
 381
 382                         /* here we have a token of some kind */
 383                         if (flag & TOKEN_REL) {
 384                                 rx_token += flag & 0x3f;
 385                                 flag >>= 6;
 386                         } else
 387                                 rx_token = read_int(f);
 388                         if (flag & 1) {
 389                                 rx_run = read_byte(f);
 390                                 rx_run += read_byte(f) << 8;
 391                                 recv_state = r_running;
 392                         }
 393                         return -1 - rx_token;
 394
 395                 case r_inflating:
 396                         rx_strm.next_out = (Bytef *)dbuf;
 397                         rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 398                         r = inflate(&rx_strm, Z_NO_FLUSH);
 399                         n = AVAIL_OUT_SIZE(CHUNK_SIZE) - rx_strm.avail_out;
 400                         if (r != Z_OK) {
 401                                 rprintf(FERROR, "inflate returned %d (%d bytes)\n", r, n);
 402                                 exit_cleanup(RERR_STREAMIO);
 403                         }
 404                         if (rx_strm.avail_in == 0)
 405                                 recv_state = r_inflated;
 406                         if (n != 0) {
 407                                 *data = dbuf;
 408                                 return n;
 409                         }
 410                         break;
 411
 412                 case r_running:
 413                         ++rx_token;
 414                         if (--rx_run == 0)
 415                                 recv_state = r_idle;
 416                         return -1 - rx_token;
 417                 }
 418         }
 419 }
 420
 421 /*
 422  * put the data corresponding to a token that we've just returned
 423  * from recv_deflated_token into the decompressor's history buffer.
 424  */
 425 static void see_deflate_token(char *buf, int len)
 426 {
 427         int r, blklen;
 428         unsigned char hdr[5];
 429
 430         rx_strm.avail_in = 0;
 431         blklen = 0;
 432         hdr[0] = 0;
 433         do {
 434                 if (rx_strm.avail_in == 0 && len != 0) {
 435                         if (blklen == 0) {
 436                                 /* Give it a fake stored-block header. */
 437                                 rx_strm.next_in = (Bytef *)hdr;
 438                                 rx_strm.avail_in = 5;
 439                                 blklen = len;
 440                                 if (blklen > 0xffff)
 441                                         blklen = 0xffff;
 442                                 hdr[1] = blklen;
 443                                 hdr[2] = blklen >> 8;
 444                                 hdr[3] = ~hdr[1];
 445                                 hdr[4] = ~hdr[2];
 446                         } else {
 447                                 rx_strm.next_in = (Bytef *)buf;
 448                                 rx_strm.avail_in = blklen;
 449                                 len -= blklen;
 450                                 blklen = 0;
 451                         }
 452                 }
 453                 rx_strm.next_out = (Bytef *)dbuf;
 454                 rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 455                 r = inflate(&rx_strm, Z_SYNC_FLUSH);
 456                 if (r != Z_OK) {
 457                         rprintf(FERROR, "inflate (token) returned %d\n", r);
 458                         exit_cleanup(RERR_STREAMIO);
 459                 }
 460         } while (len || rx_strm.avail_out == 0);
 461 }
 462
 463 /**
 464  * Transmit a verbatim buffer of length @p n followed by a token.
 465  * If token == -1 then we have reached EOF
 466  * If n == 0 then don't send a buffer
 467  */
 468 void send_token(int f,int token,struct map_struct *buf,OFF_T offset,
 469                 int n,int toklen)
 470 {
 471         if (!do_compression) {
 472                 simple_send_token(f,token,buf,offset,n);
 473         } else {
 474                 send_deflated_token(f, token, buf, offset, n, toklen);
 475         }
 476 }
 477
 478
 479 /*
 480  * receive a token or buffer from the other end. If the reurn value is >0 then
 481  * it is a data buffer of that length, and *data will point at the data.
 482  * if the return value is -i then it represents token i-1
 483  * if the return value is 0 then the end has been reached
 484  */
 485 int recv_token(int f,char **data)
 486 {
 487         int tok;
 488
 489         if (!do_compression) {
 490                 tok = simple_recv_token(f,data);
 491         } else {
 492                 tok = recv_deflated_token(f, data);
 493         }
 494         return tok;
 495 }
 496
 497 /*
 498  * look at the data corresponding to a token, if necessary
 499  */
 500 void see_token(char *data, int toklen)
 501 {
 502         if (do_compression)
 503                 see_deflate_token(data, toklen);
 504 }