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 int32 simple_recv_token(int f, char **data)
  69 {
  70         static int residue;
  71         static char *buf;
  72         int32 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                 int32 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, struct map_struct *buf,
  97                               OFF_T offset, int32 n)
  98 {
  99         if (n > 0) {
 100                 int32 len = 0;
 101                 while (len < n) {
 102                         int32 n1 = MIN(CHUNK_SIZE, n-len);
 103                         write_int(f, n1);
 104                         write_buf(f, map_ptr(buf, offset+len, n1), n1);
 105                         len += 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 /* Flag bytes in compressed stream are encoded as follows: */
 115 #define END_FLAG        0       /* that's all folks */
 116 #define TOKEN_LONG      0x20    /* followed by 32-bit token number */
 117 #define TOKENRUN_LONG   0x21    /* ditto with 16-bit run count */
 118 #define DEFLATED_DATA   0x40    /* + 6-bit high len, then low len byte */
 119 #define TOKEN_REL       0x80    /* + 6-bit relative token number */
 120 #define TOKENRUN_REL    0xc0    /* ditto with 16-bit run count */
 121
 122 #define MAX_DATA_COUNT  16383   /* fit 14 bit count into 2 bytes with flags */
 123
 124 /* zlib.h says that if we want to be able to compress something in a single
 125  * call, avail_out must be at least 0.1% larger than avail_in plus 12 bytes.
 126  * We'll add in 0.1%+16, just to be safe (and we'll avoid floating point,
 127  * to ensure that this is a compile-time value). */
 128 #define AVAIL_OUT_SIZE(avail_in_size) ((avail_in_size)*1001/1000+16)
 129
 130 /* For coding runs of tokens */
 131 static int last_token = -1;
 132 static int run_start;
 133 static int last_run_end;
 134
 135 /* Deflation state */
 136 static z_stream tx_strm;
 137
 138 /* Output buffer */
 139 static char *obuf;
 140
 141 /* We want obuf to be able to hold both MAX_DATA_COUNT+2 bytes as well as
 142  * AVAIL_OUT_SIZE(CHUNK_SIZE) bytes, so make sure that it's large enough. */
 143 #if MAX_DATA_COUNT+2 > AVAIL_OUT_SIZE(CHUNK_SIZE)
 144 #define OBUF_SIZE       (MAX_DATA_COUNT+2)
 145 #else
 146 #define OBUF_SIZE       AVAIL_OUT_SIZE(CHUNK_SIZE)
 147 #endif
 148
 149 /* Send a deflated token */
 150 static void
 151 send_deflated_token(int f, int token, struct map_struct *buf, OFF_T offset,
 152                     int32 nb, int32 toklen)
 153 {
 154         int32 n, r;
 155         static int init_done, flush_pending;
 156
 157         if (last_token == -1) {
 158                 /* initialization */
 159                 if (!init_done) {
 160                         tx_strm.next_in = NULL;
 161                         tx_strm.zalloc = NULL;
 162                         tx_strm.zfree = NULL;
 163                         if (deflateInit2(&tx_strm, compression_level,
 164                                          Z_DEFLATED, -15, 8,
 165                                          Z_DEFAULT_STRATEGY) != Z_OK) {
 166                                 rprintf(FERROR, "compression init failed\n");
 167                                 exit_cleanup(RERR_STREAMIO);
 168                         }
 169                         if ((obuf = new_array(char, OBUF_SIZE)) == NULL)
 170                                 out_of_memory("send_deflated_token");
 171                         init_done = 1;
 172                 } else
 173                         deflateReset(&tx_strm);
 174                 last_run_end = 0;
 175                 run_start = token;
 176                 flush_pending = 0;
 177
 178         } else if (last_token == -2) {
 179                 run_start = token;
 180
 181         } else if (nb != 0 || token != last_token + 1
 182                    || token >= run_start + 65536) {
 183                 /* output previous run */
 184                 r = run_start - last_run_end;
 185                 n = last_token - run_start;
 186                 if (r >= 0 && r <= 63) {
 187                         write_byte(f, (n==0? TOKEN_REL: TOKENRUN_REL) + r);
 188                 } else {
 189                         write_byte(f, (n==0? TOKEN_LONG: TOKENRUN_LONG));
 190                         write_int(f, run_start);
 191                 }
 192                 if (n != 0) {
 193                         write_byte(f, n);
 194                         write_byte(f, n >> 8);
 195                 }
 196                 last_run_end = last_token;
 197                 run_start = token;
 198         }
 199
 200         last_token = token;
 201
 202         if (nb != 0 || flush_pending) {
 203                 /* deflate the data starting at offset */
 204                 int flush = Z_NO_FLUSH;
 205                 tx_strm.avail_in = 0;
 206                 tx_strm.avail_out = 0;
 207                 do {
 208                         if (tx_strm.avail_in == 0 && nb != 0) {
 209                                 /* give it some more input */
 210                                 n = MIN(nb, CHUNK_SIZE);
 211                                 tx_strm.next_in = (Bytef *)
 212                                         map_ptr(buf, offset, n);
 213                                 tx_strm.avail_in = n;
 214                                 nb -= n;
 215                                 offset += n;
 216                         }
 217                         if (tx_strm.avail_out == 0) {
 218                                 tx_strm.next_out = (Bytef *)(obuf + 2);
 219                                 tx_strm.avail_out = MAX_DATA_COUNT;
 220                                 if (flush != Z_NO_FLUSH) {
 221                                         /*
 222                                          * We left the last 4 bytes in the
 223                                          * buffer, in case they are the
 224                                          * last 4.  Move them to the front.
 225                                          */
 226                                         memcpy(tx_strm.next_out,
 227                                                obuf+MAX_DATA_COUNT-2, 4);
 228                                         tx_strm.next_out += 4;
 229                                         tx_strm.avail_out -= 4;
 230                                 }
 231                         }
 232                         if (nb == 0 && token != -2)
 233                                 flush = Z_SYNC_FLUSH;
 234                         r = deflate(&tx_strm, flush);
 235                         if (r != Z_OK) {
 236                                 rprintf(FERROR, "deflate returned %d\n", r);
 237                                 exit_cleanup(RERR_STREAMIO);
 238                         }
 239                         if (nb == 0 || tx_strm.avail_out == 0) {
 240                                 n = MAX_DATA_COUNT - tx_strm.avail_out;
 241                                 if (flush != Z_NO_FLUSH) {
 242                                         /*
 243                                          * We have to trim off the last 4
 244                                          * bytes of output when flushing
 245                                          * (they are just 0, 0, ff, ff).
 246                                          */
 247                                         n -= 4;
 248                                 }
 249                                 if (n > 0) {
 250                                         obuf[0] = DEFLATED_DATA + (n >> 8);
 251                                         obuf[1] = n;
 252                                         write_buf(f, obuf, n+2);
 253                                 }
 254                         }
 255                 } while (nb != 0 || tx_strm.avail_out == 0);
 256                 flush_pending = token == -2;
 257         }
 258
 259         if (token == -1) {
 260                 /* end of file - clean up */
 261                 write_byte(f, END_FLAG);
 262         } else if (token != -2) {
 263                 /* add the data in the current block to the compressor's
 264                    history and hash table */
 265                 tx_strm.next_in = (Bytef *) map_ptr(buf, offset, toklen);
 266                 tx_strm.avail_in = toklen;
 267                 do {
 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) {
 272                                 rprintf(FERROR,
 273                                         "deflate on token returned %d (%d bytes left)\n",
 274                                         r, tx_strm.avail_in);
 275                                 exit_cleanup(RERR_STREAMIO);
 276                         }
 277                 } while (tx_strm.avail_in != 0);
 278         }
 279 }
 280
 281
 282 /* tells us what the receiver is in the middle of doing */
 283 static enum { r_init, r_idle, r_running, r_inflating, r_inflated } recv_state;
 284
 285 /* for inflating stuff */
 286 static z_stream rx_strm;
 287 static char *cbuf;
 288 static char *dbuf;
 289
 290 /* for decoding runs of tokens */
 291 static int32 rx_token;
 292 static int32 rx_run;
 293
 294 /* Receive a deflated token and inflate it */
 295 static int32 recv_deflated_token(int f, char **data)
 296 {
 297         static int init_done;
 298         static int saved_flag;
 299         int r, flag;
 300         int32 n;
 301
 302         for (;;) {
 303                 switch (recv_state) {
 304                 case r_init:
 305                         if (!init_done) {
 306                                 rx_strm.next_out = NULL;
 307                                 rx_strm.zalloc = NULL;
 308                                 rx_strm.zfree = NULL;
 309                                 if (inflateInit2(&rx_strm, -15) != Z_OK) {
 310                                         rprintf(FERROR, "inflate init failed\n");
 311                                         exit_cleanup(RERR_STREAMIO);
 312                                 }
 313                                 if (!(cbuf = new_array(char, MAX_DATA_COUNT))
 314                                     || !(dbuf = new_array(char, AVAIL_OUT_SIZE(CHUNK_SIZE))))
 315                                         out_of_memory("recv_deflated_token");
 316                                 init_done = 1;
 317                         } else {
 318                                 inflateReset(&rx_strm);
 319                         }
 320                         recv_state = r_idle;
 321                         rx_token = 0;
 322                         break;
 323
 324                 case r_idle:
 325                 case r_inflated:
 326                         if (saved_flag) {
 327                                 flag = saved_flag & 0xff;
 328                                 saved_flag = 0;
 329                         } else
 330                                 flag = read_byte(f);
 331                         if ((flag & 0xC0) == DEFLATED_DATA) {
 332                                 n = ((flag & 0x3f) << 8) + read_byte(f);
 333                                 read_buf(f, cbuf, n);
 334                                 rx_strm.next_in = (Bytef *)cbuf;
 335                                 rx_strm.avail_in = n;
 336                                 recv_state = r_inflating;
 337                                 break;
 338                         }
 339                         if (recv_state == r_inflated) {
 340                                 /* check previous inflated stuff ended correctly */
 341                                 rx_strm.avail_in = 0;
 342                                 rx_strm.next_out = (Bytef *)dbuf;
 343                                 rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 344                                 r = inflate(&rx_strm, Z_SYNC_FLUSH);
 345                                 n = AVAIL_OUT_SIZE(CHUNK_SIZE) - rx_strm.avail_out;
 346                                 /*
 347                                  * Z_BUF_ERROR just means no progress was
 348                                  * made, i.e. the decompressor didn't have
 349                                  * any pending output for us.
 350                                  */
 351                                 if (r != Z_OK && r != Z_BUF_ERROR) {
 352                                         rprintf(FERROR, "inflate flush returned %d (%d bytes)\n",
 353                                                 r, n);
 354                                         exit_cleanup(RERR_STREAMIO);
 355                                 }
 356                                 if (n != 0 && r != Z_BUF_ERROR) {
 357                                         /* have to return some more data and
 358                                            save the flag for later. */
 359                                         saved_flag = flag + 0x10000;
 360                                         *data = dbuf;
 361                                         return n;
 362                                 }
 363                                 /*
 364                                  * At this point the decompressor should
 365                                  * be expecting to see the 0, 0, ff, ff bytes.
 366                                  */
 367                                 if (!inflateSyncPoint(&rx_strm)) {
 368                                         rprintf(FERROR, "decompressor lost sync!\n");
 369                                         exit_cleanup(RERR_STREAMIO);
 370                                 }
 371                                 rx_strm.avail_in = 4;
 372                                 rx_strm.next_in = (Bytef *)cbuf;
 373                                 cbuf[0] = cbuf[1] = 0;
 374                                 cbuf[2] = cbuf[3] = 0xff;
 375                                 inflate(&rx_strm, Z_SYNC_FLUSH);
 376                                 recv_state = r_idle;
 377                         }
 378                         if (flag == END_FLAG) {
 379                                 /* that's all folks */
 380                                 recv_state = r_init;
 381                                 return 0;
 382                         }
 383
 384                         /* here we have a token of some kind */
 385                         if (flag & TOKEN_REL) {
 386                                 rx_token += flag & 0x3f;
 387                                 flag >>= 6;
 388                         } else
 389                                 rx_token = read_int(f);
 390                         if (flag & 1) {
 391                                 rx_run = read_byte(f);
 392                                 rx_run += read_byte(f) << 8;
 393                                 recv_state = r_running;
 394                         }
 395                         return -1 - rx_token;
 396
 397                 case r_inflating:
 398                         rx_strm.next_out = (Bytef *)dbuf;
 399                         rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 400                         r = inflate(&rx_strm, Z_NO_FLUSH);
 401                         n = AVAIL_OUT_SIZE(CHUNK_SIZE) - rx_strm.avail_out;
 402                         if (r != Z_OK) {
 403                                 rprintf(FERROR, "inflate returned %d (%d bytes)\n", r, n);
 404                                 exit_cleanup(RERR_STREAMIO);
 405                         }
 406                         if (rx_strm.avail_in == 0)
 407                                 recv_state = r_inflated;
 408                         if (n != 0) {
 409                                 *data = dbuf;
 410                                 return n;
 411                         }
 412                         break;
 413
 414                 case r_running:
 415                         ++rx_token;
 416                         if (--rx_run == 0)
 417                                 recv_state = r_idle;
 418                         return -1 - rx_token;
 419                 }
 420         }
 421 }
 422
 423 /*
 424  * put the data corresponding to a token that we've just returned
 425  * from recv_deflated_token into the decompressor's history buffer.
 426  */
 427 static void see_deflate_token(char *buf, int32 len)
 428 {
 429         int r, blklen;
 430         unsigned char hdr[5];
 431
 432         rx_strm.avail_in = 0;
 433         blklen = 0;
 434         hdr[0] = 0;
 435         do {
 436                 if (rx_strm.avail_in == 0 && len != 0) {
 437                         if (blklen == 0) {
 438                                 /* Give it a fake stored-block header. */
 439                                 rx_strm.next_in = (Bytef *)hdr;
 440                                 rx_strm.avail_in = 5;
 441                                 blklen = len;
 442                                 if (blklen > 0xffff)
 443                                         blklen = 0xffff;
 444                                 hdr[1] = blklen;
 445                                 hdr[2] = blklen >> 8;
 446                                 hdr[3] = ~hdr[1];
 447                                 hdr[4] = ~hdr[2];
 448                         } else {
 449                                 rx_strm.next_in = (Bytef *)buf;
 450                                 rx_strm.avail_in = blklen;
 451                                 len -= blklen;
 452                                 blklen = 0;
 453                         }
 454                 }
 455                 rx_strm.next_out = (Bytef *)dbuf;
 456                 rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
 457                 r = inflate(&rx_strm, Z_SYNC_FLUSH);
 458                 if (r != Z_OK) {
 459                         rprintf(FERROR, "inflate (token) returned %d\n", r);
 460                         exit_cleanup(RERR_STREAMIO);
 461                 }
 462         } while (len || rx_strm.avail_out == 0);
 463 }
 464
 465 /**
 466  * Transmit a verbatim buffer of length @p n followed by a token.
 467  * If token == -1 then we have reached EOF
 468  * If n == 0 then don't send a buffer
 469  */
 470 void send_token(int f, int token, struct map_struct *buf, OFF_T offset,
 471                 int32 n, int32 toklen)
 472 {
 473         if (!do_compression)
 474                 simple_send_token(f, token, buf, offset, n);
 475         else
 476                 send_deflated_token(f, token, buf, offset, n, toklen);
 477 }
 478
 479
 480 /*
 481  * receive a token or buffer from the other end. If the reurn value is >0 then
 482  * it is a data buffer of that length, and *data will point at the data.
 483  * if the return value is -i then it represents token i-1
 484  * if the return value is 0 then the end has been reached
 485  */
 486 int32 recv_token(int f, char **data)
 487 {
 488         int tok;
 489
 490         if (!do_compression) {
 491                 tok = simple_recv_token(f,data);
 492         } else {
 493                 tok = recv_deflated_token(f, data);
 494         }
 495         return tok;
 496 }
 497
 498 /*
 499  * look at the data corresponding to a token, if necessary
 500  */
 501 void see_token(char *data, int32 toklen)
 502 {
 503         if (do_compression)
 504                 see_deflate_token(data, toklen);
 505 }