| 1 |
// For all support, instructions and copyright go to: |
| 2 |
// http://e2guardian.org/ |
| 3 |
// Released under the GPL v2, with the OpenSSL exception described in the README file. |
| 4 |
|
| 5 |
// INCLUDES |
| 6 |
|
| 7 |
#ifdef HAVE_CONFIG_H |
| 8 |
#include "e2config.h" |
| 9 |
#endif |
| 10 |
#include "DataBuffer.hpp" |
| 11 |
#include "HTTPHeader.hpp" |
| 12 |
#include "OptionContainer.hpp" |
| 13 |
#include "Logger.hpp" |
| 14 |
|
| 15 |
#include <sys/stat.h> |
| 16 |
#include <algorithm> |
| 17 |
#include <cstdlib> |
| 18 |
#include <unistd.h> |
| 19 |
#include <zlib.h> |
| 20 |
#include <cerrno> |
| 21 |
#include <fstream> |
| 22 |
#include <sys/time.h> |
| 23 |
#include <queue> |
| 24 |
#include <istream> |
| 25 |
|
| 26 |
// DEFINES |
| 27 |
|
| 28 |
#define __E2HEADER_SENDALL 0 |
| 29 |
#define __E2HEADER_SENDFIRSTLINE 1 |
| 30 |
#define __E2HEADER_SENDREST 2 |
| 31 |
|
| 32 |
// GLOBALS |
| 33 |
|
| 34 |
extern OptionContainer o; |
| 35 |
|
| 36 |
// IMPLEMENTATION |
| 37 |
|
| 38 |
DataBuffer::DataBuffer() |
| 39 |
: data(new char[1]), data_length(0) |
| 40 |
{ |
| 41 |
data[0] = '\0'; |
| 42 |
} |
| 43 |
|
| 44 |
DataBuffer::DataBuffer(const void *indata, off_t length) //not used! PIP |
| 45 |
: data(new char[length]), data_length(length) |
| 46 |
{ |
| 47 |
memcpy(data, indata, length); |
| 48 |
} |
| 49 |
|
| 50 |
void DataBuffer::reset() |
| 51 |
{ |
| 52 |
delete[] data; |
| 53 |
delete[] compresseddata; |
| 54 |
|
| 55 |
data = new char[1]; |
| 56 |
data[0] = '\0'; |
| 57 |
|
| 58 |
|
| 59 |
compresseddata = nullptr; |
| 60 |
data_length = 0; |
| 61 |
buffer_length = 0; |
| 62 |
compressed_buffer_data_length = 0; |
| 63 |
|
| 64 |
if (tempfilefd > -1) { |
| 65 |
close(tempfilefd); |
| 66 |
if (!preservetemp) { |
| 67 |
unlink(tempfilepath.toCharArray()); |
| 68 |
} |
| 69 |
tempfilefd = -1; |
| 70 |
tempfilesize = 0; |
| 71 |
} |
| 72 |
|
| 73 |
bytesalreadysent = 0; |
| 74 |
dontsendbody = false; |
| 75 |
preservetemp = false; |
| 76 |
decompress = ""; |
| 77 |
chunked = false; |
| 78 |
icap = false; |
| 79 |
} |
| 80 |
|
| 81 |
// delete the memory block when the class is destroyed |
| 82 |
DataBuffer::~DataBuffer() |
| 83 |
{ |
| 84 |
delete[] data; |
| 85 |
if (compresseddata != nullptr) { |
| 86 |
delete[] compresseddata; |
| 87 |
compresseddata = nullptr; |
| 88 |
} |
| 89 |
if (tempfilefd > -1) { |
| 90 |
close(tempfilefd); |
| 91 |
if (!preservetemp) { |
| 92 |
unlink(tempfilepath.toCharArray()); |
| 93 |
} |
| 94 |
tempfilefd = -1; |
| 95 |
tempfilesize = 0; |
| 96 |
} |
| 97 |
} |
| 98 |
|
| 99 |
void DataBuffer::set_current_config (FOptionContainer *newfgc) { |
| 100 |
fgc = newfgc; |
| 101 |
} |
| 102 |
|
| 103 |
//} |
| 104 |
// swap back to a compressed version of the data, if one exists |
| 105 |
// also delete uncompressed version |
| 106 |
// if body was decompressed but not modified, this can save bandwidth |
| 107 |
void DataBuffer::swapbacktocompressed() |
| 108 |
{ |
| 109 |
if (compresseddata != nullptr && compressed_buffer_data_length> 0) { |
| 110 |
delete[] data; |
| 111 |
data_length = compressed_buffer_data_length; |
| 112 |
data = compresseddata; |
| 113 |
|
| 114 |
compresseddata = nullptr; |
| 115 |
compressed_buffer_data_length = 0; |
| 116 |
DEBUG_network("Compressed data size ", data_length); |
| 117 |
} |
| 118 |
} |
| 119 |
|
| 120 |
int DataBuffer::readInFromSocket(Socket *sock, int size, bool wantall, int &result) { |
| 121 |
|
| 122 |
int rc; |
| 123 |
|
| 124 |
DEBUG_trace("size:", size, " wantall:", wantall); |
| 125 |
if (size < 1) { |
| 126 |
E2LOGGER_error("read request is negative"); |
| 127 |
return -1; |
| 128 |
} |
| 129 |
|
| 130 |
if (wantall) { |
| 131 |
if (!swappedtodisk) { |
| 132 |
// if not swapped to disk and file is too large for RAM, then swap to disk |
| 133 |
if (data_length > o.content.max_content_ramcache_scan_size) { |
| 134 |
DEBUG_debug("swapping to disk"); |
| 135 |
tempfilefd = getTempFileFD(); |
| 136 |
if (tempfilefd < 0) { |
| 137 |
E2LOGGER_error("error buffering to disk so skipping disk buffering"); |
| 138 |
result = DB_TOBIG; |
| 139 |
return -1; |
| 140 |
} |
| 141 |
write(tempfilefd, data, data_length); |
| 142 |
swappedtodisk = true; |
| 143 |
tempfilesize = data_length; |
| 144 |
} |
| 145 |
} else if (tempfilesize > o.content.max_content_filecache_scan_size) { |
| 146 |
// if swapped to disk and file too large for that too, then give up |
| 147 |
DEBUG_debug("defaultdm: file too big to be scanned, halting download"); |
| 148 |
result = DB_TOBIG | DB_TOBIG_SCAN; |
| 149 |
return -1; |
| 150 |
} |
| 151 |
} else { |
| 152 |
if (data_length > o.content.max_content_filter_size) { |
| 153 |
// if we aren't downloading for virus scanning, and file too large for filtering, give up |
| 154 |
DEBUG_debug("defaultdm: file too big to be filtered, halting download"); |
| 155 |
result = DB_TOBIG | DB_TOBIG_FILTER; |
| 156 |
return -1; |
| 157 |
} |
| 158 |
} |
| 159 |
|
| 160 |
DEBUG_debug("swappedtodisk:", swappedtodisk); |
| 161 |
if (!swappedtodisk) { |
| 162 |
if (size > (buffer_length - data_length)) { |
| 163 |
if(!increase_buffer(size - (buffer_length - data_length))) { |
| 164 |
size = (buffer_length - data_length); |
| 165 |
if (size == 0) { |
| 166 |
result = DB_TOBIG; |
| 167 |
return -1; // it is too big |
| 168 |
} |
| 169 |
} |
| 170 |
} |
| 171 |
if (chunked) { |
| 172 |
DEBUG_debug("readChunk:", data_length, ",", size); |
| 173 |
rc = sock->readChunk((data + data_length), size, timeout); |
| 174 |
} else { |
| 175 |
DEBUG_debug("bufferReadFromSocket:", data_length, ",", size); |
| 176 |
rc = bufferReadFromSocket(sock, (data + data_length), size, timeout); |
| 177 |
} |
| 178 |
|
| 179 |
if (rc <= 0) { |
| 180 |
if (chunked) |
| 181 |
got_all = true; |
| 182 |
return -1; |
| 183 |
// an error occurred so end the while() |
| 184 |
// or none received so pipe iis closed or chunking has ended |
| 185 |
} else { |
| 186 |
bytes_toget -= rc; |
| 187 |
data_length += rc; |
| 188 |
data[data_length] = '\0'; |
| 189 |
} |
| 190 |
} else { |
| 191 |
if (chunked) { |
| 192 |
rc = sock->readChunk(data, buffer_length, timeout); |
| 193 |
} else { |
| 194 |
rc = bufferReadFromSocket(sock, data, |
| 195 |
// if not getting everything until connection close, grab only what is left |
| 196 |
(!geteverything && (bytes_toget < buffer_length) ? bytes_toget |
| 197 |
: buffer_length), |
| 198 |
timeout); |
| 199 |
} |
| 200 |
if (rc <= 0) { |
| 201 |
if (chunked) |
| 202 |
got_all = true; |
| 203 |
result = 0; |
| 204 |
return 0; |
| 205 |
} else { |
| 206 |
bytes_toget -= rc; |
| 207 |
write(tempfilefd, data, rc); |
| 208 |
tempfilesize += rc; |
| 209 |
DEBUG_debug("written to disk:", rc, " total:", tempfilesize); |
| 210 |
} |
| 211 |
} |
| 212 |
result = 0; |
| 213 |
DEBUG_debug("rc=", rc, " bytes_toget=", bytes_toget, " data_length=", data_length); |
| 214 |
return rc; |
| 215 |
} |
| 216 |
|
| 217 |
bool DataBuffer::increase_buffer(int extra) { |
| 218 |
int more = 65536; |
| 219 |
if (extra > more) |
| 220 |
more = extra; |
| 221 |
if ((buffer_length + more) > o.content.max_content_filter_size) { |
| 222 |
more = o.content.max_content_filter_size - buffer_length; |
| 223 |
} |
| 224 |
if (more > 0) { |
| 225 |
char *temp = new char[buffer_length + more + 1]; // replacement store |
| 226 |
temp[buffer_length + more] = '\0'; |
| 227 |
memcpy(temp, data, data_length); // copy the current data |
| 228 |
delete[] data; // delete the current data block |
| 229 |
data = temp; |
| 230 |
temp = nullptr; |
| 231 |
buffer_length += more; // update data size counter |
| 232 |
DEBUG_debug("data buffer extended by ", more, " to ", buffer_length); |
| 233 |
return true; |
| 234 |
} |
| 235 |
return false; |
| 236 |
} |
| 237 |
|
| 238 |
// a much more efficient reader that does not assume the contents of |
| 239 |
// the buffer gets filled thus reducing memcpy()ing and new()ing |
| 240 |
int DataBuffer::bufferReadFromSocket(Socket *sock, char *buffer, int size, int sockettimeout) |
| 241 |
{ |
| 242 |
int pos = 0; |
| 243 |
int rc; |
| 244 |
while (pos < size) { |
| 245 |
rc = sock->readFromSocket(&buffer[pos], size - pos, 0, sockettimeout, true); |
| 246 |
if (rc < 1) { |
| 247 |
// none recieved or an error |
| 248 |
if (pos > 0) { |
| 249 |
return pos; // some was recieved previous into buffer |
| 250 |
} |
| 251 |
return rc; // just return with the return code |
| 252 |
} |
| 253 |
pos += rc; |
| 254 |
} |
| 255 |
return size; // full buffer |
| 256 |
} |
| 257 |
|
| 258 |
// a much more efficient reader that does not assume the contents of |
| 259 |
// the buffer gets filled thus reducing memcpy()ing and new()ing. |
| 260 |
// in addition to the actual socket timeout, used for each individual read, this version |
| 261 |
// incorporates a "global" timeout within which all reads must complete. |
| 262 |
int DataBuffer::bufferReadFromSocket(Socket *sock, char *buffer, int size, int sockettimeout, int stimeout) |
| 263 |
{ |
| 264 |
DEBUG_trace(""); |
| 265 |
int pos = 0; |
| 266 |
int rc; |
| 267 |
struct timeval starttime; |
| 268 |
struct timeval nowadays; |
| 269 |
gettimeofday(&starttime, NULL); |
| 270 |
while (pos < size) { |
| 271 |
if (chunked) { |
| 272 |
rc = sock->readChunk(&buffer[pos], size - pos,sockettimeout ); |
| 273 |
} else { |
| 274 |
rc = sock->readFromSocket(&buffer[pos], size - pos, 0, sockettimeout, true); |
| 275 |
} |
| 276 |
if (rc < 1) { |
| 277 |
// none recieved or an error |
| 278 |
if (pos > 0) { |
| 279 |
return pos; // some was recieved previous into buffer |
| 280 |
} |
| 281 |
return rc; // just return with the return code |
| 282 |
} |
| 283 |
pos += rc; |
| 284 |
gettimeofday(&nowadays, NULL); |
| 285 |
if (nowadays.tv_sec - starttime.tv_sec > stimeout) { |
| 286 |
DEBUG_debug("buffered socket read more than timeout"); |
| 287 |
return pos; // just return how much got so far then |
| 288 |
} |
| 289 |
} |
| 290 |
return size; // full buffer |
| 291 |
} |
| 292 |
|
| 293 |
// make a temp file and return its FD. only currently used in DM plugins. |
| 294 |
int DataBuffer::getTempFileFD() |
| 295 |
{ |
| 296 |
if (tempfilefd > -1) { |
| 297 |
return tempfilefd; |
| 298 |
} |
| 299 |
tempfilepath = o.content.download_dir.c_str(); |
| 300 |
tempfilepath += "/tfXXXXXX"; |
| 301 |
char *tempfilepatharray = new char[tempfilepath.length() + 1]; |
| 302 |
strcpy(tempfilepatharray, tempfilepath.toCharArray()); |
| 303 |
umask(0007); // only allow access to e2g user and group |
| 304 |
if ((tempfilefd = mkstemp(tempfilepatharray)) < 0) { |
| 305 |
E2LOGGER_error("Could not create temp file to store download for scanning: ", strerror(errno)); |
| 306 |
tempfilefd = -1; |
| 307 |
tempfilepath = ""; |
| 308 |
} else { |
| 309 |
tempfilepath = tempfilepatharray; |
| 310 |
} |
| 311 |
delete[] tempfilepatharray; |
| 312 |
return tempfilefd; |
| 313 |
} |
| 314 |
|
| 315 |
// check the client's user agent, see if we have a DM plugin compatible with it, and use it to download the body of the given request |
| 316 |
bool DataBuffer::in(Socket *sock, Socket *peersock, HTTPHeader *requestheader, HTTPHeader *docheader, bool runav, int *headersent,StoryBoard &story, NaughtyFilter *cm) |
| 317 |
{ |
| 318 |
//Socket *sock = where to read from |
| 319 |
//Socket *peersock = browser to send stuff to for keeping it alive |
| 320 |
//HTTPHeader *requestheader = header client used to request |
| 321 |
//HTTPHeader *docheader = header used for sending first line of reply |
| 322 |
//bool runav = to determine if limit is av or not |
| 323 |
//int *headersent = to use to send the first line of header if needed |
| 324 |
// or to mark that the header has already been sent |
| 325 |
|
| 326 |
// so we know if we only partially downloaded from |
| 327 |
// upstream so later, if allowed, we can send the rest |
| 328 |
bool toobig = false; |
| 329 |
|
| 330 |
// match request to download manager so browsers potentially can have a prettier version |
| 331 |
// and software updates, stream clients, etc. can have a compatible version. |
| 332 |
//int rc = 0; |
| 333 |
int j = 0; |
| 334 |
// int rc = -1; |
| 335 |
DEBUG_trace(""); |
| 336 |
for (std::deque<Plugin *>::iterator i = o.plugins.dmplugins_begin; i != o.plugins.dmplugins_end; i++) { |
| 337 |
++j; |
| 338 |
if ((i + 1) == o.plugins.dmplugins_end) { |
| 339 |
DEBUG_debug("Got to final download manager so defaulting to always match."); |
| 340 |
dm_plugin = (DMPlugin *)(*i); |
| 341 |
dm_plugin->in(this, sock, peersock, requestheader, docheader, runav, headersent, &toobig); |
| 342 |
break; |
| 343 |
} else { |
| 344 |
dm_plugin = (DMPlugin *)(*i); |
| 345 |
if (story.runFunctEntry(dm_plugin->story_entry, *cm)) { |
| 346 |
DEBUG_debug("Matching download manager number: ", j); |
| 347 |
dm_plugin->in(this, sock, peersock, requestheader, docheader, runav, headersent, &toobig); |
| 348 |
break; |
| 349 |
} |
| 350 |
} |
| 351 |
j++; |
| 352 |
} |
| 353 |
return toobig; |
| 354 |
} |
| 355 |
|
| 356 |
// send the request body to the client after having been handled by a DM plugin |
| 357 |
bool DataBuffer::out(Socket *sock) |
| 358 |
{ |
| 359 |
if (dontsendbody) { |
| 360 |
DEBUG_debug("dontsendbody true; not sending"); |
| 361 |
return true; |
| 362 |
} |
| 363 |
// if (!(*sock).readyForOutput(timeout)) return false; // exceptions on timeout or error |
| 364 |
|
| 365 |
DEBUG_trace(""); |
| 366 |
if (tempfilefd > -1) { |
| 367 |
// must have been too big for ram so stream from disk in blocks |
| 368 |
DEBUG_debug("Sending ", tempfilesize - bytesalreadysent, " bytes from temp file (", bytesalreadysent, " already sent)");; |
| 369 |
off_t sent = bytesalreadysent; |
| 370 |
int rc; |
| 371 |
|
| 372 |
if (lseek(tempfilefd, bytesalreadysent, SEEK_SET) < 0) |
| 373 |
return false; |
| 374 |
int block_len; |
| 375 |
if(chunked) |
| 376 |
block_len = 4048; |
| 377 |
else |
| 378 |
block_len = data_length; |
| 379 |
|
| 380 |
while (sent < tempfilesize) { |
| 381 |
rc = read(tempfilefd, data, block_len); |
| 382 |
DEBUG_debug("reading temp file rc:", rc); |
| 383 |
if (rc < 0) { |
| 384 |
DEBUG_debug("error reading temp file so throwing exception"); |
| 385 |
return false; |
| 386 |
} |
| 387 |
if (rc == 0) { |
| 388 |
DEBUG_debug("got zero bytes reading temp file"); |
| 389 |
break; // should never happen |
| 390 |
} |
| 391 |
// as it's cached to disk the buffer must be reasonably big |
| 392 |
if(chunked) { |
| 393 |
if (!sock->writeChunk(data, rc, timeout)) |
| 394 |
return false; |
| 395 |
} else { |
| 396 |
if (!sock->writeToSocket(data, rc, 0, timeout)) { |
| 397 |
return false; |
| 398 |
} |
| 399 |
} |
| 400 |
sent += rc; |
| 401 |
DEBUG_debug("total sent from temp:", sent); |
| 402 |
} |
| 403 |
if (chunked && got_all) { |
| 404 |
String n; |
| 405 |
if (!sock->writeChunkTrailer(n)) |
| 406 |
return false; |
| 407 |
} |
| 408 |
close(tempfilefd); |
| 409 |
tempfilefd = -1; |
| 410 |
tempfilesize = 0; |
| 411 |
unlink(tempfilepath.toCharArray()); |
| 412 |
} else { |
| 413 |
off_t sent = bytesalreadysent; |
| 414 |
DEBUG_debug("Sending ", data_length, " bytes from RAM (", buffer_length, " in buffer; ", bytesalreadysent, " already sent)" ); |
| 415 |
// it's in RAM, so just send it, no streaming from disk |
| 416 |
int block_len; |
| 417 |
if(chunked) |
| 418 |
block_len = 4048; |
| 419 |
else |
| 420 |
block_len = data_length; |
| 421 |
|
| 422 |
if (data_length != 0) { |
| 423 |
while (sent < data_length) { |
| 424 |
if( block_len > (data_length - sent)) |
| 425 |
block_len = (data_length - sent); |
| 426 |
if (chunked) { |
| 427 |
if (!sock->writeChunk(data + sent, block_len, timeout)) { |
| 428 |
DEBUG_network("writeChunk failed after ", sent, " bytes"); |
| 429 |
return false; |
| 430 |
} |
| 431 |
|
| 432 |
} else { |
| 433 |
if (!sock->writeToSocket(data + sent, data_length - sent, 0, timeout)) |
| 434 |
return false; |
| 435 |
} |
| 436 |
sent += block_len; |
| 437 |
} |
| 438 |
if (chunked && got_all) { |
| 439 |
String n; |
| 440 |
if (!sock->writeChunkTrailer(n)) |
| 441 |
return false; |
| 442 |
} |
| 443 |
} else { |
| 444 |
if(chunked) { |
| 445 |
if (!sock->writeChunk(data + bytesalreadysent, 0, timeout)) |
| 446 |
return false; |
| 447 |
} else { |
| 448 |
if (!sock->writeToSocket("\r\n\r\n", 4, 0, timeout)) |
| 449 |
return false; |
| 450 |
} |
| 451 |
} |
| 452 |
DEBUG_debug("Sent ", data_length - bytesalreadysent," bytes from RAM (", buffer_length); |
| 453 |
} |
| 454 |
return true; |
| 455 |
} |
| 456 |
|
| 457 |
// zlib decompression |
| 458 |
void DataBuffer::zlibinflate(bool header) |
| 459 |
{ |
| 460 |
if (data_length < 12) { |
| 461 |
return; // it can't possibly be zlib'd |
| 462 |
} |
| 463 |
DEBUG_debug("compressed size:", data_length); |
| 464 |
|
| 465 |
#if ZLIB_VERNUM < 0x1210 |
| 466 |
#warning ************************************ |
| 467 |
#warning For gzip support you need zlib 1.2.1 |
| 468 |
#warning or later to be installed. |
| 469 |
#warning You can ignore this warning but |
| 470 |
#warning internet bandwidth may be wasted. |
| 471 |
#warning ************************************ |
| 472 |
if (header) { |
| 473 |
return; |
| 474 |
} |
| 475 |
#endif |
| 476 |
|
| 477 |
int newsize = data_length * 5; // good estimate of deflated HTML |
| 478 |
|
| 479 |
char *block = new char[newsize]; |
| 480 |
block[0] = '\0'; |
| 481 |
|
| 482 |
char *temp = NULL; |
| 483 |
|
| 484 |
off_t bytesgot = 0; |
| 485 |
int err; |
| 486 |
|
| 487 |
z_stream d_stream; |
| 488 |
d_stream.zalloc = (alloc_func)0; |
| 489 |
d_stream.zfree = (free_func)0; |
| 490 |
d_stream.opaque = (voidpf)0; |
| 491 |
d_stream.next_in = (Bytef *)data; |
| 492 |
d_stream.avail_in = data_length; |
| 493 |
d_stream.next_out = (Bytef *)block; |
| 494 |
d_stream.avail_out = newsize; |
| 495 |
|
| 496 |
// inflate either raw zlib, or possibly gzip with a header |
| 497 |
if (header) { |
| 498 |
err = inflateInit2(&d_stream, 15 + 32); |
| 499 |
} else { |
| 500 |
err = inflateInit2(&d_stream, -15); |
| 501 |
} |
| 502 |
|
| 503 |
if (err != Z_OK) { // was a problem so just return |
| 504 |
delete[] block; // don't forget to free claimed memory |
| 505 |
DEBUG_debug("bad init inflate: ", err); |
| 506 |
return; |
| 507 |
} |
| 508 |
while (true) { |
| 509 |
DEBUG_debug("inflate loop"); |
| 510 |
err = inflate(&d_stream, Z_SYNC_FLUSH); |
| 511 |
bytesgot = d_stream.total_out; |
| 512 |
if (err == Z_STREAM_END) { |
| 513 |
err = inflateEnd(&d_stream); |
| 514 |
if (err != Z_OK) { |
| 515 |
delete[] block; |
| 516 |
DEBUG_debug("bad inflateEnd: ", d_stream.msg); |
| 517 |
return; |
| 518 |
} |
| 519 |
break; |
| 520 |
} |
| 521 |
if (err != Z_OK) { // was a problem so just return |
| 522 |
delete[] block; // don't forget to free claimed memory |
| 523 |
DEBUG_debug("bad inflate: ", d_stream.msg); |
| 524 |
err = inflateEnd(&d_stream); |
| 525 |
if (err != Z_OK) { |
| 526 |
DEBUG_debug("bad inflateEnd: ", d_stream.msg); |
| 527 |
} |
| 528 |
return; |
| 529 |
} |
| 530 |
if (bytesgot > o.content.max_content_filter_size) { |
| 531 |
delete[] block; // don't forget to free claimed memory |
| 532 |
DEBUG_debug("inflated file larger than maxcontentfiltersize, not inflating further"); |
| 533 |
err = inflateEnd(&d_stream); |
| 534 |
if (err != Z_OK) { |
| 535 |
DEBUG_debug("bad inflateEnd: ", d_stream.msg); |
| 536 |
} |
| 537 |
return; |
| 538 |
} |
| 539 |
|
| 540 |
// inflation is going ok, but we don't have enough room in the output buffer |
| 541 |
newsize = bytesgot * 2; |
| 542 |
temp = new char[newsize]; |
| 543 |
memcpy(temp, block, bytesgot); |
| 544 |
delete[] block; |
| 545 |
block = temp; |
| 546 |
temp = NULL; |
| 547 |
|
| 548 |
d_stream.next_out = (Bytef *)(block + bytesgot); |
| 549 |
d_stream.avail_out = newsize - bytesgot; |
| 550 |
} |
| 551 |
|
| 552 |
compresseddata = data; |
| 553 |
compressed_buffer_data_length = data_length; // change from buffer_length |
| 554 |
buffer_length = bytesgot; |
| 555 |
DEBUG_debug("decompressed size: ", data_length); |
| 556 |
data = new char[bytesgot + 1]; |
| 557 |
data[bytesgot] = '\0'; |
| 558 |
memcpy(data, block, bytesgot); |
| 559 |
data_length = bytesgot; |
| 560 |
delete[] block; |
| 561 |
} |
| 562 |
|
| 563 |
// Does a regexp search and replace. |
| 564 |
struct newreplacement { |
| 565 |
int match; |
| 566 |
String replacement; |
| 567 |
}; |
| 568 |
bool DataBuffer::contentRegExp(FOptionContainer* &foc) |
| 569 |
{ |
| 570 |
|
| 571 |
DEBUG_debug("Starting content reg exp replace"); |
| 572 |
bool contentmodified = false; |
| 573 |
unsigned int i; |
| 574 |
unsigned int j, k, m; |
| 575 |
unsigned int s = (*foc).content_regexp_list_comp.size(); |
| 576 |
unsigned int matches; |
| 577 |
unsigned int submatch, submatches; |
| 578 |
RegExp *re; |
| 579 |
RegResult Rre; |
| 580 |
String *replacement; |
| 581 |
unsigned int replen; |
| 582 |
int sizediff; |
| 583 |
char *newblock; |
| 584 |
char *dstpos; |
| 585 |
unsigned int srcoff; |
| 586 |
unsigned int nextoffset; |
| 587 |
unsigned int matchlen; |
| 588 |
|
| 589 |
std::queue<newreplacement *> matchqueue; |
| 590 |
|
| 591 |
for (i = 0; i < s; i++) { |
| 592 |
re = &((*foc).content_regexp_list_comp[i]); |
| 593 |
if (re->match(data, Rre)) { |
| 594 |
replacement = &((*foc).content_regexp_list_rep[i]); |
| 595 |
//replen = replacement->length(); |
| 596 |
matches = Rre.numberOfMatches(); |
| 597 |
|
| 598 |
sizediff = 0; |
| 599 |
m = 0; |
| 600 |
for (j = 0; j < matches; j++) { |
| 601 |
srcoff = Rre.offset(j); |
| 602 |
matchlen = Rre.length(j); |
| 603 |
|
| 604 |
// Count matches for ()'s |
| 605 |
for (submatches = 0; j + submatches + 1 < matches; submatches++) |
| 606 |
if (Rre.offset(j + submatches + 1) + Rre.length(j + submatches + 1) > srcoff + matchlen) |
| 607 |
break; |
| 608 |
|
| 609 |
// \1 and $1 replacement |
| 610 |
|
| 611 |
// store match no. and default (empty) replacement string |
| 612 |
newreplacement *newrep = new newreplacement; |
| 613 |
newrep->match = j; |
| 614 |
newrep->replacement = ""; |
| 615 |
// iterate over regex's replacement string |
| 616 |
for (k = 0; k < replacement->length(); k++) { |
| 617 |
// look for \1..\9 and $1..$9 |
| 618 |
if (((*replacement)[k] == '\\' || (*replacement)[k] == '$') && (*replacement)[k + 1] >= '1' && (*replacement)[k + 1] <= '9') { |
| 619 |
// determine match number |
| 620 |
submatch = (*replacement)[++k] - '0'; |
| 621 |
// add submatch contents to replacement string |
| 622 |
if (submatch <= submatches) { |
| 623 |
newrep->replacement += Rre.result(j + submatch).c_str(); |
| 624 |
} |
| 625 |
} else { |
| 626 |
// unescape \\ and \$, and add other non-backreference characters |
| 627 |
if ((*replacement)[k] == '\\' && ((*replacement)[k + 1] == '\\' || (*replacement)[k + 1] == '$')) |
| 628 |
k++; |
| 629 |
newrep->replacement += replacement->subString(k, 1); |
| 630 |
} |
| 631 |
} |
| 632 |
matchqueue.push(newrep); |
| 633 |
|
| 634 |
// update size difference between original and modified content |
| 635 |
sizediff -= Rre.length(j); |
| 636 |
sizediff += newrep->replacement.length(); |
| 637 |
// skip submatches to next top level match |
| 638 |
j += submatches; |
| 639 |
m++; |
| 640 |
} |
| 641 |
|
| 642 |
// now we know eventual size of content-replaced block, allocate memory for it |
| 643 |
newblock = new char[data_length + sizediff + 1]; |
| 644 |
newblock[data_length + sizediff] = '\0'; |
| 645 |
srcoff = 0; |
| 646 |
dstpos = newblock; |
| 647 |
matches = m; |
| 648 |
|
| 649 |
DEBUG_debug("content matches:", matches); |
| 650 |
// replace top-level matches using filled-out replacement strings |
| 651 |
newreplacement *newrep; |
| 652 |
for (j = 0; j < matches; j++) { |
| 653 |
newrep = matchqueue.front(); |
| 654 |
nextoffset = Rre.offset(newrep->match); |
| 655 |
if (nextoffset > srcoff) { |
| 656 |
memcpy(dstpos, data + srcoff, nextoffset - srcoff); |
| 657 |
dstpos += nextoffset - srcoff; |
| 658 |
srcoff = nextoffset; |
| 659 |
} |
| 660 |
replen = newrep->replacement.length(); |
| 661 |
memcpy(dstpos, newrep->replacement.toCharArray(), replen); |
| 662 |
dstpos += replen; |
| 663 |
srcoff += Rre.length(newrep->match); |
| 664 |
delete newrep; |
| 665 |
matchqueue.pop(); |
| 666 |
} |
| 667 |
if (srcoff < data_length) { |
| 668 |
memcpy(dstpos, data + srcoff, data_length - srcoff); |
| 669 |
} |
| 670 |
delete[] data; |
| 671 |
data = newblock; |
| 672 |
data_length = data_length + sizediff; |
| 673 |
contentmodified = true; |
| 674 |
} |
| 675 |
} |
| 676 |
return contentmodified; |
| 677 |
} |
| 678 |
|
| 679 |
void DataBuffer::setChunked(bool ch = true) { |
| 680 |
chunked = ch; |
| 681 |
return; |
| 682 |
} |
| 683 |
|
| 684 |
void DataBuffer::setICAP(bool ch = true) { |
| 685 |
chunked = ch; |
| 686 |
icap = ch; |
| 687 |
return; |
| 688 |
} |