| The Android Open Source Project | edbf3b6 | 2009-03-03 19:31:44 -0800 | [diff] [blame] | 1 | /* | 
|  | 2 | * Copyright (C) 2007 The Android Open Source Project | 
|  | 3 | * | 
|  | 4 | * Licensed under the Apache License, Version 2.0 (the "License"); | 
|  | 5 | * you may not use this file except in compliance with the License. | 
|  | 6 | * You may obtain a copy of the License at | 
|  | 7 | * | 
|  | 8 | *      http://www.apache.org/licenses/LICENSE-2.0 | 
|  | 9 | * | 
|  | 10 | * Unless required by applicable law or agreed to in writing, software | 
|  | 11 | * distributed under the License is distributed on an "AS IS" BASIS, | 
|  | 12 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | 
|  | 13 | * See the License for the specific language governing permissions and | 
|  | 14 | * limitations under the License. | 
|  | 15 | */ | 
|  | 16 |  | 
|  | 17 | // | 
|  | 18 | // Read-only access to Zip archives, with minimal heap allocation. | 
|  | 19 | // | 
|  | 20 | #define LOG_TAG "zipro" | 
|  | 21 | //#define LOG_NDEBUG 0 | 
| Mathias Agopian | f446ba9 | 2009-06-04 13:53:57 -0700 | [diff] [blame] | 22 | #include <utils/ZipFileRO.h> | 
|  | 23 | #include <utils/Log.h> | 
|  | 24 | #include <utils/misc.h> | 
| The Android Open Source Project | edbf3b6 | 2009-03-03 19:31:44 -0800 | [diff] [blame] | 25 |  | 
|  | 26 | #include <zlib.h> | 
|  | 27 |  | 
|  | 28 | #include <string.h> | 
|  | 29 | #include <fcntl.h> | 
|  | 30 | #include <errno.h> | 
|  | 31 | #include <assert.h> | 
|  | 32 |  | 
|  | 33 | using namespace android; | 
|  | 34 |  | 
|  | 35 | /* | 
|  | 36 | * Zip file constants. | 
|  | 37 | */ | 
|  | 38 | #define kEOCDSignature      0x06054b50 | 
|  | 39 | #define kEOCDLen            22 | 
|  | 40 | #define kEOCDNumEntries     8               // offset to #of entries in file | 
|  | 41 | #define kEOCDFileOffset     16              // offset to central directory | 
|  | 42 |  | 
|  | 43 | #define kMaxCommentLen      65535           // longest possible in ushort | 
|  | 44 | #define kMaxEOCDSearch      (kMaxCommentLen + kEOCDLen) | 
|  | 45 |  | 
|  | 46 | #define kLFHSignature       0x04034b50 | 
|  | 47 | #define kLFHLen             30              // excluding variable-len fields | 
|  | 48 | #define kLFHNameLen         26              // offset to filename length | 
|  | 49 | #define kLFHExtraLen        28              // offset to extra length | 
|  | 50 |  | 
|  | 51 | #define kCDESignature       0x02014b50 | 
|  | 52 | #define kCDELen             46              // excluding variable-len fields | 
|  | 53 | #define kCDEMethod          10              // offset to compression method | 
|  | 54 | #define kCDEModWhen         12              // offset to modification timestamp | 
|  | 55 | #define kCDECRC             16              // offset to entry CRC | 
|  | 56 | #define kCDECompLen         20              // offset to compressed length | 
|  | 57 | #define kCDEUncompLen       24              // offset to uncompressed length | 
|  | 58 | #define kCDENameLen         28              // offset to filename length | 
|  | 59 | #define kCDEExtraLen        30              // offset to extra length | 
|  | 60 | #define kCDECommentLen      32              // offset to comment length | 
|  | 61 | #define kCDELocalOffset     42              // offset to local hdr | 
|  | 62 |  | 
|  | 63 | /* | 
|  | 64 | * The values we return for ZipEntryRO use 0 as an invalid value, so we | 
|  | 65 | * want to adjust the hash table index by a fixed amount.  Using a large | 
|  | 66 | * value helps insure that people don't mix & match arguments, e.g. to | 
|  | 67 | * findEntryByIndex(). | 
|  | 68 | */ | 
|  | 69 | #define kZipEntryAdj        10000 | 
|  | 70 |  | 
|  | 71 | /* | 
|  | 72 | * Convert a ZipEntryRO to a hash table index, verifying that it's in a | 
|  | 73 | * valid range. | 
|  | 74 | */ | 
|  | 75 | int ZipFileRO::entryToIndex(const ZipEntryRO entry) const | 
|  | 76 | { | 
|  | 77 | long ent = ((long) entry) - kZipEntryAdj; | 
|  | 78 | if (ent < 0 || ent >= mHashTableSize || mHashTable[ent].name == NULL) { | 
|  | 79 | LOGW("Invalid ZipEntryRO %p (%ld)\n", entry, ent); | 
|  | 80 | return -1; | 
|  | 81 | } | 
|  | 82 | return ent; | 
|  | 83 | } | 
|  | 84 |  | 
|  | 85 |  | 
|  | 86 | /* | 
|  | 87 | * Open the specified file read-only.  We memory-map the entire thing and | 
|  | 88 | * close the file before returning. | 
|  | 89 | */ | 
|  | 90 | status_t ZipFileRO::open(const char* zipFileName) | 
|  | 91 | { | 
|  | 92 | int fd = -1; | 
|  | 93 | off_t length; | 
|  | 94 |  | 
|  | 95 | assert(mFileMap == NULL); | 
|  | 96 |  | 
|  | 97 | /* | 
|  | 98 | * Open and map the specified file. | 
|  | 99 | */ | 
|  | 100 | fd = ::open(zipFileName, O_RDONLY); | 
|  | 101 | if (fd < 0) { | 
|  | 102 | LOGW("Unable to open zip '%s': %s\n", zipFileName, strerror(errno)); | 
|  | 103 | return NAME_NOT_FOUND; | 
|  | 104 | } | 
|  | 105 |  | 
|  | 106 | length = lseek(fd, 0, SEEK_END); | 
|  | 107 | if (length < 0) { | 
|  | 108 | close(fd); | 
|  | 109 | return UNKNOWN_ERROR; | 
|  | 110 | } | 
|  | 111 |  | 
|  | 112 | mFileMap = new FileMap(); | 
|  | 113 | if (mFileMap == NULL) { | 
|  | 114 | close(fd); | 
|  | 115 | return NO_MEMORY; | 
|  | 116 | } | 
|  | 117 | if (!mFileMap->create(zipFileName, fd, 0, length, true)) { | 
|  | 118 | LOGW("Unable to map '%s': %s\n", zipFileName, strerror(errno)); | 
|  | 119 | close(fd); | 
|  | 120 | return UNKNOWN_ERROR; | 
|  | 121 | } | 
|  | 122 |  | 
|  | 123 | mFd = fd; | 
|  | 124 |  | 
|  | 125 | /* | 
|  | 126 | * Got it mapped, verify it and create data structures for fast access. | 
|  | 127 | */ | 
|  | 128 | if (!parseZipArchive()) { | 
|  | 129 | mFileMap->release(); | 
|  | 130 | mFileMap = NULL; | 
|  | 131 | return UNKNOWN_ERROR; | 
|  | 132 | } | 
|  | 133 |  | 
|  | 134 | return OK; | 
|  | 135 | } | 
|  | 136 |  | 
|  | 137 | /* | 
|  | 138 | * Parse the Zip archive, verifying its contents and initializing internal | 
|  | 139 | * data structures. | 
|  | 140 | */ | 
|  | 141 | bool ZipFileRO::parseZipArchive(void) | 
|  | 142 | { | 
|  | 143 | #define CHECK_OFFSET(_off) {                                                \ | 
|  | 144 | if ((unsigned int) (_off) >= maxOffset) {                           \ | 
|  | 145 | LOGE("ERROR: bad offset %u (max %d): %s\n",                     \ | 
|  | 146 | (unsigned int) (_off), maxOffset, #_off);                   \ | 
|  | 147 | goto bail;                                                      \ | 
|  | 148 | }                                                                   \ | 
|  | 149 | } | 
|  | 150 | const unsigned char* basePtr = (const unsigned char*)mFileMap->getDataPtr(); | 
|  | 151 | const unsigned char* ptr; | 
|  | 152 | size_t length = mFileMap->getDataLength(); | 
|  | 153 | bool result = false; | 
|  | 154 | unsigned int i, numEntries, cdOffset; | 
|  | 155 | unsigned int val; | 
|  | 156 |  | 
|  | 157 | /* | 
|  | 158 | * The first 4 bytes of the file will either be the local header | 
|  | 159 | * signature for the first file (kLFHSignature) or, if the archive doesn't | 
|  | 160 | * have any files in it, the end-of-central-directory signature | 
|  | 161 | * (kEOCDSignature). | 
|  | 162 | */ | 
|  | 163 | val = get4LE(basePtr); | 
|  | 164 | if (val == kEOCDSignature) { | 
|  | 165 | LOGI("Found Zip archive, but it looks empty\n"); | 
|  | 166 | goto bail; | 
|  | 167 | } else if (val != kLFHSignature) { | 
|  | 168 | LOGV("Not a Zip archive (found 0x%08x)\n", val); | 
|  | 169 | goto bail; | 
|  | 170 | } | 
|  | 171 |  | 
|  | 172 | /* | 
|  | 173 | * Find the EOCD.  We'll find it immediately unless they have a file | 
|  | 174 | * comment. | 
|  | 175 | */ | 
|  | 176 | ptr = basePtr + length - kEOCDLen; | 
|  | 177 |  | 
|  | 178 | while (ptr >= basePtr) { | 
|  | 179 | if (*ptr == (kEOCDSignature & 0xff) && get4LE(ptr) == kEOCDSignature) | 
|  | 180 | break; | 
|  | 181 | ptr--; | 
|  | 182 | } | 
|  | 183 | if (ptr < basePtr) { | 
|  | 184 | LOGI("Could not find end-of-central-directory in Zip\n"); | 
|  | 185 | goto bail; | 
|  | 186 | } | 
|  | 187 |  | 
|  | 188 | /* | 
|  | 189 | * There are two interesting items in the EOCD block: the number of | 
|  | 190 | * entries in the file, and the file offset of the start of the | 
|  | 191 | * central directory. | 
|  | 192 | * | 
|  | 193 | * (There's actually a count of the #of entries in this file, and for | 
|  | 194 | * all files which comprise a spanned archive, but for our purposes | 
|  | 195 | * we're only interested in the current file.  Besides, we expect the | 
|  | 196 | * two to be equivalent for our stuff.) | 
|  | 197 | */ | 
|  | 198 | numEntries = get2LE(ptr + kEOCDNumEntries); | 
|  | 199 | cdOffset = get4LE(ptr + kEOCDFileOffset); | 
|  | 200 |  | 
|  | 201 | /* valid offsets are [0,EOCD] */ | 
|  | 202 | unsigned int maxOffset; | 
|  | 203 | maxOffset = (ptr - basePtr) +1; | 
|  | 204 |  | 
|  | 205 | LOGV("+++ numEntries=%d cdOffset=%d\n", numEntries, cdOffset); | 
|  | 206 | if (numEntries == 0 || cdOffset >= length) { | 
|  | 207 | LOGW("Invalid entries=%d offset=%d (len=%zd)\n", | 
|  | 208 | numEntries, cdOffset, length); | 
|  | 209 | goto bail; | 
|  | 210 | } | 
|  | 211 |  | 
|  | 212 | /* | 
|  | 213 | * Create hash table.  We have a minimum 75% load factor, possibly as | 
|  | 214 | * low as 50% after we round off to a power of 2. | 
|  | 215 | */ | 
|  | 216 | mNumEntries = numEntries; | 
|  | 217 | mHashTableSize = roundUpPower2(1 + ((numEntries * 4) / 3)); | 
|  | 218 | mHashTable = (HashEntry*) calloc(1, sizeof(HashEntry) * mHashTableSize); | 
|  | 219 |  | 
|  | 220 | /* | 
|  | 221 | * Walk through the central directory, adding entries to the hash | 
|  | 222 | * table. | 
|  | 223 | */ | 
|  | 224 | ptr = basePtr + cdOffset; | 
|  | 225 | for (i = 0; i < numEntries; i++) { | 
|  | 226 | unsigned int fileNameLen, extraLen, commentLen, localHdrOffset; | 
|  | 227 | const unsigned char* localHdr; | 
|  | 228 | unsigned int hash; | 
|  | 229 |  | 
|  | 230 | if (get4LE(ptr) != kCDESignature) { | 
|  | 231 | LOGW("Missed a central dir sig (at %d)\n", i); | 
|  | 232 | goto bail; | 
|  | 233 | } | 
|  | 234 | if (ptr + kCDELen > basePtr + length) { | 
|  | 235 | LOGW("Ran off the end (at %d)\n", i); | 
|  | 236 | goto bail; | 
|  | 237 | } | 
|  | 238 |  | 
|  | 239 | localHdrOffset = get4LE(ptr + kCDELocalOffset); | 
|  | 240 | CHECK_OFFSET(localHdrOffset); | 
|  | 241 | fileNameLen = get2LE(ptr + kCDENameLen); | 
|  | 242 | extraLen = get2LE(ptr + kCDEExtraLen); | 
|  | 243 | commentLen = get2LE(ptr + kCDECommentLen); | 
|  | 244 |  | 
|  | 245 | //LOGV("+++ %d: localHdr=%d fnl=%d el=%d cl=%d\n", | 
|  | 246 | //    i, localHdrOffset, fileNameLen, extraLen, commentLen); | 
|  | 247 | //LOGV(" '%.*s'\n", fileNameLen, ptr + kCDELen); | 
|  | 248 |  | 
|  | 249 | /* add the CDE filename to the hash table */ | 
|  | 250 | hash = computeHash((const char*)ptr + kCDELen, fileNameLen); | 
|  | 251 | addToHash((const char*)ptr + kCDELen, fileNameLen, hash); | 
|  | 252 |  | 
|  | 253 | localHdr = basePtr + localHdrOffset; | 
|  | 254 | if (get4LE(localHdr) != kLFHSignature) { | 
|  | 255 | LOGW("Bad offset to local header: %d (at %d)\n", | 
|  | 256 | localHdrOffset, i); | 
|  | 257 | goto bail; | 
|  | 258 | } | 
|  | 259 |  | 
|  | 260 | ptr += kCDELen + fileNameLen + extraLen + commentLen; | 
|  | 261 | CHECK_OFFSET(ptr - basePtr); | 
|  | 262 | } | 
|  | 263 |  | 
|  | 264 | result = true; | 
|  | 265 |  | 
|  | 266 | bail: | 
|  | 267 | return result; | 
|  | 268 | #undef CHECK_OFFSET | 
|  | 269 | } | 
|  | 270 |  | 
|  | 271 |  | 
|  | 272 | /* | 
|  | 273 | * Simple string hash function for non-null-terminated strings. | 
|  | 274 | */ | 
|  | 275 | /*static*/ unsigned int ZipFileRO::computeHash(const char* str, int len) | 
|  | 276 | { | 
|  | 277 | unsigned int hash = 0; | 
|  | 278 |  | 
|  | 279 | while (len--) | 
|  | 280 | hash = hash * 31 + *str++; | 
|  | 281 |  | 
|  | 282 | return hash; | 
|  | 283 | } | 
|  | 284 |  | 
|  | 285 | /* | 
|  | 286 | * Add a new entry to the hash table. | 
|  | 287 | */ | 
|  | 288 | void ZipFileRO::addToHash(const char* str, int strLen, unsigned int hash) | 
|  | 289 | { | 
|  | 290 | int ent = hash & (mHashTableSize-1); | 
|  | 291 |  | 
|  | 292 | /* | 
|  | 293 | * We over-allocate the table, so we're guaranteed to find an empty slot. | 
|  | 294 | */ | 
|  | 295 | while (mHashTable[ent].name != NULL) | 
|  | 296 | ent = (ent + 1) & (mHashTableSize-1); | 
|  | 297 |  | 
|  | 298 | mHashTable[ent].name = str; | 
|  | 299 | mHashTable[ent].nameLen = strLen; | 
|  | 300 | } | 
|  | 301 |  | 
|  | 302 | /* | 
|  | 303 | * Find a matching entry. | 
|  | 304 | * | 
|  | 305 | * Returns 0 if not found. | 
|  | 306 | */ | 
|  | 307 | ZipEntryRO ZipFileRO::findEntryByName(const char* fileName) const | 
|  | 308 | { | 
|  | 309 | int nameLen = strlen(fileName); | 
|  | 310 | unsigned int hash = computeHash(fileName, nameLen); | 
|  | 311 | int ent = hash & (mHashTableSize-1); | 
|  | 312 |  | 
|  | 313 | while (mHashTable[ent].name != NULL) { | 
|  | 314 | if (mHashTable[ent].nameLen == nameLen && | 
|  | 315 | memcmp(mHashTable[ent].name, fileName, nameLen) == 0) | 
|  | 316 | { | 
|  | 317 | /* match */ | 
|  | 318 | return (ZipEntryRO) (ent + kZipEntryAdj); | 
|  | 319 | } | 
|  | 320 |  | 
|  | 321 | ent = (ent + 1) & (mHashTableSize-1); | 
|  | 322 | } | 
|  | 323 |  | 
|  | 324 | return NULL; | 
|  | 325 | } | 
|  | 326 |  | 
|  | 327 | /* | 
|  | 328 | * Find the Nth entry. | 
|  | 329 | * | 
|  | 330 | * This currently involves walking through the sparse hash table, counting | 
|  | 331 | * non-empty entries.  If we need to speed this up we can either allocate | 
|  | 332 | * a parallel lookup table or (perhaps better) provide an iterator interface. | 
|  | 333 | */ | 
|  | 334 | ZipEntryRO ZipFileRO::findEntryByIndex(int idx) const | 
|  | 335 | { | 
|  | 336 | if (idx < 0 || idx >= mNumEntries) { | 
|  | 337 | LOGW("Invalid index %d\n", idx); | 
|  | 338 | return NULL; | 
|  | 339 | } | 
|  | 340 |  | 
|  | 341 | for (int ent = 0; ent < mHashTableSize; ent++) { | 
|  | 342 | if (mHashTable[ent].name != NULL) { | 
|  | 343 | if (idx-- == 0) | 
|  | 344 | return (ZipEntryRO) (ent + kZipEntryAdj); | 
|  | 345 | } | 
|  | 346 | } | 
|  | 347 |  | 
|  | 348 | return NULL; | 
|  | 349 | } | 
|  | 350 |  | 
|  | 351 | /* | 
|  | 352 | * Get the useful fields from the zip entry. | 
|  | 353 | * | 
|  | 354 | * Returns "false" if the offsets to the fields or the contents of the fields | 
|  | 355 | * appear to be bogus. | 
|  | 356 | */ | 
|  | 357 | bool ZipFileRO::getEntryInfo(ZipEntryRO entry, int* pMethod, long* pUncompLen, | 
|  | 358 | long* pCompLen, off_t* pOffset, long* pModWhen, long* pCrc32) const | 
|  | 359 | { | 
|  | 360 | int ent = entryToIndex(entry); | 
|  | 361 | if (ent < 0) | 
|  | 362 | return false; | 
|  | 363 |  | 
|  | 364 | /* | 
|  | 365 | * Recover the start of the central directory entry from the filename | 
|  | 366 | * pointer. | 
|  | 367 | */ | 
|  | 368 | const unsigned char* basePtr = (const unsigned char*)mFileMap->getDataPtr(); | 
|  | 369 | const unsigned char* ptr = (const unsigned char*) mHashTable[ent].name; | 
|  | 370 | size_t zipLength = mFileMap->getDataLength(); | 
|  | 371 |  | 
|  | 372 | ptr -= kCDELen; | 
|  | 373 |  | 
|  | 374 | int method = get2LE(ptr + kCDEMethod); | 
|  | 375 | if (pMethod != NULL) | 
|  | 376 | *pMethod = method; | 
|  | 377 |  | 
|  | 378 | if (pModWhen != NULL) | 
|  | 379 | *pModWhen = get4LE(ptr + kCDEModWhen); | 
|  | 380 | if (pCrc32 != NULL) | 
|  | 381 | *pCrc32 = get4LE(ptr + kCDECRC); | 
|  | 382 |  | 
|  | 383 | /* | 
|  | 384 | * We need to make sure that the lengths are not so large that somebody | 
|  | 385 | * trying to map the compressed or uncompressed data runs off the end | 
|  | 386 | * of the mapped region. | 
|  | 387 | */ | 
|  | 388 | unsigned long localHdrOffset = get4LE(ptr + kCDELocalOffset); | 
|  | 389 | if (localHdrOffset + kLFHLen >= zipLength) { | 
|  | 390 | LOGE("ERROR: bad local hdr offset in zip\n"); | 
|  | 391 | return false; | 
|  | 392 | } | 
|  | 393 | const unsigned char* localHdr = basePtr + localHdrOffset; | 
|  | 394 | off_t dataOffset = localHdrOffset + kLFHLen | 
|  | 395 | + get2LE(localHdr + kLFHNameLen) + get2LE(localHdr + kLFHExtraLen); | 
|  | 396 | if ((unsigned long) dataOffset >= zipLength) { | 
|  | 397 | LOGE("ERROR: bad data offset in zip\n"); | 
|  | 398 | return false; | 
|  | 399 | } | 
|  | 400 |  | 
|  | 401 | if (pCompLen != NULL) { | 
|  | 402 | *pCompLen = get4LE(ptr + kCDECompLen); | 
|  | 403 | if (*pCompLen < 0 || (size_t)(dataOffset + *pCompLen) >= zipLength) { | 
|  | 404 | LOGE("ERROR: bad compressed length in zip\n"); | 
|  | 405 | return false; | 
|  | 406 | } | 
|  | 407 | } | 
|  | 408 | if (pUncompLen != NULL) { | 
|  | 409 | *pUncompLen = get4LE(ptr + kCDEUncompLen); | 
|  | 410 | if (*pUncompLen < 0) { | 
|  | 411 | LOGE("ERROR: negative uncompressed length in zip\n"); | 
|  | 412 | return false; | 
|  | 413 | } | 
|  | 414 | if (method == kCompressStored && | 
|  | 415 | (size_t)(dataOffset + *pUncompLen) >= zipLength) | 
|  | 416 | { | 
|  | 417 | LOGE("ERROR: bad uncompressed length in zip\n"); | 
|  | 418 | return false; | 
|  | 419 | } | 
|  | 420 | } | 
|  | 421 |  | 
|  | 422 | if (pOffset != NULL) { | 
|  | 423 | *pOffset = dataOffset; | 
|  | 424 | } | 
|  | 425 | return true; | 
|  | 426 | } | 
|  | 427 |  | 
|  | 428 | /* | 
|  | 429 | * Copy the entry's filename to the buffer. | 
|  | 430 | */ | 
|  | 431 | int ZipFileRO::getEntryFileName(ZipEntryRO entry, char* buffer, int bufLen) | 
|  | 432 | const | 
|  | 433 | { | 
|  | 434 | int ent = entryToIndex(entry); | 
|  | 435 | if (ent < 0) | 
|  | 436 | return -1; | 
|  | 437 |  | 
|  | 438 | int nameLen = mHashTable[ent].nameLen; | 
|  | 439 | if (bufLen < nameLen+1) | 
|  | 440 | return nameLen+1; | 
|  | 441 |  | 
|  | 442 | memcpy(buffer, mHashTable[ent].name, nameLen); | 
|  | 443 | buffer[nameLen] = '\0'; | 
|  | 444 | return 0; | 
|  | 445 | } | 
|  | 446 |  | 
|  | 447 | /* | 
|  | 448 | * Create a new FileMap object that spans the data in "entry". | 
|  | 449 | */ | 
|  | 450 | FileMap* ZipFileRO::createEntryFileMap(ZipEntryRO entry) const | 
|  | 451 | { | 
|  | 452 | /* | 
|  | 453 | * TODO: the efficient way to do this is to modify FileMap to allow | 
|  | 454 | * sub-regions of a file to be mapped.  A reference-counting scheme | 
|  | 455 | * can manage the base memory mapping.  For now, we just create a brand | 
|  | 456 | * new mapping off of the Zip archive file descriptor. | 
|  | 457 | */ | 
|  | 458 |  | 
|  | 459 | FileMap* newMap; | 
|  | 460 | long compLen; | 
|  | 461 | off_t offset; | 
|  | 462 |  | 
|  | 463 | if (!getEntryInfo(entry, NULL, NULL, &compLen, &offset, NULL, NULL)) | 
|  | 464 | return NULL; | 
|  | 465 |  | 
|  | 466 | newMap = new FileMap(); | 
|  | 467 | if (!newMap->create(mFileMap->getFileName(), mFd, offset, compLen, true)) { | 
|  | 468 | newMap->release(); | 
|  | 469 | return NULL; | 
|  | 470 | } | 
|  | 471 |  | 
|  | 472 | return newMap; | 
|  | 473 | } | 
|  | 474 |  | 
|  | 475 | /* | 
|  | 476 | * Uncompress an entry, in its entirety, into the provided output buffer. | 
|  | 477 | * | 
|  | 478 | * This doesn't verify the data's CRC, which might be useful for | 
|  | 479 | * uncompressed data.  The caller should be able to manage it. | 
|  | 480 | */ | 
|  | 481 | bool ZipFileRO::uncompressEntry(ZipEntryRO entry, void* buffer) const | 
|  | 482 | { | 
|  | 483 | const int kSequentialMin = 32768; | 
|  | 484 | bool result = false; | 
|  | 485 | int ent = entryToIndex(entry); | 
|  | 486 | if (ent < 0) | 
|  | 487 | return -1; | 
|  | 488 |  | 
|  | 489 | const unsigned char* basePtr = (const unsigned char*)mFileMap->getDataPtr(); | 
|  | 490 | int method; | 
|  | 491 | long uncompLen, compLen; | 
|  | 492 | off_t offset; | 
|  | 493 |  | 
|  | 494 | getEntryInfo(entry, &method, &uncompLen, &compLen, &offset, NULL, NULL); | 
|  | 495 |  | 
|  | 496 | /* | 
|  | 497 | * Experiment with madvise hint.  When we want to uncompress a file, | 
|  | 498 | * we pull some stuff out of the central dir entry and then hit a | 
|  | 499 | * bunch of compressed or uncompressed data sequentially.  The CDE | 
|  | 500 | * visit will cause a limited amount of read-ahead because it's at | 
|  | 501 | * the end of the file.  We could end up doing lots of extra disk | 
|  | 502 | * access if the file we're prying open is small.  Bottom line is we | 
|  | 503 | * probably don't want to turn MADV_SEQUENTIAL on and leave it on. | 
|  | 504 | * | 
|  | 505 | * So, if the compressed size of the file is above a certain minimum | 
|  | 506 | * size, temporarily boost the read-ahead in the hope that the extra | 
|  | 507 | * pair of system calls are negated by a reduction in page faults. | 
|  | 508 | */ | 
|  | 509 | if (compLen > kSequentialMin) | 
|  | 510 | mFileMap->advise(FileMap::SEQUENTIAL); | 
|  | 511 |  | 
|  | 512 | if (method == kCompressStored) { | 
|  | 513 | memcpy(buffer, basePtr + offset, uncompLen); | 
|  | 514 | } else { | 
|  | 515 | if (!inflateBuffer(buffer, basePtr + offset, uncompLen, compLen)) | 
|  | 516 | goto bail; | 
|  | 517 | } | 
|  | 518 |  | 
|  | 519 | if (compLen > kSequentialMin) | 
|  | 520 | mFileMap->advise(FileMap::NORMAL); | 
|  | 521 |  | 
|  | 522 | result = true; | 
|  | 523 |  | 
|  | 524 | bail: | 
|  | 525 | return result; | 
|  | 526 | } | 
|  | 527 |  | 
|  | 528 | /* | 
|  | 529 | * Uncompress an entry, in its entirety, to an open file descriptor. | 
|  | 530 | * | 
|  | 531 | * This doesn't verify the data's CRC, but probably should. | 
|  | 532 | */ | 
|  | 533 | bool ZipFileRO::uncompressEntry(ZipEntryRO entry, int fd) const | 
|  | 534 | { | 
|  | 535 | bool result = false; | 
|  | 536 | int ent = entryToIndex(entry); | 
|  | 537 | if (ent < 0) | 
|  | 538 | return -1; | 
|  | 539 |  | 
|  | 540 | const unsigned char* basePtr = (const unsigned char*)mFileMap->getDataPtr(); | 
|  | 541 | int method; | 
|  | 542 | long uncompLen, compLen; | 
|  | 543 | off_t offset; | 
|  | 544 |  | 
|  | 545 | getEntryInfo(entry, &method, &uncompLen, &compLen, &offset, NULL, NULL); | 
|  | 546 |  | 
|  | 547 | if (method == kCompressStored) { | 
|  | 548 | ssize_t actual; | 
|  | 549 |  | 
|  | 550 | actual = write(fd, basePtr + offset, uncompLen); | 
|  | 551 | if (actual < 0) { | 
|  | 552 | LOGE("Write failed: %s\n", strerror(errno)); | 
|  | 553 | goto bail; | 
|  | 554 | } else if (actual != uncompLen) { | 
|  | 555 | LOGE("Partial write during uncompress (%d of %ld)\n", | 
|  | 556 | (int)actual, uncompLen); | 
|  | 557 | goto bail; | 
|  | 558 | } else { | 
|  | 559 | LOGI("+++ successful write\n"); | 
|  | 560 | } | 
|  | 561 | } else { | 
|  | 562 | if (!inflateBuffer(fd, basePtr+offset, uncompLen, compLen)) | 
|  | 563 | goto bail; | 
|  | 564 | } | 
|  | 565 |  | 
|  | 566 | result = true; | 
|  | 567 |  | 
|  | 568 | bail: | 
|  | 569 | return result; | 
|  | 570 | } | 
|  | 571 |  | 
|  | 572 | /* | 
|  | 573 | * Uncompress "deflate" data from one buffer to another. | 
|  | 574 | */ | 
|  | 575 | /*static*/ bool ZipFileRO::inflateBuffer(void* outBuf, const void* inBuf, | 
|  | 576 | long uncompLen, long compLen) | 
|  | 577 | { | 
|  | 578 | bool result = false; | 
|  | 579 | z_stream zstream; | 
|  | 580 | int zerr; | 
|  | 581 |  | 
|  | 582 | /* | 
|  | 583 | * Initialize the zlib stream struct. | 
|  | 584 | */ | 
|  | 585 | memset(&zstream, 0, sizeof(zstream)); | 
|  | 586 | zstream.zalloc = Z_NULL; | 
|  | 587 | zstream.zfree = Z_NULL; | 
|  | 588 | zstream.opaque = Z_NULL; | 
|  | 589 | zstream.next_in = (Bytef*)inBuf; | 
|  | 590 | zstream.avail_in = compLen; | 
|  | 591 | zstream.next_out = (Bytef*) outBuf; | 
|  | 592 | zstream.avail_out = uncompLen; | 
|  | 593 | zstream.data_type = Z_UNKNOWN; | 
|  | 594 |  | 
|  | 595 | /* | 
|  | 596 | * Use the undocumented "negative window bits" feature to tell zlib | 
|  | 597 | * that there's no zlib header waiting for it. | 
|  | 598 | */ | 
|  | 599 | zerr = inflateInit2(&zstream, -MAX_WBITS); | 
|  | 600 | if (zerr != Z_OK) { | 
|  | 601 | if (zerr == Z_VERSION_ERROR) { | 
|  | 602 | LOGE("Installed zlib is not compatible with linked version (%s)\n", | 
|  | 603 | ZLIB_VERSION); | 
|  | 604 | } else { | 
|  | 605 | LOGE("Call to inflateInit2 failed (zerr=%d)\n", zerr); | 
|  | 606 | } | 
|  | 607 | goto bail; | 
|  | 608 | } | 
|  | 609 |  | 
|  | 610 | /* | 
|  | 611 | * Expand data. | 
|  | 612 | */ | 
|  | 613 | zerr = inflate(&zstream, Z_FINISH); | 
|  | 614 | if (zerr != Z_STREAM_END) { | 
|  | 615 | LOGW("Zip inflate failed, zerr=%d (nIn=%p aIn=%u nOut=%p aOut=%u)\n", | 
|  | 616 | zerr, zstream.next_in, zstream.avail_in, | 
|  | 617 | zstream.next_out, zstream.avail_out); | 
|  | 618 | goto z_bail; | 
|  | 619 | } | 
|  | 620 |  | 
|  | 621 | /* paranoia */ | 
|  | 622 | if ((long) zstream.total_out != uncompLen) { | 
|  | 623 | LOGW("Size mismatch on inflated file (%ld vs %ld)\n", | 
|  | 624 | zstream.total_out, uncompLen); | 
|  | 625 | goto z_bail; | 
|  | 626 | } | 
|  | 627 |  | 
|  | 628 | result = true; | 
|  | 629 |  | 
|  | 630 | z_bail: | 
|  | 631 | inflateEnd(&zstream);        /* free up any allocated structures */ | 
|  | 632 |  | 
|  | 633 | bail: | 
|  | 634 | return result; | 
|  | 635 | } | 
|  | 636 |  | 
|  | 637 | /* | 
|  | 638 | * Uncompress "deflate" data from one buffer to an open file descriptor. | 
|  | 639 | */ | 
|  | 640 | /*static*/ bool ZipFileRO::inflateBuffer(int fd, const void* inBuf, | 
|  | 641 | long uncompLen, long compLen) | 
|  | 642 | { | 
|  | 643 | bool result = false; | 
|  | 644 | const int kWriteBufSize = 32768; | 
|  | 645 | unsigned char writeBuf[kWriteBufSize]; | 
|  | 646 | z_stream zstream; | 
|  | 647 | int zerr; | 
|  | 648 |  | 
|  | 649 | /* | 
|  | 650 | * Initialize the zlib stream struct. | 
|  | 651 | */ | 
|  | 652 | memset(&zstream, 0, sizeof(zstream)); | 
|  | 653 | zstream.zalloc = Z_NULL; | 
|  | 654 | zstream.zfree = Z_NULL; | 
|  | 655 | zstream.opaque = Z_NULL; | 
|  | 656 | zstream.next_in = (Bytef*)inBuf; | 
|  | 657 | zstream.avail_in = compLen; | 
|  | 658 | zstream.next_out = (Bytef*) writeBuf; | 
|  | 659 | zstream.avail_out = sizeof(writeBuf); | 
|  | 660 | zstream.data_type = Z_UNKNOWN; | 
|  | 661 |  | 
|  | 662 | /* | 
|  | 663 | * Use the undocumented "negative window bits" feature to tell zlib | 
|  | 664 | * that there's no zlib header waiting for it. | 
|  | 665 | */ | 
|  | 666 | zerr = inflateInit2(&zstream, -MAX_WBITS); | 
|  | 667 | if (zerr != Z_OK) { | 
|  | 668 | if (zerr == Z_VERSION_ERROR) { | 
|  | 669 | LOGE("Installed zlib is not compatible with linked version (%s)\n", | 
|  | 670 | ZLIB_VERSION); | 
|  | 671 | } else { | 
|  | 672 | LOGE("Call to inflateInit2 failed (zerr=%d)\n", zerr); | 
|  | 673 | } | 
|  | 674 | goto bail; | 
|  | 675 | } | 
|  | 676 |  | 
|  | 677 | /* | 
|  | 678 | * Loop while we have more to do. | 
|  | 679 | */ | 
|  | 680 | do { | 
|  | 681 | /* | 
|  | 682 | * Expand data. | 
|  | 683 | */ | 
|  | 684 | zerr = inflate(&zstream, Z_NO_FLUSH); | 
|  | 685 | if (zerr != Z_OK && zerr != Z_STREAM_END) { | 
|  | 686 | LOGW("zlib inflate: zerr=%d (nIn=%p aIn=%u nOut=%p aOut=%u)\n", | 
|  | 687 | zerr, zstream.next_in, zstream.avail_in, | 
|  | 688 | zstream.next_out, zstream.avail_out); | 
|  | 689 | goto z_bail; | 
|  | 690 | } | 
|  | 691 |  | 
|  | 692 | /* write when we're full or when we're done */ | 
|  | 693 | if (zstream.avail_out == 0 || | 
|  | 694 | (zerr == Z_STREAM_END && zstream.avail_out != sizeof(writeBuf))) | 
|  | 695 | { | 
|  | 696 | long writeSize = zstream.next_out - writeBuf; | 
|  | 697 | int cc = write(fd, writeBuf, writeSize); | 
|  | 698 | if (cc != (int) writeSize) { | 
|  | 699 | LOGW("write failed in inflate (%d vs %ld)\n", cc, writeSize); | 
|  | 700 | goto z_bail; | 
|  | 701 | } | 
|  | 702 |  | 
|  | 703 | zstream.next_out = writeBuf; | 
|  | 704 | zstream.avail_out = sizeof(writeBuf); | 
|  | 705 | } | 
|  | 706 | } while (zerr == Z_OK); | 
|  | 707 |  | 
|  | 708 | assert(zerr == Z_STREAM_END);       /* other errors should've been caught */ | 
|  | 709 |  | 
|  | 710 | /* paranoia */ | 
|  | 711 | if ((long) zstream.total_out != uncompLen) { | 
|  | 712 | LOGW("Size mismatch on inflated file (%ld vs %ld)\n", | 
|  | 713 | zstream.total_out, uncompLen); | 
|  | 714 | goto z_bail; | 
|  | 715 | } | 
|  | 716 |  | 
|  | 717 | result = true; | 
|  | 718 |  | 
|  | 719 | z_bail: | 
|  | 720 | inflateEnd(&zstream);        /* free up any allocated structures */ | 
|  | 721 |  | 
|  | 722 | bail: | 
|  | 723 | return result; | 
|  | 724 | } |