StringPool.cpp

   1 //
   2 // Copyright 2006 The Android Open Source Project
   3 //
   4 // Build resource files from raw assets.
   5 //
   6
   7 #include "StringPool.h"
   8 #include "ResourceTable.h"
   9
  10 #include <utils/ByteOrder.h>
  11 #include <utils/SortedVector.h>
  12
  13 #if HAVE_PRINTF_ZD
  14 #  define ZD "%zd"
  15 #  define ZD_TYPE ssize_t
  16 #else
  17 #  define ZD "%ld"
  18 #  define ZD_TYPE long
  19 #endif
  20
  21 #define NOISY(x) //x
  22
  23 void strcpy16_htod(uint16_t* dst, const uint16_t* src)
  24 {
  25     while (*src) {
  26         char16_t s = htods(*src);
  27         *dst++ = s;
  28         src++;
  29     }
  30     *dst = 0;
  31 }
  32
  33 void printStringPool(const ResStringPool* pool)
  34 {
  35     SortedVector<const void*> uniqueStrings;
  36     const size_t N = pool->size();
  37     for (size_t i=0; i<N; i++) {
  38         size_t len;
  39         if (pool->isUTF8()) {
  40             uniqueStrings.add(pool->string8At(i, &len));
  41         } else {
  42             uniqueStrings.add(pool->stringAt(i, &len));
  43         }
  44     }
  45
  46     printf("String pool of " ZD " unique %s %s strings, " ZD " entries and "
  47             ZD " styles using " ZD " bytes:\n",
  48             (ZD_TYPE)uniqueStrings.size(), pool->isUTF8() ? "UTF-8" : "UTF-16",
  49             pool->isSorted() ? "sorted" : "non-sorted",
  50             (ZD_TYPE)N, (ZD_TYPE)pool->styleCount(), (ZD_TYPE)pool->bytes());
  51
  52     const size_t NS = pool->size();
  53     for (size_t s=0; s<NS; s++) {
  54         String8 str = pool->string8ObjectAt(s);
  55         printf("String #" ZD ": %s\n", (ZD_TYPE) s, str.string());
  56     }
  57 }
  58
  59 String8 StringPool::entry::makeConfigsString() const {
  60     String8 configStr(configTypeName);
  61     if (configStr.size() > 0) configStr.append(" ");
  62     if (configs.size() > 0) {
  63         for (size_t j=0; j<configs.size(); j++) {
  64             if (j > 0) configStr.append(", ");
  65             configStr.append(configs[j].toString());
  66         }
  67     } else {
  68         configStr = "(none)";
  69     }
  70     return configStr;
  71 }
  72
  73 int StringPool::entry::compare(const entry& o) const {
  74     // Strings with styles go first, to reduce the size of the
  75     // styles array.
  76     if (hasStyles) {
  77         return o.hasStyles ? 0 : -1;
  78     }
  79     if (o.hasStyles) {
  80         return 1;
  81     }
  82     int comp = configTypeName.compare(o.configTypeName);
  83     if (comp != 0) {
  84         return comp;
  85     }
  86     const size_t LHN = configs.size();
  87     const size_t RHN = o.configs.size();
  88     size_t i=0;
  89     while (i < LHN && i < RHN) {
  90         comp = configs[i].compareLogical(o.configs[i]);
  91         if (comp != 0) {
  92             return comp;
  93         }
  94         i++;
  95     }
  96     if (LHN < RHN) return -1;
  97     else if (LHN > RHN) return 1;
  98     return 0;
  99 }
 100
 101 StringPool::StringPool(bool sorted, bool utf8)
 102     : mSorted(sorted), mUTF8(utf8), mValues(-1), mIdents(-1)
 103 {
 104 }
 105
 106 ssize_t StringPool::add(const String16& value, bool mergeDuplicates,
 107         const String8* configTypeName, const ResTable_config* config)
 108 {
 109     return add(String16(), value, mergeDuplicates, configTypeName, config);
 110 }
 111
 112 ssize_t StringPool::add(const String16& value, const Vector<entry_style_span>& spans,
 113         const String8* configTypeName, const ResTable_config* config)
 114 {
 115     ssize_t res = add(String16(), value, false, configTypeName, config);
 116     if (res >= 0) {
 117         addStyleSpans(res, spans);
 118     }
 119     return res;
 120 }
 121
 122 ssize_t StringPool::add(const String16& ident, const String16& value,
 123         bool mergeDuplicates, const String8* configTypeName, const ResTable_config* config)
 124 {
 125     if (ident.size() > 0) {
 126         ssize_t idx = mIdents.valueFor(ident);
 127         if (idx >= 0) {
 128             fprintf(stderr, "ERROR: Duplicate string identifier %s\n",
 129                     String8(mEntries[idx].value).string());
 130             return UNKNOWN_ERROR;
 131         }
 132     }
 133
 134     ssize_t vidx = mValues.indexOfKey(value);
 135     ssize_t pos = vidx >= 0 ? mValues.valueAt(vidx) : -1;
 136     ssize_t eidx = pos >= 0 ? mEntryArray.itemAt(pos) : -1;
 137     if (eidx < 0) {
 138         eidx = mEntries.add(entry(value));
 139         if (eidx < 0) {
 140             fprintf(stderr, "Failure adding string %s\n", String8(value).string());
 141             return eidx;
 142         }
 143     }
 144
 145     if (configTypeName != NULL) {
 146         entry& ent = mEntries.editItemAt(eidx);
 147         NOISY(printf("*** adding config type name %s, was %s\n",
 148                 configTypeName->string(), ent.configTypeName.string()));
 149         if (ent.configTypeName.size() <= 0) {
 150             ent.configTypeName = *configTypeName;
 151         } else if (ent.configTypeName != *configTypeName) {
 152             ent.configTypeName = " ";
 153         }
 154     }
 155
 156     if (config != NULL) {
 157         // Add this to the set of configs associated with the string.
 158         entry& ent = mEntries.editItemAt(eidx);
 159         size_t addPos;
 160         for (addPos=0; addPos<ent.configs.size(); addPos++) {
 161             int cmp = ent.configs.itemAt(addPos).compareLogical(*config);
 162             if (cmp >= 0) {
 163                 if (cmp > 0) {
 164                     NOISY(printf("*** inserting config: %s\n", config->toString().string()));
 165                     ent.configs.insertAt(*config, addPos);
 166                 }
 167                 break;
 168             }
 169         }
 170         if (addPos >= ent.configs.size()) {
 171             NOISY(printf("*** adding config: %s\n", config->toString().string()));
 172             ent.configs.add(*config);
 173         }
 174     }
 175
 176     const bool first = vidx < 0;
 177     if (first || !mergeDuplicates) {
 178         pos = mEntryArray.add(eidx);
 179         if (first) {
 180             vidx = mValues.add(value, pos);
 181         }
 182         if (!mSorted) {
 183             entry& ent = mEntries.editItemAt(eidx);
 184             ent.indices.add(pos);
 185         }
 186     }
 187
 188     if (ident.size() > 0) {
 189         mIdents.add(ident, vidx);
 190     }
 191
 192     NOISY(printf("Adding string %s to pool: pos=%d eidx=%d vidx=%d\n",
 193             String8(value).string(), pos, eidx, vidx));
 194
 195     return pos;
 196 }
 197
 198 status_t StringPool::addStyleSpan(size_t idx, const String16& name,
 199                                   uint32_t start, uint32_t end)
 200 {
 201     entry_style_span span;
 202     span.name = name;
 203     span.span.firstChar = start;
 204     span.span.lastChar = end;
 205     return addStyleSpan(idx, span);
 206 }
 207
 208 status_t StringPool::addStyleSpans(size_t idx, const Vector<entry_style_span>& spans)
 209 {
 210     const size_t N=spans.size();
 211     for (size_t i=0; i<N; i++) {
 212         status_t err = addStyleSpan(idx, spans[i]);
 213         if (err != NO_ERROR) {
 214             return err;
 215         }
 216     }
 217     return NO_ERROR;
 218 }
 219
 220 status_t StringPool::addStyleSpan(size_t idx, const entry_style_span& span)
 221 {
 222     LOG_ALWAYS_FATAL_IF(mSorted, "Can't use styles with sorted string pools.");
 223
 224     // Place blank entries in the span array up to this index.
 225     while (mEntryStyleArray.size() <= idx) {
 226         mEntryStyleArray.add();
 227     }
 228
 229     entry_style& style = mEntryStyleArray.editItemAt(idx);
 230     style.spans.add(span);
 231     mEntries.editItemAt(mEntryArray[idx]).hasStyles = true;
 232     return NO_ERROR;
 233 }
 234
 235 size_t StringPool::size() const
 236 {
 237     return mSorted ? mValues.size() : mEntryArray.size();
 238 }
 239
 240 const StringPool::entry& StringPool::entryAt(size_t idx) const
 241 {
 242     if (!mSorted) {
 243         return mEntries[mEntryArray[idx]];
 244     } else {
 245         return mEntries[mEntryArray[mValues.valueAt(idx)]];
 246     }
 247 }
 248
 249 size_t StringPool::countIdentifiers() const
 250 {
 251     return mIdents.size();
 252 }
 253
 254 int StringPool::config_sort(const size_t* lhs, const size_t* rhs, void* state)
 255 {
 256     StringPool* pool = (StringPool*)state;
 257     const entry& lhe = pool->mEntries[pool->mEntryArray[*lhs]];
 258     const entry& rhe = pool->mEntries[pool->mEntryArray[*rhs]];
 259     return lhe.compare(rhe);
 260 }
 261
 262 void StringPool::sortByConfig()
 263 {
 264     LOG_ALWAYS_FATAL_IF(mSorted, "Can't sort string pool containing identifiers.");
 265     LOG_ALWAYS_FATAL_IF(mIdents.size() > 0, "Can't sort string pool containing identifiers.");
 266     LOG_ALWAYS_FATAL_IF(mOriginalPosToNewPos.size() > 0, "Can't sort string pool after already sorted.");
 267
 268     const size_t N = mEntryArray.size();
 269
 270     // This is a vector that starts out with a 1:1 mapping to entries
 271     // in the array, which we will sort to come up with the desired order.
 272     // At that point it maps from the new position in the array to the
 273     // original position the entry appeared.
 274     Vector<size_t> newPosToOriginalPos;
 275     for (size_t i=0; i<mEntryArray.size(); i++) {
 276         newPosToOriginalPos.add(i);
 277     }
 278
 279     // Sort the array.
 280     NOISY(printf("SORTING STRINGS BY CONFIGURATION...\n"));
 281     newPosToOriginalPos.sort(config_sort, this);
 282     NOISY(printf("DONE SORTING STRINGS BY CONFIGURATION.\n"));
 283
 284     // Create the reverse mapping from the original position in the array
 285     // to the new position where it appears in the sorted array.  This is
 286     // so that clients can re-map any positions they had previously stored.
 287     mOriginalPosToNewPos = newPosToOriginalPos;
 288     for (size_t i=0; i<N; i++) {
 289         mOriginalPosToNewPos.editItemAt(newPosToOriginalPos[i]) = i;
 290     }
 291
 292 #if 0
 293     SortedVector<entry> entries;
 294
 295     for (size_t i=0; i<N; i++) {
 296         printf("#%d was %d: %s\n", i, newPosToOriginalPos[i],
 297                 mEntries[mEntryArray[newPosToOriginalPos[i]]].makeConfigsString().string());
 298         entries.add(mEntries[mEntryArray[i]]);
 299     }
 300
 301     for (size_t i=0; i<entries.size(); i++) {
 302         printf("Sorted config #%d: %s\n", i,
 303                 entries[i].makeConfigsString().string());
 304     }
 305 #endif
 306
 307     // Now we rebuild the arrays.
 308     Vector<entry> newEntries;
 309     Vector<size_t> newEntryArray;
 310     Vector<entry_style> newEntryStyleArray;
 311     DefaultKeyedVector<size_t, size_t> origOffsetToNewOffset;
 312
 313     for (size_t i=0; i<N; i++) {
 314         // We are filling in new offset 'i'; oldI is where we can find it
 315         // in the original data structure.
 316         size_t oldI = newPosToOriginalPos[i];
 317         // This is the actual entry associated with the old offset.
 318         const entry& oldEnt = mEntries[mEntryArray[oldI]];
 319         // This is the same entry the last time we added it to the
 320         // new entry array, if any.
 321         ssize_t newIndexOfOffset = origOffsetToNewOffset.indexOfKey(oldI);
 322         size_t newOffset;
 323         if (newIndexOfOffset < 0) {
 324             // This is the first time we have seen the entry, so add
 325             // it.
 326             newOffset = newEntries.add(oldEnt);
 327             newEntries.editItemAt(newOffset).indices.clear();
 328         } else {
 329             // We have seen this entry before, use the existing one
 330             // instead of adding it again.
 331             newOffset = origOffsetToNewOffset.valueAt(newIndexOfOffset);
 332         }
 333         // Update the indices to include this new position.
 334         newEntries.editItemAt(newOffset).indices.add(i);
 335         // And add the offset of the entry to the new entry array.
 336         newEntryArray.add(newOffset);
 337         // Add any old style to the new style array.
 338         if (mEntryStyleArray.size() > 0) {
 339             if (oldI < mEntryStyleArray.size()) {
 340                 newEntryStyleArray.add(mEntryStyleArray[oldI]);
 341             } else {
 342                 newEntryStyleArray.add(entry_style());
 343             }
 344         }
 345     }
 346
 347     // Now trim any entries at the end of the new style array that are
 348     // not needed.
 349     for (ssize_t i=newEntryStyleArray.size()-1; i>=0; i--) {
 350         const entry_style& style = newEntryStyleArray[i];
 351         if (style.spans.size() > 0) {
 352             // That's it.
 353             break;
 354         }
 355         // This one is not needed; remove.
 356         newEntryStyleArray.removeAt(i);
 357     }
 358
 359     // All done, install the new data structures and upate mValues with
 360     // the new positions.
 361     mEntries = newEntries;
 362     mEntryArray = newEntryArray;
 363     mEntryStyleArray = newEntryStyleArray;
 364     mValues.clear();
 365     for (size_t i=0; i<mEntries.size(); i++) {
 366         const entry& ent = mEntries[i];
 367         mValues.add(ent.value, ent.indices[0]);
 368     }
 369
 370 #if 0
 371     printf("FINAL SORTED STRING CONFIGS:\n");
 372     for (size_t i=0; i<mEntries.size(); i++) {
 373         const entry& ent = mEntries[i];
 374         printf("#" ZD " %s: %s\n", (ZD_TYPE)i, ent.makeConfigsString().string(),
 375                 String8(ent.value).string());
 376     }
 377 #endif
 378 }
 379
 380 sp<AaptFile> StringPool::createStringBlock()
 381 {
 382     sp<AaptFile> pool = new AaptFile(String8(), AaptGroupEntry(),
 383                                      String8());
 384     status_t err = writeStringBlock(pool);
 385     return err == NO_ERROR ? pool : NULL;
 386 }
 387
 388 #define ENCODE_LENGTH(str, chrsz, strSize) \
 389 { \
 390     size_t maxMask = 1 << ((chrsz*8)-1); \
 391     size_t maxSize = maxMask-1; \
 392     if (strSize > maxSize) { \
 393         *str++ = maxMask | ((strSize>>(chrsz*8))&maxSize); \
 394     } \
 395     *str++ = strSize; \
 396 }
 397
 398 status_t StringPool::writeStringBlock(const sp<AaptFile>& pool)
 399 {
 400     // Allow appending.  Sorry this is a little wacky.
 401     if (pool->getSize() > 0) {
 402         sp<AaptFile> block = createStringBlock();
 403         if (block == NULL) {
 404             return UNKNOWN_ERROR;
 405         }
 406         ssize_t res = pool->writeData(block->getData(), block->getSize());
 407         return (res >= 0) ? (status_t)NO_ERROR : res;
 408     }
 409
 410     // First we need to add all style span names to the string pool.
 411     // We do this now (instead of when the span is added) so that these
 412     // will appear at the end of the pool, not disrupting the order
 413     // our client placed their own strings in it.
 414
 415     const size_t STYLES = mEntryStyleArray.size();
 416     size_t i;
 417
 418     for (i=0; i<STYLES; i++) {
 419         entry_style& style = mEntryStyleArray.editItemAt(i);
 420         const size_t N = style.spans.size();
 421         for (size_t i=0; i<N; i++) {
 422             entry_style_span& span = style.spans.editItemAt(i);
 423             ssize_t idx = add(span.name, true);
 424             if (idx < 0) {
 425                 fprintf(stderr, "Error adding span for style tag '%s'\n",
 426                         String8(span.name).string());
 427                 return idx;
 428             }
 429             span.span.name.index = (uint32_t)idx;
 430         }
 431     }
 432
 433     const size_t ENTRIES = size();
 434
 435     // Now build the pool of unique strings.
 436
 437     const size_t STRINGS = mEntries.size();
 438     const size_t preSize = sizeof(ResStringPool_header)
 439                          + (sizeof(uint32_t)*ENTRIES)
 440                          + (sizeof(uint32_t)*STYLES);
 441     if (pool->editData(preSize) == NULL) {
 442         fprintf(stderr, "ERROR: Out of memory for string pool\n");
 443         return NO_MEMORY;
 444     }
 445
 446     const size_t charSize = mUTF8 ? sizeof(uint8_t) : sizeof(char16_t);
 447
 448     size_t strPos = 0;
 449     for (i=0; i<STRINGS; i++) {
 450         entry& ent = mEntries.editItemAt(i);
 451         const size_t strSize = (ent.value.size());
 452         const size_t lenSize = strSize > (size_t)(1<<((charSize*8)-1))-1 ?
 453             charSize*2 : charSize;
 454
 455         String8 encStr;
 456         if (mUTF8) {
 457             encStr = String8(ent.value);
 458         }
 459
 460         const size_t encSize = mUTF8 ? encStr.size() : 0;
 461         const size_t encLenSize = mUTF8 ?
 462             (encSize > (size_t)(1<<((charSize*8)-1))-1 ?
 463                 charSize*2 : charSize) : 0;
 464
 465         ent.offset = strPos;
 466
 467         const size_t totalSize = lenSize + encLenSize +
 468             ((mUTF8 ? encSize : strSize)+1)*charSize;
 469
 470         void* dat = (void*)pool->editData(preSize + strPos + totalSize);
 471         if (dat == NULL) {
 472             fprintf(stderr, "ERROR: Out of memory for string pool\n");
 473             return NO_MEMORY;
 474         }
 475         dat = (uint8_t*)dat + preSize + strPos;
 476         if (mUTF8) {
 477             uint8_t* strings = (uint8_t*)dat;
 478
 479             ENCODE_LENGTH(strings, sizeof(uint8_t), strSize)
 480
 481             ENCODE_LENGTH(strings, sizeof(uint8_t), encSize)
 482
 483             strncpy((char*)strings, encStr, encSize+1);
 484         } else {
 485             uint16_t* strings = (uint16_t*)dat;
 486
 487             ENCODE_LENGTH(strings, sizeof(uint16_t), strSize)
 488
 489             strcpy16_htod(strings, ent.value);
 490         }
 491
 492         strPos += totalSize;
 493     }
 494
 495     // Pad ending string position up to a uint32_t boundary.
 496
 497     if (strPos&0x3) {
 498         size_t padPos = ((strPos+3)&~0x3);
 499         uint8_t* dat = (uint8_t*)pool->editData(preSize + padPos);
 500         if (dat == NULL) {
 501             fprintf(stderr, "ERROR: Out of memory padding string pool\n");
 502             return NO_MEMORY;
 503         }
 504         memset(dat+preSize+strPos, 0, padPos-strPos);
 505         strPos = padPos;
 506     }
 507
 508     // Build the pool of style spans.
 509
 510     size_t styPos = strPos;
 511     for (i=0; i<STYLES; i++) {
 512         entry_style& ent = mEntryStyleArray.editItemAt(i);
 513         const size_t N = ent.spans.size();
 514         const size_t totalSize = (N*sizeof(ResStringPool_span))
 515                                + sizeof(ResStringPool_ref);
 516
 517         ent.offset = styPos-strPos;
 518         uint8_t* dat = (uint8_t*)pool->editData(preSize + styPos + totalSize);
 519         if (dat == NULL) {
 520             fprintf(stderr, "ERROR: Out of memory for string styles\n");
 521             return NO_MEMORY;
 522         }
 523         ResStringPool_span* span = (ResStringPool_span*)(dat+preSize+styPos);
 524         for (size_t i=0; i<N; i++) {
 525             span->name.index = htodl(ent.spans[i].span.name.index);
 526             span->firstChar = htodl(ent.spans[i].span.firstChar);
 527             span->lastChar = htodl(ent.spans[i].span.lastChar);
 528             span++;
 529         }
 530         span->name.index = htodl(ResStringPool_span::END);
 531
 532         styPos += totalSize;
 533     }
 534
 535     if (STYLES > 0) {
 536         // Add full terminator at the end (when reading we validate that
 537         // the end of the pool is fully terminated to simplify error
 538         // checking).
 539         size_t extra = sizeof(ResStringPool_span)-sizeof(ResStringPool_ref);
 540         uint8_t* dat = (uint8_t*)pool->editData(preSize + styPos + extra);
 541         if (dat == NULL) {
 542             fprintf(stderr, "ERROR: Out of memory for string styles\n");
 543             return NO_MEMORY;
 544         }
 545         uint32_t* p = (uint32_t*)(dat+preSize+styPos);
 546         while (extra > 0) {
 547             *p++ = htodl(ResStringPool_span::END);
 548             extra -= sizeof(uint32_t);
 549         }
 550         styPos += extra;
 551     }
 552
 553     // Write header.
 554
 555     ResStringPool_header* header =
 556         (ResStringPool_header*)pool->padData(sizeof(uint32_t));
 557     if (header == NULL) {
 558         fprintf(stderr, "ERROR: Out of memory for string pool\n");
 559         return NO_MEMORY;
 560     }
 561     memset(header, 0, sizeof(*header));
 562     header->header.type = htods(RES_STRING_POOL_TYPE);
 563     header->header.headerSize = htods(sizeof(*header));
 564     header->header.size = htodl(pool->getSize());
 565     header->stringCount = htodl(ENTRIES);
 566     header->styleCount = htodl(STYLES);
 567     if (mSorted) {
 568         header->flags |= htodl(ResStringPool_header::SORTED_FLAG);
 569     }
 570     if (mUTF8) {
 571         header->flags |= htodl(ResStringPool_header::UTF8_FLAG);
 572     }
 573     header->stringsStart = htodl(preSize);
 574     header->stylesStart = htodl(STYLES > 0 ? (preSize+strPos) : 0);
 575
 576     // Write string index array.
 577
 578     uint32_t* index = (uint32_t*)(header+1);
 579     if (mSorted) {
 580         for (i=0; i<ENTRIES; i++) {
 581             entry& ent = const_cast<entry&>(entryAt(i));
 582             ent.indices.clear();
 583             ent.indices.add(i);
 584             *index++ = htodl(ent.offset);
 585         }
 586     } else {
 587         for (i=0; i<ENTRIES; i++) {
 588             entry& ent = mEntries.editItemAt(mEntryArray[i]);
 589             *index++ = htodl(ent.offset);
 590             NOISY(printf("Writing entry #%d: \"%s\" ent=%d off=%d\n", i,
 591                     String8(ent.value).string(),
 592                     mEntryArray[i], ent.offset));
 593         }
 594     }
 595
 596     // Write style index array.
 597
 598     if (mSorted) {
 599         for (i=0; i<STYLES; i++) {
 600             LOG_ALWAYS_FATAL("Shouldn't be here!");
 601         }
 602     } else {
 603         for (i=0; i<STYLES; i++) {
 604             *index++ = htodl(mEntryStyleArray[i].offset);
 605         }
 606     }
 607
 608     return NO_ERROR;
 609 }
 610
 611 ssize_t StringPool::offsetForString(const String16& val) const
 612 {
 613     const Vector<size_t>* indices = offsetsForString(val);
 614     ssize_t res = indices != NULL && indices->size() > 0 ? indices->itemAt(0) : -1;
 615     NOISY(printf("Offset for string %s: %d (%s)\n", String8(val).string(), res,
 616             res >= 0 ? String8(mEntries[mEntryArray[res]].value).string() : String8()));
 617     return res;
 618 }
 619
 620 const Vector<size_t>* StringPool::offsetsForString(const String16& val) const
 621 {
 622     ssize_t pos = mValues.valueFor(val);
 623     if (pos < 0) {
 624         return NULL;
 625     }
 626     return &mEntries[mEntryArray[pos]].indices;
 627 }