icuSources/common/udata.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 ******************************************************************************
   5 *
   6 *   Copyright (C) 1999-2016, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 ******************************************************************************
  10 *   file name:  udata.cpp
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 1999oct25
  16 *   created by: Markus W. Scherer
  17 */
  18
  19 #include "unicode/utypes.h"  /* U_PLATFORM etc. */
  20
  21 #ifdef __GNUC__
  22 /* if gcc
  23 #define ATTRIBUTE_WEAK __attribute__ ((weak))
  24 might have to #include some other header
  25 */
  26 #endif
  27
  28 #include "unicode/putil.h"
  29 #include "unicode/udata.h"
  30 #include "unicode/uversion.h"
  31 #include "charstr.h"
  32 #include "cmemory.h"
  33 #include "cstring.h"
  34 #include "mutex.h"
  35 #include "putilimp.h"
  36 #include "uassert.h"
  37 #include "ucln_cmn.h"
  38 #include "ucmndata.h"
  39 #include "udatamem.h"
  40 #include "uhash.h"
  41 #include "umapfile.h"
  42 #include "umutex.h"
  43
  44 /***********************************************************************
  45 *
  46 *   Notes on the organization of the ICU data implementation
  47 *
  48 *      All of the public API is defined in udata.h
  49 *
  50 *      The implementation is split into several files...
  51 *
  52 *         - udata.c  (this file) contains higher level code that knows about
  53 *                     the search paths for locating data, caching opened data, etc.
  54 *
  55 *         - umapfile.c  contains the low level platform-specific code for actually loading
  56 *                     (memory mapping, file reading, whatever) data into memory.
  57 *
  58 *         - ucmndata.c  deals with the tables of contents of ICU data items within
  59 *                     an ICU common format data file.  The implementation includes
  60 *                     an abstract interface and support for multiple TOC formats.
  61 *                     All knowledge of any specific TOC format is encapsulated here.
  62 *
  63 *         - udatamem.c has code for managing UDataMemory structs.  These are little
  64 *                     descriptor objects for blocks of memory holding ICU data of
  65 *                     various types.
  66 */
  67
  68 /* configuration ---------------------------------------------------------- */
  69
  70 /* If you are excruciatingly bored turn this on .. */
  71 /* #define UDATA_DEBUG 1 */
  72 /* For debugging use of timezone data in a separate file */
  73 /* #define UDATA_TZFILES_DEBUG 1 */
  74
  75 #if defined(UDATA_DEBUG) || defined(UDATA_TZFILES_DEBUG)
  76 #   include <stdio.h>
  77 #endif
  78
  79 U_NAMESPACE_USE
  80
  81 /*
  82  *  Forward declarations
  83  */
  84 static UDataMemory *udata_findCachedData(const char *path, UErrorCode &err);
  85
  86 /***********************************************************************
  87 *
  88 *    static (Global) data
  89 *
  90 ************************************************************************/
  91
  92 /*
  93  * Pointers to the common ICU data.
  94  *
  95  * We store multiple pointers to ICU data packages and iterate through them
  96  * when looking for a data item.
  97  *
  98  * It is possible to combine this with dependency inversion:
  99  * One or more data package libraries may export
 100  * functions that each return a pointer to their piece of the ICU data,
 101  * and this file would import them as weak functions, without a
 102  * strong linker dependency from the common library on the data library.
 103  *
 104  * Then we can have applications depend on only that part of ICU's data
 105  * that they really need, reducing the size of binaries that take advantage
 106  * of this.
 107  */
 108 static UDataMemory *gCommonICUDataArray[10] = { NULL };   // Access protected by icu global mutex.
 109
 110 static u_atomic_int32_t gHaveTriedToLoadCommonData = ATOMIC_INT32_T_INITIALIZER(0);  //  See extendICUData().
 111
 112 static UHashtable  *gCommonDataCache = NULL;  /* Global hash table of opened ICU data files.  */
 113 static icu::UInitOnce gCommonDataCacheInitOnce = U_INITONCE_INITIALIZER;
 114
 115 #if U_PLATFORM_HAS_WINUWP_API == 0
 116 static UDataFileAccess  gDataFileAccess = UDATA_DEFAULT_ACCESS;  // Access not synchronized.
 117                                                                  // Modifying is documented as thread-unsafe.
 118 #else
 119 static UDataFileAccess  gDataFileAccess = UDATA_NO_FILES;        // Windows UWP looks in one spot explicitly
 120 #endif
 121
 122 static UBool U_CALLCONV
 123 udata_cleanup(void)
 124 {
 125     int32_t i;
 126
 127     if (gCommonDataCache) {             /* Delete the cache of user data mappings.  */
 128         uhash_close(gCommonDataCache);  /*   Table owns the contents, and will delete them. */
 129         gCommonDataCache = NULL;        /*   Cleanup is not thread safe.                */
 130     }
 131     gCommonDataCacheInitOnce.reset();
 132
 133     for (i = 0; i < UPRV_LENGTHOF(gCommonICUDataArray) && gCommonICUDataArray[i] != NULL; ++i) {
 134         udata_close(gCommonICUDataArray[i]);
 135         gCommonICUDataArray[i] = NULL;
 136     }
 137     gHaveTriedToLoadCommonData = 0;
 138
 139     return TRUE;                   /* Everything was cleaned up */
 140 }
 141
 142 static UBool U_CALLCONV
 143 findCommonICUDataByName(const char *inBasename, UErrorCode &err)
 144 {
 145     UBool found = FALSE;
 146     int32_t i;
 147
 148     UDataMemory  *pData = udata_findCachedData(inBasename, err);
 149     if (U_FAILURE(err) || pData == NULL)
 150         return FALSE;
 151
 152     {
 153         Mutex lock;
 154         for (i = 0; i < UPRV_LENGTHOF(gCommonICUDataArray); ++i) {
 155             if ((gCommonICUDataArray[i] != NULL) && (gCommonICUDataArray[i]->pHeader == pData->pHeader)) {
 156                 /* The data pointer is already in the array. */
 157                 found = TRUE;
 158                 break;
 159             }
 160         }
 161     }
 162     return found;
 163 }
 164
 165
 166 /*
 167  * setCommonICUData.   Set a UDataMemory to be the global ICU Data
 168  */
 169 static UBool
 170 setCommonICUData(UDataMemory *pData,     /*  The new common data.  Belongs to caller, we copy it. */
 171                  UBool       warn,       /*  If true, set USING_DEFAULT warning if ICUData was    */
 172                                          /*    changed by another thread before we got to it.     */
 173                  UErrorCode *pErr)
 174 {
 175     UDataMemory  *newCommonData = UDataMemory_createNewInstance(pErr);
 176     int32_t i;
 177     UBool didUpdate = FALSE;
 178     if (U_FAILURE(*pErr)) {
 179         return FALSE;
 180     }
 181
 182     /*  For the assignment, other threads must cleanly see either the old            */
 183     /*    or the new, not some partially initialized new.  The old can not be        */
 184     /*    deleted - someone may still have a pointer to it lying around in           */
 185     /*    their locals.                                                              */
 186     UDatamemory_assign(newCommonData, pData);
 187     umtx_lock(NULL);
 188     for (i = 0; i < UPRV_LENGTHOF(gCommonICUDataArray); ++i) {
 189         if (gCommonICUDataArray[i] == NULL) {
 190             gCommonICUDataArray[i] = newCommonData;
 191             didUpdate = TRUE;
 192             break;
 193         } else if (gCommonICUDataArray[i]->pHeader == pData->pHeader) {
 194             /* The same data pointer is already in the array. */
 195             break;
 196         }
 197     }
 198     umtx_unlock(NULL);
 199
 200     if (i == UPRV_LENGTHOF(gCommonICUDataArray) && warn) {
 201         *pErr = U_USING_DEFAULT_WARNING;
 202     }
 203     if (didUpdate) {
 204         ucln_common_registerCleanup(UCLN_COMMON_UDATA, udata_cleanup);
 205     } else {
 206         uprv_free(newCommonData);
 207     }
 208     return didUpdate;
 209 }
 210
 211 #if U_PLATFORM_HAS_WINUWP_API == 0
 212
 213 static UBool
 214 setCommonICUDataPointer(const void *pData, UBool /*warn*/, UErrorCode *pErrorCode) {
 215     UDataMemory tData;
 216     UDataMemory_init(&tData);
 217     UDataMemory_setData(&tData, pData);
 218     udata_checkCommonData(&tData, pErrorCode);
 219     return setCommonICUData(&tData, FALSE, pErrorCode);
 220 }
 221
 222 #endif
 223
 224 static const char *
 225 findBasename(const char *path) {
 226     const char *basename=uprv_strrchr(path, U_FILE_SEP_CHAR);
 227     if(basename==NULL) {
 228         return path;
 229     } else {
 230         return basename+1;
 231     }
 232 }
 233
 234 #ifdef UDATA_DEBUG
 235 static const char *
 236 packageNameFromPath(const char *path)
 237 {
 238     if((path == NULL) || (*path == 0)) {
 239         return U_ICUDATA_NAME;
 240     }
 241
 242     path = findBasename(path);
 243
 244     if((path == NULL) || (*path == 0)) {
 245         return U_ICUDATA_NAME;
 246     }
 247
 248     return path;
 249 }
 250 #endif
 251
 252 /*----------------------------------------------------------------------*
 253  *                                                                      *
 254  *   Cache for common data                                              *
 255  *      Functions for looking up or adding entries to a cache of        *
 256  *      data that has been previously opened.  Avoids a potentially     *
 257  *      expensive operation of re-opening the data for subsequent       *
 258  *      uses.                                                           *
 259  *                                                                      *
 260  *      Data remains cached for the duration of the process.            *
 261  *                                                                      *
 262  *----------------------------------------------------------------------*/
 263
 264 typedef struct DataCacheElement {
 265     char          *name;
 266     UDataMemory   *item;
 267 } DataCacheElement;
 268
 269
 270
 271 /*
 272  * Deleter function for DataCacheElements.
 273  *         udata cleanup function closes the hash table; hash table in turn calls back to
 274  *         here for each entry.
 275  */
 276 static void U_CALLCONV DataCacheElement_deleter(void *pDCEl) {
 277     DataCacheElement *p = (DataCacheElement *)pDCEl;
 278     udata_close(p->item);              /* unmaps storage */
 279     uprv_free(p->name);                /* delete the hash key string. */
 280     uprv_free(pDCEl);                  /* delete 'this'          */
 281 }
 282
 283 static void U_CALLCONV udata_initHashTable(UErrorCode &err) {
 284     U_ASSERT(gCommonDataCache == NULL);
 285     gCommonDataCache = uhash_open(uhash_hashChars, uhash_compareChars, NULL, &err);
 286     if (U_FAILURE(err)) {
 287        return;
 288     }
 289     U_ASSERT(gCommonDataCache != NULL);
 290     uhash_setValueDeleter(gCommonDataCache, DataCacheElement_deleter);
 291     ucln_common_registerCleanup(UCLN_COMMON_UDATA, udata_cleanup);
 292 }
 293
 294  /*   udata_getCacheHashTable()
 295   *     Get the hash table used to store the data cache entries.
 296   *     Lazy create it if it doesn't yet exist.
 297   */
 298 static UHashtable *udata_getHashTable(UErrorCode &err) {
 299     umtx_initOnce(gCommonDataCacheInitOnce, &udata_initHashTable, err);
 300     return gCommonDataCache;
 301 }
 302
 303
 304
 305 static UDataMemory *udata_findCachedData(const char *path, UErrorCode &err)
 306 {
 307     UHashtable        *htable;
 308     UDataMemory       *retVal = NULL;
 309     DataCacheElement  *el;
 310     const char        *baseName;
 311
 312     htable = udata_getHashTable(err);
 313     if (U_FAILURE(err)) {
 314         return NULL;
 315     }
 316
 317     baseName = findBasename(path);   /* Cache remembers only the base name, not the full path. */
 318     umtx_lock(NULL);
 319     el = (DataCacheElement *)uhash_get(htable, baseName);
 320     umtx_unlock(NULL);
 321     if (el != NULL) {
 322         retVal = el->item;
 323     }
 324 #ifdef UDATA_DEBUG
 325     fprintf(stderr, "Cache: [%s] -> %p\n", baseName, retVal);
 326 #endif
 327     return retVal;
 328 }
 329
 330
 331 static UDataMemory *udata_cacheDataItem(const char *path, UDataMemory *item, UErrorCode *pErr) {
 332     DataCacheElement *newElement;
 333     const char       *baseName;
 334     int32_t           nameLen;
 335     UHashtable       *htable;
 336     DataCacheElement *oldValue = NULL;
 337     UErrorCode        subErr = U_ZERO_ERROR;
 338
 339     htable = udata_getHashTable(*pErr);
 340     if (U_FAILURE(*pErr)) {
 341         return NULL;
 342     }
 343
 344     /* Create a new DataCacheElement - the thingy we store in the hash table -
 345      * and copy the supplied path and UDataMemoryItems into it.
 346      */
 347     newElement = (DataCacheElement *)uprv_malloc(sizeof(DataCacheElement));
 348     if (newElement == NULL) {
 349         *pErr = U_MEMORY_ALLOCATION_ERROR;
 350         return NULL;
 351     }
 352     newElement->item = UDataMemory_createNewInstance(pErr);
 353     if (U_FAILURE(*pErr)) {
 354         uprv_free(newElement);
 355         return NULL;
 356     }
 357     UDatamemory_assign(newElement->item, item);
 358
 359     baseName = findBasename(path);
 360     nameLen = (int32_t)uprv_strlen(baseName);
 361     newElement->name = (char *)uprv_malloc(nameLen+1);
 362     if (newElement->name == NULL) {
 363         *pErr = U_MEMORY_ALLOCATION_ERROR;
 364         uprv_free(newElement->item);
 365         uprv_free(newElement);
 366         return NULL;
 367     }
 368     uprv_strcpy(newElement->name, baseName);
 369
 370     /* Stick the new DataCacheElement into the hash table.
 371     */
 372     umtx_lock(NULL);
 373     oldValue = (DataCacheElement *)uhash_get(htable, path);
 374     if (oldValue != NULL) {
 375         subErr = U_USING_DEFAULT_WARNING;
 376     }
 377     else {
 378         uhash_put(
 379             htable,
 380             newElement->name,               /* Key   */
 381             newElement,                     /* Value */
 382             &subErr);
 383     }
 384     umtx_unlock(NULL);
 385
 386 #ifdef UDATA_DEBUG
 387     fprintf(stderr, "Cache: [%s] <<< %p : %s. vFunc=%p\n", newElement->name,
 388     newElement->item, u_errorName(subErr), newElement->item->vFuncs);
 389 #endif
 390
 391     if (subErr == U_USING_DEFAULT_WARNING || U_FAILURE(subErr)) {
 392         *pErr = subErr; /* copy sub err unto fillin ONLY if something happens. */
 393         uprv_free(newElement->name);
 394         uprv_free(newElement->item);
 395         uprv_free(newElement);
 396         return oldValue ? oldValue->item : NULL;
 397     }
 398
 399     return newElement->item;
 400 }
 401
 402 /*----------------------------------------------------------------------*==============
 403  *                                                                      *
 404  *  Path management.  Could be shared with other tools/etc if need be   *
 405  * later on.                                                            *
 406  *                                                                      *
 407  *----------------------------------------------------------------------*/
 408
 409 U_NAMESPACE_BEGIN
 410
 411 class UDataPathIterator
 412 {
 413 public:
 414     UDataPathIterator(const char *path, const char *pkg,
 415                       const char *item, const char *suffix, UBool doCheckLastFour,
 416                       UErrorCode *pErrorCode);
 417     const char *next(UErrorCode *pErrorCode);
 418
 419 private:
 420     const char *path;                              /* working path (u_icudata_Dir) */
 421     const char *nextPath;                          /* path following this one */
 422     const char *basename;                          /* item's basename (icudt22e_mt.res)*/
 423     const char *suffix;                            /* item suffix (can be null) */
 424
 425     uint32_t    basenameLen;                       /* length of basename */
 426
 427     CharString  itemPath;                          /* path passed in with item name */
 428     CharString  pathBuffer;                        /* output path for this it'ion */
 429     CharString  packageStub;                       /* example:  "/icudt28b". Will ignore that leaf in set paths. */
 430
 431     UBool       checkLastFour;                     /* if TRUE then allow paths such as '/foo/myapp.dat'
 432                                                     * to match, checks last 4 chars of suffix with
 433                                                     * last 4 of path, then previous chars. */
 434 };
 435
 436 /**
 437  * @param iter  The iterator to be initialized. Its current state does not matter.
 438  * @param path  The full pathname to be iterated over.  If NULL, defaults to U_ICUDATA_NAME
 439  * @param pkg   Package which is being searched for, ex "icudt28l".  Will ignore leave directories such as /icudt28l
 440  * @param item  Item to be searched for.  Can include full path, such as /a/b/foo.dat
 441  * @param suffix  Optional item suffix, if not-null (ex. ".dat") then 'path' can contain 'item' explicitly.
 442  *               Ex:   'stuff.dat' would be found in '/a/foo:/tmp/stuff.dat:/bar/baz' as item #2.
 443  *                     '/blarg/stuff.dat' would also be found.
 444  */
 445 UDataPathIterator::UDataPathIterator(const char *inPath, const char *pkg,
 446                                      const char *item, const char *inSuffix, UBool doCheckLastFour,
 447                                      UErrorCode *pErrorCode)
 448 {
 449 #ifdef UDATA_DEBUG
 450         fprintf(stderr, "SUFFIX1=%s PATH=%s\n", inSuffix, inPath);
 451 #endif
 452     /** Path **/
 453     if(inPath == NULL) {
 454         path = u_getDataDirectory();
 455     } else {
 456         path = inPath;
 457     }
 458
 459     /** Package **/
 460     if(pkg != NULL) {
 461       packageStub.append(U_FILE_SEP_CHAR, *pErrorCode).append(pkg, *pErrorCode);
 462 #ifdef UDATA_DEBUG
 463       fprintf(stderr, "STUB=%s [%d]\n", packageStub.data(), packageStub.length());
 464 #endif
 465     }
 466
 467     /** Item **/
 468     basename = findBasename(item);
 469     basenameLen = (int32_t)uprv_strlen(basename);
 470
 471     /** Item path **/
 472     if(basename == item) {
 473         nextPath = path;
 474     } else {
 475         itemPath.append(item, (int32_t)(basename-item), *pErrorCode);
 476         nextPath = itemPath.data();
 477     }
 478 #ifdef UDATA_DEBUG
 479     fprintf(stderr, "SUFFIX=%s [%p]\n", inSuffix, inSuffix);
 480 #endif
 481
 482     /** Suffix  **/
 483     if(inSuffix != NULL) {
 484         suffix = inSuffix;
 485     } else {
 486         suffix = "";
 487     }
 488
 489     checkLastFour = doCheckLastFour;
 490
 491     /* pathBuffer will hold the output path strings returned by this iterator */
 492
 493 #ifdef UDATA_DEBUG
 494     fprintf(stderr, "%p: init %s -> [path=%s], [base=%s], [suff=%s], [itempath=%s], [nextpath=%s], [checklast4=%s]\n",
 495             iter,
 496             item,
 497             path,
 498             basename,
 499             suffix,
 500             itemPath.data(),
 501             nextPath,
 502             checkLastFour?"TRUE":"false");
 503 #endif
 504 }
 505
 506 /**
 507  * Get the next path on the list.
 508  *
 509  * @param iter The Iter to be used
 510  * @param len  If set, pointer to the length of the returned path, for convenience.
 511  * @return Pointer to the next path segment, or NULL if there are no more.
 512  */
 513 const char *UDataPathIterator::next(UErrorCode *pErrorCode)
 514 {
 515     if(U_FAILURE(*pErrorCode)) {
 516         return NULL;
 517     }
 518
 519     const char *currentPath = NULL;
 520     int32_t     pathLen = 0;
 521     const char *pathBasename;
 522
 523     do
 524     {
 525         if( nextPath == NULL ) {
 526             break;
 527         }
 528         currentPath = nextPath;
 529
 530         if(nextPath == itemPath.data()) { /* we were processing item's path. */
 531             nextPath = path; /* start with regular path next tm. */
 532             pathLen = (int32_t)uprv_strlen(currentPath);
 533         } else {
 534             /* fix up next for next time */
 535             nextPath = uprv_strchr(currentPath, U_PATH_SEP_CHAR);
 536             if(nextPath == NULL) {
 537                 /* segment: entire path */
 538                 pathLen = (int32_t)uprv_strlen(currentPath);
 539             } else {
 540                 /* segment: until next segment */
 541                 pathLen = (int32_t)(nextPath - currentPath);
 542                 /* skip divider */
 543                 nextPath ++;
 544             }
 545         }
 546
 547         if(pathLen == 0) {
 548             continue;
 549         }
 550
 551 #ifdef UDATA_DEBUG
 552         fprintf(stderr, "rest of path (IDD) = %s\n", currentPath);
 553         fprintf(stderr, "                     ");
 554         {
 555             uint32_t qqq;
 556             for(qqq=0;qqq<pathLen;qqq++)
 557             {
 558                 fprintf(stderr, " ");
 559             }
 560
 561             fprintf(stderr, "^\n");
 562         }
 563 #endif
 564         pathBuffer.clear().append(currentPath, pathLen, *pErrorCode);
 565
 566         /* check for .dat files */
 567         pathBasename = findBasename(pathBuffer.data());
 568
 569         if(checkLastFour == TRUE &&
 570            (pathLen>=4) &&
 571            uprv_strncmp(pathBuffer.data() +(pathLen-4), suffix, 4)==0 && /* suffix matches */
 572            uprv_strncmp(findBasename(pathBuffer.data()), basename, basenameLen)==0  && /* base matches */
 573            uprv_strlen(pathBasename)==(basenameLen+4)) { /* base+suffix = full len */
 574
 575 #ifdef UDATA_DEBUG
 576             fprintf(stderr, "Have %s file on the path: %s\n", suffix, pathBuffer.data());
 577 #endif
 578             /* do nothing */
 579         }
 580         else
 581         {       /* regular dir path */
 582             if(pathBuffer[pathLen-1] != U_FILE_SEP_CHAR) {
 583                 if((pathLen>=4) &&
 584                    uprv_strncmp(pathBuffer.data()+(pathLen-4), ".dat", 4) == 0)
 585                 {
 586 #ifdef UDATA_DEBUG
 587                     fprintf(stderr, "skipping non-directory .dat file %s\n", pathBuffer.data());
 588 #endif
 589                     continue;
 590                 }
 591
 592                 /* Check if it is a directory with the same name as our package */
 593                 if(!packageStub.isEmpty() &&
 594                    (pathLen > packageStub.length()) &&
 595                    !uprv_strcmp(pathBuffer.data() + pathLen - packageStub.length(), packageStub.data())) {
 596 #ifdef UDATA_DEBUG
 597                   fprintf(stderr, "Found stub %s (will add package %s of len %d)\n", packageStub.data(), basename, basenameLen);
 598 #endif
 599                   pathBuffer.truncate(pathLen - packageStub.length());
 600                 }
 601                 pathBuffer.append(U_FILE_SEP_CHAR, *pErrorCode);
 602             }
 603
 604             /* + basename */
 605             pathBuffer.append(packageStub.data()+1, packageStub.length()-1, *pErrorCode);
 606
 607             if(*suffix)  /* tack on suffix */
 608             {
 609                 pathBuffer.append(suffix, *pErrorCode);
 610             }
 611         }
 612
 613 #ifdef UDATA_DEBUG
 614         fprintf(stderr, " -->  %s\n", pathBuffer.data());
 615 #endif
 616
 617         return pathBuffer.data();
 618
 619     } while(path);
 620
 621     /* fell way off the end */
 622     return NULL;
 623 }
 624
 625 U_NAMESPACE_END
 626
 627 /* ==================================================================================*/
 628
 629
 630 /*----------------------------------------------------------------------*
 631  *                                                                      *
 632  *  Add a static reference to the common data library                   *
 633  *   Unless overridden by an explicit udata_setCommonData, this will be *
 634  *      our common data.                                                *
 635  *                                                                      *
 636  *----------------------------------------------------------------------*/
 637 #if U_PLATFORM_HAS_WINUWP_API == 0 // Windows UWP Platform does not support dll icu data at this time
 638 extern "C" const DataHeader U_DATA_API U_ICUDATA_ENTRY_POINT;
 639 #endif
 640
 641 /*
 642  * This would be a good place for weak-linkage declarations of
 643  * partial-data-library access functions where each returns a pointer
 644  * to its data package, if it is linked in.
 645  */
 646 /*
 647 extern const void *uprv_getICUData_collation(void) ATTRIBUTE_WEAK;
 648 extern const void *uprv_getICUData_conversion(void) ATTRIBUTE_WEAK;
 649 */
 650
 651 /*----------------------------------------------------------------------*
 652  *                                                                      *
 653  *   openCommonData   Attempt to open a common format (.dat) file       *
 654  *                    Map it into memory (if it's not there already)    *
 655  *                    and return a UDataMemory object for it.           *
 656  *                                                                      *
 657  *                    If the requested data is already open and cached  *
 658  *                       just return the cached UDataMem object.        *
 659  *                                                                      *
 660  *----------------------------------------------------------------------*/
 661 static UDataMemory *
 662 openCommonData(const char *path,          /*  Path from OpenChoice?          */
 663                int32_t commonDataIndex,   /*  ICU Data (index >= 0) if path == NULL */
 664                UErrorCode *pErrorCode)
 665 {
 666     UDataMemory tData;
 667     const char *pathBuffer;
 668     const char *inBasename;
 669
 670     if (U_FAILURE(*pErrorCode)) {
 671         return NULL;
 672     }
 673
 674     UDataMemory_init(&tData);
 675
 676     /* ??????? TODO revisit this */
 677     if (commonDataIndex >= 0) {
 678         /* "mini-cache" for common ICU data */
 679         if(commonDataIndex >= UPRV_LENGTHOF(gCommonICUDataArray)) {
 680             return NULL;
 681         }
 682         {
 683             Mutex lock;
 684             if(gCommonICUDataArray[commonDataIndex] != NULL) {
 685                 return gCommonICUDataArray[commonDataIndex];
 686             }
 687 #if U_PLATFORM_HAS_WINUWP_API == 0 // Windows UWP Platform does not support dll icu data at this time
 688             int32_t i;
 689             for(i = 0; i < commonDataIndex; ++i) {
 690                 if(gCommonICUDataArray[i]->pHeader == &U_ICUDATA_ENTRY_POINT) {
 691                     /* The linked-in data is already in the list. */
 692                     return NULL;
 693                 }
 694             }
 695 #endif
 696         }
 697
 698         /* Add the linked-in data to the list. */
 699         /*
 700          * This is where we would check and call weakly linked partial-data-library
 701          * access functions.
 702          */
 703         /*
 704         if (uprv_getICUData_collation) {
 705             setCommonICUDataPointer(uprv_getICUData_collation(), FALSE, pErrorCode);
 706         }
 707         if (uprv_getICUData_conversion) {
 708             setCommonICUDataPointer(uprv_getICUData_conversion(), FALSE, pErrorCode);
 709         }
 710         */
 711 #if U_PLATFORM_HAS_WINUWP_API == 0 // Windows UWP Platform does not support dll icu data at this time
 712         setCommonICUDataPointer(&U_ICUDATA_ENTRY_POINT, FALSE, pErrorCode);
 713         {
 714             Mutex lock;
 715             return gCommonICUDataArray[commonDataIndex];
 716         }
 717 #endif
 718     }
 719
 720
 721     /* request is NOT for ICU Data.  */
 722
 723     /* Find the base name portion of the supplied path.   */
 724     /*   inBasename will be left pointing somewhere within the original path string.      */
 725     inBasename = findBasename(path);
 726 #ifdef UDATA_DEBUG
 727     fprintf(stderr, "inBasename = %s\n", inBasename);
 728 #endif
 729
 730     if(*inBasename==0) {
 731         /* no basename.     This will happen if the original path was a directory name,   */
 732         /*    like  "a/b/c/".   (Fallback to separate files will still work.)             */
 733 #ifdef UDATA_DEBUG
 734         fprintf(stderr, "ocd: no basename in %s, bailing.\n", path);
 735 #endif
 736         if (U_SUCCESS(*pErrorCode)) {
 737             *pErrorCode=U_FILE_ACCESS_ERROR;
 738         }
 739         return NULL;
 740     }
 741
 742    /* Is the requested common data file already open and cached?                     */
 743    /*   Note that the cache is keyed by the base name only.  The rest of the path,   */
 744    /*     if any, is not considered.                                                 */
 745     UDataMemory  *dataToReturn = udata_findCachedData(inBasename, *pErrorCode);
 746     if (dataToReturn != NULL || U_FAILURE(*pErrorCode)) {
 747         return dataToReturn;
 748     }
 749
 750     /* Requested item is not in the cache.
 751      * Hunt it down, trying all the path locations
 752      */
 753
 754     UDataPathIterator iter(u_getDataDirectory(), inBasename, path, ".dat", TRUE, pErrorCode);
 755
 756     while((UDataMemory_isLoaded(&tData)==FALSE) && (pathBuffer = iter.next(pErrorCode)) != NULL)
 757     {
 758 #ifdef UDATA_DEBUG
 759         fprintf(stderr, "ocd: trying path %s - ", pathBuffer);
 760 #endif
 761         uprv_mapFile(&tData, pathBuffer);
 762 #ifdef UDATA_DEBUG
 763         fprintf(stderr, "%s\n", UDataMemory_isLoaded(&tData)?"LOADED":"not loaded");
 764 #endif
 765     }
 766
 767 #if defined(OS390_STUBDATA) && defined(OS390BATCH)
 768     if (!UDataMemory_isLoaded(&tData)) {
 769         char ourPathBuffer[1024];
 770         /* One more chance, for extendCommonData() */
 771         uprv_strncpy(ourPathBuffer, path, 1019);
 772         ourPathBuffer[1019]=0;
 773         uprv_strcat(ourPathBuffer, ".dat");
 774         uprv_mapFile(&tData, ourPathBuffer);
 775     }
 776 #endif
 777
 778     if (U_FAILURE(*pErrorCode)) {
 779         return NULL;
 780     }
 781     if (!UDataMemory_isLoaded(&tData)) {
 782         /* no common data */
 783         *pErrorCode=U_FILE_ACCESS_ERROR;
 784         return NULL;
 785     }
 786
 787     /* we have mapped a file, check its header */
 788     udata_checkCommonData(&tData, pErrorCode);
 789
 790
 791     /* Cache the UDataMemory struct for this .dat file,
 792      *   so we won't need to hunt it down and map it again next time
 793      *   something is needed from it.                */
 794     return udata_cacheDataItem(inBasename, &tData, pErrorCode);
 795 }
 796
 797
 798 /*----------------------------------------------------------------------*
 799  *                                                                      *
 800  *   extendICUData   If the full set of ICU data was not loaded at      *
 801  *                   program startup, load it now.  This function will  *
 802  *                   be called when the lookup of an ICU data item in   *
 803  *                   the common ICU data fails.                         *
 804  *                                                                      *
 805  *                   return true if new data is loaded, false otherwise.*
 806  *                                                                      *
 807  *----------------------------------------------------------------------*/
 808 static UBool extendICUData(UErrorCode *pErr)
 809 {
 810     UDataMemory   *pData;
 811     UDataMemory   copyPData;
 812     UBool         didUpdate = FALSE;
 813
 814     /*
 815      * There is a chance for a race condition here.
 816      * Normally, ICU data is loaded from a DLL or via mmap() and
 817      * setCommonICUData() will detect if the same address is set twice.
 818      * If ICU is built with data loading via fread() then the address will
 819      * be different each time the common data is loaded and we may add
 820      * multiple copies of the data.
 821      * In this case, use a mutex to prevent the race.
 822      * Use a specific mutex to avoid nested locks of the global mutex.
 823      */
 824 #if MAP_IMPLEMENTATION==MAP_STDIO
 825     static UMutex extendICUDataMutex = U_MUTEX_INITIALIZER;
 826     umtx_lock(&extendICUDataMutex);
 827 #endif
 828     if(!umtx_loadAcquire(gHaveTriedToLoadCommonData)) {
 829         /* See if we can explicitly open a .dat file for the ICUData. */
 830         pData = openCommonData(
 831                    U_ICUDATA_NAME,            /*  "icudt20l" , for example.          */
 832                    -1,                        /*  Pretend we're not opening ICUData  */
 833                    pErr);
 834
 835         /* How about if there is no pData, eh... */
 836
 837        UDataMemory_init(&copyPData);
 838        if(pData != NULL) {
 839           UDatamemory_assign(&copyPData, pData);
 840           copyPData.map = 0;              /* The mapping for this data is owned by the hash table */
 841           copyPData.mapAddr = 0;          /*   which will unmap it when ICU is shut down.         */
 842                                           /* CommonICUData is also unmapped when ICU is shut down.*/
 843                                           /* To avoid unmapping the data twice, zero out the map  */
 844                                           /*   fields in the UDataMemory that we're assigning     */
 845                                           /*   to CommonICUData.                                  */
 846
 847           didUpdate = /* no longer using this result */
 848               setCommonICUData(&copyPData,/*  The new common data.                                */
 849                        FALSE,             /*  No warnings if write didn't happen                  */
 850                        pErr);             /*  setCommonICUData honors errors; NOP if error set    */
 851         }
 852
 853         umtx_storeRelease(gHaveTriedToLoadCommonData, 1);
 854     }
 855
 856     didUpdate = findCommonICUDataByName(U_ICUDATA_NAME, *pErr);  /* Return 'true' when a racing writes out the extended                 */
 857                                                           /* data after another thread has failed to see it (in openCommonData), so     */
 858                                                           /* extended data can be examined.                                             */
 859                                                           /* Also handles a race through here before gHaveTriedToLoadCommonData is set. */
 860
 861 #if MAP_IMPLEMENTATION==MAP_STDIO
 862     umtx_unlock(&extendICUDataMutex);
 863 #endif
 864     return didUpdate;               /* Return true if ICUData pointer was updated.   */
 865                                     /*   (Could potentialy have been done by another thread racing */
 866                                     /*   us through here, but that's fine, we still return true    */
 867                                     /*   so that current thread will also examine extended data.   */
 868 }
 869
 870 /*----------------------------------------------------------------------*
 871  *                                                                      *
 872  *   udata_setCommonData                                                *
 873  *                                                                      *
 874  *----------------------------------------------------------------------*/
 875 U_CAPI void U_EXPORT2
 876 udata_setCommonData(const void *data, UErrorCode *pErrorCode) {
 877     UDataMemory dataMemory;
 878
 879     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 880         return;
 881     }
 882
 883     if(data==NULL) {
 884         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 885         return;
 886     }
 887
 888     /* set the data pointer and test for validity */
 889     UDataMemory_init(&dataMemory);
 890     UDataMemory_setData(&dataMemory, data);
 891     udata_checkCommonData(&dataMemory, pErrorCode);
 892     if (U_FAILURE(*pErrorCode)) {return;}
 893
 894     /* we have good data */
 895     /* Set it up as the ICU Common Data.  */
 896     setCommonICUData(&dataMemory, TRUE, pErrorCode);
 897 }
 898
 899 /*---------------------------------------------------------------------------
 900  *
 901  *  udata_setAppData
 902  *
 903  *---------------------------------------------------------------------------- */
 904 U_CAPI void U_EXPORT2
 905 udata_setAppData(const char *path, const void *data, UErrorCode *err)
 906 {
 907     UDataMemory     udm;
 908
 909     if(err==NULL || U_FAILURE(*err)) {
 910         return;
 911     }
 912     if(data==NULL) {
 913         *err=U_ILLEGAL_ARGUMENT_ERROR;
 914         return;
 915     }
 916
 917     UDataMemory_init(&udm);
 918     UDataMemory_setData(&udm, data);
 919     udata_checkCommonData(&udm, err);
 920     udata_cacheDataItem(path, &udm, err);
 921 }
 922
 923 /*----------------------------------------------------------------------------*
 924  *                                                                            *
 925  *  checkDataItem     Given a freshly located/loaded data item, either        *
 926  *                    an entry in a common file or a separately loaded file,  *
 927  *                    sanity check its header, and see if the data is         *
 928  *                    acceptable to the app.                                  *
 929  *                    If the data is good, create and return a UDataMemory    *
 930  *                    object that can be returned to the application.         *
 931  *                    Return NULL on any sort of failure.                     *
 932  *                                                                            *
 933  *----------------------------------------------------------------------------*/
 934 static UDataMemory *
 935 checkDataItem
 936 (
 937  const DataHeader         *pHeader,         /* The data item to be checked.                */
 938  UDataMemoryIsAcceptable  *isAcceptable,    /* App's call-back function                    */
 939  void                     *context,         /*   pass-thru param for above.                */
 940  const char               *type,            /*   pass-thru param for above.                */
 941  const char               *name,            /*   pass-thru param for above.                */
 942  UErrorCode               *nonFatalErr,     /* Error code if this data was not acceptable  */
 943                                             /*   but openChoice should continue with       */
 944                                             /*   trying to get data from fallback path.    */
 945  UErrorCode               *fatalErr         /* Bad error, caller should return immediately */
 946  )
 947 {
 948     UDataMemory  *rDataMem = NULL;          /* the new UDataMemory, to be returned.        */
 949
 950     if (U_FAILURE(*fatalErr)) {
 951         return NULL;
 952     }
 953
 954     if(pHeader->dataHeader.magic1==0xda &&
 955         pHeader->dataHeader.magic2==0x27 &&
 956         (isAcceptable==NULL || isAcceptable(context, type, name, &pHeader->info))
 957     ) {
 958         rDataMem=UDataMemory_createNewInstance(fatalErr);
 959         if (U_FAILURE(*fatalErr)) {
 960             return NULL;
 961         }
 962         rDataMem->pHeader = pHeader;
 963     } else {
 964         /* the data is not acceptable, look further */
 965         /* If we eventually find something good, this errorcode will be */
 966         /*    cleared out.                                              */
 967         *nonFatalErr=U_INVALID_FORMAT_ERROR;
 968     }
 969     return rDataMem;
 970 }
 971
 972 /**
 973  * @return 0 if not loaded, 1 if loaded or err
 974  */
 975 static UDataMemory *doLoadFromIndividualFiles(const char *pkgName,
 976         const char *dataPath, const char *tocEntryPathSuffix,
 977             /* following arguments are the same as doOpenChoice itself */
 978             const char *path, const char *type, const char *name,
 979              UDataMemoryIsAcceptable *isAcceptable, void *context,
 980              UErrorCode *subErrorCode,
 981              UErrorCode *pErrorCode)
 982 {
 983     const char         *pathBuffer;
 984     UDataMemory         dataMemory;
 985     UDataMemory *pEntryData;
 986
 987     /* look in ind. files: package\nam.typ  ========================= */
 988     /* init path iterator for individual files */
 989     UDataPathIterator iter(dataPath, pkgName, path, tocEntryPathSuffix, FALSE, pErrorCode);
 990
 991     while((pathBuffer = iter.next(pErrorCode)) != NULL)
 992     {
 993 #ifdef UDATA_DEBUG
 994         fprintf(stderr, "UDATA: trying individual file %s\n", pathBuffer);
 995 #endif
 996         if(uprv_mapFile(&dataMemory, pathBuffer))
 997         {
 998             pEntryData = checkDataItem(dataMemory.pHeader, isAcceptable, context, type, name, subErrorCode, pErrorCode);
 999             if (pEntryData != NULL) {
1000                 /* Data is good.
1001                 *  Hand off ownership of the backing memory to the user's UDataMemory.
1002                 *  and return it.   */
1003                 pEntryData->mapAddr = dataMemory.mapAddr;
1004                 pEntryData->map     = dataMemory.map;
1005
1006 #ifdef UDATA_DEBUG
1007                 fprintf(stderr, "** Mapped file: %s\n", pathBuffer);
1008 #endif
1009                 return pEntryData;
1010             }
1011
1012             /* the data is not acceptable, or some error occured.  Either way, unmap the memory */
1013             udata_close(&dataMemory);
1014
1015             /* If we had a nasty error, bail out completely.  */
1016             if (U_FAILURE(*pErrorCode)) {
1017                 return NULL;
1018             }
1019
1020             /* Otherwise remember that we found data but didn't like it for some reason  */
1021             *subErrorCode=U_INVALID_FORMAT_ERROR;
1022         }
1023 #ifdef UDATA_DEBUG
1024         fprintf(stderr, "%s\n", UDataMemory_isLoaded(&dataMemory)?"LOADED":"not loaded");
1025 #endif
1026     }
1027     return NULL;
1028 }
1029
1030 /**
1031  * @return 0 if not loaded, 1 if loaded or err
1032  */
1033 static UDataMemory *doLoadFromCommonData(UBool isICUData, const char * /*pkgName*/,
1034         const char * /*dataPath*/, const char * /*tocEntryPathSuffix*/, const char *tocEntryName,
1035             /* following arguments are the same as doOpenChoice itself */
1036             const char *path, const char *type, const char *name,
1037              UDataMemoryIsAcceptable *isAcceptable, void *context,
1038              UErrorCode *subErrorCode,
1039              UErrorCode *pErrorCode)
1040 {
1041     UDataMemory        *pEntryData;
1042     const DataHeader   *pHeader;
1043     UDataMemory        *pCommonData;
1044     int32_t            commonDataIndex;
1045     UBool              checkedExtendedICUData = FALSE;
1046     /* try to get common data.  The loop is for platforms such as the 390 that do
1047      *  not initially load the full set of ICU data.  If the lookup of an ICU data item
1048      *  fails, the full (but slower to load) set is loaded, the and the loop repeats,
1049      *  trying the lookup again.  Once the full set of ICU data is loaded, the loop wont
1050      *  repeat because the full set will be checked the first time through.
1051      *
1052      *  The loop also handles the fallback to a .dat file if the application linked
1053      *   to the stub data library rather than a real library.
1054      */
1055     for (commonDataIndex = isICUData ? 0 : -1;;) {
1056         pCommonData=openCommonData(path, commonDataIndex, subErrorCode); /** search for pkg **/
1057
1058         if(U_SUCCESS(*subErrorCode) && pCommonData!=NULL) {
1059             int32_t length;
1060
1061             /* look up the data piece in the common data */
1062             pHeader=pCommonData->vFuncs->Lookup(pCommonData, tocEntryName, &length, subErrorCode);
1063 #ifdef UDATA_DEBUG
1064             fprintf(stderr, "%s: pHeader=%p - %s\n", tocEntryName, pHeader, u_errorName(*subErrorCode));
1065 #endif
1066
1067             if(pHeader!=NULL) {
1068                 pEntryData = checkDataItem(pHeader, isAcceptable, context, type, name, subErrorCode, pErrorCode);
1069 #ifdef UDATA_DEBUG
1070                 fprintf(stderr, "pEntryData=%p\n", pEntryData);
1071 #endif
1072                 if (U_FAILURE(*pErrorCode)) {
1073                     return NULL;
1074                 }
1075                 if (pEntryData != NULL) {
1076                     pEntryData->length = length;
1077                     return pEntryData;
1078                 }
1079             }
1080         }
1081         /* Data wasn't found.  If we were looking for an ICUData item and there is
1082          * more data available, load it and try again,
1083          * otherwise break out of this loop. */
1084         if (!isICUData) {
1085             return NULL;
1086         } else if (pCommonData != NULL) {
1087             ++commonDataIndex;  /* try the next data package */
1088         } else if ((!checkedExtendedICUData) && extendICUData(subErrorCode)) {
1089             checkedExtendedICUData = TRUE;
1090             /* try this data package slot again: it changed from NULL to non-NULL */
1091         } else {
1092             return NULL;
1093         }
1094     }
1095 }
1096
1097 /*
1098  * Identify the Time Zone resources that are subject to special override data loading.
1099  */
1100 static UBool isTimeZoneFile(const char *name, const char *type) {
1101     return ((uprv_strcmp(type, "res") == 0) &&
1102             (uprv_strcmp(name, "zoneinfo64") == 0 ||
1103              uprv_strcmp(name, "timezoneTypes") == 0 ||
1104              uprv_strcmp(name, "windowsZones") == 0 ||
1105              uprv_strcmp(name, "metaZones") == 0));
1106 }
1107
1108 /*
1109  *  A note on the ownership of Mapped Memory
1110  *
1111  *  For common format files, ownership resides with the UDataMemory object
1112  *    that lives in the cache of opened common data.  These UDataMemorys are private
1113  *    to the udata implementation, and are never seen directly by users.
1114  *
1115  *    The UDataMemory objects returned to users will have the address of some desired
1116  *    data within the mapped region, but they wont have the mapping info itself, and thus
1117  *    won't cause anything to be removed from memory when they are closed.
1118  *
1119  *  For individual data files, the UDataMemory returned to the user holds the
1120  *  information necessary to unmap the data on close.  If the user independently
1121  *  opens the same data file twice, two completely independent mappings will be made.
1122  *  (There is no cache of opened data items from individual files, only a cache of
1123  *   opened Common Data files, that is, files containing a collection of data items.)
1124  *
1125  *  For common data passed in from the user via udata_setAppData() or
1126  *  udata_setCommonData(), ownership remains with the user.
1127  *
1128  *  UDataMemory objects themselves, as opposed to the memory they describe,
1129  *  can be anywhere - heap, stack/local or global.
1130  *  They have a flag to indicate when they're heap allocated and thus
1131  *  must be deleted when closed.
1132  */
1133
1134
1135 /*----------------------------------------------------------------------------*
1136  *                                                                            *
1137  * main data loading functions                                                *
1138  *                                                                            *
1139  *----------------------------------------------------------------------------*/
1140 static UDataMemory *
1141 doOpenChoice(const char *path, const char *type, const char *name,
1142              UDataMemoryIsAcceptable *isAcceptable, void *context,
1143              UErrorCode *pErrorCode)
1144 {
1145     UDataMemory         *retVal = NULL;
1146
1147     const char         *dataPath;
1148
1149     int32_t             tocEntrySuffixIndex;
1150     const char         *tocEntryPathSuffix;
1151     UErrorCode          subErrorCode=U_ZERO_ERROR;
1152     const char         *treeChar;
1153
1154     UBool               isICUData = FALSE;
1155
1156
1157     /* Is this path ICU data? */
1158     if(path == NULL ||
1159        !strcmp(path, U_ICUDATA_ALIAS) ||  /* "ICUDATA" */
1160        !uprv_strncmp(path, U_ICUDATA_NAME U_TREE_SEPARATOR_STRING, /* "icudt26e-" */
1161                      uprv_strlen(U_ICUDATA_NAME U_TREE_SEPARATOR_STRING)) ||
1162        !uprv_strncmp(path, U_ICUDATA_ALIAS U_TREE_SEPARATOR_STRING, /* "ICUDATA-" */
1163                      uprv_strlen(U_ICUDATA_ALIAS U_TREE_SEPARATOR_STRING))) {
1164       isICUData = TRUE;
1165     }
1166
1167 #if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR)  /* Windows:  try "foo\bar" and "foo/bar" */
1168     /* remap from alternate path char to the main one */
1169     CharString altSepPath;
1170     if(path) {
1171         if(uprv_strchr(path,U_FILE_ALT_SEP_CHAR) != NULL) {
1172             altSepPath.append(path, *pErrorCode);
1173             char *p;
1174             while ((p = uprv_strchr(altSepPath.data(), U_FILE_ALT_SEP_CHAR)) != NULL) {
1175                 *p = U_FILE_SEP_CHAR;
1176             }
1177 #if defined (UDATA_DEBUG)
1178             fprintf(stderr, "Changed path from [%s] to [%s]\n", path, altSepPath.s);
1179 #endif
1180             path = altSepPath.data();
1181         }
1182     }
1183 #endif
1184
1185     CharString tocEntryName; /* entry name in tree format. ex:  'icudt28b/coll/ar.res' */
1186     CharString tocEntryPath; /* entry name in path format. ex:  'icudt28b\\coll\\ar.res' */
1187
1188     CharString pkgName;
1189     CharString treeName;
1190
1191     /* ======= Set up strings */
1192     if(path==NULL) {
1193         pkgName.append(U_ICUDATA_NAME, *pErrorCode);
1194     } else {
1195         const char *pkg;
1196         const char *first;
1197         pkg = uprv_strrchr(path, U_FILE_SEP_CHAR);
1198         first = uprv_strchr(path, U_FILE_SEP_CHAR);
1199         if(uprv_pathIsAbsolute(path) || (pkg != first)) { /* more than one slash in the path- not a tree name */
1200             /* see if this is an /absolute/path/to/package  path */
1201             if(pkg) {
1202                 pkgName.append(pkg+1, *pErrorCode);
1203             } else {
1204                 pkgName.append(path, *pErrorCode);
1205             }
1206         } else {
1207             treeChar = uprv_strchr(path, U_TREE_SEPARATOR);
1208             if(treeChar) {
1209                 treeName.append(treeChar+1, *pErrorCode); /* following '-' */
1210                 if(isICUData) {
1211                     pkgName.append(U_ICUDATA_NAME, *pErrorCode);
1212                 } else {
1213                     pkgName.append(path, (int32_t)(treeChar-path), *pErrorCode);
1214                     if (first == NULL) {
1215                         /*
1216                         This user data has no path, but there is a tree name.
1217                         Look up the correct path from the data cache later.
1218                         */
1219                         path = pkgName.data();
1220                     }
1221                 }
1222             } else {
1223                 if(isICUData) {
1224                     pkgName.append(U_ICUDATA_NAME, *pErrorCode);
1225                 } else {
1226                     pkgName.append(path, *pErrorCode);
1227                 }
1228             }
1229         }
1230     }
1231
1232 #ifdef UDATA_DEBUG
1233     fprintf(stderr, " P=%s T=%s\n", pkgName.data(), treeName.data());
1234 #endif
1235
1236     /* setting up the entry name and file name
1237      * Make up a full name by appending the type to the supplied
1238      *  name, assuming that a type was supplied.
1239      */
1240
1241     /* prepend the package */
1242     tocEntryName.append(pkgName, *pErrorCode);
1243     tocEntryPath.append(pkgName, *pErrorCode);
1244     tocEntrySuffixIndex = tocEntryName.length();
1245
1246     if(!treeName.isEmpty()) {
1247         tocEntryName.append(U_TREE_ENTRY_SEP_CHAR, *pErrorCode).append(treeName, *pErrorCode);
1248         tocEntryPath.append(U_FILE_SEP_CHAR, *pErrorCode).append(treeName, *pErrorCode);
1249     }
1250
1251     tocEntryName.append(U_TREE_ENTRY_SEP_CHAR, *pErrorCode).append(name, *pErrorCode);
1252     tocEntryPath.append(U_FILE_SEP_CHAR, *pErrorCode).append(name, *pErrorCode);
1253     if(type!=NULL && *type!=0) {
1254         tocEntryName.append(".", *pErrorCode).append(type, *pErrorCode);
1255         tocEntryPath.append(".", *pErrorCode).append(type, *pErrorCode);
1256     }
1257     tocEntryPathSuffix = tocEntryPath.data()+tocEntrySuffixIndex; /* suffix starts here */
1258
1259 #ifdef UDATA_DEBUG
1260     fprintf(stderr, " tocEntryName = %s\n", tocEntryName.data());
1261     fprintf(stderr, " tocEntryPath = %s\n", tocEntryName.data());
1262 #endif
1263
1264 #if U_PLATFORM_HAS_WINUWP_API == 0 // Windows UWP Platform does not support dll icu data at this time
1265     if(path == NULL) {
1266         path = COMMON_DATA_NAME; /* "icudt26e" */
1267     }
1268 #else
1269     // Windows UWP expects only a single data file.
1270     path = COMMON_DATA_NAME; /* "icudt26e" */
1271 #endif
1272
1273     /************************ Begin loop looking for ind. files ***************/
1274 #ifdef UDATA_DEBUG
1275     fprintf(stderr, "IND: inBasename = %s, pkg=%s\n", "(n/a)", packageNameFromPath(path));
1276 #endif
1277
1278     /* End of dealing with a null basename */
1279     dataPath = u_getDataDirectory();
1280
1281     /****    Time zone individual files override  */
1282     if (isICUData && isTimeZoneFile(name, type)) {
1283         const char *tzFilesDir = u_getTimeZoneFilesDirectory(pErrorCode);
1284         if (tzFilesDir[0] != 0) {
1285 #ifdef UDATA_DEBUG
1286             fprintf(stderr, "Trying Time Zone Files directory = %s\n", tzFilesDir);
1287 #endif
1288 #ifdef UDATA_TZFILES_DEBUG
1289             fprintf(stderr, "# dOC U_TIMEZONE_FILES_DIR:  %s\n", U_TIMEZONE_FILES_DIR);
1290 #endif
1291
1292 #if defined(U_TIMEZONE_PACKAGE)
1293             // make tztocEntryName, like tocEntryName but with our package name
1294             UErrorCode tzpkgErrorCode = U_ZERO_ERROR;
1295             CharString tztocPkgPath;
1296             tztocPkgPath.append(tzFilesDir, tzpkgErrorCode);
1297             tztocPkgPath.append(U_FILE_SEP_CHAR, tzpkgErrorCode).append(U_TIMEZONE_PACKAGE, tzpkgErrorCode);
1298             CharString tztocEntryName;
1299             tztocEntryName.append(U_TIMEZONE_PACKAGE, tzpkgErrorCode);
1300             if(!treeName.isEmpty()) {
1301                 tztocEntryName.append(U_TREE_ENTRY_SEP_CHAR, tzpkgErrorCode).append(treeName, tzpkgErrorCode);
1302             }
1303             tztocEntryName.append(U_TREE_ENTRY_SEP_CHAR, tzpkgErrorCode).append(name, tzpkgErrorCode);
1304             if(type!=NULL && *type!=0) {
1305                 tztocEntryName.append(".", tzpkgErrorCode).append(type, tzpkgErrorCode);
1306             }
1307 #ifdef UDATA_TZFILES_DEBUG
1308             fprintf(stderr, "# dOC tz pkg, doLoadFromCommonData start; U_TIMEZONE_PACKAGE: %s, tztocPkgPath.data(): %s, tztocEntryName.data(): %s, name: %s\n",
1309                             U_TIMEZONE_PACKAGE, tztocPkgPath.data(), tztocEntryName.data(), name);
1310 #endif
1311             retVal = doLoadFromCommonData(FALSE, "" /*ignored*/, "" /*ignored*/, "" /*ignored*/,
1312                             tztocEntryName.data(),  // tocEntryName, like icutz44/zoneinfo64.res
1313                             tztocPkgPath.data(),     // path =  path to pkg, like /usr/share/icu/icutz44l
1314                             type, name, isAcceptable, context, &subErrorCode, &tzpkgErrorCode);
1315 #ifdef UDATA_TZFILES_DEBUG
1316             fprintf(stderr, "# dOC tz pkg, doLoadFromCommonData end; status %d, retVal %p\n", tzpkgErrorCode, retVal);
1317 #endif
1318             if(U_SUCCESS(tzpkgErrorCode) && retVal != NULL) {
1319                 return retVal;
1320             }
1321 #endif /* defined(U_TIMEZONE_PACKAGE) */
1322             // The following assumes any timezone resources in tzFilesDir are in individual .res files
1323 #ifdef UDATA_TZFILES_DEBUG
1324             fprintf(stderr, "# dOC tz files, doLoadFromIndividualFiles start; tzFilesDir: %s, tocEntryPathSuffix: %s, name: %s\n",
1325                             tzFilesDir, tocEntryPathSuffix, name);
1326 #endif
1327             retVal = doLoadFromIndividualFiles(/* pkgName.data() */ "", tzFilesDir, tocEntryPathSuffix,
1328                             /* path */ "", type, name, isAcceptable, context, &subErrorCode, pErrorCode);
1329 #ifdef UDATA_TZFILES_DEBUG
1330             fprintf(stderr, "# dOC tz files, doLoadFromIndividualFiles end; status %d, retVal %p\n", *pErrorCode, retVal);
1331 #endif
1332             if((retVal != NULL) || U_FAILURE(*pErrorCode)) {
1333                 return retVal;
1334             }
1335         }
1336     }
1337
1338     /****    COMMON PACKAGE  - only if packages are first. */
1339     if(gDataFileAccess == UDATA_PACKAGES_FIRST) {
1340 #ifdef UDATA_DEBUG
1341         fprintf(stderr, "Trying packages (UDATA_PACKAGES_FIRST)\n");
1342 #endif
1343         /* #2 */
1344 #ifdef UDATA_TZFILES_DEBUG
1345         if (isTimeZoneFile(name, type)) {
1346             fprintf(stderr, "# dOC std common 1, doLoadFromCommonData start; U_TIMEZONE_PACKAGE: path: %s, tocEntryName.data(): %s, name: %s\n",
1347                             path, tocEntryName.data(), name);
1348         }
1349 #endif
1350         retVal = doLoadFromCommonData(isICUData,
1351                             pkgName.data(), dataPath, tocEntryPathSuffix, tocEntryName.data(),
1352                             path, type, name, isAcceptable, context, &subErrorCode, pErrorCode);
1353 #ifdef UDATA_TZFILES_DEBUG
1354         if (isTimeZoneFile(name, type)) {
1355             fprintf(stderr, "# dOC std common 1, doLoadFromCommonData end; status %d, retVal %p\n", *pErrorCode, retVal);
1356         }
1357 #endif
1358         if((retVal != NULL) || U_FAILURE(*pErrorCode)) {
1359             return retVal;
1360         }
1361     }
1362
1363     /****    INDIVIDUAL FILES  */
1364     if((gDataFileAccess==UDATA_PACKAGES_FIRST) ||
1365        (gDataFileAccess==UDATA_FILES_FIRST)) {
1366 #ifdef UDATA_DEBUG
1367         fprintf(stderr, "Trying individual files\n");
1368 #endif
1369         /* Check to make sure that there is a dataPath to iterate over */
1370         if ((dataPath && *dataPath) || !isICUData) {
1371 #ifdef UDATA_TZFILES_DEBUG
1372             if (isTimeZoneFile(name, type)) {
1373                 fprintf(stderr, "# dOC std indiv files, doLoadFromIndividualFiles start; dataPath: %s, tocEntryPathSuffix: %s, name: %s\n",
1374                             dataPath, tocEntryPathSuffix, name);
1375             }
1376 #endif
1377             retVal = doLoadFromIndividualFiles(pkgName.data(), dataPath, tocEntryPathSuffix,
1378                             path, type, name, isAcceptable, context, &subErrorCode, pErrorCode);
1379 #ifdef UDATA_TZFILES_DEBUG
1380             if (isTimeZoneFile(name, type)) {
1381                 fprintf(stderr, "# dOC std indiv files, doLoadFromIndividualFiles end; status %d, retVal %p\n", *pErrorCode, retVal);
1382             }
1383 #endif
1384             if((retVal != NULL) || U_FAILURE(*pErrorCode)) {
1385                 return retVal;
1386             }
1387         }
1388     }
1389
1390     /****    COMMON PACKAGE  */
1391     if((gDataFileAccess==UDATA_ONLY_PACKAGES) ||
1392        (gDataFileAccess==UDATA_FILES_FIRST)) {
1393 #ifdef UDATA_DEBUG
1394         fprintf(stderr, "Trying packages (UDATA_ONLY_PACKAGES || UDATA_FILES_FIRST)\n");
1395 #endif
1396 #ifdef UDATA_TZFILES_DEBUG
1397         if (isTimeZoneFile(name, type)) {
1398             fprintf(stderr, "# dOC std common 2, doLoadFromCommonData start; U_TIMEZONE_PACKAGE: path: %s, tocEntryName.data(): %s, name: %s\n",
1399                             path, tocEntryName.data(), name);
1400         }
1401 #endif
1402         retVal = doLoadFromCommonData(isICUData,
1403                             pkgName.data(), dataPath, tocEntryPathSuffix, tocEntryName.data(),
1404                             path, type, name, isAcceptable, context, &subErrorCode, pErrorCode);
1405 #ifdef UDATA_TZFILES_DEBUG
1406         if (isTimeZoneFile(name, type)) {
1407             fprintf(stderr, "# dOC std common 2, doLoadFromCommonData end; status %d, retVal %p\n", *pErrorCode, retVal);
1408         }
1409 #endif
1410         if((retVal != NULL) || U_FAILURE(*pErrorCode)) {
1411             return retVal;
1412         }
1413     }
1414
1415     /* Load from DLL.  If we haven't attempted package load, we also haven't had any chance to
1416         try a DLL (static or setCommonData/etc)  load.
1417          If we ever have a "UDATA_ONLY_FILES", add it to the or list here.  */
1418     if(gDataFileAccess==UDATA_NO_FILES) {
1419 #ifdef UDATA_DEBUG
1420         fprintf(stderr, "Trying common data (UDATA_NO_FILES)\n");
1421 #endif
1422         retVal = doLoadFromCommonData(isICUData,
1423                             pkgName.data(), "", tocEntryPathSuffix, tocEntryName.data(),
1424                             path, type, name, isAcceptable, context, &subErrorCode, pErrorCode);
1425         if((retVal != NULL) || U_FAILURE(*pErrorCode)) {
1426             return retVal;
1427         }
1428     }
1429
1430     /* data not found */
1431     if(U_SUCCESS(*pErrorCode)) {
1432         if(U_SUCCESS(subErrorCode)) {
1433             /* file not found */
1434             *pErrorCode=U_FILE_ACCESS_ERROR;
1435         } else {
1436             /* entry point not found or rejected */
1437             *pErrorCode=subErrorCode;
1438         }
1439     }
1440     return retVal;
1441 }
1442
1443
1444
1445 /* API ---------------------------------------------------------------------- */
1446
1447 U_CAPI UDataMemory * U_EXPORT2
1448 udata_open(const char *path, const char *type, const char *name,
1449            UErrorCode *pErrorCode) {
1450 #ifdef UDATA_DEBUG
1451   fprintf(stderr, "udata_open(): Opening: %s : %s . %s\n", (path?path:"NULL"), name, type);
1452     fflush(stderr);
1453 #endif
1454
1455     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1456         return NULL;
1457     } else if(name==NULL || *name==0) {
1458         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1459         return NULL;
1460     } else {
1461         return doOpenChoice(path, type, name, NULL, NULL, pErrorCode);
1462     }
1463 }
1464
1465
1466
1467 U_CAPI UDataMemory * U_EXPORT2
1468 udata_openChoice(const char *path, const char *type, const char *name,
1469                  UDataMemoryIsAcceptable *isAcceptable, void *context,
1470                  UErrorCode *pErrorCode) {
1471 #ifdef UDATA_DEBUG
1472   fprintf(stderr, "udata_openChoice(): Opening: %s : %s . %s\n", (path?path:"NULL"), name, type);
1473 #endif
1474
1475     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1476         return NULL;
1477     } else if(name==NULL || *name==0 || isAcceptable==NULL) {
1478         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1479         return NULL;
1480     } else {
1481         return doOpenChoice(path, type, name, isAcceptable, context, pErrorCode);
1482     }
1483 }
1484
1485
1486
1487 U_CAPI void U_EXPORT2
1488 udata_getInfo(UDataMemory *pData, UDataInfo *pInfo) {
1489     if(pInfo!=NULL) {
1490         if(pData!=NULL && pData->pHeader!=NULL) {
1491             const UDataInfo *info=&pData->pHeader->info;
1492             uint16_t dataInfoSize=udata_getInfoSize(info);
1493             if(pInfo->size>dataInfoSize) {
1494                 pInfo->size=dataInfoSize;
1495             }
1496             uprv_memcpy((uint16_t *)pInfo+1, (const uint16_t *)info+1, pInfo->size-2);
1497             if(info->isBigEndian!=U_IS_BIG_ENDIAN) {
1498                 /* opposite endianness */
1499                 uint16_t x=info->reservedWord;
1500                 pInfo->reservedWord=(uint16_t)((x<<8)|(x>>8));
1501             }
1502         } else {
1503             pInfo->size=0;
1504         }
1505     }
1506 }
1507
1508
1509 U_CAPI void U_EXPORT2 udata_setFileAccess(UDataFileAccess access, UErrorCode * /*status*/)
1510 {
1511     // Note: this function is documented as not thread safe.
1512     gDataFileAccess = access;
1513 }