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