| 1 | // Copyright (c) .NET Foundation and contributors. All rights reserved. Licensed under the Microsoft Reciprocal License. See LICENSE.TXT file in the project root for full license information. |
| 2 | |
| 3 | #include "precomp.h" |
| 4 | |
| 5 | |
| 6 | // Exit macros |
| 7 | #define DictExitOnLastError(x, s, ...) ExitOnLastErrorSource(DUTIL_SOURCE_DICTUTIL, x, s, __VA_ARGS__) |
| 8 | #define DictExitOnLastErrorDebugTrace(x, s, ...) ExitOnLastErrorDebugTraceSource(DUTIL_SOURCE_DICTUTIL, x, s, __VA_ARGS__) |
| 9 | #define DictExitWithLastError(x, s, ...) ExitWithLastErrorSource(DUTIL_SOURCE_DICTUTIL, x, s, __VA_ARGS__) |
| 10 | #define DictExitOnFailure(x, s, ...) ExitOnFailureSource(DUTIL_SOURCE_DICTUTIL, x, s, __VA_ARGS__) |
| 11 | #define DictExitOnRootFailure(x, s, ...) ExitOnRootFailureSource(DUTIL_SOURCE_DICTUTIL, x, s, __VA_ARGS__) |
| 12 | #define DictExitOnFailureDebugTrace(x, s, ...) ExitOnFailureDebugTraceSource(DUTIL_SOURCE_DICTUTIL, x, s, __VA_ARGS__) |
| 13 | #define DictExitOnNull(p, x, e, s, ...) ExitOnNullSource(DUTIL_SOURCE_DICTUTIL, p, x, e, s, __VA_ARGS__) |
| 14 | #define DictExitOnNullWithLastError(p, x, s, ...) ExitOnNullWithLastErrorSource(DUTIL_SOURCE_DICTUTIL, p, x, s, __VA_ARGS__) |
| 15 | #define DictExitOnNullDebugTrace(p, x, e, s, ...) ExitOnNullDebugTraceSource(DUTIL_SOURCE_DICTUTIL, p, x, e, s, __VA_ARGS__) |
| 16 | #define DictExitOnInvalidHandleWithLastError(p, x, s, ...) ExitOnInvalidHandleWithLastErrorSource(DUTIL_SOURCE_DICTUTIL, p, x, s, __VA_ARGS__) |
| 17 | #define DictExitOnWin32Error(e, x, s, ...) ExitOnWin32ErrorSource(DUTIL_SOURCE_DICTUTIL, e, x, s, __VA_ARGS__) |
| 18 | #define DictExitOnGdipFailure(g, x, s, ...) ExitOnGdipFailureSource(DUTIL_SOURCE_DICTUTIL, g, x, s, __VA_ARGS__) |
| 19 | |
| 20 | // These should all be primes, and spaced reasonably apart (currently each is about 4x the last) |
| 21 | const DWORD MAX_BUCKET_SIZES[] = { |
| 22 | 503, |
| 23 | 2017, |
| 24 | 7937, |
| 25 | 32779, |
| 26 | 131111, |
| 27 | 524341, |
| 28 | 2097709, |
| 29 | 8390857, |
| 30 | 33563437, |
| 31 | 134253719, |
| 32 | 537014927, |
| 33 | 2148059509 |
| 34 | }; |
| 35 | |
| 36 | // However many items are in the cab, let's keep the buckets at least 8 times that to avoid collisions |
| 37 | #define MAX_BUCKETS_TO_ITEMS_RATIO 8 |
| 38 | |
| 39 | enum DICT_TYPE |
| 40 | { |
| 41 | DICT_INVALID = 0, |
| 42 | DICT_EMBEDDED_KEY = 1, |
| 43 | DICT_STRING_LIST = 2 |
| 44 | }; |
| 45 | |
| 46 | struct STRINGDICT_STRUCT |
| 47 | { |
| 48 | DICT_TYPE dtType; |
| 49 | |
| 50 | // Optional flags to control the behavior of the dictionary. |
| 51 | DICT_FLAG dfFlags; |
| 52 | |
| 53 | // Index into MAX_BUCKET_SIZES (array of primes), representing number of buckets we've allocated |
| 54 | DWORD dwBucketSizeIndex; |
| 55 | |
| 56 | // Number of items currently stored in the dict buckets |
| 57 | DWORD dwNumItems; |
| 58 | |
| 59 | // Byte offset of key within bucket value, for collision checking - see |
| 60 | // comments above DictCreateEmbeddedKey() implementation for further details |
| 61 | size_t cByteOffset; |
| 62 | |
| 63 | // The actual stored buckets |
| 64 | void **ppvBuckets; |
| 65 | |
| 66 | // The actual stored items in the order they were added (used for auto freeing or enumerating) |
| 67 | void **ppvItemList; |
| 68 | |
| 69 | // Pointer to the array of items, so the caller is free to resize the array of values out from under us without harm |
| 70 | void **ppvValueArray; |
| 71 | }; |
| 72 | |
| 73 | const int STRINGDICT_HANDLE_BYTES = sizeof(STRINGDICT_STRUCT); |
| 74 | |
| 75 | static HRESULT CreateDict( |
| 76 | __out_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE* psdHandle, |
| 77 | __in DICT_TYPE dtType, |
| 78 | __in DWORD dwNumExpectedItems, |
| 79 | __in_opt void** ppvArray, |
| 80 | __in size_t cByteOffset, |
| 81 | __in DICT_FLAG dfFlags |
| 82 | ); |
| 83 | static HRESULT StringHash( |
| 84 | __in const STRINGDICT_STRUCT *psd, |
| 85 | __in DWORD dwNumBuckets, |
| 86 | __in_z LPCWSTR pszString, |
| 87 | __out DWORD *pdwHash |
| 88 | ); |
| 89 | static BOOL IsMatchExact( |
| 90 | __in const STRINGDICT_STRUCT *psd, |
| 91 | __in DWORD dwMatchIndex, |
| 92 | __in_z LPCWSTR wzOriginalString |
| 93 | ); |
| 94 | static HRESULT GetValue( |
| 95 | __in const STRINGDICT_STRUCT *psd, |
| 96 | __in_z LPCWSTR pszString, |
| 97 | __out_opt void **ppvValue |
| 98 | ); |
| 99 | static HRESULT GetInsertIndex( |
| 100 | __in const STRINGDICT_STRUCT *psd, |
| 101 | __in DWORD dwBucketCount, |
| 102 | __in void **ppvBuckets, |
| 103 | __in_z LPCWSTR pszString, |
| 104 | __out DWORD *pdwOutput |
| 105 | ); |
| 106 | static HRESULT GetIndex( |
| 107 | __in const STRINGDICT_STRUCT *psd, |
| 108 | __in_z LPCWSTR pszString, |
| 109 | __out DWORD *pdwOutput |
| 110 | ); |
| 111 | static LPCWSTR GetKey( |
| 112 | __in const STRINGDICT_STRUCT *psd, |
| 113 | __in void *pvValue |
| 114 | ); |
| 115 | static HRESULT GrowDictionary( |
| 116 | __inout STRINGDICT_STRUCT *psd |
| 117 | ); |
| 118 | // These 2 helper functions allow us to safely handle dictutil consumers resizing |
| 119 | // the value array by storing "offsets" instead of raw void *'s in our buckets. |
| 120 | static void * TranslateOffsetToValue( |
| 121 | __in const STRINGDICT_STRUCT *psd, |
| 122 | __in void *pvValue |
| 123 | ); |
| 124 | static void * TranslateValueToOffset( |
| 125 | __in const STRINGDICT_STRUCT *psd, |
| 126 | __in void *pvValue |
| 127 | ); |
| 128 | |
| 129 | // The dict will store a set of keys (as wide-char strings) and a set of values associated with those keys (as void *'s). |
| 130 | // However, to support collision checking, the key needs to be represented in the "value" object (pointed to |
| 131 | // by the void *). The "stByteOffset" parameter tells this dict the byte offset of the "key" string pointer |
| 132 | // within the "value" object. Use the offsetof() macro to fill this out. |
| 133 | // The "ppvArray" parameter gives dictutil the address of your value array. If you provide this parameter, |
| 134 | // dictutil will remember all pointer values provided as "offsets" against this array. It is only necessary to provide |
| 135 | // this parameter to dictutil if it is possible you will realloc the array. |
| 136 | // |
| 137 | // Use DictAddValue() and DictGetValue() with this dictionary type. |
| 138 | extern "C" HRESULT DAPI DictCreateWithEmbeddedKey( |
| 139 | __out_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE* psdHandle, |
| 140 | __in DWORD dwNumExpectedItems, |
| 141 | __in_opt void **ppvArray, |
| 142 | __in size_t cByteOffset, |
| 143 | __in DICT_FLAG dfFlags |
| 144 | ) |
| 145 | { |
| 146 | return CreateDict(psdHandle, DICT_EMBEDDED_KEY, dwNumExpectedItems, ppvArray, cByteOffset, dfFlags); |
| 147 | } |
| 148 | |
| 149 | // The dict will store a set of keys, with no values associated with them. Use DictAddKey() and DictKeyExists() with this dictionary type. |
| 150 | extern "C" HRESULT DAPI DictCreateStringList( |
| 151 | __out_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE* psdHandle, |
| 152 | __in DWORD dwNumExpectedItems, |
| 153 | __in DICT_FLAG dfFlags |
| 154 | ) |
| 155 | { |
| 156 | return CreateDict(psdHandle, DICT_STRING_LIST, dwNumExpectedItems, NULL, 0, dfFlags); |
| 157 | } |
| 158 | |
| 159 | extern "C" HRESULT DAPI DictCreateStringListFromArray( |
| 160 | __out_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE* psdHandle, |
| 161 | __in_ecount(cStringArray) const LPCWSTR* rgwzStringArray, |
| 162 | __in const DWORD cStringArray, |
| 163 | __in DICT_FLAG dfFlags |
| 164 | ) |
| 165 | { |
| 166 | HRESULT hr = S_OK; |
| 167 | STRINGDICT_HANDLE sd = NULL; |
| 168 | |
| 169 | hr = DictCreateStringList(&sd, cStringArray, dfFlags); |
| 170 | DictExitOnFailure(hr, "Failed to create the string dictionary."); |
| 171 | |
| 172 | for (DWORD i = 0; i < cStringArray; ++i) |
| 173 | { |
| 174 | const LPCWSTR wzKey = rgwzStringArray[i]; |
| 175 | |
| 176 | hr = DictKeyExists(sd, wzKey); |
| 177 | if (E_NOTFOUND != hr) |
| 178 | { |
| 179 | DictExitOnFailure(hr, "Failed to check the string dictionary."); |
| 180 | } |
| 181 | else |
| 182 | { |
| 183 | hr = DictAddKey(sd, wzKey); |
| 184 | DictExitOnFailure(hr, "Failed to add \"%ls\" to the string dictionary.", wzKey); |
| 185 | } |
| 186 | } |
| 187 | |
| 188 | *psdHandle = sd; |
| 189 | sd = NULL; |
| 190 | |
| 191 | LExit: |
| 192 | ReleaseDict(sd); |
| 193 | |
| 194 | return hr; |
| 195 | } |
| 196 | |
| 197 | extern "C" HRESULT DAPI DictCompareStringListToArray( |
| 198 | __in_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE sdStringList, |
| 199 | __in_ecount(cStringArray) const LPCWSTR* rgwzStringArray, |
| 200 | __in const DWORD cStringArray |
| 201 | ) |
| 202 | { |
| 203 | HRESULT hr = S_OK; |
| 204 | |
| 205 | for (DWORD i = 0; i < cStringArray; ++i) |
| 206 | { |
| 207 | hr = DictKeyExists(sdStringList, rgwzStringArray[i]); |
| 208 | if (E_NOTFOUND != hr) |
| 209 | { |
| 210 | DictExitOnFailure(hr, "Failed to check the string dictionary."); |
| 211 | ExitFunction1(hr = S_OK); |
| 212 | } |
| 213 | } |
| 214 | |
| 215 | ExitFunction1(hr = HRESULT_FROM_WIN32(ERROR_NO_MATCH)); |
| 216 | |
| 217 | LExit: |
| 218 | return hr; |
| 219 | } |
| 220 | |
| 221 | // Todo: Dict should resize itself when (number of items) exceeds (number of buckets / MAX_BUCKETS_TO_ITEMS_RATIO) |
| 222 | extern "C" HRESULT DAPI DictAddKey( |
| 223 | __in_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE sdHandle, |
| 224 | __in_z LPCWSTR pszString |
| 225 | ) |
| 226 | { |
| 227 | HRESULT hr = S_OK; |
| 228 | DWORD dwIndex = 0; |
| 229 | STRINGDICT_STRUCT *psd = static_cast<STRINGDICT_STRUCT *>(sdHandle); |
| 230 | |
| 231 | DictExitOnNull(sdHandle, hr, E_INVALIDARG, "Handle not specified while adding value to dict"); |
| 232 | DictExitOnNull(pszString, hr, E_INVALIDARG, "String not specified while adding value to dict"); |
| 233 | |
| 234 | if (psd->dwBucketSizeIndex >= countof(MAX_BUCKET_SIZES)) |
| 235 | { |
| 236 | hr = E_INVALIDARG; |
| 237 | DictExitOnFailure(hr, "Invalid dictionary - bucket size index is out of range"); |
| 238 | } |
| 239 | |
| 240 | if (DICT_STRING_LIST != psd->dtType) |
| 241 | { |
| 242 | hr = E_INVALIDARG; |
| 243 | DictExitOnFailure(hr, "Tried to add key without value to wrong dictionary type! This dictionary type is: %d", psd->dtType); |
| 244 | } |
| 245 | |
| 246 | if ((psd->dwNumItems + 1) >= MAX_BUCKET_SIZES[psd->dwBucketSizeIndex] / MAX_BUCKETS_TO_ITEMS_RATIO) |
| 247 | { |
| 248 | hr = GrowDictionary(psd); |
| 249 | if (HRESULT_FROM_WIN32(ERROR_DATABASE_FULL) == hr) |
| 250 | { |
| 251 | // If we fail to proactively grow the dictionary, don't fail unless the dictionary is completely full |
| 252 | if (psd->dwNumItems < MAX_BUCKET_SIZES[psd->dwBucketSizeIndex]) |
| 253 | { |
| 254 | hr = S_OK; |
| 255 | } |
| 256 | } |
| 257 | DictExitOnFailure(hr, "Failed to grow dictionary"); |
| 258 | } |
| 259 | |
| 260 | hr = GetInsertIndex(psd, MAX_BUCKET_SIZES[psd->dwBucketSizeIndex], psd->ppvBuckets, pszString, &dwIndex); |
| 261 | DictExitOnFailure(hr, "Failed to get index to insert into"); |
| 262 | |
| 263 | hr = MemEnsureArraySize(reinterpret_cast<void **>(&(psd->ppvItemList)), psd->dwNumItems + 1, sizeof(void *), 1000); |
| 264 | DictExitOnFailure(hr, "Failed to resize list of items in dictionary"); |
| 265 | ++psd->dwNumItems; |
| 266 | |
| 267 | hr = StrAllocString(reinterpret_cast<LPWSTR *>(&(psd->ppvBuckets[dwIndex])), pszString, 0); |
| 268 | DictExitOnFailure(hr, "Failed to allocate copy of string"); |
| 269 | |
| 270 | psd->ppvItemList[psd->dwNumItems-1] = psd->ppvBuckets[dwIndex]; |
| 271 | |
| 272 | LExit: |
| 273 | return hr; |
| 274 | } |
| 275 | |
| 276 | // Todo: Dict should resize itself when (number of items) exceeds (number of buckets / MAX_BUCKETS_TO_ITEMS_RATIO) |
| 277 | extern "C" HRESULT DAPI DictAddValue( |
| 278 | __in_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE sdHandle, |
| 279 | __in void *pvValue |
| 280 | ) |
| 281 | { |
| 282 | HRESULT hr = S_OK; |
| 283 | void *pvOffset = NULL; |
| 284 | LPCWSTR wzKey = NULL; |
| 285 | DWORD dwIndex = 0; |
| 286 | STRINGDICT_STRUCT *psd = static_cast<STRINGDICT_STRUCT *>(sdHandle); |
| 287 | |
| 288 | DictExitOnNull(sdHandle, hr, E_INVALIDARG, "Handle not specified while adding value to dict"); |
| 289 | DictExitOnNull(pvValue, hr, E_INVALIDARG, "Value not specified while adding value to dict"); |
| 290 | |
| 291 | if (psd->dwBucketSizeIndex >= countof(MAX_BUCKET_SIZES)) |
| 292 | { |
| 293 | hr = E_INVALIDARG; |
| 294 | DictExitOnFailure(hr, "Invalid dictionary - bucket size index is out of range"); |
| 295 | } |
| 296 | |
| 297 | if (DICT_EMBEDDED_KEY != psd->dtType) |
| 298 | { |
| 299 | hr = E_INVALIDARG; |
| 300 | DictExitOnFailure(hr, "Tried to add key/value pair to wrong dictionary type! This dictionary type is: %d", psd->dtType); |
| 301 | } |
| 302 | |
| 303 | wzKey = GetKey(psd, pvValue); |
| 304 | DictExitOnNull(wzKey, hr, E_INVALIDARG, "String not specified while adding value to dict"); |
| 305 | |
| 306 | if ((psd->dwNumItems + 1) >= MAX_BUCKET_SIZES[psd->dwBucketSizeIndex] / MAX_BUCKETS_TO_ITEMS_RATIO) |
| 307 | { |
| 308 | hr = GrowDictionary(psd); |
| 309 | if (HRESULT_FROM_WIN32(ERROR_DATABASE_FULL) == hr && psd->dwNumItems + 1 ) |
| 310 | { |
| 311 | // If we fail to proactively grow the dictionary, don't fail unless the dictionary is completely full |
| 312 | if (psd->dwNumItems < MAX_BUCKET_SIZES[psd->dwBucketSizeIndex]) |
| 313 | { |
| 314 | hr = S_OK; |
| 315 | } |
| 316 | } |
| 317 | DictExitOnFailure(hr, "Failed to grow dictionary"); |
| 318 | } |
| 319 | |
| 320 | hr = GetInsertIndex(psd, MAX_BUCKET_SIZES[psd->dwBucketSizeIndex], psd->ppvBuckets, wzKey, &dwIndex); |
| 321 | DictExitOnFailure(hr, "Failed to get index to insert into"); |
| 322 | |
| 323 | hr = MemEnsureArraySize(reinterpret_cast<void **>(&(psd->ppvItemList)), psd->dwNumItems + 1, sizeof(void *), 1000); |
| 324 | DictExitOnFailure(hr, "Failed to resize list of items in dictionary"); |
| 325 | ++psd->dwNumItems; |
| 326 | |
| 327 | pvOffset = TranslateValueToOffset(psd, pvValue); |
| 328 | psd->ppvBuckets[dwIndex] = pvOffset; |
| 329 | psd->ppvItemList[psd->dwNumItems-1] = pvOffset; |
| 330 | |
| 331 | LExit: |
| 332 | return hr; |
| 333 | } |
| 334 | |
| 335 | extern "C" HRESULT DAPI DictGetValue( |
| 336 | __in_bcount(STRINGDICT_HANDLE_BYTES) C_STRINGDICT_HANDLE sdHandle, |
| 337 | __in_z LPCWSTR pszString, |
| 338 | __out void **ppvValue |
| 339 | ) |
| 340 | { |
| 341 | HRESULT hr = S_OK; |
| 342 | |
| 343 | DictExitOnNull(sdHandle, hr, E_INVALIDARG, "Handle not specified while searching dict"); |
| 344 | DictExitOnNull(pszString, hr, E_INVALIDARG, "String not specified while searching dict"); |
| 345 | |
| 346 | const STRINGDICT_STRUCT *psd = static_cast<const STRINGDICT_STRUCT *>(sdHandle); |
| 347 | |
| 348 | if (DICT_EMBEDDED_KEY != psd->dtType) |
| 349 | { |
| 350 | hr = E_INVALIDARG; |
| 351 | DictExitOnFailure(hr, "Tried to lookup value in wrong dictionary type! This dictionary type is: %d", psd->dtType); |
| 352 | } |
| 353 | |
| 354 | hr = GetValue(psd, pszString, ppvValue); |
| 355 | if (E_NOTFOUND == hr) |
| 356 | { |
| 357 | ExitFunction(); |
| 358 | } |
| 359 | DictExitOnFailure(hr, "Failed to call internal GetValue()"); |
| 360 | |
| 361 | LExit: |
| 362 | return hr; |
| 363 | } |
| 364 | |
| 365 | extern "C" HRESULT DAPI DictKeyExists( |
| 366 | __in_bcount(STRINGDICT_HANDLE_BYTES) C_STRINGDICT_HANDLE sdHandle, |
| 367 | __in_z LPCWSTR pszString |
| 368 | ) |
| 369 | { |
| 370 | HRESULT hr = S_OK; |
| 371 | |
| 372 | DictExitOnNull(sdHandle, hr, E_INVALIDARG, "Handle not specified while searching dict"); |
| 373 | DictExitOnNull(pszString, hr, E_INVALIDARG, "String not specified while searching dict"); |
| 374 | |
| 375 | const STRINGDICT_STRUCT *psd = static_cast<const STRINGDICT_STRUCT *>(sdHandle); |
| 376 | |
| 377 | // This works with either type of dictionary |
| 378 | hr = GetValue(psd, pszString, NULL); |
| 379 | if (E_NOTFOUND == hr) |
| 380 | { |
| 381 | ExitFunction(); |
| 382 | } |
| 383 | DictExitOnFailure(hr, "Failed to call internal GetValue()"); |
| 384 | |
| 385 | LExit: |
| 386 | return hr; |
| 387 | } |
| 388 | |
| 389 | extern "C" void DAPI DictDestroy( |
| 390 | __in_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE sdHandle |
| 391 | ) |
| 392 | { |
| 393 | DWORD i; |
| 394 | |
| 395 | STRINGDICT_STRUCT *psd = static_cast<STRINGDICT_STRUCT *>(sdHandle); |
| 396 | |
| 397 | if (DICT_STRING_LIST == psd->dtType) |
| 398 | { |
| 399 | for (i = 0; i < psd->dwNumItems; ++i) |
| 400 | { |
| 401 | ReleaseStr(reinterpret_cast<LPWSTR>(psd->ppvItemList[i])); |
| 402 | } |
| 403 | } |
| 404 | |
| 405 | ReleaseMem(psd->ppvItemList); |
| 406 | ReleaseMem(psd->ppvBuckets); |
| 407 | ReleaseMem(psd); |
| 408 | } |
| 409 | |
| 410 | static HRESULT CreateDict( |
| 411 | __out_bcount(STRINGDICT_HANDLE_BYTES) STRINGDICT_HANDLE* psdHandle, |
| 412 | __in DICT_TYPE dtType, |
| 413 | __in DWORD dwNumExpectedItems, |
| 414 | __in_opt void** ppvArray, |
| 415 | __in size_t cByteOffset, |
| 416 | __in DICT_FLAG dfFlags |
| 417 | ) |
| 418 | { |
| 419 | HRESULT hr = S_OK; |
| 420 | |
| 421 | DictExitOnNull(psdHandle, hr, E_INVALIDARG, "Handle not specified while creating dict."); |
| 422 | |
| 423 | // Allocate the handle |
| 424 | *psdHandle = static_cast<STRINGDICT_HANDLE>(MemAlloc(sizeof(STRINGDICT_STRUCT), TRUE)); |
| 425 | DictExitOnNull(*psdHandle, hr, E_OUTOFMEMORY, "Failed to allocate dictionary object."); |
| 426 | |
| 427 | STRINGDICT_STRUCT* psd = static_cast<STRINGDICT_STRUCT*>(*psdHandle); |
| 428 | |
| 429 | // Fill out the new handle's values |
| 430 | psd->dtType = dtType; |
| 431 | psd->dfFlags = dfFlags; |
| 432 | psd->cByteOffset = cByteOffset; |
| 433 | psd->ppvValueArray = ppvArray; |
| 434 | |
| 435 | // Make psd->dwBucketSizeIndex point to the appropriate spot in the prime |
| 436 | // array based on expected number of items and items to buckets ratio |
| 437 | // Careful: the "-1" in "countof(MAX_BUCKET_SIZES)-1" ensures we don't end |
| 438 | // this loop past the end of the array! |
| 439 | while (psd->dwBucketSizeIndex < (countof(MAX_BUCKET_SIZES) - 1) && |
| 440 | MAX_BUCKET_SIZES[psd->dwBucketSizeIndex] < dwNumExpectedItems * MAX_BUCKETS_TO_ITEMS_RATIO) |
| 441 | { |
| 442 | ++psd->dwBucketSizeIndex; |
| 443 | } |
| 444 | |
| 445 | hr = MemAllocArray(reinterpret_cast<LPVOID*>(&psd->ppvBuckets), sizeof(void*), MAX_BUCKET_SIZES[psd->dwBucketSizeIndex]); |
| 446 | DictExitOnFailure(hr, "Failed to allocate buckets for dictionary."); |
| 447 | |
| 448 | if (dwNumExpectedItems) |
| 449 | { |
| 450 | hr = MemAllocArray(reinterpret_cast<LPVOID*>(&psd->ppvItemList), sizeof(void*), dwNumExpectedItems); |
| 451 | DictExitOnFailure(hr, "Failed to pre-allocate item list for dictionary."); |
| 452 | } |
| 453 | |
| 454 | LExit: |
| 455 | return hr; |
| 456 | } |
| 457 | |
| 458 | static HRESULT StringHash( |
| 459 | __in const STRINGDICT_STRUCT *psd, |
| 460 | __in DWORD dwNumBuckets, |
| 461 | __in_z LPCWSTR pszString, |
| 462 | __out DWORD *pdwHash |
| 463 | ) |
| 464 | { |
| 465 | HRESULT hr = S_OK; |
| 466 | LPCWSTR wzKey = NULL; |
| 467 | LPWSTR sczNewKey = NULL; |
| 468 | DWORD result = 0; |
| 469 | |
| 470 | if (DICT_FLAG_CASEINSENSITIVE & psd->dfFlags) |
| 471 | { |
| 472 | hr = StrAllocStringToUpperInvariant(&sczNewKey, pszString, 0); |
| 473 | DictExitOnFailure(hr, "Failed to convert the string to upper-case."); |
| 474 | |
| 475 | wzKey = sczNewKey; |
| 476 | } |
| 477 | else |
| 478 | { |
| 479 | wzKey = pszString; |
| 480 | } |
| 481 | |
| 482 | while (*wzKey) |
| 483 | { |
| 484 | result = ~(*wzKey++ * 509) + result * 65599; |
| 485 | } |
| 486 | |
| 487 | *pdwHash = result % dwNumBuckets; |
| 488 | |
| 489 | LExit: |
| 490 | ReleaseStr(sczNewKey); |
| 491 | |
| 492 | return hr; |
| 493 | } |
| 494 | |
| 495 | static BOOL IsMatchExact( |
| 496 | __in const STRINGDICT_STRUCT *psd, |
| 497 | __in DWORD dwMatchIndex, |
| 498 | __in_z LPCWSTR wzOriginalString |
| 499 | ) |
| 500 | { |
| 501 | LPCWSTR wzMatchString = GetKey(psd, TranslateOffsetToValue(psd, psd->ppvBuckets[dwMatchIndex])); |
| 502 | DWORD dwFlags = 0; |
| 503 | |
| 504 | if (DICT_FLAG_CASEINSENSITIVE & psd->dfFlags) |
| 505 | { |
| 506 | dwFlags |= NORM_IGNORECASE; |
| 507 | } |
| 508 | |
| 509 | if (CSTR_EQUAL == ::CompareStringW(LOCALE_INVARIANT, dwFlags, wzOriginalString, -1, wzMatchString, -1)) |
| 510 | { |
| 511 | return TRUE; |
| 512 | } |
| 513 | |
| 514 | return FALSE; |
| 515 | } |
| 516 | |
| 517 | static HRESULT GetValue( |
| 518 | __in const STRINGDICT_STRUCT *psd, |
| 519 | __in_z LPCWSTR pszString, |
| 520 | __out_opt void **ppvValue |
| 521 | ) |
| 522 | { |
| 523 | HRESULT hr = S_OK; |
| 524 | DWORD dwOriginalIndexCandidate = 0; |
| 525 | void *pvCandidateValue = NULL; |
| 526 | DWORD dwIndex = 0; |
| 527 | |
| 528 | DictExitOnNull(psd, hr, E_INVALIDARG, "Handle not specified while searching dict"); |
| 529 | DictExitOnNull(pszString, hr, E_INVALIDARG, "String not specified while searching dict"); |
| 530 | |
| 531 | if (psd->dwBucketSizeIndex >= countof(MAX_BUCKET_SIZES)) |
| 532 | { |
| 533 | hr = E_INVALIDARG; |
| 534 | DictExitOnFailure(hr, "Invalid dictionary - bucket size index is out of range"); |
| 535 | } |
| 536 | |
| 537 | hr = StringHash(psd, MAX_BUCKET_SIZES[psd->dwBucketSizeIndex], pszString, &dwOriginalIndexCandidate); |
| 538 | DictExitOnFailure(hr, "Failed to hash the string."); |
| 539 | |
| 540 | DWORD dwIndexCandidate = dwOriginalIndexCandidate; |
| 541 | |
| 542 | pvCandidateValue = TranslateOffsetToValue(psd, psd->ppvBuckets[dwIndexCandidate]); |
| 543 | |
| 544 | // If no match exists in the dict |
| 545 | if (NULL == pvCandidateValue) |
| 546 | { |
| 547 | if (NULL != ppvValue) |
| 548 | { |
| 549 | *ppvValue = NULL; |
| 550 | } |
| 551 | ExitFunction1(hr = E_NOTFOUND); |
| 552 | } |
| 553 | |
| 554 | hr = GetIndex(psd, pszString, &dwIndex); |
| 555 | if (E_NOTFOUND == hr) |
| 556 | { |
| 557 | ExitFunction(); |
| 558 | } |
| 559 | DictExitOnFailure(hr, "Failed to find index to get"); |
| 560 | |
| 561 | if (NULL != ppvValue) |
| 562 | { |
| 563 | *ppvValue = TranslateOffsetToValue(psd, psd->ppvBuckets[dwIndex]); |
| 564 | } |
| 565 | |
| 566 | LExit: |
| 567 | if (FAILED(hr) && NULL != ppvValue) |
| 568 | { |
| 569 | *ppvValue = NULL; |
| 570 | } |
| 571 | |
| 572 | return hr; |
| 573 | } |
| 574 | |
| 575 | static HRESULT GetInsertIndex( |
| 576 | __in const STRINGDICT_STRUCT *psd, |
| 577 | __in DWORD dwBucketCount, |
| 578 | __in void **ppvBuckets, |
| 579 | __in_z LPCWSTR pszString, |
| 580 | __out DWORD *pdwOutput |
| 581 | ) |
| 582 | { |
| 583 | HRESULT hr = S_OK; |
| 584 | DWORD dwOriginalIndexCandidate = 0; |
| 585 | |
| 586 | hr = StringHash(psd, dwBucketCount, pszString, &dwOriginalIndexCandidate); |
| 587 | DictExitOnFailure(hr, "Failed to hash the string."); |
| 588 | |
| 589 | DWORD dwIndexCandidate = dwOriginalIndexCandidate; |
| 590 | |
| 591 | // If we collide, keep iterating forward from our intended position, even wrapping around to zero, until we find an empty bucket |
| 592 | #pragma prefast(push) |
| 593 | #pragma prefast(disable:26007) |
| 594 | while (NULL != ppvBuckets[dwIndexCandidate]) |
| 595 | #pragma prefast(pop) |
| 596 | { |
| 597 | ++dwIndexCandidate; |
| 598 | |
| 599 | // If we got to the end of the array, wrap around to zero index |
| 600 | if (dwIndexCandidate >= dwBucketCount) |
| 601 | { |
| 602 | dwIndexCandidate = 0; |
| 603 | } |
| 604 | |
| 605 | // If we wrapped all the way back around to our original index, the dict is full - throw an error |
| 606 | if (dwIndexCandidate == dwOriginalIndexCandidate) |
| 607 | { |
| 608 | // The dict table is full - this error seems to be a reasonably close match |
| 609 | hr = HRESULT_FROM_WIN32(ERROR_DATABASE_FULL); |
| 610 | DictExitOnRootFailure(hr, "Failed to add item '%ls' to dict table because dict table is full of items", pszString); |
| 611 | } |
| 612 | } |
| 613 | |
| 614 | *pdwOutput = dwIndexCandidate; |
| 615 | |
| 616 | LExit: |
| 617 | return hr; |
| 618 | } |
| 619 | |
| 620 | static HRESULT GetIndex( |
| 621 | __in const STRINGDICT_STRUCT *psd, |
| 622 | __in_z LPCWSTR pszString, |
| 623 | __out DWORD *pdwOutput |
| 624 | ) |
| 625 | { |
| 626 | HRESULT hr = S_OK; |
| 627 | DWORD dwOriginalIndexCandidate = 0; |
| 628 | |
| 629 | if (psd->dwBucketSizeIndex >= countof(MAX_BUCKET_SIZES)) |
| 630 | { |
| 631 | hr = E_INVALIDARG; |
| 632 | DictExitOnFailure(hr, "Invalid dictionary - bucket size index is out of range"); |
| 633 | } |
| 634 | |
| 635 | hr = StringHash(psd, MAX_BUCKET_SIZES[psd->dwBucketSizeIndex], pszString, &dwOriginalIndexCandidate); |
| 636 | DictExitOnFailure(hr, "Failed to hash the string."); |
| 637 | |
| 638 | DWORD dwIndexCandidate = dwOriginalIndexCandidate; |
| 639 | |
| 640 | while (!IsMatchExact(psd, dwIndexCandidate, pszString)) |
| 641 | { |
| 642 | ++dwIndexCandidate; |
| 643 | |
| 644 | // If we got to the end of the array, wrap around to zero index |
| 645 | if (dwIndexCandidate >= MAX_BUCKET_SIZES[psd->dwBucketSizeIndex]) |
| 646 | { |
| 647 | dwIndexCandidate = 0; |
| 648 | } |
| 649 | |
| 650 | // If no match exists in the dict |
| 651 | if (NULL == psd->ppvBuckets[dwIndexCandidate]) |
| 652 | { |
| 653 | ExitFunction1(hr = E_NOTFOUND); |
| 654 | } |
| 655 | |
| 656 | // If we wrapped all the way back around to our original index, the dict is full and we found nothing, so return as such |
| 657 | if (dwIndexCandidate == dwOriginalIndexCandidate) |
| 658 | { |
| 659 | ExitFunction1(hr = E_NOTFOUND); |
| 660 | } |
| 661 | } |
| 662 | |
| 663 | *pdwOutput = dwIndexCandidate; |
| 664 | |
| 665 | LExit: |
| 666 | return hr; |
| 667 | } |
| 668 | |
| 669 | static LPCWSTR GetKey( |
| 670 | __in const STRINGDICT_STRUCT *psd, |
| 671 | __in void *pvValue |
| 672 | ) |
| 673 | { |
| 674 | const BYTE *lpByte = reinterpret_cast<BYTE *>(pvValue); |
| 675 | |
| 676 | if (DICT_EMBEDDED_KEY == psd->dtType) |
| 677 | { |
| 678 | void *pvKey = reinterpret_cast<void *>(reinterpret_cast<BYTE *>(pvValue) + psd->cByteOffset); |
| 679 | |
| 680 | #pragma prefast(push) |
| 681 | #pragma prefast(disable:26010) |
| 682 | return *(reinterpret_cast<LPCWSTR *>(pvKey)); |
| 683 | #pragma prefast(pop) |
| 684 | } |
| 685 | else |
| 686 | { |
| 687 | return (reinterpret_cast<LPCWSTR>(lpByte)); |
| 688 | } |
| 689 | } |
| 690 | |
| 691 | static HRESULT GrowDictionary( |
| 692 | __inout STRINGDICT_STRUCT *psd |
| 693 | ) |
| 694 | { |
| 695 | HRESULT hr = S_OK; |
| 696 | DWORD dwInsertIndex = 0; |
| 697 | LPCWSTR wzKey = NULL; |
| 698 | DWORD dwNewBucketSizeIndex = 0; |
| 699 | size_t cbAllocSize = 0; |
| 700 | void **ppvNewBuckets = NULL; |
| 701 | |
| 702 | dwNewBucketSizeIndex = psd->dwBucketSizeIndex + 1; |
| 703 | |
| 704 | if (dwNewBucketSizeIndex >= countof(MAX_BUCKET_SIZES)) |
| 705 | { |
| 706 | ExitFunction1(hr = HRESULT_FROM_WIN32(ERROR_DATABASE_FULL)); |
| 707 | } |
| 708 | |
| 709 | hr = ::SizeTMult(sizeof(void *), MAX_BUCKET_SIZES[dwNewBucketSizeIndex], &cbAllocSize); |
| 710 | DictExitOnFailure(hr, "Overflow while calculating allocation size to grow dictionary"); |
| 711 | |
| 712 | ppvNewBuckets = static_cast<void**>(MemAlloc(cbAllocSize, TRUE)); |
| 713 | DictExitOnNull(ppvNewBuckets, hr, E_OUTOFMEMORY, "Failed to allocate %u buckets while growing dictionary", MAX_BUCKET_SIZES[dwNewBucketSizeIndex]); |
| 714 | |
| 715 | for (DWORD i = 0; i < psd->dwNumItems; ++i) |
| 716 | { |
| 717 | wzKey = GetKey(psd, TranslateOffsetToValue(psd, psd->ppvItemList[i])); |
| 718 | DictExitOnNull(wzKey, hr, E_INVALIDARG, "String not specified in existing dict value"); |
| 719 | |
| 720 | hr = GetInsertIndex(psd, MAX_BUCKET_SIZES[dwNewBucketSizeIndex], ppvNewBuckets, wzKey, &dwInsertIndex); |
| 721 | DictExitOnFailure(hr, "Failed to get index to insert into"); |
| 722 | |
| 723 | ppvNewBuckets[dwInsertIndex] = psd->ppvItemList[i]; |
| 724 | } |
| 725 | |
| 726 | psd->dwBucketSizeIndex = dwNewBucketSizeIndex; |
| 727 | ReleaseMem(psd->ppvBuckets); |
| 728 | psd->ppvBuckets = ppvNewBuckets; |
| 729 | ppvNewBuckets = NULL; |
| 730 | |
| 731 | LExit: |
| 732 | ReleaseMem(ppvNewBuckets); |
| 733 | |
| 734 | return hr; |
| 735 | } |
| 736 | |
| 737 | static void * TranslateOffsetToValue( |
| 738 | __in const STRINGDICT_STRUCT *psd, |
| 739 | __in void *pvValue |
| 740 | ) |
| 741 | { |
| 742 | if (NULL == pvValue) |
| 743 | { |
| 744 | return NULL; |
| 745 | } |
| 746 | |
| 747 | // All offsets are stored as (real offset + 1), so subtract 1 to get back to the real value |
| 748 | if (NULL != psd->ppvValueArray) |
| 749 | { |
| 750 | return reinterpret_cast<void *>(reinterpret_cast<DWORD_PTR>(pvValue) + reinterpret_cast<DWORD_PTR>(*psd->ppvValueArray) - 1); |
| 751 | } |
| 752 | else |
| 753 | { |
| 754 | return pvValue; |
| 755 | } |
| 756 | } |
| 757 | |
| 758 | static void * TranslateValueToOffset( |
| 759 | __in const STRINGDICT_STRUCT *psd, |
| 760 | __in void *pvValue |
| 761 | ) |
| 762 | { |
| 763 | if (NULL != psd->ppvValueArray) |
| 764 | { |
| 765 | // 0 has a special meaning - we don't want offset 0 into the array to have NULL for the offset - so add 1 to avoid this issue |
| 766 | return reinterpret_cast<void *>(reinterpret_cast<DWORD_PTR>(pvValue) - reinterpret_cast<DWORD_PTR>(*psd->ppvValueArray) + 1); |
| 767 | } |
| 768 | else |
| 769 | { |
| 770 | return pvValue; |
| 771 | } |
| 772 | } |