BsDataStream.cpp 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487
  1. #include "BsDataStream.h"
  2. #include "BsDebug.h"
  3. #include "BsException.h"
  4. #include <codecvt>
  5. namespace BansheeEngine
  6. {
  7. const UINT32 DataStream::StreamTempSize = 128;
  8. /**
  9. * @brief Checks does the provided buffer has an UTF32 byte order mark
  10. * in little endian order.
  11. */
  12. bool isUTF32LE(const UINT8* buffer)
  13. {
  14. return buffer[0] == 0xFF && buffer[1] == 0xFE && buffer[2] == 0x00 && buffer[3] == 0x00;
  15. }
  16. /**
  17. * @brief Checks does the provided buffer has an UTF32 byte order mark
  18. * in big endian order.
  19. */
  20. bool isUTF32BE(const UINT8* buffer)
  21. {
  22. return buffer[0] == 0x00 && buffer[1] == 0x00 && buffer[2] == 0xFE && buffer[3] == 0xFF;
  23. }
  24. /**
  25. * @brief Checks does the provided buffer has an UTF16 byte order mark
  26. * in little endian order.
  27. */
  28. bool isUTF16LE(const UINT8* buffer)
  29. {
  30. return buffer[0] == 0xFF && buffer[1] == 0xFE;
  31. }
  32. /**
  33. * @brief Checks does the provided buffer has an UTF16 byte order mark
  34. * in big endian order.
  35. */
  36. bool isUTF16BE(const UINT8* buffer)
  37. {
  38. return buffer[0] == 0xFE && buffer[1] == 0xFF;
  39. }
  40. /**
  41. * @brief Checks does the provided buffer has an UTF8 byte order mark.
  42. */
  43. bool isUTF8(const UINT8* buffer)
  44. {
  45. return (buffer[0] == 0xEF && buffer[1] == 0xBB && buffer[2] == 0xBF);
  46. }
  47. template <typename T> DataStream& DataStream::operator>> (T& val)
  48. {
  49. read(static_cast<void*>(&val), sizeof(T));
  50. return *this;
  51. }
  52. void DataStream::writeString(const String& string, StringEncoding encoding)
  53. {
  54. if (encoding == StringEncoding::UTF16)
  55. {
  56. const std::codecvt_mode convMode = (std::codecvt_mode)(std::generate_header);
  57. typedef std::codecvt_utf8_utf16<char, 1114111, convMode> UTF8ToUTF16Conv;
  58. std::wstring_convert<UTF8ToUTF16Conv, char> conversion("?");
  59. std::string encodedString = conversion.from_bytes(string.c_str());
  60. write(encodedString.data(), encodedString.length());
  61. }
  62. else
  63. {
  64. write(string.data(), string.length());
  65. }
  66. }
  67. void DataStream::writeString(const WString& string, StringEncoding encoding)
  68. {
  69. if (encoding == StringEncoding::UTF16)
  70. {
  71. const std::codecvt_mode convMode = (std::codecvt_mode)(std::generate_header | std::little_endian);
  72. typedef std::codecvt_utf16<wchar_t, 1114111, convMode> WCharToUTF16Conv;
  73. std::wstring_convert<WCharToUTF16Conv, wchar_t> conversion("?");
  74. std::string encodedString = conversion.to_bytes(string.c_str());
  75. write(encodedString.data(), encodedString.length());
  76. }
  77. else
  78. {
  79. const std::codecvt_mode convMode = (std::codecvt_mode)(std::generate_header);
  80. typedef std::codecvt_utf8<wchar_t, 1114111, convMode> WCharToUTF8Conv;
  81. std::wstring_convert<WCharToUTF8Conv, wchar_t> conversion("?");
  82. std::string encodedString = conversion.to_bytes(string.c_str());
  83. write(encodedString.data(), encodedString.length());
  84. }
  85. }
  86. String DataStream::getAsString()
  87. {
  88. // Read the entire buffer - ideally in one read, but if the size of
  89. // the buffer is unknown, do multiple fixed size reads.
  90. size_t bufSize = (mSize > 0 ? mSize : 4096);
  91. std::stringstream::char_type* tempBuffer = (std::stringstream::char_type*)bs_alloc((UINT32)bufSize);
  92. // Ensure read from begin of stream
  93. seek(0);
  94. std::stringstream result;
  95. while (!eof())
  96. {
  97. size_t numReadBytes = read(tempBuffer, bufSize);
  98. result.write(tempBuffer, numReadBytes);
  99. }
  100. free(tempBuffer);
  101. std::string string = result.str();
  102. UINT32 readBytes = (UINT32)string.size();
  103. if (readBytes >= 4)
  104. {
  105. if (isUTF32LE((UINT8*)string.data()))
  106. {
  107. const std::codecvt_mode convMode = (std::codecvt_mode)(std::consume_header | std::little_endian);
  108. typedef std::codecvt_utf8<char32_t, 1114111, convMode> utf8utf32;
  109. std::wstring_convert<utf8utf32, char32_t> conversion("?");
  110. char32_t* start = (char32_t*)string.data();
  111. char32_t* end = (start + (string.size() - 1) / 4);
  112. return conversion.to_bytes(start, end).c_str();
  113. }
  114. else if (isUTF32BE((UINT8*)string.data()))
  115. {
  116. const std::codecvt_mode convMode = (std::codecvt_mode)(std::consume_header);
  117. typedef std::codecvt_utf8<char32_t, 1114111, convMode> utf8utf32;
  118. std::wstring_convert<utf8utf32, char32_t> conversion("?");
  119. char32_t* start = (char32_t*)string.data();
  120. char32_t* end = (start + (string.size() - 1) / 4);
  121. return conversion.to_bytes(start, end).c_str();
  122. }
  123. }
  124. if (readBytes >= 3)
  125. {
  126. if (isUTF8((UINT8*)string.data()))
  127. {
  128. return string.c_str() + 3;
  129. }
  130. }
  131. if (readBytes >= 2)
  132. {
  133. if (isUTF16LE((UINT8*)string.data()))
  134. {
  135. const std::codecvt_mode convMode = (std::codecvt_mode)(std::little_endian);
  136. typedef std::codecvt_utf8_utf16<char16_t, 1114111, convMode> utf8utf16;
  137. std::wstring_convert<utf8utf16, char16_t> conversion("?");
  138. char16_t* start = (char16_t*)(string.c_str() + 2); // Bug?: std::consume_header seems to be ignored so I manually remove the header
  139. return conversion.to_bytes(start).c_str();
  140. }
  141. else if (isUTF16BE((UINT8*)string.data()))
  142. {
  143. const std::codecvt_mode convMode = (std::codecvt_mode)(0);
  144. typedef std::codecvt_utf8_utf16<char16_t, 1114111, convMode> utf8utf16;
  145. // Bug?: Regardless of not providing the std::little_endian flag it seems that is how the data is read
  146. // so I manually flip it
  147. UINT32 numChars = (UINT32)(string.size() - 2) / 2;
  148. for (UINT32 i = 0; i < numChars; i++)
  149. std::swap(string[i * 2 + 0], string[i * 2 + 1]);
  150. std::wstring_convert<utf8utf16, char16_t> conversion("?");
  151. char16_t* start = (char16_t*)(string.c_str() + 2); // Bug?: std::consume_header seems to be ignored so I manually remove the header
  152. return conversion.to_bytes(start).c_str();
  153. }
  154. }
  155. return string.c_str();
  156. }
  157. WString DataStream::getAsWString()
  158. {
  159. // Read the entire buffer - ideally in one read, but if the size of
  160. // the buffer is unknown, do multiple fixed size reads.
  161. size_t bufSize = (mSize > 0 ? mSize : 4096);
  162. std::stringstream::char_type* tempBuffer = (std::stringstream::char_type*)bs_alloc((UINT32)bufSize);
  163. // Ensure read from begin of stream
  164. seek(0);
  165. std::stringstream result;
  166. while (!eof())
  167. {
  168. size_t numReadBytes = read(tempBuffer, bufSize);
  169. result.write(tempBuffer, numReadBytes);
  170. }
  171. free(tempBuffer);
  172. std::string string = result.str();
  173. UINT32 readBytes = (UINT32)string.size();
  174. if (readBytes >= 4)
  175. {
  176. if (isUTF32LE((UINT8*)string.data()))
  177. {
  178. // Not supported
  179. }
  180. else if (isUTF32BE((UINT8*)string.data()))
  181. {
  182. // Not supported
  183. }
  184. }
  185. if (readBytes >= 3)
  186. {
  187. if (isUTF8((UINT8*)string.data()))
  188. {
  189. const std::codecvt_mode convMode = (std::codecvt_mode)(std::consume_header);
  190. typedef std::codecvt_utf8<wchar_t, 1114111, convMode> wcharutf8;
  191. std::wstring_convert<wcharutf8> conversion("?");
  192. return conversion.from_bytes(string).c_str();
  193. }
  194. }
  195. if (readBytes >= 2)
  196. {
  197. if (isUTF16LE((UINT8*)string.data()))
  198. {
  199. const std::codecvt_mode convMode = (std::codecvt_mode)(std::consume_header | std::little_endian);
  200. typedef std::codecvt_utf16<wchar_t, 1114111, convMode> wcharutf16;
  201. std::wstring_convert<wcharutf16> conversion("?");
  202. return conversion.from_bytes(string).c_str();
  203. }
  204. else if (isUTF16BE((UINT8*)string.data()))
  205. {
  206. const std::codecvt_mode convMode = (std::codecvt_mode)(std::consume_header);
  207. typedef std::codecvt_utf16<wchar_t, 1114111, convMode> wcharutf16;
  208. std::wstring_convert<wcharutf16> conversion("?");
  209. return conversion.from_bytes(string).c_str();
  210. }
  211. }
  212. {
  213. const std::codecvt_mode convMode = (std::codecvt_mode)(std::consume_header);
  214. typedef std::codecvt_utf8<wchar_t, 1114111, convMode> wcharutf8;
  215. std::wstring_convert<wcharutf8> conversion("?");
  216. return conversion.from_bytes(string).c_str();
  217. }
  218. }
  219. MemoryDataStream::MemoryDataStream(void* memory, size_t inSize)
  220. : DataStream(READ | WRITE), mData(nullptr)
  221. {
  222. mData = mPos = static_cast<UINT8*>(memory);
  223. mSize = inSize;
  224. mEnd = mData + mSize;
  225. assert(mEnd >= mPos);
  226. }
  227. MemoryDataStream::MemoryDataStream(DataStream& sourceStream)
  228. : DataStream(READ | WRITE), mData(nullptr)
  229. {
  230. // Copy data from incoming stream
  231. mSize = sourceStream.size();
  232. mData = (UINT8*)bs_alloc((UINT32)mSize);
  233. mPos = mData;
  234. mEnd = mData + sourceStream.read(mData, mSize);
  235. assert(mEnd >= mPos);
  236. }
  237. MemoryDataStream::MemoryDataStream(const DataStreamPtr& sourceStream)
  238. :DataStream(READ | WRITE), mData(nullptr)
  239. {
  240. // Copy data from incoming stream
  241. mSize = sourceStream->size();
  242. mData = (UINT8*)bs_alloc((UINT32)mSize);
  243. mPos = mData;
  244. mEnd = mData + sourceStream->read(mData, mSize);
  245. assert(mEnd >= mPos);
  246. }
  247. MemoryDataStream::~MemoryDataStream()
  248. {
  249. close();
  250. }
  251. size_t MemoryDataStream::read(void* buf, size_t count)
  252. {
  253. size_t cnt = count;
  254. if (mPos + cnt > mEnd)
  255. cnt = mEnd - mPos;
  256. if (cnt == 0)
  257. return 0;
  258. assert (cnt <= count);
  259. memcpy(buf, mPos, cnt);
  260. mPos += cnt;
  261. return cnt;
  262. }
  263. size_t MemoryDataStream::write(const void* buf, size_t count)
  264. {
  265. size_t written = 0;
  266. if (isWriteable())
  267. {
  268. written = count;
  269. if (mPos + written > mEnd)
  270. written = mEnd - mPos;
  271. if (written == 0)
  272. return 0;
  273. memcpy(mPos, buf, written);
  274. mPos += written;
  275. }
  276. return written;
  277. }
  278. void MemoryDataStream::skip(size_t count)
  279. {
  280. size_t newpos = (size_t)( (mPos - mData) + count );
  281. assert(mData + newpos <= mEnd);
  282. mPos = mData + newpos;
  283. }
  284. void MemoryDataStream::seek(size_t pos)
  285. {
  286. assert(mData + pos <= mEnd);
  287. mPos = mData + pos;
  288. }
  289. size_t MemoryDataStream::tell() const
  290. {
  291. return mPos - mData;
  292. }
  293. bool MemoryDataStream::eof() const
  294. {
  295. return mPos >= mEnd;
  296. }
  297. void MemoryDataStream::close()
  298. {
  299. if (mData != nullptr)
  300. {
  301. bs_free(mData);
  302. mData = nullptr;
  303. }
  304. }
  305. FileDataStream::FileDataStream(std::shared_ptr<std::ifstream> s, bool freeOnClose)
  306. : DataStream(READ), mpInStream(s), mpFStreamRO(s), mpFStream(0), mFreeOnClose(freeOnClose)
  307. {
  308. mpInStream->seekg(0, std::ios_base::end);
  309. mSize = (size_t)mpInStream->tellg();
  310. mpInStream->seekg(0, std::ios_base::beg);
  311. determineAccess();
  312. }
  313. FileDataStream::FileDataStream(std::shared_ptr<std::ifstream> s, size_t inSize, bool freeOnClose)
  314. : DataStream(READ), mpInStream(s), mpFStreamRO(s), mpFStream(0), mFreeOnClose(freeOnClose)
  315. {
  316. mSize = inSize;
  317. determineAccess();
  318. }
  319. FileDataStream::FileDataStream(std::shared_ptr<std::fstream> s, bool freeOnClose)
  320. : DataStream(READ | WRITE), mpInStream(s), mpFStreamRO(0), mpFStream(s), mFreeOnClose(freeOnClose)
  321. {
  322. mpInStream->seekg(0, std::ios_base::end);
  323. mSize = (size_t)mpInStream->tellg();
  324. mpInStream->seekg(0, std::ios_base::beg);
  325. determineAccess();
  326. }
  327. FileDataStream::FileDataStream(std::shared_ptr<std::fstream> s, size_t inSize, bool freeOnClose)
  328. : DataStream(READ | WRITE), mpInStream(s), mpFStreamRO(0), mpFStream(s), mFreeOnClose(freeOnClose)
  329. {
  330. mSize = inSize;
  331. determineAccess();
  332. }
  333. void FileDataStream::determineAccess()
  334. {
  335. mAccess = 0;
  336. if (mpInStream)
  337. mAccess |= READ;
  338. if (mpFStream)
  339. mAccess |= WRITE;
  340. }
  341. FileDataStream::~FileDataStream()
  342. {
  343. close();
  344. }
  345. size_t FileDataStream::read(void* buf, size_t count)
  346. {
  347. mpInStream->read(static_cast<char*>(buf), static_cast<std::streamsize>(count));
  348. return (size_t)mpInStream->gcount();
  349. }
  350. size_t FileDataStream::write(const void* buf, size_t count)
  351. {
  352. size_t written = 0;
  353. if (isWriteable() && mpFStream)
  354. {
  355. mpFStream->write(static_cast<const char*>(buf), static_cast<std::streamsize>(count));
  356. written = count;
  357. }
  358. return written;
  359. }
  360. void FileDataStream::skip(size_t count)
  361. {
  362. mpInStream->clear(); // Clear fail status in case eof was set
  363. mpInStream->seekg(static_cast<std::ifstream::pos_type>(count), std::ios::cur);
  364. }
  365. void FileDataStream::seek(size_t pos)
  366. {
  367. mpInStream->clear(); // Clear fail status in case eof was set
  368. mpInStream->seekg(static_cast<std::streamoff>(pos), std::ios::beg);
  369. }
  370. size_t FileDataStream::tell() const
  371. {
  372. mpInStream->clear(); // Clear fail status in case eof was set
  373. return (size_t)mpInStream->tellg();
  374. }
  375. bool FileDataStream::eof() const
  376. {
  377. return mpInStream->eof();
  378. }
  379. void FileDataStream::close()
  380. {
  381. if (mpInStream)
  382. {
  383. if (mpFStreamRO)
  384. mpFStreamRO->close();
  385. if (mpFStream)
  386. {
  387. mpFStream->flush();
  388. mpFStream->close();
  389. }
  390. if (mFreeOnClose)
  391. {
  392. mpInStream = nullptr;
  393. mpFStreamRO = nullptr;
  394. mpFStream = nullptr;
  395. }
  396. }
  397. }
  398. }