sysencoding.inc 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536
  1. {$ifndef VER2_4}
  2. { TEncoding }
  3. class function TEncoding.GetANSI: TEncoding;
  4. begin
  5. if not Assigned(FStandardEncodings[seAnsi]) then
  6. FStandardEncodings[seAnsi] := TMBCSEncoding.Create(DefaultSystemCodePage);
  7. Result := FStandardEncodings[seAnsi];
  8. end;
  9. class function TEncoding.GetASCII: TEncoding;
  10. begin
  11. if not Assigned(FStandardEncodings[seAscii]) then
  12. FStandardEncodings[seAscii] := TMBCSEncoding.Create(CP_ASCII);
  13. Result := FStandardEncodings[seAscii];
  14. end;
  15. class function TEncoding.GetBigEndianUnicode: TEncoding;
  16. begin
  17. if not Assigned(FStandardEncodings[seBigEndianUnicode]) then
  18. FStandardEncodings[seBigEndianUnicode] := TBigEndianUnicodeEncoding.Create;
  19. Result := FStandardEncodings[seBigEndianUnicode];
  20. end;
  21. class function TEncoding.GetDefault: TEncoding;
  22. begin
  23. Result := GetANSI;
  24. end;
  25. class function TEncoding.GetUnicode: TEncoding;
  26. begin
  27. if not Assigned(FStandardEncodings[seUnicode]) then
  28. FStandardEncodings[seUnicode] := TUnicodeEncoding.Create;
  29. Result := FStandardEncodings[seUnicode];
  30. end;
  31. class function TEncoding.GetUTF7: TEncoding;
  32. begin
  33. if not Assigned(FStandardEncodings[seUTF7]) then
  34. FStandardEncodings[seUTF7] := TUTF7Encoding.Create;
  35. Result := FStandardEncodings[seUTF7];
  36. end;
  37. class function TEncoding.GetUTF8: TEncoding;
  38. begin
  39. if not Assigned(FStandardEncodings[seUTF8]) then
  40. FStandardEncodings[seUTF8] := TUTF8Encoding.Create;
  41. Result := FStandardEncodings[seUTF8];
  42. end;
  43. class procedure TEncoding.FreeEncodings;
  44. var
  45. E: TStandardEncoding;
  46. begin
  47. for E := Low(FStandardEncodings) to High(FStandardEncodings) do
  48. FStandardEncodings[E].Free;
  49. end;
  50. class constructor TEncoding.Create;
  51. var
  52. E: TStandardEncoding;
  53. begin
  54. for E := Low(FStandardEncodings) to High(FStandardEncodings) do
  55. FStandardEncodings[E] := nil;
  56. end;
  57. class destructor TEncoding.Destroy;
  58. begin
  59. FreeEncodings;
  60. end;
  61. function TEncoding.Clone: TEncoding;
  62. begin
  63. Result := nil;
  64. end;
  65. class function TEncoding.Convert(Source, Destination: TEncoding;
  66. const Bytes: TBytes): TBytes;
  67. begin
  68. Result := Destination.GetBytes(Source.GetChars(Bytes));
  69. end;
  70. class function TEncoding.Convert(Source, Destination: TEncoding;
  71. const Bytes: TBytes; StartIndex, Count: Integer): TBytes;
  72. begin
  73. Result := Destination.GetBytes(Source.GetChars(Bytes, StartIndex, Count));
  74. end;
  75. class function TEncoding.IsStandardEncoding(AEncoding: TEncoding): Boolean;
  76. var
  77. Encoding: TEncoding;
  78. begin
  79. if Assigned(AEncoding) then
  80. for Encoding in FStandardEncodings do
  81. if Encoding = AEncoding then
  82. Exit(True);
  83. Result := False;
  84. end;
  85. class function TEncoding.GetBufferEncoding(const Buffer: TBytes; var AEncoding: TEncoding): Integer;
  86. begin
  87. Result := GetBufferEncoding(Buffer, AEncoding, Default);
  88. end;
  89. class function TEncoding.GetBufferEncoding(const Buffer: TBytes;
  90. var AEncoding: TEncoding; ADefaultEncoding: TEncoding): Integer;
  91. function CheckEncoding(AEncoding: TEncoding; out ByteCount: Integer): Boolean;
  92. var
  93. Preamble: TBytes;
  94. begin
  95. Preamble := AEncoding.GetPreamble;
  96. ByteCount := Length(Preamble);
  97. Result := (Length(Buffer) >= ByteCount) and (ByteCount > 0);
  98. if Result then
  99. Result := CompareMem(@Preamble[0], @Buffer[0], ByteCount);
  100. end;
  101. begin
  102. if Assigned(AEncoding) then
  103. begin
  104. if not CheckEncoding(AEncoding, Result) then
  105. Result := 0;
  106. end
  107. else
  108. if CheckEncoding(Unicode, Result) then
  109. AEncoding := Unicode
  110. else
  111. if CheckEncoding(BigEndianUnicode, Result) then
  112. AEncoding := BigEndianUnicode
  113. else
  114. if CheckEncoding(UTF8, Result) then
  115. AEncoding := UTF8
  116. else
  117. begin
  118. AEncoding := ADefaultEncoding;
  119. Result := 0;
  120. end;
  121. end;
  122. function TEncoding.GetByteCount(const Chars: TUnicodeCharArray): Integer;
  123. begin
  124. Result := GetByteCount(Chars, 0, Length(Chars));
  125. end;
  126. function TEncoding.GetByteCount(const Chars: TUnicodeCharArray; CharIndex,
  127. CharCount: Integer): Integer;
  128. begin
  129. Result := GetByteCount(@Chars[CharIndex], CharCount);
  130. end;
  131. function TEncoding.GetByteCount(const S: UnicodeString): Integer;
  132. begin
  133. Result := GetByteCount(PUnicodeChar(S), Length(S));
  134. end;
  135. function TEncoding.GetByteCount(const S: UnicodeString; CharIndex, CharCount: Integer): Integer;
  136. begin
  137. Result := GetByteCount(@S[CharIndex], CharCount);
  138. end;
  139. function TEncoding.GetBytes(const Chars: TUnicodeCharArray): TBytes;
  140. begin
  141. SetLength(Result, GetByteCount(Chars));
  142. GetBytes(@Chars[0], Length(Chars), @Result[0], Length(Result));
  143. end;
  144. function TEncoding.GetBytes(const Chars: TUnicodeCharArray; CharIndex,
  145. CharCount: Integer): TBytes;
  146. begin
  147. SetLength(Result, GetByteCount(Chars, CharIndex, CharCount));
  148. GetBytes(@Chars[CharIndex], CharCount, @Result[0], Length(Result));
  149. end;
  150. function TEncoding.GetBytes(const Chars: TUnicodeCharArray; CharIndex,
  151. CharCount: Integer; const Bytes: TBytes; ByteIndex: Integer): Integer;
  152. begin
  153. Result := GetBytes(@Chars[CharIndex], CharCount, @Bytes[ByteIndex], Length(Bytes) - ByteIndex);
  154. end;
  155. function TEncoding.GetBytes(const S: UnicodeString): TBytes;
  156. begin
  157. SetLength(Result, GetByteCount(S));
  158. GetBytes(@S[1], Length(S), @Result[0], Length(Result));
  159. end;
  160. function TEncoding.GetBytes(const S: UnicodeString; CharIndex, CharCount: Integer;
  161. const Bytes: TBytes; ByteIndex: Integer): Integer;
  162. begin
  163. Result := GetBytes(@S[CharIndex], CharCount, @Bytes[ByteIndex], Length(Bytes) - ByteIndex);
  164. end;
  165. function TEncoding.GetCharCount(const Bytes: TBytes): Integer;
  166. begin
  167. Result := GetCharCount(@Bytes[0], Length(Bytes));
  168. end;
  169. function TEncoding.GetCharCount(const Bytes: TBytes; ByteIndex,
  170. ByteCount: Integer): Integer;
  171. begin
  172. Result := GetCharCount(@Bytes[ByteIndex], ByteCount);
  173. end;
  174. function TEncoding.GetChars(const Bytes: TBytes): TUnicodeCharArray;
  175. begin
  176. SetLength(Result, GetCharCount(Bytes));
  177. GetChars(@Bytes[0], Length(Bytes), @Result[0], Length(Result));
  178. end;
  179. function TEncoding.GetChars(const Bytes: TBytes; ByteIndex, ByteCount: Integer): TUnicodeCharArray;
  180. begin
  181. SetLength(Result, GetCharCount(Bytes, ByteIndex, ByteCount));
  182. GetChars(@Bytes[ByteIndex], ByteCount, @Result[0], Length(Result));
  183. end;
  184. function TEncoding.GetChars(const Bytes: TBytes; ByteIndex, ByteCount: Integer;
  185. const Chars: TUnicodeCharArray; CharIndex: Integer): Integer;
  186. begin
  187. Result := GetChars(@Bytes[ByteIndex], ByteCount, @Chars[CharIndex], Length(Chars) - CharIndex);
  188. end;
  189. class function TEncoding.GetEncoding(CodePage: Integer): TEncoding;
  190. begin
  191. case CodePage of
  192. CP_UTF16: Result := TUnicodeEncoding.Create;
  193. CP_UTF16BE: Result := TBigEndianUnicodeEncoding.Create;
  194. CP_UTF7: Result := TUTF7Encoding.Create;
  195. CP_UTF8: Result := TUTF8Encoding.Create;
  196. else
  197. Result := TMBCSEncoding.Create(CodePage);
  198. end;
  199. end;
  200. class function TEncoding.GetEncoding(const EncodingName: UnicodeString): TEncoding;
  201. var
  202. ACodePage: TSystemCodePage;
  203. begin
  204. ACodePage := CodePageNameToCodePage(EncodingName);
  205. if ACodePage = $FFFF then
  206. raise EEncodingError.CreateFmt(SNotValidCodePageName, [EncodingName]);
  207. Result := TMBCSEncoding.Create(ACodePage);
  208. end;
  209. function TEncoding.GetString(const Bytes: TBytes): UnicodeString;
  210. var
  211. Chars: TUnicodeCharArray;
  212. begin
  213. Chars := GetChars(Bytes);
  214. SetString(Result, PUnicodeChar(Chars), Length(Chars));
  215. end;
  216. function TEncoding.GetString(const Bytes: TBytes; ByteIndex, ByteCount: Integer): UnicodeString;
  217. var
  218. Chars: TUnicodeCharArray;
  219. begin
  220. Chars := GetChars(Bytes, ByteIndex, ByteCount);
  221. SetString(Result, PUnicodeChar(Chars), Length(Chars));
  222. end;
  223. { TMBCSEncoding }
  224. function TMBCSEncoding.GetByteCount(Chars: PUnicodeChar; CharCount: Integer): Integer;
  225. var
  226. S: RawByteString;
  227. begin
  228. widestringmanager.Unicode2AnsiMoveProc(Chars, S, CodePage, CharCount);
  229. Result := Length(S);
  230. end;
  231. function TMBCSEncoding.GetBytes(Chars: PUnicodeChar; CharCount: Integer; Bytes: PByte;
  232. ByteCount: Integer): Integer;
  233. var
  234. S: RawByteString;
  235. begin
  236. widestringmanager.Unicode2AnsiMoveProc(Chars, S, CodePage, CharCount);
  237. Result := Length(S);
  238. if ByteCount < Result then
  239. Result := ByteCount;
  240. if Result > 0 then
  241. Move(S[1], Bytes[0], Result);
  242. end;
  243. function TMBCSEncoding.GetCharCount(Bytes: PByte; ByteCount: Integer): Integer;
  244. var
  245. U: UnicodeString;
  246. begin
  247. widestringmanager.Ansi2UnicodeMoveProc(PChar(Bytes), CodePage, U, ByteCount);
  248. Result := Length(U);
  249. end;
  250. function TMBCSEncoding.GetChars(Bytes: PByte; ByteCount: Integer; Chars: PUnicodeChar;
  251. CharCount: Integer): Integer;
  252. var
  253. U: UnicodeString;
  254. begin
  255. widestringmanager.Ansi2UnicodeMoveProc(PChar(Bytes), CodePage, U, ByteCount);
  256. Result := Length(U);
  257. if CharCount < Result then
  258. Result := CharCount;
  259. if Result > 0 then
  260. Move(U[1], Chars[0], Result * SizeOf(UnicodeChar));
  261. end;
  262. function TMBCSEncoding.GetCodePage: Cardinal;
  263. begin
  264. Result := FCodePage;
  265. end;
  266. function TMBCSEncoding.GetEncodingName: UnicodeString;
  267. begin
  268. Result := CodePageToCodePageName(CodePage);
  269. end;
  270. constructor TMBCSEncoding.Create;
  271. begin
  272. Create(DefaultSystemCodePage, 0, 0);
  273. end;
  274. constructor TMBCSEncoding.Create(ACodePage: Integer);
  275. begin
  276. Create(ACodePage, 0, 0);
  277. end;
  278. constructor TMBCSEncoding.Create(ACodePage, MBToWCharFlags,
  279. WCharToMBFlags: Integer);
  280. begin
  281. FCodePage := ACodePage;
  282. FMBToWCharFlags := MBToWCharFlags;
  283. FWCharToMBFlags := WCharToMBFlags;
  284. end;
  285. function TMBCSEncoding.Clone: TEncoding;
  286. begin
  287. Result := TMBCSEncoding.Create(FCodePage, FMBToWCharFlags, FWCharToMBFlags);
  288. end;
  289. function TMBCSEncoding.GetMaxByteCount(CharCount: Integer): Integer;
  290. begin
  291. Result := CharCount;
  292. end;
  293. function TMBCSEncoding.GetMaxCharCount(ByteCount: Integer): Integer;
  294. begin
  295. Result := ByteCount;
  296. end;
  297. function TMBCSEncoding.GetPreamble: TBytes;
  298. begin
  299. Result := nil;
  300. end;
  301. { TUTF7Encoding }
  302. constructor TUTF7Encoding.Create;
  303. begin
  304. inherited Create(CP_UTF7);
  305. FIsSingleByte := False;
  306. end;
  307. function TUTF7Encoding.Clone: TEncoding;
  308. begin
  309. Result := TUTF7Encoding.Create;
  310. end;
  311. function TUTF7Encoding.GetMaxByteCount(CharCount: Integer): Integer;
  312. begin
  313. Result := CharCount * 3 + 2;
  314. end;
  315. function TUTF7Encoding.GetMaxCharCount(ByteCount: Integer): Integer;
  316. begin
  317. Result := ByteCount;
  318. end;
  319. { TUTF8Encoding }
  320. constructor TUTF8Encoding.Create;
  321. begin
  322. inherited Create(CP_UTF8);
  323. FIsSingleByte := False;
  324. end;
  325. function TUTF8Encoding.Clone: TEncoding;
  326. begin
  327. Result := TUTF8Encoding.Create;
  328. end;
  329. function TUTF8Encoding.GetMaxByteCount(CharCount: Integer): Integer;
  330. begin
  331. Result := CharCount * 3;
  332. end;
  333. function TUTF8Encoding.GetMaxCharCount(ByteCount: Integer): Integer;
  334. begin
  335. Result := ByteCount;
  336. end;
  337. function TUTF8Encoding.GetPreamble: TBytes;
  338. begin
  339. SetLength(Result, 3);
  340. Result[0] := $EF;
  341. Result[1] := $BB;
  342. Result[2] := $BF;
  343. end;
  344. { TUnicodeEncoding }
  345. function TUnicodeEncoding.GetByteCount(Chars: PUnicodeChar; CharCount: Integer): Integer;
  346. begin
  347. Result := CharCount * SizeOf(UnicodeChar);
  348. end;
  349. function TUnicodeEncoding.GetBytes(Chars: PUnicodeChar; CharCount: Integer;
  350. Bytes: PByte; ByteCount: Integer): Integer;
  351. begin
  352. Result := CharCount * SizeOf(UnicodeChar);
  353. if ByteCount < Result then
  354. Result := ByteCount;
  355. if Result > 0 then
  356. Move(Chars[0], Bytes[0], Result);
  357. end;
  358. function TUnicodeEncoding.GetCharCount(Bytes: PByte; ByteCount: Integer): Integer;
  359. begin
  360. Result := ByteCount div SizeOf(UnicodeChar);
  361. end;
  362. function TUnicodeEncoding.GetChars(Bytes: PByte; ByteCount: Integer;
  363. Chars: PUnicodeChar; CharCount: Integer): Integer;
  364. begin
  365. Result := ByteCount div 2;
  366. if CharCount < Result then
  367. Result := CharCount;
  368. Move(Bytes[0], Chars[0], Result * SizeOf(UnicodeChar));
  369. end;
  370. function TUnicodeEncoding.GetCodePage: Cardinal;
  371. begin
  372. Result := CP_UTF16;
  373. end;
  374. function TUnicodeEncoding.GetEncodingName: UnicodeString;
  375. begin
  376. Result := CodePageToCodePageName(CodePage);
  377. end;
  378. constructor TUnicodeEncoding.Create;
  379. begin
  380. inherited Create;
  381. FIsSingleByte := False;
  382. FMaxCharSize := SizeOf(UnicodeChar);
  383. end;
  384. function TUnicodeEncoding.Clone: TEncoding;
  385. begin
  386. Result := TUnicodeEncoding.Create;
  387. end;
  388. function TUnicodeEncoding.GetMaxByteCount(CharCount: Integer): Integer;
  389. begin
  390. Result := CharCount * SizeOf(UnicodeChar);
  391. end;
  392. function TUnicodeEncoding.GetMaxCharCount(ByteCount: Integer): Integer;
  393. begin
  394. Result := ByteCount div SizeOf(UnicodeChar);
  395. end;
  396. function TUnicodeEncoding.GetPreamble: TBytes;
  397. begin
  398. SetLength(Result, 2);
  399. Result[0] := $FF;
  400. Result[1] := $FE;
  401. end;
  402. { TBigEndianUnicodeEncoding }
  403. function TBigEndianUnicodeEncoding.GetBytes(Chars: PUnicodeChar; CharCount: Integer;
  404. Bytes: PByte; ByteCount: Integer): Integer;
  405. var
  406. LastByte: PByte;
  407. begin
  408. Result := CharCount * SizeOf(UnicodeChar);
  409. if ByteCount < Result then
  410. Result := ByteCount;
  411. LastByte := @Bytes[Result];
  412. while Bytes < LastByte do
  413. begin
  414. Bytes^ := Hi(Word(Chars^));
  415. inc(Bytes);
  416. if Bytes < LastByte then
  417. Bytes^ := Lo(Word(Chars^));
  418. inc(Bytes);
  419. inc(Chars);
  420. end;
  421. end;
  422. function TBigEndianUnicodeEncoding.GetChars(Bytes: PByte; ByteCount: Integer;
  423. Chars: PUnicodeChar; CharCount: Integer): Integer;
  424. var
  425. LastChar: PUnicodeChar;
  426. begin
  427. Result := ByteCount div SizeOf(UnicodeChar);
  428. if CharCount < Result then
  429. Result := CharCount;
  430. LastChar := @Chars[Result];
  431. while Chars <= LastChar do
  432. begin
  433. Chars^ := UnicodeChar(Bytes[1] + Bytes[0] shl 8);
  434. inc(Bytes, SizeOf(UnicodeChar));
  435. inc(Chars);
  436. end;
  437. end;
  438. function TBigEndianUnicodeEncoding.GetCodePage: Cardinal;
  439. begin
  440. Result := CP_UTF16BE;
  441. end;
  442. function TBigEndianUnicodeEncoding.GetEncodingName: UnicodeString;
  443. begin
  444. Result := CodePageToCodePageName(CodePage);
  445. end;
  446. function TBigEndianUnicodeEncoding.Clone: TEncoding;
  447. begin
  448. Result := TBigEndianUnicodeEncoding.Create;
  449. end;
  450. function TBigEndianUnicodeEncoding.GetPreamble: TBytes;
  451. begin
  452. SetLength(Result, 2);
  453. Result[0] := $FE;
  454. Result[1] := $FF;
  455. end;
  456. {$endif VER2_4}