Version 3.6.0.2, tag libreoffice-3.6.0.2
[LibreOffice.git] / sal / textenc / convertiso2022kr.cxx
blobd4e51ea26710e4f532cd224e6a27214451b62d86
1 /* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
2 /*************************************************************************
4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
6 * Copyright 2000, 2010 Oracle and/or its affiliates.
8 * OpenOffice.org - a multi-platform office productivity suite
10 * This file is part of OpenOffice.org.
12 * OpenOffice.org is free software: you can redistribute it and/or modify
13 * it under the terms of the GNU Lesser General Public License version 3
14 * only, as published by the Free Software Foundation.
16 * OpenOffice.org is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 * GNU Lesser General Public License version 3 for more details
20 * (a copy is included in the LICENSE file that accompanied this code).
22 * You should have received a copy of the GNU Lesser General Public License
23 * version 3 along with OpenOffice.org. If not, see
24 * <http://www.openoffice.org/license.html>
25 * for a copy of the LGPLv3 License.
27 ************************************************************************/
29 #include "sal/config.h"
31 #include "rtl/textcvt.h"
32 #include "sal/types.h"
34 #include "context.hxx"
35 #include "converter.hxx"
36 #include "convertiso2022kr.hxx"
37 #include "tenchelp.hxx"
38 #include "unichars.hxx"
40 namespace {
42 enum ImplIso2022KrToUnicodeState // order is important:
44 IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII,
45 IMPL_ISO_2022_KR_TO_UNICODE_STATE_1001,
46 IMPL_ISO_2022_KR_TO_UNICODE_STATE_1001_2,
47 IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC,
48 IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC_DOLLAR,
49 IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC_DOLLAR_RPAREN
52 struct ImplIso2022KrToUnicodeContext
54 ImplIso2022KrToUnicodeState m_eState;
55 sal_uInt32 m_nRow;
58 enum ImplUnicodeToIso2022KrSet
60 IMPL_UNICODE_TO_ISO_2022_KR_SET_NONE,
61 IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII,
62 IMPL_UNICODE_TO_ISO_2022_KR_SET_1001
65 struct ImplUnicodeToIso2022KrContext
67 sal_Unicode m_nHighSurrogate;
68 ImplUnicodeToIso2022KrSet m_eSet;
73 void * ImplCreateIso2022KrToUnicodeContext()
75 ImplIso2022KrToUnicodeContext * pContext =
76 new ImplIso2022KrToUnicodeContext;
77 pContext->m_eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII;
78 return pContext;
81 void ImplResetIso2022KrToUnicodeContext(void * pContext)
83 if (pContext)
84 static_cast< ImplIso2022KrToUnicodeContext * >(pContext)->m_eState
85 = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII;
88 void ImplDestroyIso2022KrToUnicodeContext(void * pContext)
90 delete static_cast< ImplIso2022KrToUnicodeContext * >(pContext);
93 sal_Size ImplConvertIso2022KrToUnicode(void const * pData,
94 void * pContext,
95 char const * pSrcBuf,
96 sal_Size nSrcBytes,
97 sal_Unicode * pDestBuf,
98 sal_Size nDestChars,
99 sal_uInt32 nFlags,
100 sal_uInt32 * pInfo,
101 sal_Size * pSrcCvtBytes)
103 ImplDBCSToUniLeadTab const * pKsX1001Data
104 = static_cast< ImplIso2022KrConverterData const * >(pData)->
105 m_pKsX1001ToUnicodeData;
106 ImplIso2022KrToUnicodeState eState
107 = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII;
108 sal_uInt32 nRow = 0;
109 sal_uInt32 nInfo = 0;
110 sal_Size nConverted = 0;
111 sal_Unicode * pDestBufPtr = pDestBuf;
112 sal_Unicode * pDestBufEnd = pDestBuf + nDestChars;
114 if (pContext)
116 eState = static_cast< ImplIso2022KrToUnicodeContext * >(pContext)->m_eState;
117 nRow = static_cast< ImplIso2022KrToUnicodeContext * >(pContext)->m_nRow;
120 for (; nConverted < nSrcBytes; ++nConverted)
122 bool bUndefined = true;
123 sal_uInt32 nChar = *(sal_uChar const *) pSrcBuf++;
124 switch (eState)
126 case IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII:
127 if (nChar == 0x0E) // SO
128 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_1001;
129 else if (nChar == 0x1B) // ESC
130 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC;
131 else if (nChar < 0x80)
132 if (pDestBufPtr != pDestBufEnd)
133 *pDestBufPtr++ = (sal_Unicode) nChar;
134 else
135 goto no_output;
136 else
138 bUndefined = false;
139 goto bad_input;
141 break;
143 case IMPL_ISO_2022_KR_TO_UNICODE_STATE_1001:
144 if (nChar == 0x0F) // SI
145 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII;
146 else if (nChar >= 0x21 && nChar <= 0x7E)
148 nRow = nChar + 0x80;
149 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_1001_2;
151 else
153 bUndefined = false;
154 goto bad_input;
156 break;
158 case IMPL_ISO_2022_KR_TO_UNICODE_STATE_1001_2:
159 if (nChar >= 0x21 && nChar <= 0x7E)
161 sal_uInt16 nUnicode = 0;
162 sal_uInt32 nFirst = pKsX1001Data[nRow].mnTrailStart;
163 nChar += 0x80;
164 if (nChar >= nFirst && nChar <= pKsX1001Data[nRow].mnTrailEnd)
165 nUnicode = pKsX1001Data[nRow].
166 mpToUniTrailTab[nChar - nFirst];
167 if (nUnicode != 0)
168 if (pDestBufPtr != pDestBufEnd)
170 *pDestBufPtr++ = (sal_Unicode) nUnicode;
171 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_1001;
173 else
174 goto no_output;
175 else
176 goto bad_input;
178 else
180 bUndefined = false;
181 goto bad_input;
183 break;
185 case IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC:
186 if (nChar == 0x24) // $
187 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC_DOLLAR;
188 else
190 bUndefined = false;
191 goto bad_input;
193 break;
195 case IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC_DOLLAR:
196 if (nChar == 0x29) // )
197 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC_DOLLAR_RPAREN;
198 else
200 bUndefined = false;
201 goto bad_input;
203 break;
205 case IMPL_ISO_2022_KR_TO_UNICODE_STATE_ESC_DOLLAR_RPAREN:
206 if (nChar == 0x43) // C
207 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII;
208 else
210 bUndefined = false;
211 goto bad_input;
213 break;
215 continue;
217 bad_input:
218 switch (sal::detail::textenc::handleBadInputTextToUnicodeConversion(
219 bUndefined, true, 0, nFlags, &pDestBufPtr, pDestBufEnd,
220 &nInfo))
222 case sal::detail::textenc::BAD_INPUT_STOP:
223 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII;
224 break;
226 case sal::detail::textenc::BAD_INPUT_CONTINUE:
227 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII;
228 continue;
230 case sal::detail::textenc::BAD_INPUT_NO_OUTPUT:
231 goto no_output;
233 break;
235 no_output:
236 --pSrcBuf;
237 nInfo |= RTL_TEXTTOUNICODE_INFO_DESTBUFFERTOSMALL;
238 break;
241 if (eState > IMPL_ISO_2022_KR_TO_UNICODE_STATE_1001
242 && (nInfo & (RTL_TEXTTOUNICODE_INFO_ERROR
243 | RTL_TEXTTOUNICODE_INFO_DESTBUFFERTOSMALL))
244 == 0)
246 if ((nFlags & RTL_TEXTTOUNICODE_FLAGS_FLUSH) == 0)
247 nInfo |= RTL_TEXTTOUNICODE_INFO_SRCBUFFERTOSMALL;
248 else
249 switch (sal::detail::textenc::handleBadInputTextToUnicodeConversion(
250 false, true, 0, nFlags, &pDestBufPtr, pDestBufEnd,
251 &nInfo))
253 case sal::detail::textenc::BAD_INPUT_STOP:
254 case sal::detail::textenc::BAD_INPUT_CONTINUE:
255 eState = IMPL_ISO_2022_KR_TO_UNICODE_STATE_ASCII;
256 break;
258 case sal::detail::textenc::BAD_INPUT_NO_OUTPUT:
259 nInfo |= RTL_TEXTTOUNICODE_INFO_DESTBUFFERTOSMALL;
260 break;
264 if (pContext)
266 static_cast< ImplIso2022KrToUnicodeContext * >(pContext)->m_eState = eState;
267 static_cast< ImplIso2022KrToUnicodeContext * >(pContext)->m_nRow = nRow;
269 if (pInfo)
270 *pInfo = nInfo;
271 if (pSrcCvtBytes)
272 *pSrcCvtBytes = nConverted;
274 return pDestBufPtr - pDestBuf;
277 void * ImplCreateUnicodeToIso2022KrContext()
279 ImplUnicodeToIso2022KrContext * pContext =
280 new ImplUnicodeToIso2022KrContext;
281 pContext->m_nHighSurrogate = 0;
282 pContext->m_eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_NONE;
283 return pContext;
286 void ImplResetUnicodeToIso2022KrContext(void * pContext)
288 if (pContext)
290 static_cast< ImplUnicodeToIso2022KrContext * >(pContext)->m_nHighSurrogate = 0;
291 static_cast< ImplUnicodeToIso2022KrContext * >(pContext)->m_eSet
292 = IMPL_UNICODE_TO_ISO_2022_KR_SET_NONE;
296 void ImplDestroyUnicodeToIso2022KrContext(void * pContext)
298 delete static_cast< ImplUnicodeToIso2022KrContext * >(pContext);
301 sal_Size ImplConvertUnicodeToIso2022Kr(void const * pData,
302 void * pContext,
303 sal_Unicode const * pSrcBuf,
304 sal_Size nSrcChars,
305 char * pDestBuf,
306 sal_Size nDestBytes,
307 sal_uInt32 nFlags,
308 sal_uInt32 * pInfo,
309 sal_Size * pSrcCvtChars)
311 ImplUniToDBCSHighTab const * pKsX1001Data
312 = static_cast< ImplIso2022KrConverterData const * >(pData)->
313 m_pUnicodeToKsX1001Data;
314 sal_Unicode nHighSurrogate = 0;
315 ImplUnicodeToIso2022KrSet eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_NONE;
316 sal_uInt32 nInfo = 0;
317 sal_Size nConverted = 0;
318 char * pDestBufPtr = pDestBuf;
319 char * pDestBufEnd = pDestBuf + nDestBytes;
320 bool bWritten;
322 if (pContext)
324 nHighSurrogate
325 = static_cast< ImplUnicodeToIso2022KrContext * >(pContext)->m_nHighSurrogate;
326 eSet = static_cast< ImplUnicodeToIso2022KrContext * >(pContext)->m_eSet;
329 if (eSet == IMPL_UNICODE_TO_ISO_2022_KR_SET_NONE)
331 if (pDestBufEnd - pDestBufPtr >= 4)
333 *pDestBufPtr++ = 0x1B; // ESC
334 *pDestBufPtr++ = 0x24; // $
335 *pDestBufPtr++ = 0x29; // )
336 *pDestBufPtr++ = 0x43; // C
337 eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII;
339 else
340 nInfo |= RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL;
343 if ((nInfo & RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL) == 0)
344 for (; nConverted < nSrcChars; ++nConverted)
346 bool bUndefined = true;
347 sal_uInt32 nChar = *pSrcBuf++;
348 if (nHighSurrogate == 0)
350 if (ImplIsHighSurrogate(nChar))
352 nHighSurrogate = (sal_Unicode) nChar;
353 continue;
356 else if (ImplIsLowSurrogate(nChar))
357 nChar = ImplCombineSurrogates(nHighSurrogate, nChar);
358 else
360 bUndefined = false;
361 goto bad_input;
364 if (ImplIsLowSurrogate(nChar) || ImplIsNoncharacter(nChar))
366 bUndefined = false;
367 goto bad_input;
370 if (nChar == 0x0A || nChar == 0x0D) // LF, CR
372 if (eSet == IMPL_UNICODE_TO_ISO_2022_KR_SET_1001)
374 if (pDestBufPtr != pDestBufEnd)
376 *pDestBufPtr++ = 0x0F; // SI
377 eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII;
379 else
380 goto no_output;
382 if (pDestBufPtr != pDestBufEnd)
383 *pDestBufPtr++ = static_cast< char >(nChar);
384 else
385 goto no_output;
387 else if (nChar == 0x0E || nChar == 0x0F || nChar == 0x1B)
388 goto bad_input;
389 else if (nChar < 0x80)
391 if (eSet == IMPL_UNICODE_TO_ISO_2022_KR_SET_1001)
393 if (pDestBufPtr != pDestBufEnd)
395 *pDestBufPtr++ = 0x0F; // SI
396 eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII;
398 else
399 goto no_output;
401 if (pDestBufPtr != pDestBufEnd)
402 *pDestBufPtr++ = static_cast< char >(nChar);
403 else
404 goto no_output;
406 else
408 sal_uInt16 nBytes = 0;
409 sal_uInt32 nIndex1 = nChar >> 8;
410 if (nIndex1 < 0x100)
412 sal_uInt32 nIndex2 = nChar & 0xFF;
413 sal_uInt32 nFirst = pKsX1001Data[nIndex1].mnLowStart;
414 if (nIndex2 >= nFirst
415 && nIndex2 <= pKsX1001Data[nIndex1].mnLowEnd)
416 nBytes = pKsX1001Data[nIndex1].
417 mpToUniTrailTab[nIndex2 - nFirst];
419 if (nBytes != 0)
421 if (eSet == IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII)
423 if (pDestBufPtr != pDestBufEnd)
425 *pDestBufPtr++ = 0x0E; // SO
426 eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_1001;
428 else
429 goto no_output;
431 if (pDestBufEnd - pDestBufPtr >= 2)
433 *pDestBufPtr++ = static_cast< char >((nBytes >> 8) & 0x7F);
434 *pDestBufPtr++ = static_cast< char >(nBytes & 0x7F);
436 else
437 goto no_output;
439 else
440 goto bad_input;
442 nHighSurrogate = 0;
443 continue;
445 bad_input:
446 switch (sal::detail::textenc::handleBadInputUnicodeToTextConversion(
447 bUndefined, nChar, nFlags, &pDestBufPtr, pDestBufEnd,
448 &nInfo, "\x0F" /* SI */,
449 eSet == IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII ? 0 : 1,
450 &bWritten))
452 case sal::detail::textenc::BAD_INPUT_STOP:
453 nHighSurrogate = 0;
454 break;
456 case sal::detail::textenc::BAD_INPUT_CONTINUE:
457 if (bWritten)
458 eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII;
459 nHighSurrogate = 0;
460 continue;
462 case sal::detail::textenc::BAD_INPUT_NO_OUTPUT:
463 goto no_output;
465 break;
467 no_output:
468 --pSrcBuf;
469 nInfo |= RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL;
470 break;
473 if ((nInfo & (RTL_UNICODETOTEXT_INFO_ERROR
474 | RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL))
475 == 0)
477 bool bFlush = true;
478 if (nHighSurrogate != 0)
480 if ((nFlags & RTL_UNICODETOTEXT_FLAGS_FLUSH) != 0)
481 nInfo |= RTL_UNICODETOTEXT_INFO_SRCBUFFERTOSMALL;
482 else
483 switch (sal::detail::textenc::handleBadInputUnicodeToTextConversion(
484 false, 0, nFlags, &pDestBufPtr, pDestBufEnd, &nInfo,
485 "\x0F" /* SI */,
486 (eSet == IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII
487 ? 0 : 1),
488 &bWritten))
490 case sal::detail::textenc::BAD_INPUT_STOP:
491 nHighSurrogate = 0;
492 bFlush = false;
493 break;
495 case sal::detail::textenc::BAD_INPUT_CONTINUE:
496 if (bWritten)
497 eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII;
498 nHighSurrogate = 0;
499 break;
501 case sal::detail::textenc::BAD_INPUT_NO_OUTPUT:
502 nInfo |= RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL;
503 break;
506 if (bFlush
507 && eSet == IMPL_UNICODE_TO_ISO_2022_KR_SET_1001
508 && (nFlags & RTL_UNICODETOTEXT_FLAGS_FLUSH) != 0)
510 if (pDestBufPtr != pDestBufEnd)
512 *pDestBufPtr++ = 0x0F; // SI
513 eSet = IMPL_UNICODE_TO_ISO_2022_KR_SET_ASCII;
515 else
516 nInfo |= RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL;
520 if (pContext)
522 static_cast< ImplUnicodeToIso2022KrContext * >(pContext)->m_nHighSurrogate
523 = nHighSurrogate;
524 static_cast< ImplUnicodeToIso2022KrContext * >(pContext)->m_eSet = eSet;
526 if (pInfo)
527 *pInfo = nInfo;
528 if (pSrcCvtChars)
529 *pSrcCvtChars = nConverted;
531 return pDestBufPtr - pDestBuf;
534 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */