• file
• revisions
• annotate
• diff
• comparison
• raw

intl/icu/source/common/ucnvscsu.c@fc2d59ddac77 (annotated)

intl/icu/source/common/ucnvscsu.c

Wed, 31 Dec 2014 07:22:50 +0100

author

Michael Schloh von Bennewitz <michael@schloh.com>

date

Wed, 31 Dec 2014 07:22:50 +0100

branch

TOR_BUG_3246

changeset 4

fc2d59ddac77

permissions

-rw-r--r--

Correct previous dual key logic pending first delivery installment.

michael@0	1	/*
michael@0	2	******************************************************************************
michael@0	3	*
michael@0	4	* Copyright (C) 2000-2011, International Business Machines
michael@0	5	* Corporation and others. All Rights Reserved.
michael@0	6	*
michael@0	7	******************************************************************************
michael@0	8	* file name: ucnvscsu.c
michael@0	9	* encoding: US-ASCII
michael@0	10	* tab size: 8 (not used)
michael@0	11	* indentation:4
michael@0	12	*
michael@0	13	* created on: 2000nov18
michael@0	14	* created by: Markus W. Scherer
michael@0	15	*
michael@0	16	* This is an implementation of the Standard Compression Scheme for Unicode
michael@0	17	* as defined in http://www.unicode.org/unicode/reports/tr6/ .
michael@0	18	* Reserved commands and window settings are treated as illegal sequences and
michael@0	19	* will result in callback calls.
michael@0	20	*/
michael@0	21
michael@0	22	#include "unicode/utypes.h"
michael@0	23
michael@0	24	#if !UCONFIG_NO_CONVERSION
michael@0	25
michael@0	26	#include "unicode/ucnv.h"
michael@0	27	#include "unicode/ucnv_cb.h"
michael@0	28	#include "unicode/utf16.h"
michael@0	29	#include "ucnv_bld.h"
michael@0	30	#include "ucnv_cnv.h"
michael@0	31	#include "cmemory.h"
michael@0	32
michael@0	33	/* SCSU definitions --------------------------------------------------------- */
michael@0	34
michael@0	35	/* SCSU command byte values */
michael@0	36	enum {
michael@0	37	SQ0=0x01, /* Quote from window pair 0 */
michael@0	38	SQ7=0x08, /* Quote from window pair 7 */
michael@0	39	SDX=0x0B, /* Define a window as extended */
michael@0	40	Srs=0x0C, /* reserved */
michael@0	41	SQU=0x0E, /* Quote a single Unicode character */
michael@0	42	SCU=0x0F, /* Change to Unicode mode */
michael@0	43	SC0=0x10, /* Select window 0 */
michael@0	44	SC7=0x17, /* Select window 7 */
michael@0	45	SD0=0x18, /* Define and select window 0 */
michael@0	46	SD7=0x1F, /* Define and select window 7 */
michael@0	47
michael@0	48	UC0=0xE0, /* Select window 0 */
michael@0	49	UC7=0xE7, /* Select window 7 */
michael@0	50	UD0=0xE8, /* Define and select window 0 */
michael@0	51	UD7=0xEF, /* Define and select window 7 */
michael@0	52	UQU=0xF0, /* Quote a single Unicode character */
michael@0	53	UDX=0xF1, /* Define a Window as extended */
michael@0	54	Urs=0xF2 /* reserved */
michael@0	55	};
michael@0	56
michael@0	57	enum {
michael@0	58	/*
michael@0	59	* Unicode code points from 3400 to E000 are not adressible by
michael@0	60	* dynamic window, since in these areas no short run alphabets are
michael@0	61	* found. Therefore add gapOffset to all values from gapThreshold.
michael@0	62	*/
michael@0	63	gapThreshold=0x68,
michael@0	64	gapOffset=0xAC00,
michael@0	65
michael@0	66	/* values between reservedStart and fixedThreshold are reserved */
michael@0	67	reservedStart=0xA8,
michael@0	68
michael@0	69	/* use table of predefined fixed offsets for values from fixedThreshold */
michael@0	70	fixedThreshold=0xF9
michael@0	71	};
michael@0	72
michael@0	73	/* constant offsets for the 8 static windows */
michael@0	74	static const uint32_t staticOffsets[8]={
michael@0	75	0x0000, /* ASCII for quoted tags */
michael@0	76	0x0080, /* Latin - 1 Supplement (for access to punctuation) */
michael@0	77	0x0100, /* Latin Extended-A */
michael@0	78	0x0300, /* Combining Diacritical Marks */
michael@0	79	0x2000, /* General Punctuation */
michael@0	80	0x2080, /* Currency Symbols */
michael@0	81	0x2100, /* Letterlike Symbols and Number Forms */
michael@0	82	0x3000 /* CJK Symbols and punctuation */
michael@0	83	};
michael@0	84
michael@0	85	/* initial offsets for the 8 dynamic (sliding) windows */
michael@0	86	static const uint32_t initialDynamicOffsets[8]={
michael@0	87	0x0080, /* Latin-1 */
michael@0	88	0x00C0, /* Latin Extended A */
michael@0	89	0x0400, /* Cyrillic */
michael@0	90	0x0600, /* Arabic */
michael@0	91	0x0900, /* Devanagari */
michael@0	92	0x3040, /* Hiragana */
michael@0	93	0x30A0, /* Katakana */
michael@0	94	0xFF00 /* Fullwidth ASCII */
michael@0	95	};
michael@0	96
michael@0	97	/* Table of fixed predefined Offsets */
michael@0	98	static const uint32_t fixedOffsets[]={
michael@0	99	/* 0xF9 */ 0x00C0, /* Latin-1 Letters + half of Latin Extended A */
michael@0	100	/* 0xFA */ 0x0250, /* IPA extensions */
michael@0	101	/* 0xFB */ 0x0370, /* Greek */
michael@0	102	/* 0xFC */ 0x0530, /* Armenian */
michael@0	103	/* 0xFD */ 0x3040, /* Hiragana */
michael@0	104	/* 0xFE */ 0x30A0, /* Katakana */
michael@0	105	/* 0xFF */ 0xFF60 /* Halfwidth Katakana */
michael@0	106	};
michael@0	107
michael@0	108	/* state values */
michael@0	109	enum {
michael@0	110	readCommand,
michael@0	111	quotePairOne,
michael@0	112	quotePairTwo,
michael@0	113	quoteOne,
michael@0	114	definePairOne,
michael@0	115	definePairTwo,
michael@0	116	defineOne
michael@0	117	};
michael@0	118
michael@0	119	typedef struct SCSUData {
michael@0	120	/* dynamic window offsets, intitialize to default values from initialDynamicOffsets */
michael@0	121	uint32_t toUDynamicOffsets[8];
michael@0	122	uint32_t fromUDynamicOffsets[8];
michael@0	123
michael@0	124	/* state machine state - toUnicode */
michael@0	125	UBool toUIsSingleByteMode;
michael@0	126	uint8_t toUState;
michael@0	127	int8_t toUQuoteWindow, toUDynamicWindow;
michael@0	128	uint8_t toUByteOne;
michael@0	129	uint8_t toUPadding[3];
michael@0	130
michael@0	131	/* state machine state - fromUnicode */
michael@0	132	UBool fromUIsSingleByteMode;
michael@0	133	int8_t fromUDynamicWindow;
michael@0	134
michael@0	135	/*
michael@0	136	* windowUse[] keeps track of the use of the dynamic windows:
michael@0	137	* At nextWindowUseIndex there is the least recently used window,
michael@0	138	* and the following windows (in a wrapping manner) are more and more
michael@0	139	* recently used.
michael@0	140	* At nextWindowUseIndex-1 there is the most recently used window.
michael@0	141	*/
michael@0	142	uint8_t locale;
michael@0	143	int8_t nextWindowUseIndex;
michael@0	144	int8_t windowUse[8];
michael@0	145	} SCSUData;
michael@0	146
michael@0	147	static const int8_t initialWindowUse[8]={ 7, 0, 3, 2, 4, 5, 6, 1 };
michael@0	148	static const int8_t initialWindowUse_ja[8]={ 3, 2, 4, 1, 0, 7, 5, 6 };
michael@0	149
michael@0	150	enum {
michael@0	151	lGeneric, l_ja
michael@0	152	};
michael@0	153
michael@0	154	/* SCSU setup functions ----------------------------------------------------- */
michael@0	155
michael@0	156	static void
michael@0	157	_SCSUReset(UConverter *cnv, UConverterResetChoice choice) {
michael@0	158	SCSUData *scsu=(SCSUData *)cnv->extraInfo;
michael@0	159
michael@0	160	if(choice<=UCNV_RESET_TO_UNICODE) {
michael@0	161	/* reset toUnicode */
michael@0	162	uprv_memcpy(scsu->toUDynamicOffsets, initialDynamicOffsets, 32);
michael@0	163
michael@0	164	scsu->toUIsSingleByteMode=TRUE;
michael@0	165	scsu->toUState=readCommand;
michael@0	166	scsu->toUQuoteWindow=scsu->toUDynamicWindow=0;
michael@0	167	scsu->toUByteOne=0;
michael@0	168
michael@0	169	cnv->toULength=0;
michael@0	170	}
michael@0	171	if(choice!=UCNV_RESET_TO_UNICODE) {
michael@0	172	/* reset fromUnicode */
michael@0	173	uprv_memcpy(scsu->fromUDynamicOffsets, initialDynamicOffsets, 32);
michael@0	174
michael@0	175	scsu->fromUIsSingleByteMode=TRUE;
michael@0	176	scsu->fromUDynamicWindow=0;
michael@0	177
michael@0	178	scsu->nextWindowUseIndex=0;
michael@0	179	switch(scsu->locale) {
michael@0	180	case l_ja:
michael@0	181	uprv_memcpy(scsu->windowUse, initialWindowUse_ja, 8);
michael@0	182	break;
michael@0	183	default:
michael@0	184	uprv_memcpy(scsu->windowUse, initialWindowUse, 8);
michael@0	185	break;
michael@0	186	}
michael@0	187
michael@0	188	cnv->fromUChar32=0;
michael@0	189	}
michael@0	190	}
michael@0	191
michael@0	192	static void
michael@0	193	_SCSUOpen(UConverter *cnv,
michael@0	194	UConverterLoadArgs *pArgs,
michael@0	195	UErrorCode *pErrorCode) {
michael@0	196	const char *locale=pArgs->locale;
michael@0	197	if(pArgs->onlyTestIsLoadable) {
michael@0	198	return;
michael@0	199	}
michael@0	200	cnv->extraInfo=uprv_malloc(sizeof(SCSUData));
michael@0	201	if(cnv->extraInfo!=NULL) {
michael@0	202	if(locale!=NULL && locale[0]=='j' && locale[1]=='a' && (locale[2]==0 || locale[2]=='_')) {
michael@0	203	((SCSUData *)cnv->extraInfo)->locale=l_ja;
michael@0	204	} else {
michael@0	205	((SCSUData *)cnv->extraInfo)->locale=lGeneric;
michael@0	206	}
michael@0	207	_SCSUReset(cnv, UCNV_RESET_BOTH);
michael@0	208	} else {
michael@0	209	*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
michael@0	210	}
michael@0	211
michael@0	212	/* Set the substitution character U+fffd as a Unicode string. */
michael@0	213	cnv->subUChars[0]=0xfffd;
michael@0	214	cnv->subCharLen=-1;
michael@0	215	}
michael@0	216
michael@0	217	static void
michael@0	218	_SCSUClose(UConverter *cnv) {
michael@0	219	if(cnv->extraInfo!=NULL) {
michael@0	220	if(!cnv->isExtraLocal) {
michael@0	221	uprv_free(cnv->extraInfo);
michael@0	222	}
michael@0	223	cnv->extraInfo=NULL;
michael@0	224	}
michael@0	225	}
michael@0	226
michael@0	227	/* SCSU-to-Unicode conversion functions ------------------------------------- */
michael@0	228
michael@0	229	static void
michael@0	230	_SCSUToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs,
michael@0	231	UErrorCode *pErrorCode) {
michael@0	232	UConverter *cnv;
michael@0	233	SCSUData *scsu;
michael@0	234	const uint8_t *source, *sourceLimit;
michael@0	235	UChar *target;
michael@0	236	const UChar *targetLimit;
michael@0	237	int32_t *offsets;
michael@0	238	UBool isSingleByteMode;
michael@0	239	uint8_t state, byteOne;
michael@0	240	int8_t quoteWindow, dynamicWindow;
michael@0	241
michael@0	242	int32_t sourceIndex, nextSourceIndex;
michael@0	243
michael@0	244	uint8_t b;
michael@0	245
michael@0	246	/* set up the local pointers */
michael@0	247	cnv=pArgs->converter;
michael@0	248	scsu=(SCSUData *)cnv->extraInfo;
michael@0	249
michael@0	250	source=(const uint8_t *)pArgs->source;
michael@0	251	sourceLimit=(const uint8_t *)pArgs->sourceLimit;
michael@0	252	target=pArgs->target;
michael@0	253	targetLimit=pArgs->targetLimit;
michael@0	254	offsets=pArgs->offsets;
michael@0	255
michael@0	256	/* get the state machine state */
michael@0	257	isSingleByteMode=scsu->toUIsSingleByteMode;
michael@0	258	state=scsu->toUState;
michael@0	259	quoteWindow=scsu->toUQuoteWindow;
michael@0	260	dynamicWindow=scsu->toUDynamicWindow;
michael@0	261	byteOne=scsu->toUByteOne;
michael@0	262
michael@0	263	/* sourceIndex=-1 if the current character began in the previous buffer */
michael@0	264	sourceIndex=state==readCommand ? 0 : -1;
michael@0	265	nextSourceIndex=0;
michael@0	266
michael@0	267	/*
michael@0	268	* conversion "loop"
michael@0	269	*
michael@0	270	* For performance, this is not a normal C loop.
michael@0	271	* Instead, there are two code blocks for the two SCSU modes.
michael@0	272	* The function branches to either one, and a change of the mode is done with a goto to
michael@0	273	* the other branch.
michael@0	274	*
michael@0	275	* Each branch has two conventional loops:
michael@0	276	* - a fast-path loop for the most common codes in the mode
michael@0	277	* - a loop for all other codes in the mode
michael@0	278	* When the fast-path runs into a code that it cannot handle, its loop ends and it
michael@0	279	* runs into the following loop to handle the other codes.
michael@0	280	* The end of the input or output buffer is also handled by the slower loop.
michael@0	281	* The slow loop jumps (goto) to the fast-path loop again as soon as possible.
michael@0	282	*
michael@0	283	* The callback handling is done by returning with an error code.
michael@0	284	* The conversion framework actually calls the callback function.
michael@0	285	*/
michael@0	286	if(isSingleByteMode) {
michael@0	287	/* fast path for single-byte mode */
michael@0	288	if(state==readCommand) {
michael@0	289	fastSingle:
michael@0	290	while(source<sourceLimit && target<targetLimit && (b=*source)>=0x20) {
michael@0	291	++source;
michael@0	292	++nextSourceIndex;
michael@0	293	if(b<=0x7f) {
michael@0	294	/* write US-ASCII graphic character or DEL */
michael@0	295	*target++=(UChar)b;
michael@0	296	if(offsets!=NULL) {
michael@0	297	*offsets++=sourceIndex;
michael@0	298	}
michael@0	299	} else {
michael@0	300	/* write from dynamic window */
michael@0	301	uint32_t c=scsu->toUDynamicOffsets[dynamicWindow]+(b&0x7f);
michael@0	302	if(c<=0xffff) {
michael@0	303	*target++=(UChar)c;
michael@0	304	if(offsets!=NULL) {
michael@0	305	*offsets++=sourceIndex;
michael@0	306	}
michael@0	307	} else {
michael@0	308	/* output surrogate pair */
michael@0	309	*target++=(UChar)(0xd7c0+(c>>10));
michael@0	310	if(target<targetLimit) {
michael@0	311	*target++=(UChar)(0xdc00|(c&0x3ff));
michael@0	312	if(offsets!=NULL) {
michael@0	313	*offsets++=sourceIndex;
michael@0	314	*offsets++=sourceIndex;
michael@0	315	}
michael@0	316	} else {
michael@0	317	/* target overflow */
michael@0	318	if(offsets!=NULL) {
michael@0	319	*offsets++=sourceIndex;
michael@0	320	}
michael@0	321	cnv->UCharErrorBuffer[0]=(UChar)(0xdc00|(c&0x3ff));
michael@0	322	cnv->UCharErrorBufferLength=1;
michael@0	323	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	324	goto endloop;
michael@0	325	}
michael@0	326	}
michael@0	327	}
michael@0	328	sourceIndex=nextSourceIndex;
michael@0	329	}
michael@0	330	}
michael@0	331
michael@0	332	/* normal state machine for single-byte mode, minus handling for what fastSingle covers */
michael@0	333	singleByteMode:
michael@0	334	while(source<sourceLimit) {
michael@0	335	if(target>=targetLimit) {
michael@0	336	/* target is full */
michael@0	337	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	338	break;
michael@0	339	}
michael@0	340	b=*source++;
michael@0	341	++nextSourceIndex;
michael@0	342	switch(state) {
michael@0	343	case readCommand:
michael@0	344	/* redundant conditions are commented out */
michael@0	345	/* here: b<0x20 because otherwise we would be in fastSingle */
michael@0	346	if((1UL<<b)&0x2601 /* binary 0010 0110 0000 0001, check for b==0xd || b==0xa || b==9 || b==0 */) {
michael@0	347	/* CR/LF/TAB/NUL */
michael@0	348	*target++=(UChar)b;
michael@0	349	if(offsets!=NULL) {
michael@0	350	*offsets++=sourceIndex;
michael@0	351	}
michael@0	352	sourceIndex=nextSourceIndex;
michael@0	353	goto fastSingle;
michael@0	354	} else if(SC0<=b) {
michael@0	355	if(b<=SC7) {
michael@0	356	dynamicWindow=(int8_t)(b-SC0);
michael@0	357	sourceIndex=nextSourceIndex;
michael@0	358	goto fastSingle;
michael@0	359	} else /* if(SD0<=b && b<=SD7) */ {
michael@0	360	dynamicWindow=(int8_t)(b-SD0);
michael@0	361	state=defineOne;
michael@0	362	}
michael@0	363	} else if(/* SQ0<=b && */ b<=SQ7) {
michael@0	364	quoteWindow=(int8_t)(b-SQ0);
michael@0	365	state=quoteOne;
michael@0	366	} else if(b==SDX) {
michael@0	367	state=definePairOne;
michael@0	368	} else if(b==SQU) {
michael@0	369	state=quotePairOne;
michael@0	370	} else if(b==SCU) {
michael@0	371	sourceIndex=nextSourceIndex;
michael@0	372	isSingleByteMode=FALSE;
michael@0	373	goto fastUnicode;
michael@0	374	} else /* Srs */ {
michael@0	375	/* callback(illegal) */
michael@0	376	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	377	cnv->toUBytes[0]=b;
michael@0	378	cnv->toULength=1;
michael@0	379	goto endloop;
michael@0	380	}
michael@0	381
michael@0	382	/* store the first byte of a multibyte sequence in toUBytes[] */
michael@0	383	cnv->toUBytes[0]=b;
michael@0	384	cnv->toULength=1;
michael@0	385	break;
michael@0	386	case quotePairOne:
michael@0	387	byteOne=b;
michael@0	388	cnv->toUBytes[1]=b;
michael@0	389	cnv->toULength=2;
michael@0	390	state=quotePairTwo;
michael@0	391	break;
michael@0	392	case quotePairTwo:
michael@0	393	*target++=(UChar)((byteOne<<8)|b);
michael@0	394	if(offsets!=NULL) {
michael@0	395	*offsets++=sourceIndex;
michael@0	396	}
michael@0	397	sourceIndex=nextSourceIndex;
michael@0	398	state=readCommand;
michael@0	399	goto fastSingle;
michael@0	400	case quoteOne:
michael@0	401	if(b<0x80) {
michael@0	402	/* all static offsets are in the BMP */
michael@0	403	*target++=(UChar)(staticOffsets[quoteWindow]+b);
michael@0	404	if(offsets!=NULL) {
michael@0	405	*offsets++=sourceIndex;
michael@0	406	}
michael@0	407	} else {
michael@0	408	/* write from dynamic window */
michael@0	409	uint32_t c=scsu->toUDynamicOffsets[quoteWindow]+(b&0x7f);
michael@0	410	if(c<=0xffff) {
michael@0	411	*target++=(UChar)c;
michael@0	412	if(offsets!=NULL) {
michael@0	413	*offsets++=sourceIndex;
michael@0	414	}
michael@0	415	} else {
michael@0	416	/* output surrogate pair */
michael@0	417	*target++=(UChar)(0xd7c0+(c>>10));
michael@0	418	if(target<targetLimit) {
michael@0	419	*target++=(UChar)(0xdc00|(c&0x3ff));
michael@0	420	if(offsets!=NULL) {
michael@0	421	*offsets++=sourceIndex;
michael@0	422	*offsets++=sourceIndex;
michael@0	423	}
michael@0	424	} else {
michael@0	425	/* target overflow */
michael@0	426	if(offsets!=NULL) {
michael@0	427	*offsets++=sourceIndex;
michael@0	428	}
michael@0	429	cnv->UCharErrorBuffer[0]=(UChar)(0xdc00|(c&0x3ff));
michael@0	430	cnv->UCharErrorBufferLength=1;
michael@0	431	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	432	goto endloop;
michael@0	433	}
michael@0	434	}
michael@0	435	}
michael@0	436	sourceIndex=nextSourceIndex;
michael@0	437	state=readCommand;
michael@0	438	goto fastSingle;
michael@0	439	case definePairOne:
michael@0	440	dynamicWindow=(int8_t)((b>>5)&7);
michael@0	441	byteOne=(uint8_t)(b&0x1f);
michael@0	442	cnv->toUBytes[1]=b;
michael@0	443	cnv->toULength=2;
michael@0	444	state=definePairTwo;
michael@0	445	break;
michael@0	446	case definePairTwo:
michael@0	447	scsu->toUDynamicOffsets[dynamicWindow]=0x10000+(byteOne<<15UL | b<<7UL);
michael@0	448	sourceIndex=nextSourceIndex;
michael@0	449	state=readCommand;
michael@0	450	goto fastSingle;
michael@0	451	case defineOne:
michael@0	452	if(b==0) {
michael@0	453	/* callback(illegal): Reserved window offset value 0 */
michael@0	454	cnv->toUBytes[1]=b;
michael@0	455	cnv->toULength=2;
michael@0	456	goto endloop;
michael@0	457	} else if(b<gapThreshold) {
michael@0	458	scsu->toUDynamicOffsets[dynamicWindow]=b<<7UL;
michael@0	459	} else if((uint8_t)(b-gapThreshold)<(reservedStart-gapThreshold)) {
michael@0	460	scsu->toUDynamicOffsets[dynamicWindow]=(b<<7UL)+gapOffset;
michael@0	461	} else if(b>=fixedThreshold) {
michael@0	462	scsu->toUDynamicOffsets[dynamicWindow]=fixedOffsets[b-fixedThreshold];
michael@0	463	} else {
michael@0	464	/* callback(illegal): Reserved window offset value 0xa8..0xf8 */
michael@0	465	cnv->toUBytes[1]=b;
michael@0	466	cnv->toULength=2;
michael@0	467	goto endloop;
michael@0	468	}
michael@0	469	sourceIndex=nextSourceIndex;
michael@0	470	state=readCommand;
michael@0	471	goto fastSingle;
michael@0	472	}
michael@0	473	}
michael@0	474	} else {
michael@0	475	/* fast path for Unicode mode */
michael@0	476	if(state==readCommand) {
michael@0	477	fastUnicode:
michael@0	478	while(source+1<sourceLimit && target<targetLimit && (uint8_t)((b=*source)-UC0)>(Urs-UC0)) {
michael@0	479	*target++=(UChar)((b<<8)|source[1]);
michael@0	480	if(offsets!=NULL) {
michael@0	481	*offsets++=sourceIndex;
michael@0	482	}
michael@0	483	sourceIndex=nextSourceIndex;
michael@0	484	nextSourceIndex+=2;
michael@0	485	source+=2;
michael@0	486	}
michael@0	487	}
michael@0	488
michael@0	489	/* normal state machine for Unicode mode */
michael@0	490	/* unicodeByteMode: */
michael@0	491	while(source<sourceLimit) {
michael@0	492	if(target>=targetLimit) {
michael@0	493	/* target is full */
michael@0	494	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	495	break;
michael@0	496	}
michael@0	497	b=*source++;
michael@0	498	++nextSourceIndex;
michael@0	499	switch(state) {
michael@0	500	case readCommand:
michael@0	501	if((uint8_t)(b-UC0)>(Urs-UC0)) {
michael@0	502	byteOne=b;
michael@0	503	cnv->toUBytes[0]=b;
michael@0	504	cnv->toULength=1;
michael@0	505	state=quotePairTwo;
michael@0	506	} else if(/* UC0<=b && */ b<=UC7) {
michael@0	507	dynamicWindow=(int8_t)(b-UC0);
michael@0	508	sourceIndex=nextSourceIndex;
michael@0	509	isSingleByteMode=TRUE;
michael@0	510	goto fastSingle;
michael@0	511	} else if(/* UD0<=b && */ b<=UD7) {
michael@0	512	dynamicWindow=(int8_t)(b-UD0);
michael@0	513	isSingleByteMode=TRUE;
michael@0	514	cnv->toUBytes[0]=b;
michael@0	515	cnv->toULength=1;
michael@0	516	state=defineOne;
michael@0	517	goto singleByteMode;
michael@0	518	} else if(b==UDX) {
michael@0	519	isSingleByteMode=TRUE;
michael@0	520	cnv->toUBytes[0]=b;
michael@0	521	cnv->toULength=1;
michael@0	522	state=definePairOne;
michael@0	523	goto singleByteMode;
michael@0	524	} else if(b==UQU) {
michael@0	525	cnv->toUBytes[0]=b;
michael@0	526	cnv->toULength=1;
michael@0	527	state=quotePairOne;
michael@0	528	} else /* Urs */ {
michael@0	529	/* callback(illegal) */
michael@0	530	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	531	cnv->toUBytes[0]=b;
michael@0	532	cnv->toULength=1;
michael@0	533	goto endloop;
michael@0	534	}
michael@0	535	break;
michael@0	536	case quotePairOne:
michael@0	537	byteOne=b;
michael@0	538	cnv->toUBytes[1]=b;
michael@0	539	cnv->toULength=2;
michael@0	540	state=quotePairTwo;
michael@0	541	break;
michael@0	542	case quotePairTwo:
michael@0	543	*target++=(UChar)((byteOne<<8)|b);
michael@0	544	if(offsets!=NULL) {
michael@0	545	*offsets++=sourceIndex;
michael@0	546	}
michael@0	547	sourceIndex=nextSourceIndex;
michael@0	548	state=readCommand;
michael@0	549	goto fastUnicode;
michael@0	550	}
michael@0	551	}
michael@0	552	}
michael@0	553	endloop:
michael@0	554
michael@0	555	/* set the converter state back into UConverter */
michael@0	556	if(U_FAILURE(*pErrorCode) && *pErrorCode!=U_BUFFER_OVERFLOW_ERROR) {
michael@0	557	/* reset to deal with the next character */
michael@0	558	state=readCommand;
michael@0	559	} else if(state==readCommand) {
michael@0	560	/* not in a multi-byte sequence, reset toULength */
michael@0	561	cnv->toULength=0;
michael@0	562	}
michael@0	563	scsu->toUIsSingleByteMode=isSingleByteMode;
michael@0	564	scsu->toUState=state;
michael@0	565	scsu->toUQuoteWindow=quoteWindow;
michael@0	566	scsu->toUDynamicWindow=dynamicWindow;
michael@0	567	scsu->toUByteOne=byteOne;
michael@0	568
michael@0	569	/* write back the updated pointers */
michael@0	570	pArgs->source=(const char *)source;
michael@0	571	pArgs->target=target;
michael@0	572	pArgs->offsets=offsets;
michael@0	573	return;
michael@0	574	}
michael@0	575
michael@0	576	/*
michael@0	577	* Identical to _SCSUToUnicodeWithOffsets but without offset handling.
michael@0	578	* If a change is made in the original function, then either
michael@0	579	* change this function the same way or
michael@0	580	* re-copy the original function and remove the variables
michael@0	581	* offsets, sourceIndex, and nextSourceIndex.
michael@0	582	*/
michael@0	583	static void
michael@0	584	_SCSUToUnicode(UConverterToUnicodeArgs *pArgs,
michael@0	585	UErrorCode *pErrorCode) {
michael@0	586	UConverter *cnv;
michael@0	587	SCSUData *scsu;
michael@0	588	const uint8_t *source, *sourceLimit;
michael@0	589	UChar *target;
michael@0	590	const UChar *targetLimit;
michael@0	591	UBool isSingleByteMode;
michael@0	592	uint8_t state, byteOne;
michael@0	593	int8_t quoteWindow, dynamicWindow;
michael@0	594
michael@0	595	uint8_t b;
michael@0	596
michael@0	597	/* set up the local pointers */
michael@0	598	cnv=pArgs->converter;
michael@0	599	scsu=(SCSUData *)cnv->extraInfo;
michael@0	600
michael@0	601	source=(const uint8_t *)pArgs->source;
michael@0	602	sourceLimit=(const uint8_t *)pArgs->sourceLimit;
michael@0	603	target=pArgs->target;
michael@0	604	targetLimit=pArgs->targetLimit;
michael@0	605
michael@0	606	/* get the state machine state */
michael@0	607	isSingleByteMode=scsu->toUIsSingleByteMode;
michael@0	608	state=scsu->toUState;
michael@0	609	quoteWindow=scsu->toUQuoteWindow;
michael@0	610	dynamicWindow=scsu->toUDynamicWindow;
michael@0	611	byteOne=scsu->toUByteOne;
michael@0	612
michael@0	613	/*
michael@0	614	* conversion "loop"
michael@0	615	*
michael@0	616	* For performance, this is not a normal C loop.
michael@0	617	* Instead, there are two code blocks for the two SCSU modes.
michael@0	618	* The function branches to either one, and a change of the mode is done with a goto to
michael@0	619	* the other branch.
michael@0	620	*
michael@0	621	* Each branch has two conventional loops:
michael@0	622	* - a fast-path loop for the most common codes in the mode
michael@0	623	* - a loop for all other codes in the mode
michael@0	624	* When the fast-path runs into a code that it cannot handle, its loop ends and it
michael@0	625	* runs into the following loop to handle the other codes.
michael@0	626	* The end of the input or output buffer is also handled by the slower loop.
michael@0	627	* The slow loop jumps (goto) to the fast-path loop again as soon as possible.
michael@0	628	*
michael@0	629	* The callback handling is done by returning with an error code.
michael@0	630	* The conversion framework actually calls the callback function.
michael@0	631	*/
michael@0	632	if(isSingleByteMode) {
michael@0	633	/* fast path for single-byte mode */
michael@0	634	if(state==readCommand) {
michael@0	635	fastSingle:
michael@0	636	while(source<sourceLimit && target<targetLimit && (b=*source)>=0x20) {
michael@0	637	++source;
michael@0	638	if(b<=0x7f) {
michael@0	639	/* write US-ASCII graphic character or DEL */
michael@0	640	*target++=(UChar)b;
michael@0	641	} else {
michael@0	642	/* write from dynamic window */
michael@0	643	uint32_t c=scsu->toUDynamicOffsets[dynamicWindow]+(b&0x7f);
michael@0	644	if(c<=0xffff) {
michael@0	645	*target++=(UChar)c;
michael@0	646	} else {
michael@0	647	/* output surrogate pair */
michael@0	648	*target++=(UChar)(0xd7c0+(c>>10));
michael@0	649	if(target<targetLimit) {
michael@0	650	*target++=(UChar)(0xdc00|(c&0x3ff));
michael@0	651	} else {
michael@0	652	/* target overflow */
michael@0	653	cnv->UCharErrorBuffer[0]=(UChar)(0xdc00|(c&0x3ff));
michael@0	654	cnv->UCharErrorBufferLength=1;
michael@0	655	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	656	goto endloop;
michael@0	657	}
michael@0	658	}
michael@0	659	}
michael@0	660	}
michael@0	661	}
michael@0	662
michael@0	663	/* normal state machine for single-byte mode, minus handling for what fastSingle covers */
michael@0	664	singleByteMode:
michael@0	665	while(source<sourceLimit) {
michael@0	666	if(target>=targetLimit) {
michael@0	667	/* target is full */
michael@0	668	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	669	break;
michael@0	670	}
michael@0	671	b=*source++;
michael@0	672	switch(state) {
michael@0	673	case readCommand:
michael@0	674	/* redundant conditions are commented out */
michael@0	675	/* here: b<0x20 because otherwise we would be in fastSingle */
michael@0	676	if((1UL<<b)&0x2601 /* binary 0010 0110 0000 0001, check for b==0xd || b==0xa || b==9 || b==0 */) {
michael@0	677	/* CR/LF/TAB/NUL */
michael@0	678	*target++=(UChar)b;
michael@0	679	goto fastSingle;
michael@0	680	} else if(SC0<=b) {
michael@0	681	if(b<=SC7) {
michael@0	682	dynamicWindow=(int8_t)(b-SC0);
michael@0	683	goto fastSingle;
michael@0	684	} else /* if(SD0<=b && b<=SD7) */ {
michael@0	685	dynamicWindow=(int8_t)(b-SD0);
michael@0	686	state=defineOne;
michael@0	687	}
michael@0	688	} else if(/* SQ0<=b && */ b<=SQ7) {
michael@0	689	quoteWindow=(int8_t)(b-SQ0);
michael@0	690	state=quoteOne;
michael@0	691	} else if(b==SDX) {
michael@0	692	state=definePairOne;
michael@0	693	} else if(b==SQU) {
michael@0	694	state=quotePairOne;
michael@0	695	} else if(b==SCU) {
michael@0	696	isSingleByteMode=FALSE;
michael@0	697	goto fastUnicode;
michael@0	698	} else /* Srs */ {
michael@0	699	/* callback(illegal) */
michael@0	700	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	701	cnv->toUBytes[0]=b;
michael@0	702	cnv->toULength=1;
michael@0	703	goto endloop;
michael@0	704	}
michael@0	705
michael@0	706	/* store the first byte of a multibyte sequence in toUBytes[] */
michael@0	707	cnv->toUBytes[0]=b;
michael@0	708	cnv->toULength=1;
michael@0	709	break;
michael@0	710	case quotePairOne:
michael@0	711	byteOne=b;
michael@0	712	cnv->toUBytes[1]=b;
michael@0	713	cnv->toULength=2;
michael@0	714	state=quotePairTwo;
michael@0	715	break;
michael@0	716	case quotePairTwo:
michael@0	717	*target++=(UChar)((byteOne<<8)|b);
michael@0	718	state=readCommand;
michael@0	719	goto fastSingle;
michael@0	720	case quoteOne:
michael@0	721	if(b<0x80) {
michael@0	722	/* all static offsets are in the BMP */
michael@0	723	*target++=(UChar)(staticOffsets[quoteWindow]+b);
michael@0	724	} else {
michael@0	725	/* write from dynamic window */
michael@0	726	uint32_t c=scsu->toUDynamicOffsets[quoteWindow]+(b&0x7f);
michael@0	727	if(c<=0xffff) {
michael@0	728	*target++=(UChar)c;
michael@0	729	} else {
michael@0	730	/* output surrogate pair */
michael@0	731	*target++=(UChar)(0xd7c0+(c>>10));
michael@0	732	if(target<targetLimit) {
michael@0	733	*target++=(UChar)(0xdc00|(c&0x3ff));
michael@0	734	} else {
michael@0	735	/* target overflow */
michael@0	736	cnv->UCharErrorBuffer[0]=(UChar)(0xdc00|(c&0x3ff));
michael@0	737	cnv->UCharErrorBufferLength=1;
michael@0	738	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	739	goto endloop;
michael@0	740	}
michael@0	741	}
michael@0	742	}
michael@0	743	state=readCommand;
michael@0	744	goto fastSingle;
michael@0	745	case definePairOne:
michael@0	746	dynamicWindow=(int8_t)((b>>5)&7);
michael@0	747	byteOne=(uint8_t)(b&0x1f);
michael@0	748	cnv->toUBytes[1]=b;
michael@0	749	cnv->toULength=2;
michael@0	750	state=definePairTwo;
michael@0	751	break;
michael@0	752	case definePairTwo:
michael@0	753	scsu->toUDynamicOffsets[dynamicWindow]=0x10000+(byteOne<<15UL | b<<7UL);
michael@0	754	state=readCommand;
michael@0	755	goto fastSingle;
michael@0	756	case defineOne:
michael@0	757	if(b==0) {
michael@0	758	/* callback(illegal): Reserved window offset value 0 */
michael@0	759	cnv->toUBytes[1]=b;
michael@0	760	cnv->toULength=2;
michael@0	761	goto endloop;
michael@0	762	} else if(b<gapThreshold) {
michael@0	763	scsu->toUDynamicOffsets[dynamicWindow]=b<<7UL;
michael@0	764	} else if((uint8_t)(b-gapThreshold)<(reservedStart-gapThreshold)) {
michael@0	765	scsu->toUDynamicOffsets[dynamicWindow]=(b<<7UL)+gapOffset;
michael@0	766	} else if(b>=fixedThreshold) {
michael@0	767	scsu->toUDynamicOffsets[dynamicWindow]=fixedOffsets[b-fixedThreshold];
michael@0	768	} else {
michael@0	769	/* callback(illegal): Reserved window offset value 0xa8..0xf8 */
michael@0	770	cnv->toUBytes[1]=b;
michael@0	771	cnv->toULength=2;
michael@0	772	goto endloop;
michael@0	773	}
michael@0	774	state=readCommand;
michael@0	775	goto fastSingle;
michael@0	776	}
michael@0	777	}
michael@0	778	} else {
michael@0	779	/* fast path for Unicode mode */
michael@0	780	if(state==readCommand) {
michael@0	781	fastUnicode:
michael@0	782	while(source+1<sourceLimit && target<targetLimit && (uint8_t)((b=*source)-UC0)>(Urs-UC0)) {
michael@0	783	*target++=(UChar)((b<<8)|source[1]);
michael@0	784	source+=2;
michael@0	785	}
michael@0	786	}
michael@0	787
michael@0	788	/* normal state machine for Unicode mode */
michael@0	789	/* unicodeByteMode: */
michael@0	790	while(source<sourceLimit) {
michael@0	791	if(target>=targetLimit) {
michael@0	792	/* target is full */
michael@0	793	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	794	break;
michael@0	795	}
michael@0	796	b=*source++;
michael@0	797	switch(state) {
michael@0	798	case readCommand:
michael@0	799	if((uint8_t)(b-UC0)>(Urs-UC0)) {
michael@0	800	byteOne=b;
michael@0	801	cnv->toUBytes[0]=b;
michael@0	802	cnv->toULength=1;
michael@0	803	state=quotePairTwo;
michael@0	804	} else if(/* UC0<=b && */ b<=UC7) {
michael@0	805	dynamicWindow=(int8_t)(b-UC0);
michael@0	806	isSingleByteMode=TRUE;
michael@0	807	goto fastSingle;
michael@0	808	} else if(/* UD0<=b && */ b<=UD7) {
michael@0	809	dynamicWindow=(int8_t)(b-UD0);
michael@0	810	isSingleByteMode=TRUE;
michael@0	811	cnv->toUBytes[0]=b;
michael@0	812	cnv->toULength=1;
michael@0	813	state=defineOne;
michael@0	814	goto singleByteMode;
michael@0	815	} else if(b==UDX) {
michael@0	816	isSingleByteMode=TRUE;
michael@0	817	cnv->toUBytes[0]=b;
michael@0	818	cnv->toULength=1;
michael@0	819	state=definePairOne;
michael@0	820	goto singleByteMode;
michael@0	821	} else if(b==UQU) {
michael@0	822	cnv->toUBytes[0]=b;
michael@0	823	cnv->toULength=1;
michael@0	824	state=quotePairOne;
michael@0	825	} else /* Urs */ {
michael@0	826	/* callback(illegal) */
michael@0	827	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	828	cnv->toUBytes[0]=b;
michael@0	829	cnv->toULength=1;
michael@0	830	goto endloop;
michael@0	831	}
michael@0	832	break;
michael@0	833	case quotePairOne:
michael@0	834	byteOne=b;
michael@0	835	cnv->toUBytes[1]=b;
michael@0	836	cnv->toULength=2;
michael@0	837	state=quotePairTwo;
michael@0	838	break;
michael@0	839	case quotePairTwo:
michael@0	840	*target++=(UChar)((byteOne<<8)|b);
michael@0	841	state=readCommand;
michael@0	842	goto fastUnicode;
michael@0	843	}
michael@0	844	}
michael@0	845	}
michael@0	846	endloop:
michael@0	847
michael@0	848	/* set the converter state back into UConverter */
michael@0	849	if(U_FAILURE(*pErrorCode) && *pErrorCode!=U_BUFFER_OVERFLOW_ERROR) {
michael@0	850	/* reset to deal with the next character */
michael@0	851	state=readCommand;
michael@0	852	} else if(state==readCommand) {
michael@0	853	/* not in a multi-byte sequence, reset toULength */
michael@0	854	cnv->toULength=0;
michael@0	855	}
michael@0	856	scsu->toUIsSingleByteMode=isSingleByteMode;
michael@0	857	scsu->toUState=state;
michael@0	858	scsu->toUQuoteWindow=quoteWindow;
michael@0	859	scsu->toUDynamicWindow=dynamicWindow;
michael@0	860	scsu->toUByteOne=byteOne;
michael@0	861
michael@0	862	/* write back the updated pointers */
michael@0	863	pArgs->source=(const char *)source;
michael@0	864	pArgs->target=target;
michael@0	865	return;
michael@0	866	}
michael@0	867
michael@0	868	/* SCSU-from-Unicode conversion functions ----------------------------------- */
michael@0	869
michael@0	870	/*
michael@0	871	* This SCSU Encoder is fairly simple but uses all SCSU commands to achieve
michael@0	872	* reasonable results. The lookahead is minimal.
michael@0	873	* Many cases are simple:
michael@0	874	* A character fits directly into the current mode, a dynamic or static window,
michael@0	875	* or is not compressible. These cases are tested first.
michael@0	876	* Real compression heuristics are applied to the rest, in code branches for
michael@0	877	* single/Unicode mode and BMP/supplementary code points.
michael@0	878	* The heuristics used here are extremely simple.
michael@0	879	*/
michael@0	880
michael@0	881	/* get the number of the window that this character is in, or -1 */
michael@0	882	static int8_t
michael@0	883	getWindow(const uint32_t offsets[8], uint32_t c) {
michael@0	884	int i;
michael@0	885	for(i=0; i<8; ++i) {
michael@0	886	if((uint32_t)(c-offsets[i])<=0x7f) {
michael@0	887	return (int8_t)(i);
michael@0	888	}
michael@0	889	}
michael@0	890	return -1;
michael@0	891	}
michael@0	892
michael@0	893	/* is the character in the dynamic window starting at the offset, or in the direct-encoded range? */
michael@0	894	static UBool
michael@0	895	isInOffsetWindowOrDirect(uint32_t offset, uint32_t c) {
michael@0	896	return (UBool)(c<=offset+0x7f &&
michael@0	897	(c>=offset || (c<=0x7f &&
michael@0	898	(c>=0x20 || (1UL<<c)&0x2601))));
michael@0	899	/* binary 0010 0110 0000 0001,
michael@0	900	check for b==0xd || b==0xa || b==9 || b==0 */
michael@0	901	}
michael@0	902
michael@0	903	/*
michael@0	904	* getNextDynamicWindow returns the next dynamic window to be redefined
michael@0	905	*/
michael@0	906	static int8_t
michael@0	907	getNextDynamicWindow(SCSUData *scsu) {
michael@0	908	int8_t window=scsu->windowUse[scsu->nextWindowUseIndex];
michael@0	909	if(++scsu->nextWindowUseIndex==8) {
michael@0	910	scsu->nextWindowUseIndex=0;
michael@0	911	}
michael@0	912	return window;
michael@0	913	}
michael@0	914
michael@0	915	/*
michael@0	916	* useDynamicWindow() adjusts
michael@0	917	* windowUse[] and nextWindowUseIndex for the algorithm to choose
michael@0	918	* the next dynamic window to be defined;
michael@0	919	* a subclass may override it and provide its own algorithm.
michael@0	920	*/
michael@0	921	static void
michael@0	922	useDynamicWindow(SCSUData *scsu, int8_t window) {
michael@0	923	/*
michael@0	924	* move the existing window, which just became the most recently used one,
michael@0	925	* up in windowUse[] to nextWindowUseIndex-1
michael@0	926	*/
michael@0	927
michael@0	928	/* first, find the index of the window - backwards to favor the more recently used windows */
michael@0	929	int i, j;
michael@0	930
michael@0	931	i=scsu->nextWindowUseIndex;
michael@0	932	do {
michael@0	933	if(--i<0) {
michael@0	934	i=7;
michael@0	935	}
michael@0	936	} while(scsu->windowUse[i]!=window);
michael@0	937
michael@0	938	/* now copy each windowUse[i+1] to [i] */
michael@0	939	j=i+1;
michael@0	940	if(j==8) {
michael@0	941	j=0;
michael@0	942	}
michael@0	943	while(j!=scsu->nextWindowUseIndex) {
michael@0	944	scsu->windowUse[i]=scsu->windowUse[j];
michael@0	945	i=j;
michael@0	946	if(++j==8) { j=0; }
michael@0	947	}
michael@0	948
michael@0	949	/* finally, set the window into the most recently used index */
michael@0	950	scsu->windowUse[i]=window;
michael@0	951	}
michael@0	952
michael@0	953	/*
michael@0	954	* calculate the offset and the code for a dynamic window that contains the character
michael@0	955	* takes fixed offsets into account
michael@0	956	* the offset of the window is stored in the offset variable,
michael@0	957	* the code is returned
michael@0	958	*
michael@0	959	* return offset code: -1 none <=0xff code for SDn/UDn else code for SDX/UDX, subtract 0x200 to get the true code
michael@0	960	*/
michael@0	961	static int
michael@0	962	getDynamicOffset(uint32_t c, uint32_t *pOffset) {
michael@0	963	int i;
michael@0	964
michael@0	965	for(i=0; i<7; ++i) {
michael@0	966	if((uint32_t)(c-fixedOffsets[i])<=0x7f) {
michael@0	967	*pOffset=fixedOffsets[i];
michael@0	968	return 0xf9+i;
michael@0	969	}
michael@0	970	}
michael@0	971
michael@0	972	if(c<0x80) {
michael@0	973	/* No dynamic window for US-ASCII. */
michael@0	974	return -1;
michael@0	975	} else if(c<0x3400 ||
michael@0	976	(uint32_t)(c-0x10000)<(0x14000-0x10000) ||
michael@0	977	(uint32_t)(c-0x1d000)<=(0x1ffff-0x1d000)
michael@0	978	) {
michael@0	979	/* This character is in a code range for a "small", i.e., reasonably windowable, script. */
michael@0	980	*pOffset=c&0x7fffff80;
michael@0	981	return (int)(c>>7);
michael@0	982	} else if(0xe000<=c && c!=0xfeff && c<0xfff0) {
michael@0	983	/* For these characters we need to take the gapOffset into account. */
michael@0	984	*pOffset=c&0x7fffff80;
michael@0	985	return (int)((c-gapOffset)>>7);
michael@0	986	} else {
michael@0	987	return -1;
michael@0	988	}
michael@0	989	}
michael@0	990
michael@0	991	/*
michael@0	992	* Idea for compression:
michael@0	993	* - save SCSUData and other state before really starting work
michael@0	994	* - at endloop, see if compression could be better with just unicode mode
michael@0	995	* - don't do this if a callback has been called
michael@0	996	* - if unicode mode would be smaller, then override the results with it - may need SCU at the beginning
michael@0	997	* - different buffer handling!
michael@0	998	*
michael@0	999	* Drawback or need for corrective handling:
michael@0	1000	* it is desirable to encode U+feff as SQU fe ff for the SCSU signature, and
michael@0	1001	* it is desirable to start a document in US-ASCII/Latin-1 for as long as possible
michael@0	1002	* not only for compression but also for HTML/XML documents with following charset/encoding announcers.
michael@0	1003	*
michael@0	1004	* How to achieve both?
michael@0	1005	* - Only replace the result after an SDX or SCU?
michael@0	1006	*/
michael@0	1007
michael@0	1008	static void
michael@0	1009	_SCSUFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs,
michael@0	1010	UErrorCode *pErrorCode) {
michael@0	1011	UConverter *cnv;
michael@0	1012	SCSUData *scsu;
michael@0	1013	const UChar *source, *sourceLimit;
michael@0	1014	uint8_t *target;
michael@0	1015	int32_t targetCapacity;
michael@0	1016	int32_t *offsets;
michael@0	1017
michael@0	1018	UBool isSingleByteMode;
michael@0	1019	uint8_t dynamicWindow;
michael@0	1020	uint32_t currentOffset;
michael@0	1021
michael@0	1022	uint32_t c, delta;
michael@0	1023
michael@0	1024	int32_t sourceIndex, nextSourceIndex;
michael@0	1025
michael@0	1026	int32_t length;
michael@0	1027
michael@0	1028	/* variables for compression heuristics */
michael@0	1029	uint32_t offset;
michael@0	1030	UChar lead, trail;
michael@0	1031	int code;
michael@0	1032	int8_t window;
michael@0	1033
michael@0	1034	/* set up the local pointers */
michael@0	1035	cnv=pArgs->converter;
michael@0	1036	scsu=(SCSUData *)cnv->extraInfo;
michael@0	1037
michael@0	1038	/* set up the local pointers */
michael@0	1039	source=pArgs->source;
michael@0	1040	sourceLimit=pArgs->sourceLimit;
michael@0	1041	target=(uint8_t *)pArgs->target;
michael@0	1042	targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target);
michael@0	1043	offsets=pArgs->offsets;
michael@0	1044
michael@0	1045	/* get the state machine state */
michael@0	1046	isSingleByteMode=scsu->fromUIsSingleByteMode;
michael@0	1047	dynamicWindow=scsu->fromUDynamicWindow;
michael@0	1048	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1049
michael@0	1050	c=cnv->fromUChar32;
michael@0	1051
michael@0	1052	/* sourceIndex=-1 if the current character began in the previous buffer */
michael@0	1053	sourceIndex= c==0 ? 0 : -1;
michael@0	1054	nextSourceIndex=0;
michael@0	1055
michael@0	1056	/* similar conversion "loop" as in toUnicode */
michael@0	1057	loop:
michael@0	1058	if(isSingleByteMode) {
michael@0	1059	if(c!=0 && targetCapacity>0) {
michael@0	1060	goto getTrailSingle;
michael@0	1061	}
michael@0	1062
michael@0	1063	/* state machine for single-byte mode */
michael@0	1064	/* singleByteMode: */
michael@0	1065	while(source<sourceLimit) {
michael@0	1066	if(targetCapacity<=0) {
michael@0	1067	/* target is full */
michael@0	1068	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	1069	break;
michael@0	1070	}
michael@0	1071	c=*source++;
michael@0	1072	++nextSourceIndex;
michael@0	1073
michael@0	1074	if((c-0x20)<=0x5f) {
michael@0	1075	/* pass US-ASCII graphic character through */
michael@0	1076	*target++=(uint8_t)c;
michael@0	1077	if(offsets!=NULL) {
michael@0	1078	*offsets++=sourceIndex;
michael@0	1079	}
michael@0	1080	--targetCapacity;
michael@0	1081	} else if(c<0x20) {
michael@0	1082	if((1UL<<c)&0x2601 /* binary 0010 0110 0000 0001, check for b==0xd || b==0xa || b==9 || b==0 */) {
michael@0	1083	/* CR/LF/TAB/NUL */
michael@0	1084	*target++=(uint8_t)c;
michael@0	1085	if(offsets!=NULL) {
michael@0	1086	*offsets++=sourceIndex;
michael@0	1087	}
michael@0	1088	--targetCapacity;
michael@0	1089	} else {
michael@0	1090	/* quote C0 control character */
michael@0	1091	c|=SQ0<<8;
michael@0	1092	length=2;
michael@0	1093	goto outputBytes;
michael@0	1094	}
michael@0	1095	} else if((delta=c-currentOffset)<=0x7f) {
michael@0	1096	/* use the current dynamic window */
michael@0	1097	*target++=(uint8_t)(delta|0x80);
michael@0	1098	if(offsets!=NULL) {
michael@0	1099	*offsets++=sourceIndex;
michael@0	1100	}
michael@0	1101	--targetCapacity;
michael@0	1102	} else if(U16_IS_SURROGATE(c)) {
michael@0	1103	if(U16_IS_SURROGATE_LEAD(c)) {
michael@0	1104	getTrailSingle:
michael@0	1105	lead=(UChar)c;
michael@0	1106	if(source<sourceLimit) {
michael@0	1107	/* test the following code unit */
michael@0	1108	trail=*source;
michael@0	1109	if(U16_IS_TRAIL(trail)) {
michael@0	1110	++source;
michael@0	1111	++nextSourceIndex;
michael@0	1112	c=U16_GET_SUPPLEMENTARY(c, trail);
michael@0	1113	/* convert this surrogate code point */
michael@0	1114	/* exit this condition tree */
michael@0	1115	} else {
michael@0	1116	/* this is an unmatched lead code unit (1st surrogate) */
michael@0	1117	/* callback(illegal) */
michael@0	1118	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	1119	goto endloop;
michael@0	1120	}
michael@0	1121	} else {
michael@0	1122	/* no more input */
michael@0	1123	break;
michael@0	1124	}
michael@0	1125	} else {
michael@0	1126	/* this is an unmatched trail code unit (2nd surrogate) */
michael@0	1127	/* callback(illegal) */
michael@0	1128	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	1129	goto endloop;
michael@0	1130	}
michael@0	1131
michael@0	1132	/* compress supplementary character U+10000..U+10ffff */
michael@0	1133	if((delta=c-currentOffset)<=0x7f) {
michael@0	1134	/* use the current dynamic window */
michael@0	1135	*target++=(uint8_t)(delta|0x80);
michael@0	1136	if(offsets!=NULL) {
michael@0	1137	*offsets++=sourceIndex;
michael@0	1138	}
michael@0	1139	--targetCapacity;
michael@0	1140	} else if((window=getWindow(scsu->fromUDynamicOffsets, c))>=0) {
michael@0	1141	/* there is a dynamic window that contains this character, change to it */
michael@0	1142	dynamicWindow=window;
michael@0	1143	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1144	useDynamicWindow(scsu, dynamicWindow);
michael@0	1145	c=((uint32_t)(SC0+dynamicWindow)<<8)|(c-currentOffset)|0x80;
michael@0	1146	length=2;
michael@0	1147	goto outputBytes;
michael@0	1148	} else if((code=getDynamicOffset(c, &offset))>=0) {
michael@0	1149	/* might check if there are more characters in this window to come */
michael@0	1150	/* define an extended window with this character */
michael@0	1151	code-=0x200;
michael@0	1152	dynamicWindow=getNextDynamicWindow(scsu);
michael@0	1153	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset;
michael@0	1154	useDynamicWindow(scsu, dynamicWindow);
michael@0	1155	c=((uint32_t)SDX<<24)|((uint32_t)dynamicWindow<<21)|((uint32_t)code<<8)|(c-currentOffset)|0x80;
michael@0	1156	length=4;
michael@0	1157	goto outputBytes;
michael@0	1158	} else {
michael@0	1159	/* change to Unicode mode and output this (lead, trail) pair */
michael@0	1160	isSingleByteMode=FALSE;
michael@0	1161	*target++=(uint8_t)SCU;
michael@0	1162	if(offsets!=NULL) {
michael@0	1163	*offsets++=sourceIndex;
michael@0	1164	}
michael@0	1165	--targetCapacity;
michael@0	1166	c=((uint32_t)lead<<16)|trail;
michael@0	1167	length=4;
michael@0	1168	goto outputBytes;
michael@0	1169	}
michael@0	1170	} else if(c<0xa0) {
michael@0	1171	/* quote C1 control character */
michael@0	1172	c=(c&0x7f)|(SQ0+1)<<8; /* SQ0+1==SQ1 */
michael@0	1173	length=2;
michael@0	1174	goto outputBytes;
michael@0	1175	} else if(c==0xfeff || c>=0xfff0) {
michael@0	1176	/* quote signature character=byte order mark and specials */
michael@0	1177	c|=SQU<<16;
michael@0	1178	length=3;
michael@0	1179	goto outputBytes;
michael@0	1180	} else {
michael@0	1181	/* compress all other BMP characters */
michael@0	1182	if((window=getWindow(scsu->fromUDynamicOffsets, c))>=0) {
michael@0	1183	/* there is a window defined that contains this character - switch to it or quote from it? */
michael@0	1184	if(source>=sourceLimit || isInOffsetWindowOrDirect(scsu->fromUDynamicOffsets[window], *source)) {
michael@0	1185	/* change to dynamic window */
michael@0	1186	dynamicWindow=window;
michael@0	1187	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1188	useDynamicWindow(scsu, dynamicWindow);
michael@0	1189	c=((uint32_t)(SC0+dynamicWindow)<<8)|(c-currentOffset)|0x80;
michael@0	1190	length=2;
michael@0	1191	goto outputBytes;
michael@0	1192	} else {
michael@0	1193	/* quote from dynamic window */
michael@0	1194	c=((uint32_t)(SQ0+window)<<8)|(c-scsu->fromUDynamicOffsets[window])|0x80;
michael@0	1195	length=2;
michael@0	1196	goto outputBytes;
michael@0	1197	}
michael@0	1198	} else if((window=getWindow(staticOffsets, c))>=0) {
michael@0	1199	/* quote from static window */
michael@0	1200	c=((uint32_t)(SQ0+window)<<8)|(c-staticOffsets[window]);
michael@0	1201	length=2;
michael@0	1202	goto outputBytes;
michael@0	1203	} else if((code=getDynamicOffset(c, &offset))>=0) {
michael@0	1204	/* define a dynamic window with this character */
michael@0	1205	dynamicWindow=getNextDynamicWindow(scsu);
michael@0	1206	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset;
michael@0	1207	useDynamicWindow(scsu, dynamicWindow);
michael@0	1208	c=((uint32_t)(SD0+dynamicWindow)<<16)|((uint32_t)code<<8)|(c-currentOffset)|0x80;
michael@0	1209	length=3;
michael@0	1210	goto outputBytes;
michael@0	1211	} else if((uint32_t)(c-0x3400)<(0xd800-0x3400) &&
michael@0	1212	(source>=sourceLimit || (uint32_t)(*source-0x3400)<(0xd800-0x3400))
michael@0	1213	) {
michael@0	1214	/*
michael@0	1215	* this character is not compressible (a BMP ideograph or similar);
michael@0	1216	* switch to Unicode mode if this is the last character in the block
michael@0	1217	* or there is at least one more ideograph following immediately
michael@0	1218	*/
michael@0	1219	isSingleByteMode=FALSE;
michael@0	1220	c|=SCU<<16;
michael@0	1221	length=3;
michael@0	1222	goto outputBytes;
michael@0	1223	} else {
michael@0	1224	/* quote Unicode */
michael@0	1225	c|=SQU<<16;
michael@0	1226	length=3;
michael@0	1227	goto outputBytes;
michael@0	1228	}
michael@0	1229	}
michael@0	1230
michael@0	1231	/* normal end of conversion: prepare for a new character */
michael@0	1232	c=0;
michael@0	1233	sourceIndex=nextSourceIndex;
michael@0	1234	}
michael@0	1235	} else {
michael@0	1236	if(c!=0 && targetCapacity>0) {
michael@0	1237	goto getTrailUnicode;
michael@0	1238	}
michael@0	1239
michael@0	1240	/* state machine for Unicode mode */
michael@0	1241	/* unicodeByteMode: */
michael@0	1242	while(source<sourceLimit) {
michael@0	1243	if(targetCapacity<=0) {
michael@0	1244	/* target is full */
michael@0	1245	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	1246	break;
michael@0	1247	}
michael@0	1248	c=*source++;
michael@0	1249	++nextSourceIndex;
michael@0	1250
michael@0	1251	if((uint32_t)(c-0x3400)<(0xd800-0x3400)) {
michael@0	1252	/* not compressible, write character directly */
michael@0	1253	if(targetCapacity>=2) {
michael@0	1254	*target++=(uint8_t)(c>>8);
michael@0	1255	*target++=(uint8_t)c;
michael@0	1256	if(offsets!=NULL) {
michael@0	1257	*offsets++=sourceIndex;
michael@0	1258	*offsets++=sourceIndex;
michael@0	1259	}
michael@0	1260	targetCapacity-=2;
michael@0	1261	} else {
michael@0	1262	length=2;
michael@0	1263	goto outputBytes;
michael@0	1264	}
michael@0	1265	} else if((uint32_t)(c-0x3400)>=(0xf300-0x3400) /* c<0x3400 || c>=0xf300 */) {
michael@0	1266	/* compress BMP character if the following one is not an uncompressible ideograph */
michael@0	1267	if(!(source<sourceLimit && (uint32_t)(*source-0x3400)<(0xd800-0x3400))) {
michael@0	1268	if(((uint32_t)(c-0x30)<10 || (uint32_t)(c-0x61)<26 || (uint32_t)(c-0x41)<26)) {
michael@0	1269	/* ASCII digit or letter */
michael@0	1270	isSingleByteMode=TRUE;
michael@0	1271	c|=((uint32_t)(UC0+dynamicWindow)<<8)|c;
michael@0	1272	length=2;
michael@0	1273	goto outputBytes;
michael@0	1274	} else if((window=getWindow(scsu->fromUDynamicOffsets, c))>=0) {
michael@0	1275	/* there is a dynamic window that contains this character, change to it */
michael@0	1276	isSingleByteMode=TRUE;
michael@0	1277	dynamicWindow=window;
michael@0	1278	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1279	useDynamicWindow(scsu, dynamicWindow);
michael@0	1280	c=((uint32_t)(UC0+dynamicWindow)<<8)|(c-currentOffset)|0x80;
michael@0	1281	length=2;
michael@0	1282	goto outputBytes;
michael@0	1283	} else if((code=getDynamicOffset(c, &offset))>=0) {
michael@0	1284	/* define a dynamic window with this character */
michael@0	1285	isSingleByteMode=TRUE;
michael@0	1286	dynamicWindow=getNextDynamicWindow(scsu);
michael@0	1287	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset;
michael@0	1288	useDynamicWindow(scsu, dynamicWindow);
michael@0	1289	c=((uint32_t)(UD0+dynamicWindow)<<16)|((uint32_t)code<<8)|(c-currentOffset)|0x80;
michael@0	1290	length=3;
michael@0	1291	goto outputBytes;
michael@0	1292	}
michael@0	1293	}
michael@0	1294
michael@0	1295	/* don't know how to compress this character, just write it directly */
michael@0	1296	length=2;
michael@0	1297	goto outputBytes;
michael@0	1298	} else if(c<0xe000) {
michael@0	1299	/* c is a surrogate */
michael@0	1300	if(U16_IS_SURROGATE_LEAD(c)) {
michael@0	1301	getTrailUnicode:
michael@0	1302	lead=(UChar)c;
michael@0	1303	if(source<sourceLimit) {
michael@0	1304	/* test the following code unit */
michael@0	1305	trail=*source;
michael@0	1306	if(U16_IS_TRAIL(trail)) {
michael@0	1307	++source;
michael@0	1308	++nextSourceIndex;
michael@0	1309	c=U16_GET_SUPPLEMENTARY(c, trail);
michael@0	1310	/* convert this surrogate code point */
michael@0	1311	/* exit this condition tree */
michael@0	1312	} else {
michael@0	1313	/* this is an unmatched lead code unit (1st surrogate) */
michael@0	1314	/* callback(illegal) */
michael@0	1315	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	1316	goto endloop;
michael@0	1317	}
michael@0	1318	} else {
michael@0	1319	/* no more input */
michael@0	1320	break;
michael@0	1321	}
michael@0	1322	} else {
michael@0	1323	/* this is an unmatched trail code unit (2nd surrogate) */
michael@0	1324	/* callback(illegal) */
michael@0	1325	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	1326	goto endloop;
michael@0	1327	}
michael@0	1328
michael@0	1329	/* compress supplementary character */
michael@0	1330	if( (window=getWindow(scsu->fromUDynamicOffsets, c))>=0 &&
michael@0	1331	!(source<sourceLimit && (uint32_t)(*source-0x3400)<(0xd800-0x3400))
michael@0	1332	) {
michael@0	1333	/*
michael@0	1334	* there is a dynamic window that contains this character and
michael@0	1335	* the following character is not uncompressible,
michael@0	1336	* change to the window
michael@0	1337	*/
michael@0	1338	isSingleByteMode=TRUE;
michael@0	1339	dynamicWindow=window;
michael@0	1340	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1341	useDynamicWindow(scsu, dynamicWindow);
michael@0	1342	c=((uint32_t)(UC0+dynamicWindow)<<8)|(c-currentOffset)|0x80;
michael@0	1343	length=2;
michael@0	1344	goto outputBytes;
michael@0	1345	} else if(source<sourceLimit && lead==*source && /* too lazy to check trail in same window as source[1] */
michael@0	1346	(code=getDynamicOffset(c, &offset))>=0
michael@0	1347	) {
michael@0	1348	/* two supplementary characters in (probably) the same window - define an extended one */
michael@0	1349	isSingleByteMode=TRUE;
michael@0	1350	code-=0x200;
michael@0	1351	dynamicWindow=getNextDynamicWindow(scsu);
michael@0	1352	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset;
michael@0	1353	useDynamicWindow(scsu, dynamicWindow);
michael@0	1354	c=((uint32_t)UDX<<24)|((uint32_t)dynamicWindow<<21)|((uint32_t)code<<8)|(c-currentOffset)|0x80;
michael@0	1355	length=4;
michael@0	1356	goto outputBytes;
michael@0	1357	} else {
michael@0	1358	/* don't know how to compress this character, just write it directly */
michael@0	1359	c=((uint32_t)lead<<16)|trail;
michael@0	1360	length=4;
michael@0	1361	goto outputBytes;
michael@0	1362	}
michael@0	1363	} else /* 0xe000<=c<0xf300 */ {
michael@0	1364	/* quote to avoid SCSU tags */
michael@0	1365	c|=UQU<<16;
michael@0	1366	length=3;
michael@0	1367	goto outputBytes;
michael@0	1368	}
michael@0	1369
michael@0	1370	/* normal end of conversion: prepare for a new character */
michael@0	1371	c=0;
michael@0	1372	sourceIndex=nextSourceIndex;
michael@0	1373	}
michael@0	1374	}
michael@0	1375	endloop:
michael@0	1376
michael@0	1377	/* set the converter state back into UConverter */
michael@0	1378	scsu->fromUIsSingleByteMode=isSingleByteMode;
michael@0	1379	scsu->fromUDynamicWindow=dynamicWindow;
michael@0	1380
michael@0	1381	cnv->fromUChar32=c;
michael@0	1382
michael@0	1383	/* write back the updated pointers */
michael@0	1384	pArgs->source=source;
michael@0	1385	pArgs->target=(char *)target;
michael@0	1386	pArgs->offsets=offsets;
michael@0	1387	return;
michael@0	1388
michael@0	1389	outputBytes:
michael@0	1390	/* write the output character bytes from c and length [code copied from ucnvmbcs.c] */
michael@0	1391	/* from the first if in the loop we know that targetCapacity>0 */
michael@0	1392	if(length<=targetCapacity) {
michael@0	1393	if(offsets==NULL) {
michael@0	1394	switch(length) {
michael@0	1395	/* each branch falls through to the next one */
michael@0	1396	case 4:
michael@0	1397	*target++=(uint8_t)(c>>24);
michael@0	1398	case 3: /*fall through*/
michael@0	1399	*target++=(uint8_t)(c>>16);
michael@0	1400	case 2: /*fall through*/
michael@0	1401	*target++=(uint8_t)(c>>8);
michael@0	1402	case 1: /*fall through*/
michael@0	1403	*target++=(uint8_t)c;
michael@0	1404	default:
michael@0	1405	/* will never occur */
michael@0	1406	break;
michael@0	1407	}
michael@0	1408	} else {
michael@0	1409	switch(length) {
michael@0	1410	/* each branch falls through to the next one */
michael@0	1411	case 4:
michael@0	1412	*target++=(uint8_t)(c>>24);
michael@0	1413	*offsets++=sourceIndex;
michael@0	1414	case 3: /*fall through*/
michael@0	1415	*target++=(uint8_t)(c>>16);
michael@0	1416	*offsets++=sourceIndex;
michael@0	1417	case 2: /*fall through*/
michael@0	1418	*target++=(uint8_t)(c>>8);
michael@0	1419	*offsets++=sourceIndex;
michael@0	1420	case 1: /*fall through*/
michael@0	1421	*target++=(uint8_t)c;
michael@0	1422	*offsets++=sourceIndex;
michael@0	1423	default:
michael@0	1424	/* will never occur */
michael@0	1425	break;
michael@0	1426	}
michael@0	1427	}
michael@0	1428	targetCapacity-=length;
michael@0	1429
michael@0	1430	/* normal end of conversion: prepare for a new character */
michael@0	1431	c=0;
michael@0	1432	sourceIndex=nextSourceIndex;
michael@0	1433	goto loop;
michael@0	1434	} else {
michael@0	1435	uint8_t *p;
michael@0	1436
michael@0	1437	/*
michael@0	1438	* We actually do this backwards here:
michael@0	1439	* In order to save an intermediate variable, we output
michael@0	1440	* first to the overflow buffer what does not fit into the
michael@0	1441	* regular target.
michael@0	1442	*/
michael@0	1443	/* we know that 0<=targetCapacity<length<=4 */
michael@0	1444	/* targetCapacity==0 when SCU+supplementary where SCU used up targetCapacity==1 */
michael@0	1445	length-=targetCapacity;
michael@0	1446	p=(uint8_t *)cnv->charErrorBuffer;
michael@0	1447	switch(length) {
michael@0	1448	/* each branch falls through to the next one */
michael@0	1449	case 4:
michael@0	1450	*p++=(uint8_t)(c>>24);
michael@0	1451	case 3: /*fall through*/
michael@0	1452	*p++=(uint8_t)(c>>16);
michael@0	1453	case 2: /*fall through*/
michael@0	1454	*p++=(uint8_t)(c>>8);
michael@0	1455	case 1: /*fall through*/
michael@0	1456	*p=(uint8_t)c;
michael@0	1457	default:
michael@0	1458	/* will never occur */
michael@0	1459	break;
michael@0	1460	}
michael@0	1461	cnv->charErrorBufferLength=(int8_t)length;
michael@0	1462
michael@0	1463	/* now output what fits into the regular target */
michael@0	1464	c>>=8*length; /* length was reduced by targetCapacity */
michael@0	1465	switch(targetCapacity) {
michael@0	1466	/* each branch falls through to the next one */
michael@0	1467	case 3:
michael@0	1468	*target++=(uint8_t)(c>>16);
michael@0	1469	if(offsets!=NULL) {
michael@0	1470	*offsets++=sourceIndex;
michael@0	1471	}
michael@0	1472	case 2: /*fall through*/
michael@0	1473	*target++=(uint8_t)(c>>8);
michael@0	1474	if(offsets!=NULL) {
michael@0	1475	*offsets++=sourceIndex;
michael@0	1476	}
michael@0	1477	case 1: /*fall through*/
michael@0	1478	*target++=(uint8_t)c;
michael@0	1479	if(offsets!=NULL) {
michael@0	1480	*offsets++=sourceIndex;
michael@0	1481	}
michael@0	1482	default:
michael@0	1483	break;
michael@0	1484	}
michael@0	1485
michael@0	1486	/* target overflow */
michael@0	1487	targetCapacity=0;
michael@0	1488	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	1489	c=0;
michael@0	1490	goto endloop;
michael@0	1491	}
michael@0	1492	}
michael@0	1493
michael@0	1494	/*
michael@0	1495	* Identical to _SCSUFromUnicodeWithOffsets but without offset handling.
michael@0	1496	* If a change is made in the original function, then either
michael@0	1497	* change this function the same way or
michael@0	1498	* re-copy the original function and remove the variables
michael@0	1499	* offsets, sourceIndex, and nextSourceIndex.
michael@0	1500	*/
michael@0	1501	static void
michael@0	1502	_SCSUFromUnicode(UConverterFromUnicodeArgs *pArgs,
michael@0	1503	UErrorCode *pErrorCode) {
michael@0	1504	UConverter *cnv;
michael@0	1505	SCSUData *scsu;
michael@0	1506	const UChar *source, *sourceLimit;
michael@0	1507	uint8_t *target;
michael@0	1508	int32_t targetCapacity;
michael@0	1509
michael@0	1510	UBool isSingleByteMode;
michael@0	1511	uint8_t dynamicWindow;
michael@0	1512	uint32_t currentOffset;
michael@0	1513
michael@0	1514	uint32_t c, delta;
michael@0	1515
michael@0	1516	int32_t length;
michael@0	1517
michael@0	1518	/* variables for compression heuristics */
michael@0	1519	uint32_t offset;
michael@0	1520	UChar lead, trail;
michael@0	1521	int code;
michael@0	1522	int8_t window;
michael@0	1523
michael@0	1524	/* set up the local pointers */
michael@0	1525	cnv=pArgs->converter;
michael@0	1526	scsu=(SCSUData *)cnv->extraInfo;
michael@0	1527
michael@0	1528	/* set up the local pointers */
michael@0	1529	source=pArgs->source;
michael@0	1530	sourceLimit=pArgs->sourceLimit;
michael@0	1531	target=(uint8_t *)pArgs->target;
michael@0	1532	targetCapacity=(int32_t)(pArgs->targetLimit-pArgs->target);
michael@0	1533
michael@0	1534	/* get the state machine state */
michael@0	1535	isSingleByteMode=scsu->fromUIsSingleByteMode;
michael@0	1536	dynamicWindow=scsu->fromUDynamicWindow;
michael@0	1537	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1538
michael@0	1539	c=cnv->fromUChar32;
michael@0	1540
michael@0	1541	/* similar conversion "loop" as in toUnicode */
michael@0	1542	loop:
michael@0	1543	if(isSingleByteMode) {
michael@0	1544	if(c!=0 && targetCapacity>0) {
michael@0	1545	goto getTrailSingle;
michael@0	1546	}
michael@0	1547
michael@0	1548	/* state machine for single-byte mode */
michael@0	1549	/* singleByteMode: */
michael@0	1550	while(source<sourceLimit) {
michael@0	1551	if(targetCapacity<=0) {
michael@0	1552	/* target is full */
michael@0	1553	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	1554	break;
michael@0	1555	}
michael@0	1556	c=*source++;
michael@0	1557
michael@0	1558	if((c-0x20)<=0x5f) {
michael@0	1559	/* pass US-ASCII graphic character through */
michael@0	1560	*target++=(uint8_t)c;
michael@0	1561	--targetCapacity;
michael@0	1562	} else if(c<0x20) {
michael@0	1563	if((1UL<<c)&0x2601 /* binary 0010 0110 0000 0001, check for b==0xd || b==0xa || b==9 || b==0 */) {
michael@0	1564	/* CR/LF/TAB/NUL */
michael@0	1565	*target++=(uint8_t)c;
michael@0	1566	--targetCapacity;
michael@0	1567	} else {
michael@0	1568	/* quote C0 control character */
michael@0	1569	c|=SQ0<<8;
michael@0	1570	length=2;
michael@0	1571	goto outputBytes;
michael@0	1572	}
michael@0	1573	} else if((delta=c-currentOffset)<=0x7f) {
michael@0	1574	/* use the current dynamic window */
michael@0	1575	*target++=(uint8_t)(delta|0x80);
michael@0	1576	--targetCapacity;
michael@0	1577	} else if(U16_IS_SURROGATE(c)) {
michael@0	1578	if(U16_IS_SURROGATE_LEAD(c)) {
michael@0	1579	getTrailSingle:
michael@0	1580	lead=(UChar)c;
michael@0	1581	if(source<sourceLimit) {
michael@0	1582	/* test the following code unit */
michael@0	1583	trail=*source;
michael@0	1584	if(U16_IS_TRAIL(trail)) {
michael@0	1585	++source;
michael@0	1586	c=U16_GET_SUPPLEMENTARY(c, trail);
michael@0	1587	/* convert this surrogate code point */
michael@0	1588	/* exit this condition tree */
michael@0	1589	} else {
michael@0	1590	/* this is an unmatched lead code unit (1st surrogate) */
michael@0	1591	/* callback(illegal) */
michael@0	1592	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	1593	goto endloop;
michael@0	1594	}
michael@0	1595	} else {
michael@0	1596	/* no more input */
michael@0	1597	break;
michael@0	1598	}
michael@0	1599	} else {
michael@0	1600	/* this is an unmatched trail code unit (2nd surrogate) */
michael@0	1601	/* callback(illegal) */
michael@0	1602	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	1603	goto endloop;
michael@0	1604	}
michael@0	1605
michael@0	1606	/* compress supplementary character U+10000..U+10ffff */
michael@0	1607	if((delta=c-currentOffset)<=0x7f) {
michael@0	1608	/* use the current dynamic window */
michael@0	1609	*target++=(uint8_t)(delta|0x80);
michael@0	1610	--targetCapacity;
michael@0	1611	} else if((window=getWindow(scsu->fromUDynamicOffsets, c))>=0) {
michael@0	1612	/* there is a dynamic window that contains this character, change to it */
michael@0	1613	dynamicWindow=window;
michael@0	1614	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1615	useDynamicWindow(scsu, dynamicWindow);
michael@0	1616	c=((uint32_t)(SC0+dynamicWindow)<<8)|(c-currentOffset)|0x80;
michael@0	1617	length=2;
michael@0	1618	goto outputBytes;
michael@0	1619	} else if((code=getDynamicOffset(c, &offset))>=0) {
michael@0	1620	/* might check if there are more characters in this window to come */
michael@0	1621	/* define an extended window with this character */
michael@0	1622	code-=0x200;
michael@0	1623	dynamicWindow=getNextDynamicWindow(scsu);
michael@0	1624	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset;
michael@0	1625	useDynamicWindow(scsu, dynamicWindow);
michael@0	1626	c=((uint32_t)SDX<<24)|((uint32_t)dynamicWindow<<21)|((uint32_t)code<<8)|(c-currentOffset)|0x80;
michael@0	1627	length=4;
michael@0	1628	goto outputBytes;
michael@0	1629	} else {
michael@0	1630	/* change to Unicode mode and output this (lead, trail) pair */
michael@0	1631	isSingleByteMode=FALSE;
michael@0	1632	*target++=(uint8_t)SCU;
michael@0	1633	--targetCapacity;
michael@0	1634	c=((uint32_t)lead<<16)|trail;
michael@0	1635	length=4;
michael@0	1636	goto outputBytes;
michael@0	1637	}
michael@0	1638	} else if(c<0xa0) {
michael@0	1639	/* quote C1 control character */
michael@0	1640	c=(c&0x7f)|(SQ0+1)<<8; /* SQ0+1==SQ1 */
michael@0	1641	length=2;
michael@0	1642	goto outputBytes;
michael@0	1643	} else if(c==0xfeff || c>=0xfff0) {
michael@0	1644	/* quote signature character=byte order mark and specials */
michael@0	1645	c|=SQU<<16;
michael@0	1646	length=3;
michael@0	1647	goto outputBytes;
michael@0	1648	} else {
michael@0	1649	/* compress all other BMP characters */
michael@0	1650	if((window=getWindow(scsu->fromUDynamicOffsets, c))>=0) {
michael@0	1651	/* there is a window defined that contains this character - switch to it or quote from it? */
michael@0	1652	if(source>=sourceLimit || isInOffsetWindowOrDirect(scsu->fromUDynamicOffsets[window], *source)) {
michael@0	1653	/* change to dynamic window */
michael@0	1654	dynamicWindow=window;
michael@0	1655	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1656	useDynamicWindow(scsu, dynamicWindow);
michael@0	1657	c=((uint32_t)(SC0+dynamicWindow)<<8)|(c-currentOffset)|0x80;
michael@0	1658	length=2;
michael@0	1659	goto outputBytes;
michael@0	1660	} else {
michael@0	1661	/* quote from dynamic window */
michael@0	1662	c=((uint32_t)(SQ0+window)<<8)|(c-scsu->fromUDynamicOffsets[window])|0x80;
michael@0	1663	length=2;
michael@0	1664	goto outputBytes;
michael@0	1665	}
michael@0	1666	} else if((window=getWindow(staticOffsets, c))>=0) {
michael@0	1667	/* quote from static window */
michael@0	1668	c=((uint32_t)(SQ0+window)<<8)|(c-staticOffsets[window]);
michael@0	1669	length=2;
michael@0	1670	goto outputBytes;
michael@0	1671	} else if((code=getDynamicOffset(c, &offset))>=0) {
michael@0	1672	/* define a dynamic window with this character */
michael@0	1673	dynamicWindow=getNextDynamicWindow(scsu);
michael@0	1674	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset;
michael@0	1675	useDynamicWindow(scsu, dynamicWindow);
michael@0	1676	c=((uint32_t)(SD0+dynamicWindow)<<16)|((uint32_t)code<<8)|(c-currentOffset)|0x80;
michael@0	1677	length=3;
michael@0	1678	goto outputBytes;
michael@0	1679	} else if((uint32_t)(c-0x3400)<(0xd800-0x3400) &&
michael@0	1680	(source>=sourceLimit || (uint32_t)(*source-0x3400)<(0xd800-0x3400))
michael@0	1681	) {
michael@0	1682	/*
michael@0	1683	* this character is not compressible (a BMP ideograph or similar);
michael@0	1684	* switch to Unicode mode if this is the last character in the block
michael@0	1685	* or there is at least one more ideograph following immediately
michael@0	1686	*/
michael@0	1687	isSingleByteMode=FALSE;
michael@0	1688	c|=SCU<<16;
michael@0	1689	length=3;
michael@0	1690	goto outputBytes;
michael@0	1691	} else {
michael@0	1692	/* quote Unicode */
michael@0	1693	c|=SQU<<16;
michael@0	1694	length=3;
michael@0	1695	goto outputBytes;
michael@0	1696	}
michael@0	1697	}
michael@0	1698
michael@0	1699	/* normal end of conversion: prepare for a new character */
michael@0	1700	c=0;
michael@0	1701	}
michael@0	1702	} else {
michael@0	1703	if(c!=0 && targetCapacity>0) {
michael@0	1704	goto getTrailUnicode;
michael@0	1705	}
michael@0	1706
michael@0	1707	/* state machine for Unicode mode */
michael@0	1708	/* unicodeByteMode: */
michael@0	1709	while(source<sourceLimit) {
michael@0	1710	if(targetCapacity<=0) {
michael@0	1711	/* target is full */
michael@0	1712	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	1713	break;
michael@0	1714	}
michael@0	1715	c=*source++;
michael@0	1716
michael@0	1717	if((uint32_t)(c-0x3400)<(0xd800-0x3400)) {
michael@0	1718	/* not compressible, write character directly */
michael@0	1719	if(targetCapacity>=2) {
michael@0	1720	*target++=(uint8_t)(c>>8);
michael@0	1721	*target++=(uint8_t)c;
michael@0	1722	targetCapacity-=2;
michael@0	1723	} else {
michael@0	1724	length=2;
michael@0	1725	goto outputBytes;
michael@0	1726	}
michael@0	1727	} else if((uint32_t)(c-0x3400)>=(0xf300-0x3400) /* c<0x3400 || c>=0xf300 */) {
michael@0	1728	/* compress BMP character if the following one is not an uncompressible ideograph */
michael@0	1729	if(!(source<sourceLimit && (uint32_t)(*source-0x3400)<(0xd800-0x3400))) {
michael@0	1730	if(((uint32_t)(c-0x30)<10 || (uint32_t)(c-0x61)<26 || (uint32_t)(c-0x41)<26)) {
michael@0	1731	/* ASCII digit or letter */
michael@0	1732	isSingleByteMode=TRUE;
michael@0	1733	c|=((uint32_t)(UC0+dynamicWindow)<<8)|c;
michael@0	1734	length=2;
michael@0	1735	goto outputBytes;
michael@0	1736	} else if((window=getWindow(scsu->fromUDynamicOffsets, c))>=0) {
michael@0	1737	/* there is a dynamic window that contains this character, change to it */
michael@0	1738	isSingleByteMode=TRUE;
michael@0	1739	dynamicWindow=window;
michael@0	1740	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1741	useDynamicWindow(scsu, dynamicWindow);
michael@0	1742	c=((uint32_t)(UC0+dynamicWindow)<<8)|(c-currentOffset)|0x80;
michael@0	1743	length=2;
michael@0	1744	goto outputBytes;
michael@0	1745	} else if((code=getDynamicOffset(c, &offset))>=0) {
michael@0	1746	/* define a dynamic window with this character */
michael@0	1747	isSingleByteMode=TRUE;
michael@0	1748	dynamicWindow=getNextDynamicWindow(scsu);
michael@0	1749	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset;
michael@0	1750	useDynamicWindow(scsu, dynamicWindow);
michael@0	1751	c=((uint32_t)(UD0+dynamicWindow)<<16)|((uint32_t)code<<8)|(c-currentOffset)|0x80;
michael@0	1752	length=3;
michael@0	1753	goto outputBytes;
michael@0	1754	}
michael@0	1755	}
michael@0	1756
michael@0	1757	/* don't know how to compress this character, just write it directly */
michael@0	1758	length=2;
michael@0	1759	goto outputBytes;
michael@0	1760	} else if(c<0xe000) {
michael@0	1761	/* c is a surrogate */
michael@0	1762	if(U16_IS_SURROGATE_LEAD(c)) {
michael@0	1763	getTrailUnicode:
michael@0	1764	lead=(UChar)c;
michael@0	1765	if(source<sourceLimit) {
michael@0	1766	/* test the following code unit */
michael@0	1767	trail=*source;
michael@0	1768	if(U16_IS_TRAIL(trail)) {
michael@0	1769	++source;
michael@0	1770	c=U16_GET_SUPPLEMENTARY(c, trail);
michael@0	1771	/* convert this surrogate code point */
michael@0	1772	/* exit this condition tree */
michael@0	1773	} else {
michael@0	1774	/* this is an unmatched lead code unit (1st surrogate) */
michael@0	1775	/* callback(illegal) */
michael@0	1776	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	1777	goto endloop;
michael@0	1778	}
michael@0	1779	} else {
michael@0	1780	/* no more input */
michael@0	1781	break;
michael@0	1782	}
michael@0	1783	} else {
michael@0	1784	/* this is an unmatched trail code unit (2nd surrogate) */
michael@0	1785	/* callback(illegal) */
michael@0	1786	*pErrorCode=U_ILLEGAL_CHAR_FOUND;
michael@0	1787	goto endloop;
michael@0	1788	}
michael@0	1789
michael@0	1790	/* compress supplementary character */
michael@0	1791	if( (window=getWindow(scsu->fromUDynamicOffsets, c))>=0 &&
michael@0	1792	!(source<sourceLimit && (uint32_t)(*source-0x3400)<(0xd800-0x3400))
michael@0	1793	) {
michael@0	1794	/*
michael@0	1795	* there is a dynamic window that contains this character and
michael@0	1796	* the following character is not uncompressible,
michael@0	1797	* change to the window
michael@0	1798	*/
michael@0	1799	isSingleByteMode=TRUE;
michael@0	1800	dynamicWindow=window;
michael@0	1801	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow];
michael@0	1802	useDynamicWindow(scsu, dynamicWindow);
michael@0	1803	c=((uint32_t)(UC0+dynamicWindow)<<8)|(c-currentOffset)|0x80;
michael@0	1804	length=2;
michael@0	1805	goto outputBytes;
michael@0	1806	} else if(source<sourceLimit && lead==*source && /* too lazy to check trail in same window as source[1] */
michael@0	1807	(code=getDynamicOffset(c, &offset))>=0
michael@0	1808	) {
michael@0	1809	/* two supplementary characters in (probably) the same window - define an extended one */
michael@0	1810	isSingleByteMode=TRUE;
michael@0	1811	code-=0x200;
michael@0	1812	dynamicWindow=getNextDynamicWindow(scsu);
michael@0	1813	currentOffset=scsu->fromUDynamicOffsets[dynamicWindow]=offset;
michael@0	1814	useDynamicWindow(scsu, dynamicWindow);
michael@0	1815	c=((uint32_t)UDX<<24)|((uint32_t)dynamicWindow<<21)|((uint32_t)code<<8)|(c-currentOffset)|0x80;
michael@0	1816	length=4;
michael@0	1817	goto outputBytes;
michael@0	1818	} else {
michael@0	1819	/* don't know how to compress this character, just write it directly */
michael@0	1820	c=((uint32_t)lead<<16)|trail;
michael@0	1821	length=4;
michael@0	1822	goto outputBytes;
michael@0	1823	}
michael@0	1824	} else /* 0xe000<=c<0xf300 */ {
michael@0	1825	/* quote to avoid SCSU tags */
michael@0	1826	c|=UQU<<16;
michael@0	1827	length=3;
michael@0	1828	goto outputBytes;
michael@0	1829	}
michael@0	1830
michael@0	1831	/* normal end of conversion: prepare for a new character */
michael@0	1832	c=0;
michael@0	1833	}
michael@0	1834	}
michael@0	1835	endloop:
michael@0	1836
michael@0	1837	/* set the converter state back into UConverter */
michael@0	1838	scsu->fromUIsSingleByteMode=isSingleByteMode;
michael@0	1839	scsu->fromUDynamicWindow=dynamicWindow;
michael@0	1840
michael@0	1841	cnv->fromUChar32=c;
michael@0	1842
michael@0	1843	/* write back the updated pointers */
michael@0	1844	pArgs->source=source;
michael@0	1845	pArgs->target=(char *)target;
michael@0	1846	return;
michael@0	1847
michael@0	1848	outputBytes:
michael@0	1849	/* write the output character bytes from c and length [code copied from ucnvmbcs.c] */
michael@0	1850	/* from the first if in the loop we know that targetCapacity>0 */
michael@0	1851	if(length<=targetCapacity) {
michael@0	1852	switch(length) {
michael@0	1853	/* each branch falls through to the next one */
michael@0	1854	case 4:
michael@0	1855	*target++=(uint8_t)(c>>24);
michael@0	1856	case 3: /*fall through*/
michael@0	1857	*target++=(uint8_t)(c>>16);
michael@0	1858	case 2: /*fall through*/
michael@0	1859	*target++=(uint8_t)(c>>8);
michael@0	1860	case 1: /*fall through*/
michael@0	1861	*target++=(uint8_t)c;
michael@0	1862	default:
michael@0	1863	/* will never occur */
michael@0	1864	break;
michael@0	1865	}
michael@0	1866	targetCapacity-=length;
michael@0	1867
michael@0	1868	/* normal end of conversion: prepare for a new character */
michael@0	1869	c=0;
michael@0	1870	goto loop;
michael@0	1871	} else {
michael@0	1872	uint8_t *p;
michael@0	1873
michael@0	1874	/*
michael@0	1875	* We actually do this backwards here:
michael@0	1876	* In order to save an intermediate variable, we output
michael@0	1877	* first to the overflow buffer what does not fit into the
michael@0	1878	* regular target.
michael@0	1879	*/
michael@0	1880	/* we know that 0<=targetCapacity<length<=4 */
michael@0	1881	/* targetCapacity==0 when SCU+supplementary where SCU used up targetCapacity==1 */
michael@0	1882	length-=targetCapacity;
michael@0	1883	p=(uint8_t *)cnv->charErrorBuffer;
michael@0	1884	switch(length) {
michael@0	1885	/* each branch falls through to the next one */
michael@0	1886	case 4:
michael@0	1887	*p++=(uint8_t)(c>>24);
michael@0	1888	case 3: /*fall through*/
michael@0	1889	*p++=(uint8_t)(c>>16);
michael@0	1890	case 2: /*fall through*/
michael@0	1891	*p++=(uint8_t)(c>>8);
michael@0	1892	case 1: /*fall through*/
michael@0	1893	*p=(uint8_t)c;
michael@0	1894	default:
michael@0	1895	/* will never occur */
michael@0	1896	break;
michael@0	1897	}
michael@0	1898	cnv->charErrorBufferLength=(int8_t)length;
michael@0	1899
michael@0	1900	/* now output what fits into the regular target */
michael@0	1901	c>>=8*length; /* length was reduced by targetCapacity */
michael@0	1902	switch(targetCapacity) {
michael@0	1903	/* each branch falls through to the next one */
michael@0	1904	case 3:
michael@0	1905	*target++=(uint8_t)(c>>16);
michael@0	1906	case 2: /*fall through*/
michael@0	1907	*target++=(uint8_t)(c>>8);
michael@0	1908	case 1: /*fall through*/
michael@0	1909	*target++=(uint8_t)c;
michael@0	1910	default:
michael@0	1911	break;
michael@0	1912	}
michael@0	1913
michael@0	1914	/* target overflow */
michael@0	1915	targetCapacity=0;
michael@0	1916	*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0	1917	c=0;
michael@0	1918	goto endloop;
michael@0	1919	}
michael@0	1920	}
michael@0	1921
michael@0	1922	/* miscellaneous ------------------------------------------------------------ */
michael@0	1923
michael@0	1924	static const char *
michael@0	1925	_SCSUGetName(const UConverter *cnv) {
michael@0	1926	SCSUData *scsu=(SCSUData *)cnv->extraInfo;
michael@0	1927
michael@0	1928	switch(scsu->locale) {
michael@0	1929	case l_ja:
michael@0	1930	return "SCSU,locale=ja";
michael@0	1931	default:
michael@0	1932	return "SCSU";
michael@0	1933	}
michael@0	1934	}
michael@0	1935
michael@0	1936	/* structure for SafeClone calculations */
michael@0	1937	struct cloneSCSUStruct
michael@0	1938	{
michael@0	1939	UConverter cnv;
michael@0	1940	SCSUData mydata;
michael@0	1941	};
michael@0	1942
michael@0	1943	static UConverter *
michael@0	1944	_SCSUSafeClone(const UConverter *cnv,
michael@0	1945	void *stackBuffer,
michael@0	1946	int32_t *pBufferSize,
michael@0	1947	UErrorCode *status)
michael@0	1948	{
michael@0	1949	struct cloneSCSUStruct * localClone;
michael@0	1950	int32_t bufferSizeNeeded = sizeof(struct cloneSCSUStruct);
michael@0	1951
michael@0	1952	if (U_FAILURE(*status)){
michael@0	1953	return 0;
michael@0	1954	}
michael@0	1955
michael@0	1956	if (*pBufferSize == 0){ /* 'preflighting' request - set needed size into *pBufferSize */
michael@0	1957	*pBufferSize = bufferSizeNeeded;
michael@0	1958	return 0;
michael@0	1959	}
michael@0	1960
michael@0	1961	localClone = (struct cloneSCSUStruct *)stackBuffer;
michael@0	1962	/* ucnv.c/ucnv_safeClone() copied the main UConverter already */
michael@0	1963
michael@0	1964	uprv_memcpy(&localClone->mydata, cnv->extraInfo, sizeof(SCSUData));
michael@0	1965	localClone->cnv.extraInfo = &localClone->mydata;
michael@0	1966	localClone->cnv.isExtraLocal = TRUE;
michael@0	1967
michael@0	1968	return &localClone->cnv;
michael@0	1969	}
michael@0	1970
michael@0	1971
michael@0	1972	static const UConverterImpl _SCSUImpl={
michael@0	1973	UCNV_SCSU,
michael@0	1974
michael@0	1975	NULL,
michael@0	1976	NULL,
michael@0	1977
michael@0	1978	_SCSUOpen,
michael@0	1979	_SCSUClose,
michael@0	1980	_SCSUReset,
michael@0	1981
michael@0	1982	_SCSUToUnicode,
michael@0	1983	_SCSUToUnicodeWithOffsets,
michael@0	1984	_SCSUFromUnicode,
michael@0	1985	_SCSUFromUnicodeWithOffsets,
michael@0	1986	NULL,
michael@0	1987
michael@0	1988	NULL,
michael@0	1989	_SCSUGetName,
michael@0	1990	NULL,
michael@0	1991	_SCSUSafeClone,
michael@0	1992	ucnv_getCompleteUnicodeSet
michael@0	1993	};
michael@0	1994
michael@0	1995	static const UConverterStaticData _SCSUStaticData={
michael@0	1996	sizeof(UConverterStaticData),
michael@0	1997	"SCSU",
michael@0	1998	1212, /* CCSID for SCSU */
michael@0	1999	UCNV_IBM, UCNV_SCSU,
michael@0	2000	1, 3, /* one UChar generates at least 1 byte and at most 3 bytes */
michael@0	2001	/*
michael@0	2002	* The subchar here is ignored because _SCSUOpen() sets U+fffd as a Unicode
michael@0	2003	* substitution string.
michael@0	2004	*/
michael@0	2005	{ 0x0e, 0xff, 0xfd, 0 }, 3,
michael@0	2006	FALSE, FALSE,
michael@0	2007	0,
michael@0	2008	0,
michael@0	2009	{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */
michael@0	2010	};
michael@0	2011
michael@0	2012	const UConverterSharedData _SCSUData={
michael@0	2013	sizeof(UConverterSharedData), ~((uint32_t)0),
michael@0	2014	NULL, NULL, &_SCSUStaticData, FALSE, &_SCSUImpl,
michael@0	2015	0
michael@0	2016	};
michael@0	2017
michael@0	2018	#endif