]> git.saurik.com Git - apple/icu.git/blame - icuSources/common/ucnvisci.c
projects / apple / icu.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
ICU-8.11.4.tar.gz
[apple/icu.git] / icuSources / common / ucnvisci.c
CommitLineData
b75a7d8f
A		 1/*
2**********************************************************************
73c04bcf 3* Copyright (C) 2000-2006, International Business Machines
b75a7d8f
A		 4* Corporation and others. All Rights Reserved.
5**********************************************************************
6* file name: ucnvisci.c
7* encoding: US-ASCII
8* tab size: 8 (not used)
9* indentation:4
10*
11* created on: 2001JUN26
12* created by: Ram Viswanadha
13*
14* Date Name Description
15* 24/7/2001 Ram Added support for EXT character handling
16*/
17
18#include "unicode/utypes.h"
19
374ca955 20#if !UCONFIG_NO_CONVERSION && !UCONFIG_NO_LEGACY_CONVERSION
b75a7d8f
A		 21
22#include "cmemory.h"
b75a7d8f
A		 23#include "ucnv_bld.h"
24#include "unicode/ucnv.h"
25#include "ucnv_cnv.h"
26#include "unicode/ucnv_cb.h"
27#include "unicode/uset.h"
28#include "cstring.h"
29
30#define UCNV_OPTIONS_VERSION_MASK 0xf
31#define NUKTA 0x093c
32#define HALANT 0x094d
33#define ZWNJ 0x200c /* Zero Width Non Joiner */
34#define ZWJ 0x200d /* Zero width Joiner */
35#define INVALID_CHAR 0xffff
36#define ATR 0xEF /* Attribute code */
37#define EXT 0xF0 /* Extension code */
38#define DANDA 0x0964
39#define DOUBLE_DANDA 0x0965
40#define ISCII_NUKTA 0xE9
41#define ISCII_HALANT 0xE8
42#define ISCII_DANDA 0xEA
43#define ISCII_INV 0xD9
44#define INDIC_BLOCK_BEGIN 0x0900
45#define INDIC_BLOCK_END 0x0D7F
46#define INDIC_RANGE (INDIC_BLOCK_END - INDIC_BLOCK_BEGIN)
47#define VOCALLIC_RR 0x0931
48#define LF 0x0A
49#define ASCII_END 0xA0
50#define NO_CHAR_MARKER 0xFFFE
51#define TELUGU_DELTA DELTA * TELUGU
52#define DEV_ABBR_SIGN 0x0970
53#define DEV_ANUDATTA 0x0952
54#define EXT_RANGE_BEGIN 0xA1
55#define EXT_RANGE_END 0xEE
56
57
58typedef enum {
59 DEVANAGARI =0,
60 BENGALI,
61 GURMUKHI,
62 GUJARATI,
63 ORIYA,
64 TAMIL,
65 TELUGU,
66 KANNADA,
67 MALAYALAM,
68 DELTA=0x80
69}UniLang;
70
71
72/**
73c04bcf 73 * Enumeration for switching code pages if <ATR>+<one of below values>
b75a7d8f
A		 74 * is encountered
75 */
76typedef enum {
73c04bcf
A		 77 DEF = 0x40,
78 RMN = 0x41,
79 DEV = 0x42,
80 BNG = 0x43,
81 TML = 0x44,
82 TLG = 0x45,
83 ASM = 0x46,
84 ORI = 0x47,
85 KND = 0x48,
86 MLM = 0x49,
87 GJR = 0x4A,
88 PNJ = 0x4B,
89 ARB = 0x71,
90 PES = 0x72,
91 URD = 0x73,
92 SND = 0x74,
93 KSM = 0x75,
94 PST = 0x76
b75a7d8f
A		 95}ISCIILang;
96
97typedef enum{
98 DEV_MASK =0x80,
99 PNJ_MASK =0x40,
100 GJR_MASK =0x20,
101 ORI_MASK =0x10,
102 BNG_MASK =0x08,
103 KND_MASK =0x04,
104 MLM_MASK =0x02,
105 TML_MASK =0x01,
106 ZERO =0x00
107}MaskEnum;
108
109typedef struct{
110 UChar contextCharToUnicode; /* previous Unicode codepoint for contextual analysis */
111 UChar contextCharFromUnicode; /* previous Unicode codepoint for contextual analysis */
112 uint16_t defDeltaToUnicode; /* delta for switching to default state when DEF is encountered */
113 uint16_t currentDeltaFromUnicode;/* current delta in Indic block */
114 uint16_t currentDeltaToUnicode; /* current delta in Indic block */
115 MaskEnum currentMaskFromUnicode; /* mask for current state in toUnicode */
116 MaskEnum currentMaskToUnicode; /* mask for current state in toUnicode */
117 MaskEnum defMaskToUnicode; /* mask for default state in toUnicode */
374ca955 118 UBool isFirstBuffer; /* boolean for fromUnicode to see if we need to announce the first script */
73c04bcf 119 UBool resetToDefaultToUnicode; /* boolean for reseting to default delta and mask when a newline is encountered*/
b75a7d8f
A		 120 char name[30];
121}UConverterDataISCII;
122
73c04bcf
A		 123typedef struct LookupDataStruct
124{
125 UniLang uniLang;
126 MaskEnum maskEnum;
127 ISCIILang isciiLang;
128} LookupDataStruct;
129
130static const LookupDataStruct lookupInitialData[]={
b75a7d8f
A		 131 { DEVANAGARI, DEV_MASK, DEV },
132 { BENGALI, BNG_MASK, BNG },
133 { GURMUKHI, PNJ_MASK, PNJ },
134 { GUJARATI, GJR_MASK, GJR },
135 { ORIYA, ORI_MASK, ORI },
136 { TAMIL, TML_MASK, TML },
137 { TELUGU, KND_MASK, TLG },
138 { KANNADA, KND_MASK, KND },
139 { MALAYALAM, MLM_MASK, MLM }
140};
141
142static void
143_ISCIIOpen(UConverter *cnv, const char *name,const char *locale,uint32_t options, UErrorCode *errorCode){
144 cnv->extraInfo = uprv_malloc (sizeof (UConverterDataISCII));
145
146 if(cnv->extraInfo != NULL) {
147 int32_t len=0;
148 UConverterDataISCII *converterData=(UConverterDataISCII *) cnv->extraInfo;
149 converterData->contextCharToUnicode=NO_CHAR_MARKER;
150 cnv->toUnicodeStatus = missingCharMarker;
151 converterData->contextCharFromUnicode=0x0000;
73c04bcf 152 converterData->resetToDefaultToUnicode=FALSE;
b75a7d8f
A		 153 /* check if the version requested is supported */
154 if((options & UCNV_OPTIONS_VERSION_MASK) < 9){
155 /* initialize state variables */
156 converterData->currentDeltaFromUnicode=converterData->currentDeltaToUnicode=
157 converterData->defDeltaToUnicode=
73c04bcf 158 (uint16_t)(lookupInitialData[options & UCNV_OPTIONS_VERSION_MASK].uniLang * DELTA);
b75a7d8f
A		 159
160 converterData->currentMaskFromUnicode = converterData->currentMaskToUnicode =
73c04bcf 161 converterData->defMaskToUnicode=lookupInitialData[options & UCNV_OPTIONS_VERSION_MASK].maskEnum;
b75a7d8f
A		 162
163 converterData->isFirstBuffer=TRUE;
164 uprv_strcpy(converterData->name,"ISCII,version=");
165 len = (int32_t)uprv_strlen(converterData->name);
166 converterData->name[len]= (char)((options & UCNV_OPTIONS_VERSION_MASK) + '0');
167 converterData->name[len+1]=0;
168 }else{
169 uprv_free(cnv->extraInfo);
170 cnv->extraInfo = NULL;
171 *errorCode = U_ILLEGAL_ARGUMENT_ERROR;
172 }
173
174 }else{
175 *errorCode =U_MEMORY_ALLOCATION_ERROR;
176 }
177}
178static void
179_ISCIIClose(UConverter *cnv){
180 if(cnv->extraInfo!=NULL) {
181 if(!cnv->isExtraLocal) {
182 uprv_free(cnv->extraInfo);
183 }
184 cnv->extraInfo=NULL;
185 }
186}
187
188static const char*
189_ISCIIgetName(const UConverter* cnv){
190 if(cnv->extraInfo){
191 UConverterDataISCII* myData= (UConverterDataISCII*)cnv->extraInfo;
192 return myData->name;
193 }
194 return NULL;
195}
196
197static void
198_ISCIIReset(UConverter *cnv, UConverterResetChoice choice){
199 UConverterDataISCII* data =(UConverterDataISCII *) (cnv->extraInfo);
200 if(choice<=UCNV_RESET_TO_UNICODE) {
201 cnv->toUnicodeStatus = missingCharMarker;
202 cnv->mode=0;
203 data->currentDeltaToUnicode=data->defDeltaToUnicode;
204 data->currentMaskToUnicode = data->defMaskToUnicode;
205 data->contextCharToUnicode=NO_CHAR_MARKER;
206 }
207 if(choice!=UCNV_RESET_TO_UNICODE) {
374ca955 208 cnv->fromUChar32=0x0000;
b75a7d8f 209 data->contextCharFromUnicode=0x00;
73c04bcf 210 data->currentMaskFromUnicode=data->defMaskToUnicode;
b75a7d8f 211 data->currentDeltaFromUnicode=data->defDeltaToUnicode;
374ca955 212 data->isFirstBuffer=TRUE;
73c04bcf 213 data->resetToDefaultToUnicode=FALSE;
b75a7d8f 214 }
b75a7d8f
A		 215}
216
217/**
218 * The values in validity table are indexed by the lower bits of Unicode
219 * range 0x0900 - 0x09ff. The values have a structure like:
220 * ---------------------------------------------------------------
221 * | DEV | PNJ | GJR | ORI | BNG | TLG | MLM | TML |
222 * | | | | | ASM | KND | | |
223 * ---------------------------------------------------------------
224 * If a code point is valid in a particular script
225 * then that bit is turned on
226 *
227 * Unicode does not distinguish between Bengali and Assamese so we use 1 bit for
228 * to represent these languages
229 *
230 * Telugu and Kannada have same codepoints except for Vocallic_RR which we special case
231 * and combine and use 1 bit to represent these languages.
232 *
233 * TODO: It is probably easier to understand and maintain to change this
234 * to use uint16_t and give each of the 9 Unicode/script blocks its own bit.
235 */
236
237static const uint8_t validityTable[128] = {
238/* This state table is tool generated please donot edit unless you know exactly what you are doing */
239/*ISCII:Valid:Unicode */
240/*0xa0 : 0x00: 0x900 */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
241/*0xa1 : 0xb8: 0x901 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO ,
242/*0xa2 : 0xfe: 0x902 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
243/*0xa3 : 0xbf: 0x903 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
244/*0x00 : 0x00: 0x904 */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
245/*0xa4 : 0xff: 0x905 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
246/*0xa5 : 0xff: 0x906 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
247/*0xa6 : 0xff: 0x907 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
248/*0xa7 : 0xff: 0x908 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
249/*0xa8 : 0xff: 0x909 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
250/*0xa9 : 0xff: 0x90a */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
251/*0xaa : 0xfe: 0x90b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
252/*0x00 : 0x00: 0x90c */ DEV_MASK + ZERO + ZERO + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
253/*0xae : 0x80: 0x90d */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
254/*0xab : 0x87: 0x90e */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + KND_MASK + MLM_MASK + TML_MASK ,
255/*0xac : 0xff: 0x90f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
256/*0xad : 0xff: 0x910 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
257/*0xb2 : 0x80: 0x911 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
258/*0xaf : 0x87: 0x912 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + KND_MASK + MLM_MASK + TML_MASK ,
259/*0xb0 : 0xff: 0x913 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
260/*0xb1 : 0xff: 0x914 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
261/*0xb3 : 0xff: 0x915 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
262/*0xb4 : 0xfe: 0x916 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
263/*0xb5 : 0xfe: 0x917 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
264/*0xb6 : 0xfe: 0x918 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
265/*0xb7 : 0xff: 0x919 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
266/*0xb8 : 0xff: 0x91a */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
267/*0xb9 : 0xfe: 0x91b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
268/*0xba : 0xff: 0x91c */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
269/*0xbb : 0xfe: 0x91d */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
270/*0xbc : 0xff: 0x91e */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
271/*0xbd : 0xff: 0x91f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
272/*0xbe : 0xfe: 0x920 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
273/*0xbf : 0xfe: 0x921 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
274/*0xc0 : 0xfe: 0x922 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
275/*0xc1 : 0xff: 0x923 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
276/*0xc2 : 0xff: 0x924 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
277/*0xc3 : 0xfe: 0x925 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
278/*0xc4 : 0xfe: 0x926 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
279/*0xc5 : 0xfe: 0x927 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
280/*0xc6 : 0xff: 0x928 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
281/*0xc7 : 0x81: 0x929 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + TML_MASK ,
282/*0xc8 : 0xff: 0x92a */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
283/*0xc9 : 0xfe: 0x92b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
284/*0xca : 0xfe: 0x92c */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
285/*0xcb : 0xfe: 0x92d */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
286/*0xcc : 0xfe: 0x92e */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
287/*0xcd : 0xff: 0x92f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
288/*0xcf : 0xff: 0x930 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
289/*0xd0 : 0x87: 0x931 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + MLM_MASK + TML_MASK ,
290/*0xd1 : 0xff: 0x932 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
291/*0xd2 : 0xb7: 0x933 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + ZERO + KND_MASK + MLM_MASK + TML_MASK ,
292/*0xd3 : 0x83: 0x934 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + MLM_MASK + TML_MASK ,
293/*0xd4 : 0xff: 0x935 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
294/*0xd5 : 0xfe: 0x936 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
295/*0xd6 : 0xbf: 0x937 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
296/*0xd7 : 0xff: 0x938 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
297/*0xd8 : 0xff: 0x939 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
298/*0x00 : 0x00: 0x93A */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
299/*0x00 : 0x00: 0x93B */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
300/*0xe9 : 0xda: 0x93c */ DEV_MASK + PNJ_MASK + ZERO + ORI_MASK + BNG_MASK + ZERO + MLM_MASK + ZERO ,
301/*0x00 : 0x00: 0x93d */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
302/*0xda : 0xff: 0x93e */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
303/*0xdb : 0xff: 0x93f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
304/*0xdc : 0xff: 0x940 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
305/*0xdd : 0xff: 0x941 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
306/*0xde : 0xff: 0x942 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
307/*0xdf : 0xbe: 0x943 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
308/*0x00 : 0x00: 0x944 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
309/*0xe3 : 0x80: 0x945 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
310/*0xe0 : 0x87: 0x946 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + KND_MASK + MLM_MASK + TML_MASK ,
311/*0xe1 : 0xff: 0x947 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
312/*0xe2 : 0xff: 0x948 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
313/*0xe7 : 0x80: 0x949 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
314/*0xe4 : 0x87: 0x94a */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + KND_MASK + MLM_MASK + TML_MASK ,
315/*0xe5 : 0xff: 0x94b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
316/*0xe6 : 0xff: 0x94c */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
317/*0xe8 : 0xff: 0x94d */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
318/*0xec : 0x00: 0x94e */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
319/*0xed : 0x00: 0x94f */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
320/*0x00 : 0x00: 0x950 */ DEV_MASK + ZERO + GJR_MASK + ZERO + ZERO + ZERO + ZERO + ZERO ,
321/*0x00 : 0x00: 0x951 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
322/*0x00 : 0x00: 0x952 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
323/*0x00 : 0x00: 0x953 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
324/*0x00 : 0x00: 0x954 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
325/*0x00 : 0x00: 0x955 */ ZERO + ZERO + ZERO + ZERO + ZERO + KND_MASK + ZERO + ZERO ,
326/*0x00 : 0x00: 0x956 */ ZERO + ZERO + ZERO + ORI_MASK + ZERO + KND_MASK + ZERO + ZERO ,
327/*0x00 : 0x00: 0x957 */ ZERO + ZERO + ZERO + ORI_MASK + ZERO + ZERO + MLM_MASK + ZERO ,
328/*0x00 : 0x00: 0x958 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
329/*0x00 : 0x00: 0x959 */ DEV_MASK + PNJ_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
330/*0x00 : 0x00: 0x95a */ DEV_MASK + PNJ_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
331/*0x00 : 0x00: 0x95b */ DEV_MASK + PNJ_MASK + ZERO + ORI_MASK + ZERO + ZERO + ZERO + ZERO ,
332/*0x00 : 0x00: 0x95c */ DEV_MASK + PNJ_MASK + ZERO + ZERO + BNG_MASK + ZERO + ZERO + ZERO ,
333/*0x00 : 0x00: 0x95d */ DEV_MASK + ZERO + ZERO + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO ,
334/*0x00 : 0x00: 0x95e */ DEV_MASK + PNJ_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
335/*0xce : 0x98: 0x95f */ DEV_MASK + ZERO + ZERO + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO ,
336/*0x00 : 0x00: 0x960 */ DEV_MASK + ZERO + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
337/*0x00 : 0x00: 0x961 */ DEV_MASK + ZERO + ZERO + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + ZERO ,
338/*0x00 : 0x00: 0x962 */ DEV_MASK + ZERO + ZERO + ZERO + BNG_MASK + ZERO + ZERO + ZERO ,
339/*0x00 : 0x00: 0x963 */ DEV_MASK + ZERO + ZERO + ZERO + BNG_MASK + ZERO + ZERO + ZERO ,
340/*0xea : 0xf8: 0x964 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO ,
341/*0xeaea : 0x00: 0x965*/ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + ZERO + ZERO + ZERO ,
342/*0xf1 : 0xff: 0x966 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
343/*0xf2 : 0xff: 0x967 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
344/*0xf3 : 0xff: 0x968 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
345/*0xf4 : 0xff: 0x969 */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
346/*0xf5 : 0xff: 0x96a */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
347/*0xf6 : 0xff: 0x96b */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
348/*0xf7 : 0xff: 0x96c */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
349/*0xf8 : 0xff: 0x96d */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
350/*0xf9 : 0xff: 0x96e */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
351/*0xfa : 0xff: 0x96f */ DEV_MASK + PNJ_MASK + GJR_MASK + ORI_MASK + BNG_MASK + KND_MASK + MLM_MASK + TML_MASK ,
352/*0x00 : 0x80: 0x970 */ DEV_MASK + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO ,
353
354/*
355 * The length of the array is 128 to provide values for 0x900..0x97f.
356 * The last 15 entries for 0x971..0x97f of the validity table are all zero
357 * because no Indic script uses such Unicode code points.
358 */
359/*0x00 : 0x00: 0x9yz */ ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO + ZERO
360};
361
362static const uint16_t fromUnicodeTable[128]={
363 0x00a0 ,/* 0x0900 */
364 0x00a1 ,/* 0x0901 */
365 0x00a2 ,/* 0x0902 */
366 0x00a3 ,/* 0x0903 */
367 0xFFFF ,/* 0x0904 */
368 0x00a4 ,/* 0x0905 */
369 0x00a5 ,/* 0x0906 */
370 0x00a6 ,/* 0x0907 */
371 0x00a7 ,/* 0x0908 */
372 0x00a8 ,/* 0x0909 */
373 0x00a9 ,/* 0x090a */
374 0x00aa ,/* 0x090b */
375 0xA6E9 ,/* 0x090c */
376 0x00ae ,/* 0x090d */
377 0x00ab ,/* 0x090e */
378 0x00ac ,/* 0x090f */
379 0x00ad ,/* 0x0910 */
380 0x00b2 ,/* 0x0911 */
381 0x00af ,/* 0x0912 */
382 0x00b0 ,/* 0x0913 */
383 0x00b1 ,/* 0x0914 */
384 0x00b3 ,/* 0x0915 */
385 0x00b4 ,/* 0x0916 */
386 0x00b5 ,/* 0x0917 */
387 0x00b6 ,/* 0x0918 */
388 0x00b7 ,/* 0x0919 */
389 0x00b8 ,/* 0x091a */
390 0x00b9 ,/* 0x091b */
391 0x00ba ,/* 0x091c */
392 0x00bb ,/* 0x091d */
393 0x00bc ,/* 0x091e */
394 0x00bd ,/* 0x091f */
395 0x00be ,/* 0x0920 */
396 0x00bf ,/* 0x0921 */
397 0x00c0 ,/* 0x0922 */
398 0x00c1 ,/* 0x0923 */
399 0x00c2 ,/* 0x0924 */
400 0x00c3 ,/* 0x0925 */
401 0x00c4 ,/* 0x0926 */
402 0x00c5 ,/* 0x0927 */
403 0x00c6 ,/* 0x0928 */
404 0x00c7 ,/* 0x0929 */
405 0x00c8 ,/* 0x092a */
406 0x00c9 ,/* 0x092b */
407 0x00ca ,/* 0x092c */
408 0x00cb ,/* 0x092d */
409 0x00cc ,/* 0x092e */
410 0x00cd ,/* 0x092f */
411 0x00cf ,/* 0x0930 */
412 0x00d0 ,/* 0x0931 */
413 0x00d1 ,/* 0x0932 */
414 0x00d2 ,/* 0x0933 */
415 0x00d3 ,/* 0x0934 */
416 0x00d4 ,/* 0x0935 */
417 0x00d5 ,/* 0x0936 */
418 0x00d6 ,/* 0x0937 */
419 0x00d7 ,/* 0x0938 */
420 0x00d8 ,/* 0x0939 */
421 0xFFFF ,/* 0x093A */
422 0xFFFF ,/* 0x093B */
423 0x00e9 ,/* 0x093c */
424 0xEAE9 ,/* 0x093d */
425 0x00da ,/* 0x093e */
426 0x00db ,/* 0x093f */
427 0x00dc ,/* 0x0940 */
428 0x00dd ,/* 0x0941 */
429 0x00de ,/* 0x0942 */
430 0x00df ,/* 0x0943 */
431 0xDFE9 ,/* 0x0944 */
432 0x00e3 ,/* 0x0945 */
433 0x00e0 ,/* 0x0946 */
434 0x00e1 ,/* 0x0947 */
435 0x00e2 ,/* 0x0948 */
436 0x00e7 ,/* 0x0949 */
437 0x00e4 ,/* 0x094a */
438 0x00e5 ,/* 0x094b */
439 0x00e6 ,/* 0x094c */
440 0x00e8 ,/* 0x094d */
441 0x00ec ,/* 0x094e */
442 0x00ed ,/* 0x094f */
443 0xA1E9 ,/* 0x0950 */ /* OM Symbol */
444 0xFFFF ,/* 0x0951 */
445 0xF0B8 ,/* 0x0952 */
446 0xFFFF ,/* 0x0953 */
447 0xFFFF ,/* 0x0954 */
448 0xFFFF ,/* 0x0955 */
449 0xFFFF ,/* 0x0956 */
450 0xFFFF ,/* 0x0957 */
451 0xb3e9 ,/* 0x0958 */
452 0xb4e9 ,/* 0x0959 */
453 0xb5e9 ,/* 0x095a */
454 0xbae9 ,/* 0x095b */
455 0xbfe9 ,/* 0x095c */
456 0xC0E9 ,/* 0x095d */
457 0xc9e9 ,/* 0x095e */
458 0x00ce ,/* 0x095f */
459 0xAAe9 ,/* 0x0960 */
460 0xA7E9 ,/* 0x0961 */
461 0xDBE9 ,/* 0x0962 */
462 0xDCE9 ,/* 0x0963 */
463 0x00ea ,/* 0x0964 */
464 0xeaea ,/* 0x0965 */
465 0x00f1 ,/* 0x0966 */
466 0x00f2 ,/* 0x0967 */
467 0x00f3 ,/* 0x0968 */
468 0x00f4 ,/* 0x0969 */
469 0x00f5 ,/* 0x096a */
470 0x00f6 ,/* 0x096b */
471 0x00f7 ,/* 0x096c */
472 0x00f8 ,/* 0x096d */
473 0x00f9 ,/* 0x096e */
474 0x00fa ,/* 0x096f */
475 0xF0BF ,/* 0x0970 */
476 0xFFFF ,/* 0x0971 */
477 0xFFFF ,/* 0x0972 */
478 0xFFFF ,/* 0x0973 */
479 0xFFFF ,/* 0x0974 */
480 0xFFFF ,/* 0x0975 */
481 0xFFFF ,/* 0x0976 */
482 0xFFFF ,/* 0x0977 */
483 0xFFFF ,/* 0x0978 */
484 0xFFFF ,/* 0x0979 */
485 0xFFFF ,/* 0x097a */
486 0xFFFF ,/* 0x097b */
487 0xFFFF ,/* 0x097c */
488 0xFFFF ,/* 0x097d */
489 0xFFFF ,/* 0x097e */
490 0xFFFF ,/* 0x097f */
491};
492static const uint16_t toUnicodeTable[256]={
493 0x0000,/* 0x00 */
494 0x0001,/* 0x01 */
495 0x0002,/* 0x02 */
496 0x0003,/* 0x03 */
497 0x0004,/* 0x04 */
498 0x0005,/* 0x05 */
499 0x0006,/* 0x06 */
500 0x0007,/* 0x07 */
501 0x0008,/* 0x08 */
502 0x0009,/* 0x09 */
503 0x000a,/* 0x0a */
504 0x000b,/* 0x0b */
505 0x000c,/* 0x0c */
506 0x000d,/* 0x0d */
507 0x000e,/* 0x0e */
508 0x000f,/* 0x0f */
509 0x0010,/* 0x10 */
510 0x0011,/* 0x11 */
511 0x0012,/* 0x12 */
512 0x0013,/* 0x13 */
513 0x0014,/* 0x14 */
514 0x0015,/* 0x15 */
515 0x0016,/* 0x16 */
516 0x0017,/* 0x17 */
517 0x0018,/* 0x18 */
518 0x0019,/* 0x19 */
519 0x001a,/* 0x1a */
520 0x001b,/* 0x1b */
521 0x001c,/* 0x1c */
522 0x001d,/* 0x1d */
523 0x001e,/* 0x1e */
524 0x001f,/* 0x1f */
525 0x0020,/* 0x20 */
526 0x0021,/* 0x21 */
527 0x0022,/* 0x22 */
528 0x0023,/* 0x23 */
529 0x0024,/* 0x24 */
530 0x0025,/* 0x25 */
531 0x0026,/* 0x26 */
532 0x0027,/* 0x27 */
533 0x0028,/* 0x28 */
534 0x0029,/* 0x29 */
535 0x002a,/* 0x2a */
536 0x002b,/* 0x2b */
537 0x002c,/* 0x2c */
538 0x002d,/* 0x2d */
539 0x002e,/* 0x2e */
540 0x002f,/* 0x2f */
541 0x0030,/* 0x30 */
542 0x0031,/* 0x31 */
543 0x0032,/* 0x32 */
544 0x0033,/* 0x33 */
545 0x0034,/* 0x34 */
546 0x0035,/* 0x35 */
547 0x0036,/* 0x36 */
548 0x0037,/* 0x37 */
549 0x0038,/* 0x38 */
550 0x0039,/* 0x39 */
551 0x003A,/* 0x3A */
552 0x003B,/* 0x3B */
553 0x003c,/* 0x3c */
554 0x003d,/* 0x3d */
555 0x003e,/* 0x3e */
556 0x003f,/* 0x3f */
557 0x0040,/* 0x40 */
558 0x0041,/* 0x41 */
559 0x0042,/* 0x42 */
560 0x0043,/* 0x43 */
561 0x0044,/* 0x44 */
562 0x0045,/* 0x45 */
563 0x0046,/* 0x46 */
564 0x0047,/* 0x47 */
565 0x0048,/* 0x48 */
566 0x0049,/* 0x49 */
567 0x004a,/* 0x4a */
568 0x004b,/* 0x4b */
569 0x004c,/* 0x4c */
570 0x004d,/* 0x4d */
571 0x004e,/* 0x4e */
572 0x004f,/* 0x4f */
573 0x0050,/* 0x50 */
574 0x0051,/* 0x51 */
575 0x0052,/* 0x52 */
576 0x0053,/* 0x53 */
577 0x0054,/* 0x54 */
578 0x0055,/* 0x55 */
579 0x0056,/* 0x56 */
580 0x0057,/* 0x57 */
581 0x0058,/* 0x58 */
582 0x0059,/* 0x59 */
583 0x005a,/* 0x5a */
584 0x005b,/* 0x5b */
585 0x005c,/* 0x5c */
586 0x005d,/* 0x5d */
587 0x005e,/* 0x5e */
588 0x005f,/* 0x5f */
589 0x0060,/* 0x60 */
590 0x0061,/* 0x61 */
591 0x0062,/* 0x62 */
592 0x0063,/* 0x63 */
593 0x0064,/* 0x64 */
594 0x0065,/* 0x65 */
595 0x0066,/* 0x66 */
596 0x0067,/* 0x67 */
597 0x0068,/* 0x68 */
598 0x0069,/* 0x69 */
599 0x006a,/* 0x6a */
600 0x006b,/* 0x6b */
601 0x006c,/* 0x6c */
602 0x006d,/* 0x6d */
603 0x006e,/* 0x6e */
604 0x006f,/* 0x6f */
605 0x0070,/* 0x70 */
606 0x0071,/* 0x71 */
607 0x0072,/* 0x72 */
608 0x0073,/* 0x73 */
609 0x0074,/* 0x74 */
610 0x0075,/* 0x75 */
611 0x0076,/* 0x76 */
612 0x0077,/* 0x77 */
613 0x0078,/* 0x78 */
614 0x0079,/* 0x79 */
615 0x007a,/* 0x7a */
616 0x007b,/* 0x7b */
617 0x007c,/* 0x7c */
618 0x007d,/* 0x7d */
619 0x007e,/* 0x7e */
620 0x007f,/* 0x7f */
621 0x0080,/* 0x80 */
622 0x0081,/* 0x81 */
623 0x0082,/* 0x82 */
624 0x0083,/* 0x83 */
625 0x0084,/* 0x84 */
626 0x0085,/* 0x85 */
627 0x0086,/* 0x86 */
628 0x0087,/* 0x87 */
629 0x0088,/* 0x88 */
630 0x0089,/* 0x89 */
631 0x008a,/* 0x8a */
632 0x008b,/* 0x8b */
633 0x008c,/* 0x8c */
634 0x008d,/* 0x8d */
635 0x008e,/* 0x8e */
636 0x008f,/* 0x8f */
637 0x0090,/* 0x90 */
638 0x0091,/* 0x91 */
639 0x0092,/* 0x92 */
640 0x0093,/* 0x93 */
641 0x0094,/* 0x94 */
642 0x0095,/* 0x95 */
643 0x0096,/* 0x96 */
644 0x0097,/* 0x97 */
645 0x0098,/* 0x98 */
646 0x0099,/* 0x99 */
647 0x009a,/* 0x9a */
648 0x009b,/* 0x9b */
649 0x009c,/* 0x9c */
650 0x009d,/* 0x9d */
651 0x009e,/* 0x9e */
652 0x009f,/* 0x9f */
653 0x00A0,/* 0xa0 */
654 0x0901,/* 0xa1 */
655 0x0902,/* 0xa2 */
656 0x0903,/* 0xa3 */
657 0x0905,/* 0xa4 */
658 0x0906,/* 0xa5 */
659 0x0907,/* 0xa6 */
660 0x0908,/* 0xa7 */
661 0x0909,/* 0xa8 */
662 0x090a,/* 0xa9 */
663 0x090b,/* 0xaa */
664 0x090e,/* 0xab */
665 0x090f,/* 0xac */
666 0x0910,/* 0xad */
667 0x090d,/* 0xae */
668 0x0912,/* 0xaf */
669 0x0913,/* 0xb0 */
670 0x0914,/* 0xb1 */
671 0x0911,/* 0xb2 */
672 0x0915,/* 0xb3 */
673 0x0916,/* 0xb4 */
674 0x0917,/* 0xb5 */
675 0x0918,/* 0xb6 */
676 0x0919,/* 0xb7 */
677 0x091a,/* 0xb8 */
678 0x091b,/* 0xb9 */
679 0x091c,/* 0xba */
680 0x091d,/* 0xbb */
681 0x091e,/* 0xbc */
682 0x091f,/* 0xbd */
683 0x0920,/* 0xbe */
684 0x0921,/* 0xbf */
685 0x0922,/* 0xc0 */
686 0x0923,/* 0xc1 */
687 0x0924,/* 0xc2 */
688 0x0925,/* 0xc3 */
689 0x0926,/* 0xc4 */
690 0x0927,/* 0xc5 */
691 0x0928,/* 0xc6 */
692 0x0929,/* 0xc7 */
693 0x092a,/* 0xc8 */
694 0x092b,/* 0xc9 */
695 0x092c,/* 0xca */
696 0x092d,/* 0xcb */
697 0x092e,/* 0xcc */
698 0x092f,/* 0xcd */
699 0x095f,/* 0xce */
700 0x0930,/* 0xcf */
701 0x0931,/* 0xd0 */
702 0x0932,/* 0xd1 */
703 0x0933,/* 0xd2 */
704 0x0934,/* 0xd3 */
705 0x0935,/* 0xd4 */
706 0x0936,/* 0xd5 */
707 0x0937,/* 0xd6 */
708 0x0938,/* 0xd7 */
709 0x0939,/* 0xd8 */
710 0x200D,/* 0xd9 */
711 0x093e,/* 0xda */
712 0x093f,/* 0xdb */
713 0x0940,/* 0xdc */
714 0x0941,/* 0xdd */
715 0x0942,/* 0xde */
716 0x0943,/* 0xdf */
717 0x0946,/* 0xe0 */
718 0x0947,/* 0xe1 */
719 0x0948,/* 0xe2 */
720 0x0945,/* 0xe3 */
721 0x094a,/* 0xe4 */
722 0x094b,/* 0xe5 */
723 0x094c,/* 0xe6 */
724 0x0949,/* 0xe7 */
725 0x094d,/* 0xe8 */
726 0x093c,/* 0xe9 */
727 0x0964,/* 0xea */
728 0xFFFF,/* 0xeb */
729 0xFFFF,/* 0xec */
730 0xFFFF,/* 0xed */
731 0xFFFF,/* 0xee */
732 0xFFFF,/* 0xef */
733 0xFFFF,/* 0xf0 */
734 0x0966,/* 0xf1 */
735 0x0967,/* 0xf2 */
736 0x0968,/* 0xf3 */
737 0x0969,/* 0xf4 */
738 0x096a,/* 0xf5 */
739 0x096b,/* 0xf6 */
740 0x096c,/* 0xf7 */
741 0x096d,/* 0xf8 */
742 0x096e,/* 0xf9 */
743 0x096f,/* 0xfa */
744 0xFFFF,/* 0xfb */
745 0xFFFF,/* 0xfc */
746 0xFFFF,/* 0xfd */
747 0xFFFF,/* 0xfe */
748 0xFFFF /* 0xff */
749};
750
751static const uint16_t nuktaSpecialCases[][2]={
752 { 16 /*length of array*/ , 0 },
753 { 0xA6 , 0x090c },
754 { 0xEA , 0x093D },
755 { 0xDF , 0x0944 },
756 { 0xA1 , 0x0950 },
757 { 0xb3 , 0x0958 },
758 { 0xb4 , 0x0959 },
759 { 0xb5 , 0x095a },
760 { 0xba , 0x095b },
761 { 0xbf , 0x095c },
762 { 0xC0 , 0x095d },
763 { 0xc9 , 0x095e },
764 { 0xAA , 0x0960 },
765 { 0xA7 , 0x0961 },
766 { 0xDB , 0x0962 },
767 { 0xDC , 0x0963 },
768};
769
770#define WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,targetByteUnit,err){ \
771 /* write the targetUniChar to target */ \
772 if(target <targetLimit){ \
773 if(targetByteUnit <= 0xFF){ \
774 *(target)++ = (uint8_t)(targetByteUnit); \
775 if(offsets){ \
776 *(offsets++) = (int32_t)(source - args->source-1); \
777 } \
778 }else{ \
779 *(target)++ = (uint8_t)(targetByteUnit>>8); \
780 if(offsets){ \
781 *(offsets++) = (int32_t)(source - args->source-1); \
782 } \
783 if(target < targetLimit){ \
784 *(target)++ = (uint8_t) targetByteUnit; \
785 if(offsets){ \
786 *(offsets++) = (int32_t)(source - args->source-1); \
787 } \
788 }else{ \
789 args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = \
790 (uint8_t) (targetByteUnit); \
791 *err = U_BUFFER_OVERFLOW_ERROR; \
792 } \
793 } \
794 }else{ \
795 if(targetByteUnit & 0xFF00){ \
796 args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = \
797 (uint8_t) (targetByteUnit >>8); \
798 } \
799 args->converter->charErrorBuffer[args->converter->charErrorBufferLength++] = \
800 (uint8_t) (targetByteUnit); \
801 *err = U_BUFFER_OVERFLOW_ERROR; \
802 } \
803}
804
805/* Rules:
806 * Explicit Halant :
807 * <HALANT> + <ZWNJ>
808 * Soft Halant :
809 * <HALANT> + <ZWJ>
810 */
811
812static void
813UConverter_fromUnicode_ISCII_OFFSETS_LOGIC (UConverterFromUnicodeArgs * args,
814 UErrorCode * err){
815 const UChar *source = args->source;
816 const UChar *sourceLimit = args->sourceLimit;
817 unsigned char *target = (unsigned char *) args->target;
818 unsigned char *targetLimit = (unsigned char *) args->targetLimit;
819 int32_t* offsets = args->offsets;
820 uint32_t targetByteUnit = 0x0000;
821 UChar32 sourceChar = 0x0000;
b75a7d8f
A		 822 UConverterDataISCII *converterData;
823 uint16_t newDelta=0;
824 uint16_t range = 0;
825 UBool deltaChanged = FALSE;
826
827 if ((args->converter == NULL) || (args->targetLimit < args->target) || (args->sourceLimit < args->source)){
828 *err = U_ILLEGAL_ARGUMENT_ERROR;
829 return;
830 }
831 /* initialize data */
832 converterData=(UConverterDataISCII*)args->converter->extraInfo;
b75a7d8f
A		 833 newDelta=converterData->currentDeltaFromUnicode;
834 range = (uint16_t)(newDelta/DELTA);
835
374ca955 836 if((sourceChar = args->converter->fromUChar32)!=0) {
b75a7d8f
A		 837 goto getTrail;
838 }
839
840 /*writing the char to the output stream */
841 while(source < sourceLimit){
842
843 targetByteUnit = missingCharMarker;
844
845 sourceChar = *source++;
846
847 /*check if input is in ASCII and C0 control codes range*/
848 if (sourceChar <= ASCII_END) {
849 WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,sourceChar,err);
850 if(U_FAILURE(*err)){
851 break;
852 }
853 if(sourceChar == LF){
854 targetByteUnit = ATR<<8;
73c04bcf 855 targetByteUnit += (uint8_t) lookupInitialData[range].isciiLang;
b75a7d8f
A		 856 args->converter->fromUnicodeStatus=sourceChar;
857 /* now append ATR and language code */
858 WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,targetByteUnit,err);
859 if(U_FAILURE(*err)){
860 break;
861 }
862 }
863 continue;
864 }
865 switch(sourceChar){
866 case ZWNJ:
867 /* contextChar has HALANT */
868 if(converterData->contextCharFromUnicode){
869 converterData->contextCharFromUnicode = 0x00;
870 targetByteUnit = ISCII_HALANT;
871 }else{
872 /* consume ZWNJ and continue */
873 converterData->contextCharFromUnicode = 0x00;
874 continue;
875 }
876 break;
877 case ZWJ:
878 /* contextChar has HALANT */
879 if(converterData->contextCharFromUnicode){
880 targetByteUnit = ISCII_NUKTA;
881 }else{
882 targetByteUnit =ISCII_INV;
883 }
884 converterData->contextCharFromUnicode = 0x00;
885 break;
886 default:
887 /* is the sourceChar in the INDIC_RANGE? */
888 if((uint16_t)(INDIC_BLOCK_END-sourceChar) <= INDIC_RANGE){
889 /* Danda and Double Danda are valid in Northern scripts.. since Unicode
890 * does not include these codepoints in all Northern scrips we need to
891 * filter them out
892 */
893 if(sourceChar!= DANDA && sourceChar != DOUBLE_DANDA){
894 /* find out to which block the souceChar belongs*/
895 range =(uint16_t)((sourceChar-INDIC_BLOCK_BEGIN)/DELTA);
896 newDelta =(uint16_t)(range*DELTA);
897
898 /* Now are we in the same block as the previous? */
899 if(newDelta!= converterData->currentDeltaFromUnicode || converterData->isFirstBuffer){
900 converterData->currentDeltaFromUnicode = newDelta;
73c04bcf 901 converterData->currentMaskFromUnicode = lookupInitialData[range].maskEnum;
b75a7d8f
A		 902 deltaChanged =TRUE;
903 converterData->isFirstBuffer=FALSE;
904 }
905 /* Normalize all Indic codepoints to Devanagari and map them to ISCII */
906 /* now subtract the new delta from sourceChar*/
907 sourceChar -= converterData->currentDeltaFromUnicode ;
908 }
909
910 /* get the target byte unit */
911 targetByteUnit=fromUnicodeTable[(uint8_t)sourceChar];
912
913 /* is the code point valid in current script? */
914 if((validityTable[(uint8_t)sourceChar] & converterData->currentMaskFromUnicode)==0){
915 /* Vocallic RR is assigne in ISCII Telugu and Unicode */
916 if(converterData->currentDeltaFromUnicode!=(TELUGU_DELTA) && sourceChar!=VOCALLIC_RR){
917 targetByteUnit=missingCharMarker;
918 }
919 }
920
921 if(deltaChanged){
922 /* we are in a script block which is different than
923 * previous sourceChar's script block write ATR and language codes
924 */
925 uint16_t temp=0;
926 temp =(uint16_t)(ATR<<8);
73c04bcf 927 temp += (uint16_t)((uint8_t) lookupInitialData[range].isciiLang);
b75a7d8f
A		 928 /* reset */
929 deltaChanged=FALSE;
930 /* now append ATR and language code */
931 WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,temp,err);
932 if(U_FAILURE(*err)){
933 break;
934 }
935 }
936 }
937 /* reset context char */
938 converterData->contextCharFromUnicode = 0x00;
939 break;
940 }
941
942
943 if(targetByteUnit != missingCharMarker){
944 if(targetByteUnit==ISCII_HALANT){
945 converterData->contextCharFromUnicode = (UChar)targetByteUnit;
946 }
947 /* write targetByteUnit to target*/
948 WRITE_TO_TARGET_FROM_U(args,offsets,source,target,targetLimit,targetByteUnit,err);
949 if(U_FAILURE(*err)){
950 break;
951 }
952 }
953 else{
374ca955 954 /* oops.. the code point is unassigned */
b75a7d8f
A		 955 /*check if the char is a First surrogate*/
956 if(UTF_IS_SURROGATE(sourceChar)) {
957 if(UTF_IS_SURROGATE_FIRST(sourceChar)) {
b75a7d8f
A		 958getTrail:
959 /*look ahead to find the trail surrogate*/
960 if(source < sourceLimit) {
961 /* test the following code unit */
962 UChar trail= (*source);
963 if(UTF_IS_SECOND_SURROGATE(trail)) {
964 source++;
374ca955 965 sourceChar=UTF16_GET_PAIR_VALUE(sourceChar, trail);
b75a7d8f
A		 966 *err =U_INVALID_CHAR_FOUND;
967 /* convert this surrogate code point */
968 /* exit this condition tree */
969 } else {
970 /* this is an unmatched lead code unit (1st surrogate) */
971 /* callback(illegal) */
b75a7d8f
A		 972 *err=U_ILLEGAL_CHAR_FOUND;
973 }
974 } else {
975 /* no more input */
976 *err = U_ZERO_ERROR;
b75a7d8f
A		 977 }
978 } else {
979 /* this is an unmatched trail code unit (2nd surrogate) */
980 /* callback(illegal) */
b75a7d8f
A		 981 *err=U_ILLEGAL_CHAR_FOUND;
982 }
374ca955
A		 983 } else {
984 /* callback(unassigned) for a BMP code point */
985 *err = U_INVALID_CHAR_FOUND;
b75a7d8f 986 }
b75a7d8f 987
374ca955
A		 988 args->converter->fromUChar32=sourceChar;
989 break;
b75a7d8f 990 }
b75a7d8f
A		 991 }/* end while(mySourceIndex<mySourceLength) */
992
b75a7d8f
A		 993 /*save the state and return */
994 args->source = source;
995 args->target = (char*)target;
996}
997
998static const int32_t lookupTable[][2]={
999 { ZERO, ZERO }, /*DEFALT*/
1000 { ZERO, ZERO }, /*ROMAN*/
1001 { DEVANAGARI, DEV_MASK },
1002 { BENGALI, BNG_MASK },
1003 { TAMIL, TML_MASK },
1004 { TELUGU, KND_MASK },
1005 { BENGALI, BNG_MASK },
1006 { ORIYA, ORI_MASK },
1007 { KANNADA, KND_MASK },
73c04bcf 1008 { MALAYALAM, MLM_MASK },
b75a7d8f 1009 { GUJARATI, GJR_MASK },
73c04bcf
A		 1010 { GURMUKHI, PNJ_MASK }
1011
b75a7d8f
A		 1012};
1013
1014#define WRITE_TO_TARGET_TO_U(args,source,target,offsets,offset,targetUniChar,delta, err){\
1015 /* add offset to current Indic Block */ \
1016 if(targetUniChar>ASCII_END && \
1017 targetUniChar != ZWJ && \
1018 targetUniChar != ZWNJ && \
1019 targetUniChar != DANDA && \
1020 targetUniChar != DOUBLE_DANDA){ \
1021 \
1022 targetUniChar+=(uint16_t)(delta); \
1023 } \
1024 /* now write the targetUniChar */ \
1025 if(target<args->targetLimit){ \
1026 *(target)++ = (UChar)targetUniChar; \
1027 if(offsets){ \
1028 *(offsets)++ = (int32_t)(offset); \
1029 } \
1030 }else{ \
1031 args->converter->UCharErrorBuffer[args->converter->UCharErrorBufferLength++] = \
1032 (UChar)targetUniChar; \
1033 *err = U_BUFFER_OVERFLOW_ERROR; \
1034 } \
1035}
1036
1037#define GET_MAPPING(sourceChar,targetUniChar,data){ \
1038 targetUniChar = toUnicodeTable[(sourceChar)] ; \
1039 /* is the code point valid in current script? */ \
1040 if(sourceChar> ASCII_END && \
1041 (validityTable[(uint8_t)targetUniChar] & data->currentMaskToUnicode)==0){ \
1042 /* Vocallic RR is assigne in ISCII Telugu and Unicode */ \
1043 if(data->currentDeltaToUnicode!=(TELUGU_DELTA) && \
1044 targetUniChar!=VOCALLIC_RR){ \
1045 targetUniChar=missingCharMarker; \
1046 } \
1047 } \
1048}
1049
1050/***********
1051 * Rules for ISCII to Unicode converter
1052 * ISCII is stateful encoding. To convert ISCII bytes to Unicode,
1053 * which has both precomposed and decomposed forms characters
1054 * pre-context and post-context need to be considered.
1055 *
1056 * Post context
1057 * i) ATR : Attribute code is used to declare the font and script switching.
1058 * Currently we only switch scripts and font codes consumed without generating an error
1059 * ii) EXT : Extention code is used to declare switching to Sanskrit and for obscure,
1060 * obsolete characters
1061 * Pre context
1062 * i) Halant: if preceeded by a halant then it is a explicit halant
1063 * ii) Nukta :
1064 * a) if preceeded by a halant then it is a soft halant
1065 * b) if preceeded by specific consonants and the ligatures have pre-composed
1066 * characters in Unicode then convert to pre-composed characters
1067 * iii) Danda: If Danda is preceeded by a Danda then convert to Double Danda
1068 *
1069 */
1070
1071static void
1072UConverter_toUnicode_ISCII_OFFSETS_LOGIC(UConverterToUnicodeArgs *args,
1073 UErrorCode* err){
1074 const char *source = ( char *) args->source;
1075 UChar *target = args->target;
1076 const char *sourceLimit = args->sourceLimit;
1077 const UChar* targetLimit = args->targetLimit;
1078 uint32_t targetUniChar = 0x0000;
1079 uint8_t sourceChar = 0x0000;
1080 UConverterDataISCII* data;
b75a7d8f
A		 1081 UChar32* toUnicodeStatus=NULL;
1082 UChar* contextCharToUnicode = NULL;
1083
1084 if ((args->converter == NULL) || (target < args->target) || (source < args->source)){
1085 *err = U_ILLEGAL_ARGUMENT_ERROR;
1086 return;
1087 }
1088
1089 data = (UConverterDataISCII*)(args->converter->extraInfo);
1090 contextCharToUnicode = &data->contextCharToUnicode; /* contains previous ISCII codepoint visited */
1091 toUnicodeStatus = (UChar32*)&args->converter->toUnicodeStatus;/* contains the mapping to Unicode of the above codepoint*/
1092
1093 while(source<sourceLimit){
1094
1095 targetUniChar = missingCharMarker;
1096
1097 if(target < targetLimit){
1098 sourceChar = (unsigned char)*(source)++;
1099
1100 /* look at the post-context preform special processing */
1101 if(*contextCharToUnicode==ATR){
1102
1103 /* If we have ATR in *contextCharToUnicode then we need to change our
1104 * state to the Indic Script specified by sourceChar
1105 */
1106
1107 /* check if the sourceChar is supported script range*/
1108 if((uint8_t)(PNJ-sourceChar)<=PNJ-DEV){
1109 data->currentDeltaToUnicode =
1110 (uint16_t)(lookupTable[sourceChar & 0x0F][0] * DELTA);
1111 data->currentMaskToUnicode =
73c04bcf 1112 (MaskEnum)lookupTable[sourceChar & 0x0F][1] ;
b75a7d8f
A		 1113 }
1114 else if(sourceChar==DEF){
1115 /* switch back to default */
1116 data->currentDeltaToUnicode = data->defDeltaToUnicode;
1117 data->currentMaskToUnicode = data->defMaskToUnicode;
1118 }else{
b75a7d8f
A		 1119 if((sourceChar >= 0x21 && sourceChar <= 0x3F)){
1120 /* these are display codes consume and continue */
1121 }else{
1122 *err =U_ILLEGAL_CHAR_FOUND;
1123 /* reset */
1124 *contextCharToUnicode=NO_CHAR_MARKER;
b75a7d8f
A		 1125 goto CALLBACK;
1126 }
b75a7d8f
A		 1127 }
1128
1129 /* reset */
1130 *contextCharToUnicode=NO_CHAR_MARKER;
1131
1132 continue;
1133
1134 }else if(*contextCharToUnicode==EXT){
1135 /* check if sourceChar is in 0xA1-0xEE range */
1136 if((uint8_t) (EXT_RANGE_END - sourceChar) <= (EXT_RANGE_END - EXT_RANGE_BEGIN)){
1137 /* We currently support only Anudatta and Devanagari abbreviation sign */
1138 if(sourceChar==0xBF || sourceChar == 0xB8){
1139 targetUniChar = (sourceChar==0xBF) ? DEV_ABBR_SIGN : DEV_ANUDATTA;
1140
1141 /* find out if the mapping is valid in this state */
1142 if(validityTable[(uint8_t)targetUniChar] & data->currentMaskToUnicode){
1143
1144 *contextCharToUnicode= NO_CHAR_MARKER;
1145
1146 /* write to target */
1147 WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),
1148 targetUniChar,data->currentDeltaToUnicode,err);
1149
1150 continue;
1151 }
1152 }
1153 /* byte unit is unassigned */
1154 targetUniChar = missingCharMarker;
1155 *err= U_INVALID_CHAR_FOUND;
b75a7d8f
A		 1156 }else{
1157 /* only 0xA1 - 0xEE are legal after EXT char */
1158 *contextCharToUnicode= NO_CHAR_MARKER;
b75a7d8f
A		 1159 *err = U_ILLEGAL_CHAR_FOUND;
1160 }
1161 goto CALLBACK;
1162 }else if(*contextCharToUnicode==ISCII_INV){
1163 if(sourceChar==ISCII_HALANT){
1164 targetUniChar = 0x0020; /* replace with space accoding to Indic FAQ */
1165 }else{
1166 targetUniChar = ZWJ;
1167 }
1168
1169 /* write to target */
1170 WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),
1171 targetUniChar,data->currentDeltaToUnicode,err);
1172 /* reset */
1173 *contextCharToUnicode=NO_CHAR_MARKER;
1174 }
1175
1176 /* look at the pre-context and perform special processing */
1177 switch(sourceChar){
1178 case ISCII_INV:
1179 case EXT: /*falls through*/
1180 case ATR:
1181 *contextCharToUnicode = (UChar)sourceChar;
1182
1183 if(*toUnicodeStatus != missingCharMarker){
1184
1185 WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),
1186 *toUnicodeStatus,data->currentDeltaToUnicode,err);
1187 *toUnicodeStatus = missingCharMarker;
1188 }
1189 continue;
1190 case ISCII_DANDA:
1191 /* handle double danda*/
1192 if(*contextCharToUnicode== ISCII_DANDA){
1193 targetUniChar = DOUBLE_DANDA;
1194 /* clear the context */
1195 *contextCharToUnicode = NO_CHAR_MARKER;
1196 *toUnicodeStatus = missingCharMarker;
1197 }else{
1198 GET_MAPPING(sourceChar,targetUniChar,data);
1199 *contextCharToUnicode = sourceChar;
1200 }
1201 break;
1202 case ISCII_HALANT:
1203 /* handle explicit halant */
1204 if(*contextCharToUnicode == ISCII_HALANT){
1205 targetUniChar = ZWNJ;
1206 /* clear the context */
1207 *contextCharToUnicode = NO_CHAR_MARKER;
1208 }else{
1209 GET_MAPPING(sourceChar,targetUniChar,data);
1210 *contextCharToUnicode = sourceChar;
1211 }
1212 break;
73c04bcf
A		 1213 case 0x0A:
1214 /* fall through */
1215 case 0x0D:
1216 data->resetToDefaultToUnicode = TRUE;
1217 GET_MAPPING(sourceChar,targetUniChar,data);
1218 *contextCharToUnicode = sourceChar;
1219 break;
1220
b75a7d8f
A		 1221 case ISCII_NUKTA:
1222 /* handle soft halant */
1223 if(*contextCharToUnicode == ISCII_HALANT){
1224 targetUniChar = ZWJ;
1225 /* clear the context */
1226 *contextCharToUnicode = NO_CHAR_MARKER;
1227 break;
1228 }else{
1229 /* try to handle <CHAR> + ISCII_NUKTA special mappings */
1230 int i=1;
1231 UBool found =FALSE;
1232 for( ;i<nuktaSpecialCases[0][0];i++){
1233 if(nuktaSpecialCases[i][0]==(uint8_t)*contextCharToUnicode){
1234 targetUniChar=nuktaSpecialCases[i][1];
1235 found =TRUE;
1236 break;
1237 }
1238 }
1239 if(found){
1240 /* find out if the mapping is valid in this state */
1241 if(validityTable[(uint8_t)targetUniChar] & data->currentMaskToUnicode){
73c04bcf 1242 /*targetUniChar += data->currentDeltaToUnicode ;*/
b75a7d8f
A		 1243 *contextCharToUnicode= NO_CHAR_MARKER;
1244 *toUnicodeStatus = missingCharMarker;
1245 break;
1246 }
1247 /* else fall through to default */
1248 }
1249 /* else fall through to default */
1250 }
1251 default:
1252 GET_MAPPING(sourceChar,targetUniChar,data);
1253 *contextCharToUnicode = sourceChar;
1254 break;
1255 }
1256
1257
1258 if(*toUnicodeStatus != missingCharMarker){
1259 /* write the previously mapped codepoint */
1260 WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source-args->source -2),
1261 *toUnicodeStatus,data->currentDeltaToUnicode,err);
1262 *toUnicodeStatus = missingCharMarker;
1263 }
1264
1265
1266 if(targetUniChar != missingCharMarker ){
1267 /* now save the targetUniChar for delayed write */
1268 *toUnicodeStatus = (UChar) targetUniChar;
73c04bcf
A		 1269 if(data->resetToDefaultToUnicode==TRUE){
1270 data->currentDeltaToUnicode = data->defDeltaToUnicode;
1271 data->currentMaskToUnicode = data->defMaskToUnicode;
1272 data->resetToDefaultToUnicode=FALSE;
1273 }
b75a7d8f
A		 1274 }else{
1275
1276 /* we reach here only if targetUniChar == missingCharMarker
1277 * so assign codes to reason and err
1278 */
b75a7d8f
A		 1279 *err = U_INVALID_CHAR_FOUND;
1280CALLBACK:
374ca955
A		 1281 args->converter->toUBytes[0] = (uint8_t) sourceChar;
1282 args->converter->toULength = 1;
1283 break;
b75a7d8f
A		 1284 }
1285
1286 }
1287 else{
1288 *err =U_BUFFER_OVERFLOW_ERROR;
1289 break;
1290 }
1291 }
374ca955
A		 1292
1293 if(U_SUCCESS(*err) && args->flush && source == sourceLimit) {
1294 /* end of the input stream */
1295 UConverter *cnv = args->converter;
1296
1297 if(*contextCharToUnicode==ATR || *contextCharToUnicode==EXT || *contextCharToUnicode==ISCII_INV){
1298 /* set toUBytes[] */
1299 cnv->toUBytes[0] = (uint8_t)*contextCharToUnicode;
1300 cnv->toULength = 1;
1301
1302 /* avoid looping on truncated sequences */
1303 *contextCharToUnicode = NO_CHAR_MARKER;
b75a7d8f 1304 }else{
374ca955
A		 1305 cnv->toULength = 0;
1306 }
1307
1308 if(*toUnicodeStatus != missingCharMarker) {
1309 /* output a remaining target character */
b75a7d8f
A		 1310 WRITE_TO_TARGET_TO_U(args,source,target,args->offsets,(source - args->source -1),
1311 *toUnicodeStatus,data->currentDeltaToUnicode,err);
374ca955 1312 *toUnicodeStatus = missingCharMarker;
b75a7d8f 1313 }
b75a7d8f 1314 }
374ca955 1315
b75a7d8f
A		 1316 args->target = target;
1317 args->source = source;
1318}
1319
1320/* structure for SafeClone calculations */
374ca955 1321struct cloneISCIIStruct
b75a7d8f
A		 1322{
1323 UConverter cnv;
1324 UConverterDataISCII mydata;
1325};
1326
1327
1328static UConverter *
1329_ISCII_SafeClone(const UConverter *cnv,
1330 void *stackBuffer,
1331 int32_t *pBufferSize,
1332 UErrorCode *status)
1333{
374ca955
A		 1334 struct cloneISCIIStruct * localClone;
1335 int32_t bufferSizeNeeded = sizeof(struct cloneISCIIStruct);
b75a7d8f
A		 1336
1337 if (U_FAILURE(*status)){
1338 return 0;
1339 }
1340
1341 if (*pBufferSize == 0){ /* 'preflighting' request - set needed size into *pBufferSize */
1342 *pBufferSize = bufferSizeNeeded;
1343 return 0;
1344 }
1345
374ca955 1346 localClone = (struct cloneISCIIStruct *)stackBuffer;
73c04bcf 1347 /* ucnv.c/ucnv_safeClone() copied the main UConverter already */
b75a7d8f
A		 1348
1349 uprv_memcpy(&localClone->mydata, cnv->extraInfo, sizeof(UConverterDataISCII));
1350 localClone->cnv.extraInfo = &localClone->mydata;
1351 localClone->cnv.isExtraLocal = TRUE;
1352
1353 return &localClone->cnv;
1354}
1355
1356static void
1357_ISCIIGetUnicodeSet(const UConverter *cnv,
73c04bcf 1358 const USetAdder *sa,
b75a7d8f
A		 1359 UConverterUnicodeSet which,
1360 UErrorCode *pErrorCode)
1361{
1362 int32_t idx, script;
1363 uint8_t mask;
1364
1365 /* Since all ISCII versions allow switching to other ISCII
1366 scripts, we add all roundtrippable characters to this set. */
374ca955 1367 sa->addRange(sa->set, 0, ASCII_END);
b75a7d8f 1368 for (script = DEVANAGARI; script <= MALAYALAM; script++) {
73c04bcf 1369 mask = (uint8_t)(lookupInitialData[script].maskEnum);
b75a7d8f
A		 1370 for (idx = 0; idx < DELTA; idx++) {
1371 if (validityTable[idx] & mask) {
374ca955 1372 sa->add(sa->set, idx + (script * DELTA) + INDIC_BLOCK_BEGIN);
b75a7d8f
A		 1373 }
1374 }
1375 }
374ca955
A		 1376 sa->add(sa->set, DANDA);
1377 sa->add(sa->set, DOUBLE_DANDA);
1378 sa->add(sa->set, ZWNJ);
1379 sa->add(sa->set, ZWJ);
b75a7d8f
A		 1380}
1381
1382static const UConverterImpl _ISCIIImpl={
1383
1384 UCNV_ISCII,
1385
1386 NULL,
1387 NULL,
1388
1389 _ISCIIOpen,
1390 _ISCIIClose,
1391 _ISCIIReset,
1392
1393 UConverter_toUnicode_ISCII_OFFSETS_LOGIC,
1394 UConverter_toUnicode_ISCII_OFFSETS_LOGIC,
1395 UConverter_fromUnicode_ISCII_OFFSETS_LOGIC,
1396 UConverter_fromUnicode_ISCII_OFFSETS_LOGIC,
1397 NULL,
1398
1399 NULL,
1400 _ISCIIgetName,
1401 NULL,
1402 _ISCII_SafeClone,
1403 _ISCIIGetUnicodeSet
1404};
1405
1406static const UConverterStaticData _ISCIIStaticData={
1407 sizeof(UConverterStaticData),
1408 "ISCII",
1409 0,
1410 UCNV_IBM,
1411 UCNV_ISCII,
1412 1,
1413 4,
1414 { 0x1a, 0, 0, 0 },
1415 0x1,
1416 FALSE,
1417 FALSE,
1418 0x0,
1419 0x0,
1420 { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 }, /* reserved */
1421
1422};
1423
1424const UConverterSharedData _ISCIIData={
1425 sizeof(UConverterSharedData),
1426 ~((uint32_t) 0),
1427 NULL,
1428 NULL,
1429 &_ISCIIStaticData,
1430 FALSE,
1431 &_ISCIIImpl,
1432 0
1433};
1434
1435#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */
mirror of ICU