1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/ucnv_err.c Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,406 @@
1.4 +/*
1.5 + *****************************************************************************
1.6 + *
1.7 + * Copyright (C) 1998-2007, International Business Machines
1.8 + * Corporation and others. All Rights Reserved.
1.9 + *
1.10 + *****************************************************************************
1.11 + *
1.12 + * ucnv_err.c
1.13 + * Implements error behaviour functions called by T_UConverter_{from,to}Unicode
1.14 + *
1.15 + *
1.16 +* Change history:
1.17 +*
1.18 +* 06/29/2000 helena Major rewrite of the callback APIs.
1.19 +*/
1.20 +
1.21 +#include "unicode/utypes.h"
1.22 +
1.23 +#if !UCONFIG_NO_CONVERSION
1.24 +
1.25 +#include "unicode/ucnv_err.h"
1.26 +#include "unicode/ucnv_cb.h"
1.27 +#include "ucnv_cnv.h"
1.28 +#include "cmemory.h"
1.29 +#include "unicode/ucnv.h"
1.30 +#include "ustrfmt.h"
1.31 +
1.32 +#define VALUE_STRING_LENGTH 32
1.33 +/*Magic # 32 = 4(number of char in value string) * 8(max number of bytes per char for any converter) */
1.34 +#define UNICODE_PERCENT_SIGN_CODEPOINT 0x0025
1.35 +#define UNICODE_U_CODEPOINT 0x0055
1.36 +#define UNICODE_X_CODEPOINT 0x0058
1.37 +#define UNICODE_RS_CODEPOINT 0x005C
1.38 +#define UNICODE_U_LOW_CODEPOINT 0x0075
1.39 +#define UNICODE_X_LOW_CODEPOINT 0x0078
1.40 +#define UNICODE_AMP_CODEPOINT 0x0026
1.41 +#define UNICODE_HASH_CODEPOINT 0x0023
1.42 +#define UNICODE_SEMICOLON_CODEPOINT 0x003B
1.43 +#define UNICODE_PLUS_CODEPOINT 0x002B
1.44 +#define UNICODE_LEFT_CURLY_CODEPOINT 0x007B
1.45 +#define UNICODE_RIGHT_CURLY_CODEPOINT 0x007D
1.46 +#define UNICODE_SPACE_CODEPOINT 0x0020
1.47 +#define UCNV_PRV_ESCAPE_ICU 0
1.48 +#define UCNV_PRV_ESCAPE_C 'C'
1.49 +#define UCNV_PRV_ESCAPE_XML_DEC 'D'
1.50 +#define UCNV_PRV_ESCAPE_XML_HEX 'X'
1.51 +#define UCNV_PRV_ESCAPE_JAVA 'J'
1.52 +#define UCNV_PRV_ESCAPE_UNICODE 'U'
1.53 +#define UCNV_PRV_ESCAPE_CSS2 'S'
1.54 +#define UCNV_PRV_STOP_ON_ILLEGAL 'i'
1.55 +
1.56 +/*Function Pointer STOPS at the ILLEGAL_SEQUENCE */
1.57 +U_CAPI void U_EXPORT2
1.58 +UCNV_FROM_U_CALLBACK_STOP (
1.59 + const void *context,
1.60 + UConverterFromUnicodeArgs *fromUArgs,
1.61 + const UChar* codeUnits,
1.62 + int32_t length,
1.63 + UChar32 codePoint,
1.64 + UConverterCallbackReason reason,
1.65 + UErrorCode * err)
1.66 +{
1.67 + /* the caller must have set the error code accordingly */
1.68 + return;
1.69 +}
1.70 +
1.71 +
1.72 +/*Function Pointer STOPS at the ILLEGAL_SEQUENCE */
1.73 +U_CAPI void U_EXPORT2
1.74 +UCNV_TO_U_CALLBACK_STOP (
1.75 + const void *context,
1.76 + UConverterToUnicodeArgs *toUArgs,
1.77 + const char* codePoints,
1.78 + int32_t length,
1.79 + UConverterCallbackReason reason,
1.80 + UErrorCode * err)
1.81 +{
1.82 + /* the caller must have set the error code accordingly */
1.83 + return;
1.84 +}
1.85 +
1.86 +U_CAPI void U_EXPORT2
1.87 +UCNV_FROM_U_CALLBACK_SKIP (
1.88 + const void *context,
1.89 + UConverterFromUnicodeArgs *fromUArgs,
1.90 + const UChar* codeUnits,
1.91 + int32_t length,
1.92 + UChar32 codePoint,
1.93 + UConverterCallbackReason reason,
1.94 + UErrorCode * err)
1.95 +{
1.96 + if (reason <= UCNV_IRREGULAR)
1.97 + {
1.98 + if (context == NULL || (*((char*)context) == UCNV_PRV_STOP_ON_ILLEGAL && reason == UCNV_UNASSIGNED))
1.99 + {
1.100 + *err = U_ZERO_ERROR;
1.101 + }
1.102 + /* else the caller must have set the error code accordingly. */
1.103 + }
1.104 + /* else ignore the reset, close and clone calls. */
1.105 +}
1.106 +
1.107 +U_CAPI void U_EXPORT2
1.108 +UCNV_FROM_U_CALLBACK_SUBSTITUTE (
1.109 + const void *context,
1.110 + UConverterFromUnicodeArgs *fromArgs,
1.111 + const UChar* codeUnits,
1.112 + int32_t length,
1.113 + UChar32 codePoint,
1.114 + UConverterCallbackReason reason,
1.115 + UErrorCode * err)
1.116 +{
1.117 + if (reason <= UCNV_IRREGULAR)
1.118 + {
1.119 + if (context == NULL || (*((char*)context) == UCNV_PRV_STOP_ON_ILLEGAL && reason == UCNV_UNASSIGNED))
1.120 + {
1.121 + *err = U_ZERO_ERROR;
1.122 + ucnv_cbFromUWriteSub(fromArgs, 0, err);
1.123 + }
1.124 + /* else the caller must have set the error code accordingly. */
1.125 + }
1.126 + /* else ignore the reset, close and clone calls. */
1.127 +}
1.128 +
1.129 +/*uses uprv_itou to get a unicode escape sequence of the offensive sequence,
1.130 + *uses a clean copy (resetted) of the converter, to convert that unicode
1.131 + *escape sequence to the target codepage (if conversion failure happens then
1.132 + *we revert to substituting with subchar)
1.133 + */
1.134 +U_CAPI void U_EXPORT2
1.135 +UCNV_FROM_U_CALLBACK_ESCAPE (
1.136 + const void *context,
1.137 + UConverterFromUnicodeArgs *fromArgs,
1.138 + const UChar *codeUnits,
1.139 + int32_t length,
1.140 + UChar32 codePoint,
1.141 + UConverterCallbackReason reason,
1.142 + UErrorCode * err)
1.143 +{
1.144 +
1.145 + UChar valueString[VALUE_STRING_LENGTH];
1.146 + int32_t valueStringLength = 0;
1.147 + int32_t i = 0;
1.148 +
1.149 + const UChar *myValueSource = NULL;
1.150 + UErrorCode err2 = U_ZERO_ERROR;
1.151 + UConverterFromUCallback original = NULL;
1.152 + const void *originalContext;
1.153 +
1.154 + UConverterFromUCallback ignoredCallback = NULL;
1.155 + const void *ignoredContext;
1.156 +
1.157 + if (reason > UCNV_IRREGULAR)
1.158 + {
1.159 + return;
1.160 + }
1.161 +
1.162 + ucnv_setFromUCallBack (fromArgs->converter,
1.163 + (UConverterFromUCallback) UCNV_FROM_U_CALLBACK_SUBSTITUTE,
1.164 + NULL,
1.165 + &original,
1.166 + &originalContext,
1.167 + &err2);
1.168 +
1.169 + if (U_FAILURE (err2))
1.170 + {
1.171 + *err = err2;
1.172 + return;
1.173 + }
1.174 + if(context==NULL)
1.175 + {
1.176 + while (i < length)
1.177 + {
1.178 + valueString[valueStringLength++] = (UChar) UNICODE_PERCENT_SIGN_CODEPOINT; /* adding % */
1.179 + valueString[valueStringLength++] = (UChar) UNICODE_U_CODEPOINT; /* adding U */
1.180 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[i++], 16, 4);
1.181 + }
1.182 + }
1.183 + else
1.184 + {
1.185 + switch(*((char*)context))
1.186 + {
1.187 + case UCNV_PRV_ESCAPE_JAVA:
1.188 + while (i < length)
1.189 + {
1.190 + valueString[valueStringLength++] = (UChar) UNICODE_RS_CODEPOINT; /* adding \ */
1.191 + valueString[valueStringLength++] = (UChar) UNICODE_U_LOW_CODEPOINT; /* adding u */
1.192 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[i++], 16, 4);
1.193 + }
1.194 + break;
1.195 +
1.196 + case UCNV_PRV_ESCAPE_C:
1.197 + valueString[valueStringLength++] = (UChar) UNICODE_RS_CODEPOINT; /* adding \ */
1.198 +
1.199 + if(length==2){
1.200 + valueString[valueStringLength++] = (UChar) UNICODE_U_CODEPOINT; /* adding U */
1.201 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 16, 8);
1.202 +
1.203 + }
1.204 + else{
1.205 + valueString[valueStringLength++] = (UChar) UNICODE_U_LOW_CODEPOINT; /* adding u */
1.206 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[0], 16, 4);
1.207 + }
1.208 + break;
1.209 +
1.210 + case UCNV_PRV_ESCAPE_XML_DEC:
1.211 +
1.212 + valueString[valueStringLength++] = (UChar) UNICODE_AMP_CODEPOINT; /* adding & */
1.213 + valueString[valueStringLength++] = (UChar) UNICODE_HASH_CODEPOINT; /* adding # */
1.214 + if(length==2){
1.215 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 10, 0);
1.216 + }
1.217 + else{
1.218 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[0], 10, 0);
1.219 + }
1.220 + valueString[valueStringLength++] = (UChar) UNICODE_SEMICOLON_CODEPOINT; /* adding ; */
1.221 + break;
1.222 +
1.223 + case UCNV_PRV_ESCAPE_XML_HEX:
1.224 +
1.225 + valueString[valueStringLength++] = (UChar) UNICODE_AMP_CODEPOINT; /* adding & */
1.226 + valueString[valueStringLength++] = (UChar) UNICODE_HASH_CODEPOINT; /* adding # */
1.227 + valueString[valueStringLength++] = (UChar) UNICODE_X_LOW_CODEPOINT; /* adding x */
1.228 + if(length==2){
1.229 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 16, 0);
1.230 + }
1.231 + else{
1.232 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[0], 16, 0);
1.233 + }
1.234 + valueString[valueStringLength++] = (UChar) UNICODE_SEMICOLON_CODEPOINT; /* adding ; */
1.235 + break;
1.236 +
1.237 + case UCNV_PRV_ESCAPE_UNICODE:
1.238 + valueString[valueStringLength++] = (UChar) UNICODE_LEFT_CURLY_CODEPOINT; /* adding { */
1.239 + valueString[valueStringLength++] = (UChar) UNICODE_U_CODEPOINT; /* adding U */
1.240 + valueString[valueStringLength++] = (UChar) UNICODE_PLUS_CODEPOINT; /* adding + */
1.241 + if (length == 2) {
1.242 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 16, 4);
1.243 + } else {
1.244 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[0], 16, 4);
1.245 + }
1.246 + valueString[valueStringLength++] = (UChar) UNICODE_RIGHT_CURLY_CODEPOINT; /* adding } */
1.247 + break;
1.248 +
1.249 + case UCNV_PRV_ESCAPE_CSS2:
1.250 + valueString[valueStringLength++] = (UChar) UNICODE_RS_CODEPOINT; /* adding \ */
1.251 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, codePoint, 16, 0);
1.252 + /* Always add space character, becase the next character might be whitespace,
1.253 + which would erroneously be considered the termination of the escape sequence. */
1.254 + valueString[valueStringLength++] = (UChar) UNICODE_SPACE_CODEPOINT;
1.255 + break;
1.256 +
1.257 + default:
1.258 + while (i < length)
1.259 + {
1.260 + valueString[valueStringLength++] = (UChar) UNICODE_PERCENT_SIGN_CODEPOINT; /* adding % */
1.261 + valueString[valueStringLength++] = (UChar) UNICODE_U_CODEPOINT; /* adding U */
1.262 + valueStringLength += uprv_itou (valueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint16_t)codeUnits[i++], 16, 4);
1.263 + }
1.264 + }
1.265 + }
1.266 + myValueSource = valueString;
1.267 +
1.268 + /* reset the error */
1.269 + *err = U_ZERO_ERROR;
1.270 +
1.271 + ucnv_cbFromUWriteUChars(fromArgs, &myValueSource, myValueSource+valueStringLength, 0, err);
1.272 +
1.273 + ucnv_setFromUCallBack (fromArgs->converter,
1.274 + original,
1.275 + originalContext,
1.276 + &ignoredCallback,
1.277 + &ignoredContext,
1.278 + &err2);
1.279 + if (U_FAILURE (err2))
1.280 + {
1.281 + *err = err2;
1.282 + return;
1.283 + }
1.284 +
1.285 + return;
1.286 +}
1.287 +
1.288 +
1.289 +
1.290 +U_CAPI void U_EXPORT2
1.291 +UCNV_TO_U_CALLBACK_SKIP (
1.292 + const void *context,
1.293 + UConverterToUnicodeArgs *toArgs,
1.294 + const char* codeUnits,
1.295 + int32_t length,
1.296 + UConverterCallbackReason reason,
1.297 + UErrorCode * err)
1.298 +{
1.299 + if (reason <= UCNV_IRREGULAR)
1.300 + {
1.301 + if (context == NULL || (*((char*)context) == UCNV_PRV_STOP_ON_ILLEGAL && reason == UCNV_UNASSIGNED))
1.302 + {
1.303 + *err = U_ZERO_ERROR;
1.304 + }
1.305 + /* else the caller must have set the error code accordingly. */
1.306 + }
1.307 + /* else ignore the reset, close and clone calls. */
1.308 +}
1.309 +
1.310 +U_CAPI void U_EXPORT2
1.311 +UCNV_TO_U_CALLBACK_SUBSTITUTE (
1.312 + const void *context,
1.313 + UConverterToUnicodeArgs *toArgs,
1.314 + const char* codeUnits,
1.315 + int32_t length,
1.316 + UConverterCallbackReason reason,
1.317 + UErrorCode * err)
1.318 +{
1.319 + if (reason <= UCNV_IRREGULAR)
1.320 + {
1.321 + if (context == NULL || (*((char*)context) == UCNV_PRV_STOP_ON_ILLEGAL && reason == UCNV_UNASSIGNED))
1.322 + {
1.323 + *err = U_ZERO_ERROR;
1.324 + ucnv_cbToUWriteSub(toArgs,0,err);
1.325 + }
1.326 + /* else the caller must have set the error code accordingly. */
1.327 + }
1.328 + /* else ignore the reset, close and clone calls. */
1.329 +}
1.330 +
1.331 +/*uses uprv_itou to get a unicode escape sequence of the offensive sequence,
1.332 + *and uses that as the substitution sequence
1.333 + */
1.334 +U_CAPI void U_EXPORT2
1.335 +UCNV_TO_U_CALLBACK_ESCAPE (
1.336 + const void *context,
1.337 + UConverterToUnicodeArgs *toArgs,
1.338 + const char* codeUnits,
1.339 + int32_t length,
1.340 + UConverterCallbackReason reason,
1.341 + UErrorCode * err)
1.342 +{
1.343 + UChar uniValueString[VALUE_STRING_LENGTH];
1.344 + int32_t valueStringLength = 0;
1.345 + int32_t i = 0;
1.346 +
1.347 + if (reason > UCNV_IRREGULAR)
1.348 + {
1.349 + return;
1.350 + }
1.351 +
1.352 + if(context==NULL)
1.353 + {
1.354 + while (i < length)
1.355 + {
1.356 + uniValueString[valueStringLength++] = (UChar) UNICODE_PERCENT_SIGN_CODEPOINT; /* adding % */
1.357 + uniValueString[valueStringLength++] = (UChar) UNICODE_X_CODEPOINT; /* adding X */
1.358 + valueStringLength += uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t) codeUnits[i++], 16, 2);
1.359 + }
1.360 + }
1.361 + else
1.362 + {
1.363 + switch(*((char*)context))
1.364 + {
1.365 + case UCNV_PRV_ESCAPE_XML_DEC:
1.366 + while (i < length)
1.367 + {
1.368 + uniValueString[valueStringLength++] = (UChar) UNICODE_AMP_CODEPOINT; /* adding & */
1.369 + uniValueString[valueStringLength++] = (UChar) UNICODE_HASH_CODEPOINT; /* adding # */
1.370 + valueStringLength += uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t)codeUnits[i++], 10, 0);
1.371 + uniValueString[valueStringLength++] = (UChar) UNICODE_SEMICOLON_CODEPOINT; /* adding ; */
1.372 + }
1.373 + break;
1.374 +
1.375 + case UCNV_PRV_ESCAPE_XML_HEX:
1.376 + while (i < length)
1.377 + {
1.378 + uniValueString[valueStringLength++] = (UChar) UNICODE_AMP_CODEPOINT; /* adding & */
1.379 + uniValueString[valueStringLength++] = (UChar) UNICODE_HASH_CODEPOINT; /* adding # */
1.380 + uniValueString[valueStringLength++] = (UChar) UNICODE_X_LOW_CODEPOINT; /* adding x */
1.381 + valueStringLength += uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t)codeUnits[i++], 16, 0);
1.382 + uniValueString[valueStringLength++] = (UChar) UNICODE_SEMICOLON_CODEPOINT; /* adding ; */
1.383 + }
1.384 + break;
1.385 + case UCNV_PRV_ESCAPE_C:
1.386 + while (i < length)
1.387 + {
1.388 + uniValueString[valueStringLength++] = (UChar) UNICODE_RS_CODEPOINT; /* adding \ */
1.389 + uniValueString[valueStringLength++] = (UChar) UNICODE_X_LOW_CODEPOINT; /* adding x */
1.390 + valueStringLength += uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t)codeUnits[i++], 16, 2);
1.391 + }
1.392 + break;
1.393 + default:
1.394 + while (i < length)
1.395 + {
1.396 + uniValueString[valueStringLength++] = (UChar) UNICODE_PERCENT_SIGN_CODEPOINT; /* adding % */
1.397 + uniValueString[valueStringLength++] = (UChar) UNICODE_X_CODEPOINT; /* adding X */
1.398 + uprv_itou (uniValueString + valueStringLength, VALUE_STRING_LENGTH - valueStringLength, (uint8_t) codeUnits[i++], 16, 2);
1.399 + valueStringLength += 2;
1.400 + }
1.401 + }
1.402 + }
1.403 + /* reset the error */
1.404 + *err = U_ZERO_ERROR;
1.405 +
1.406 + ucnv_cbToUWriteUChars(toArgs, uniValueString, valueStringLength, 0, err);
1.407 +}
1.408 +
1.409 +#endif
