/*
** Return UTF-16 encoded English language explanation of the most recent
** error.
*/
const void *sqlite3_errmsg16(sqlite3 *db){
/* Because all the characters in the string are in the unicode
** range 0x00-0xFF, if we pad the big-endian string with a
** zero byte, we can obtain the little-endian string with
** &big_endian[1].
*/
static const char outOfMemBe[] = {
0, 'o', 0, 'u', 0, 't', 0, ' ',
0, 'o', 0, 'f', 0, ' ',
0, 'm', 0, 'e', 0, 'm', 0, 'o', 0, 'r', 0, 'y', 0, 0, 0
};
static const char misuseBe [] = {
0, 'l', 0, 'i', 0, 'b', 0, 'r', 0, 'a', 0, 'r', 0, 'y', 0, ' ',
0, 'r', 0, 'o', 0, 'u', 0, 't', 0, 'i', 0, 'n', 0, 'e', 0, ' ',
0, 'c', 0, 'a', 0, 'l', 0, 'l', 0, 'e', 0, 'd', 0, ' ',
0, 'o', 0, 'u', 0, 't', 0, ' ',
0, 'o', 0, 'f', 0, ' ',
0, 's', 0, 'e', 0, 'q', 0, 'u', 0, 'e', 0, 'n', 0, 'c', 0, 'e', 0, 0, 0
};
const void *z;
if( sqlite3_malloc_failed ){
return (void *)(&outOfMemBe[SQLITE_UTF16NATIVE==SQLITE_UTF16LE?1:0]);
}
if( sqlite3SafetyCheck(db) || db->errCode==SQLITE_MISUSE ){
return (void *)(&misuseBe[SQLITE_UTF16NATIVE==SQLITE_UTF16LE?1:0]);
}
z = sqlite3_value_text16(db->pErr);
if( z==0 ){
sqlite3ValueSetStr(db->pErr, -1, sqlite3ErrStr(db->errCode),
SQLITE_UTF8, SQLITE_STATIC);
z = sqlite3_value_text16(db->pErr);
}
return z;
}
/*
** Implementation of SQLite REGEXP operator. This scalar function takes
** two arguments. The first is a regular expression pattern to compile
** the second is a string to match against that pattern. If either
** argument is an SQL NULL, then NULL Is returned. Otherwise, the result
** is 1 if the string matches the pattern, or 0 otherwise.
**
** SQLite maps the regexp() function to the regexp() operator such
** that the following two are equivalent:
**
** zString REGEXP zPattern
** regexp(zPattern, zString)
**
** Uses the following ICU regexp APIs:
**
** uregex_open()
** uregex_matches()
** uregex_close()
*/
static void icuRegexpFunc(sqlite3_context *p, int nArg, sqlite3_value **apArg){
UErrorCode status = U_ZERO_ERROR;
URegularExpression *pExpr;
UBool res;
const UChar *zString = sqlite3_value_text16(apArg[1]);
(void)nArg; /* Unused parameter */
/* If the left hand side of the regexp operator is NULL,
** then the result is also NULL.
*/
if( !zString ){
return;
}
pExpr = sqlite3_get_auxdata(p, 0);
if( !pExpr ){
const UChar *zPattern = sqlite3_value_text16(apArg[0]);
if( !zPattern ){
return;
}
pExpr = uregex_open(zPattern, -1, 0, 0, &status);
if( U_SUCCESS(status) ){
sqlite3_set_auxdata(p, 0, pExpr, icuRegexpDelete);
}else{
assert(!pExpr);
icuFunctionError(p, "uregex_open", status);
return;
}
}
/* Configure the text that the regular expression operates on. */
uregex_setText(pExpr, zString, -1, &status);
if( !U_SUCCESS(status) ){
icuFunctionError(p, "uregex_setText", status);
return;
}
/* Attempt the match */
res = uregex_matches(pExpr, 0, &status);
if( !U_SUCCESS(status) ){
icuFunctionError(p, "uregex_matches", status);
return;
}
/* Set the text that the regular expression operates on to a NULL
** pointer. This is not really necessary, but it is tidier than
** leaving the regular expression object configured with an invalid
** pointer after this function returns.
*/
uregex_setText(pExpr, 0, 0, &status);
/* Return 1 or 0. */
sqlite3_result_int(p, res ? 1 : 0);
}
/*
** This function takes two arguments. It performance UTF-8/16 type
** conversions on the first argument then returns a copy of the second
** argument.
**
** This function is used in cases such as the following:
**
** SELECT test_isolation(x,x) FROM t1;
**
** We want to verify that the type conversions that occur on the
** first argument do not invalidate the second argument.
*/
static void test_isolation(
sqlite3_context *pCtx,
int nArg,
sqlite3_value **argv
){
#ifndef SQLITE_OMIT_UTF16
sqlite3_value_text16(argv[0]);
sqlite3_value_text(argv[0]);
sqlite3_value_text16(argv[0]);
sqlite3_value_text(argv[0]);
#endif
sqlite3_result_value(pCtx, argv[1]);
}
SQLValue SQLiteStatement::getColumnValue(int col)
{
ASSERT(col >= 0);
if (!m_statement)
if (prepareAndStep() != SQLITE_ROW)
return SQLValue();
if (columnCount() <= col)
return SQLValue();
// SQLite is typed per value. optional column types are
// "(mostly) ignored"
sqlite3_value* value = sqlite3_column_value(m_statement, col);
switch (sqlite3_value_type(value)) {
case SQLITE_INTEGER: // SQLValue and JS don't represent integers, so use FLOAT -case
case SQLITE_FLOAT:
return SQLValue(sqlite3_value_double(value));
case SQLITE_BLOB: // SQLValue and JS don't represent blobs, so use TEXT -case
case SQLITE_TEXT: {
const UChar* string = reinterpret_cast<const UChar*>(sqlite3_value_text16(value));
unsigned length = WTF::lengthOfNullTerminatedString(string);
return SQLValue(StringImpl::create8BitIfPossible(string, length));
}
case SQLITE_NULL:
return SQLValue();
default:
break;
}
ASSERT_NOT_REACHED();
return SQLValue();
}
/*
** Implementations of scalar functions for case mapping - upper() and
** lower(). Function upper() converts its input to upper-case (ABC).
** Function lower() converts to lower-case (abc).
**
** ICU provides two types of case mapping, "general" case mapping and
** "language specific". Refer to ICU documentation for the differences
** between the two.
**
** To utilise "general" case mapping, the upper() or lower() scalar
** functions are invoked with one argument:
**
** upper('ABC') -> 'abc'
** lower('abc') -> 'ABC'
**
** To access ICU "language specific" case mapping, upper() or lower()
** should be invoked with two arguments. The second argument is the name
** of the locale to use. Passing an empty string ("") or SQL NULL value
** as the second argument is the same as invoking the 1 argument version
** of upper() or lower().
**
** lower('I', 'en_us') -> 'i'
** lower('I', 'tr_tr') -> '\u131' (small dotless i)
**
** http://www.icu-project.org/userguide/posix.html#case_mappings
*/
static void icuCaseFunc16(sqlite3_context *p, int nArg, sqlite3_value **apArg){
const UChar *zInput; /* Pointer to input string */
UChar *zOutput = 0; /* Pointer to output buffer */
int nInput; /* Size of utf-16 input string in bytes */
int nOut; /* Size of output buffer in bytes */
int cnt;
int bToUpper; /* True for toupper(), false for tolower() */
UErrorCode status;
const char *zLocale = 0;
assert(nArg==1 || nArg==2);
bToUpper = (sqlite3_user_data(p)!=0);
if( nArg==2 ){
zLocale = (const char *)sqlite3_value_text(apArg[1]);
}
zInput = sqlite3_value_text16(apArg[0]);
if( !zInput ){
return;
}
nOut = nInput = sqlite3_value_bytes16(apArg[0]);
if( nOut==0 ){
sqlite3_result_text16(p, "", 0, SQLITE_STATIC);
return;
}
for(cnt=0; cnt<2; cnt++){
UChar *zNew = sqlite3_realloc(zOutput, nOut);
if( zNew==0 ){
sqlite3_free(zOutput);
sqlite3_result_error_nomem(p);
return;
}
zOutput = zNew;
status = U_ZERO_ERROR;
if( bToUpper ){
nOut = 2*u_strToUpper(zOutput,nOut/2,zInput,nInput/2,zLocale,&status);
}else{
nOut = 2*u_strToLower(zOutput,nOut/2,zInput,nInput/2,zLocale,&status);
}
if( U_SUCCESS(status) ){
sqlite3_result_text16(p, zOutput, nOut, xFree);
}else if( status==U_BUFFER_OVERFLOW_ERROR ){
assert( cnt==0 );
continue;
}else{
icuFunctionError(p, bToUpper ? "u_strToUpper" : "u_strToLower", status);
}
return;
}
assert( 0 ); /* Unreachable */
}
ikptr
ik_sqlite3_value_text16 (ikptr s_value, ikpcb * pcb)
{
#ifdef HAVE_SQLITE3_VALUE_TEXT16
sqlite3_value * value = IK_SQLITE_VALUE(s_value);
const void * data;
int bytes;
bytes = sqlite3_value_bytes16(value);
data = sqlite3_value_text16(value);
return ika_bytevector_from_memory_block(pcb, data, bytes);
#else
feature_failure(__func__);
#endif
}
static void getUnicodeResult(const RtlFieldInfo *field, sqlite3_value *val, size32_t &chars, UChar * &result)
{
assertex(val);
if (isNull(val))
{
NullFieldProcessor p(field);
rtlUnicodeToUnicodeX(chars, result, p.resultChars, p.unicodeResult);
return;
}
if (sqlite3_value_type(val) != SQLITE_TEXT)
typeError("string", field);
const UChar *text = (const UChar *) sqlite3_value_text16(val);
int bytes = sqlite3_value_bytes16(val);
unsigned numchars = bytes / sizeof(UChar);
rtlUnicodeToUnicodeX(chars, result, numchars, text);
}
/*
** UTF-16 implementation of the substr()
*/
void sqlite3utf16Substr(
sqlite3_context *context,
int argc,
sqlite3_value **argv
){
int y, z;
unsigned char const *zStr;
unsigned char const *zStrEnd;
unsigned char const *zStart;
unsigned char const *zEnd;
int i;
zStr = (unsigned char const *)sqlite3_value_text16(argv[0]);
zStrEnd = &zStr[sqlite3_value_bytes16(argv[0])];
y = sqlite3_value_int(argv[1]);
z = sqlite3_value_int(argv[2]);
if( y>0 ){
y = y-1;
zStart = zStr;
if( SQLITE_UTF16BE==SQLITE_UTF16NATIVE ){
for(i=0; i<y && zStart<zStrEnd; i++) SKIP_UTF16BE(zStart);
}else{
for(i=0; i<y && zStart<zStrEnd; i++) SKIP_UTF16LE(zStart);
}
}else{
zStart = zStrEnd;
if( SQLITE_UTF16BE==SQLITE_UTF16NATIVE ){
for(i=y; i<0 && zStart>zStr; i++) RSKIP_UTF16BE(zStart);
}else{
for(i=y; i<0 && zStart>zStr; i++) RSKIP_UTF16LE(zStart);
}
for(; i<0; i++) z -= 1;
}
zEnd = zStart;
if( SQLITE_UTF16BE==SQLITE_UTF16NATIVE ){
for(i=0; i<z && zEnd<zStrEnd; i++) SKIP_UTF16BE(zEnd);
}else{
for(i=0; i<z && zEnd<zStrEnd; i++) SKIP_UTF16LE(zEnd);
}
sqlite3_result_text16(context, zStart, zEnd-zStart, SQLITE_TRANSIENT);
}
// Called each time a custom function is evaluated.
static void sqliteCustomFunctionCallback(sqlite3_context *context,
int argc, sqlite3_value **argv) {
JNIEnv* env = AndroidRuntime::getJNIEnv();
// Get the callback function object.
// Create a new local reference to it in case the callback tries to do something
// dumb like unregister the function (thereby destroying the global ref) while it is running.
jobject functionObjGlobal = reinterpret_cast<jobject>(sqlite3_user_data(context));
jobject functionObj = env->NewLocalRef(functionObjGlobal);
jobjectArray argsArray = env->NewObjectArray(argc, gStringClassInfo.clazz, NULL);
if (argsArray) {
for (int i = 0; i < argc; i++) {
const jchar* arg = static_cast<const jchar*>(sqlite3_value_text16(argv[i]));
if (!arg) {
ALOGW("NULL argument in custom_function_callback. This should not happen.");
} else {
size_t argLen = sqlite3_value_bytes16(argv[i]) / sizeof(jchar);
jstring argStr = env->NewString(arg, argLen);
if (!argStr) {
goto error; // out of memory error
}
env->SetObjectArrayElement(argsArray, i, argStr);
env->DeleteLocalRef(argStr);
}
}
// TODO: Support functions that return values.
env->CallVoidMethod(functionObj,
gSQLiteCustomFunctionClassInfo.dispatchCallback, argsArray);
error:
env->DeleteLocalRef(argsArray);
}
env->DeleteLocalRef(functionObj);
if (env->ExceptionCheck()) {
ALOGE("An exception was thrown by custom SQLite function.");
LOGE_EX(env);
env->ExceptionClear();
}
}
/*
** Implementations of scalar functions for case mapping - upper() and
** lower(). Function upper() converts its input to upper-case (ABC).
** Function lower() converts to lower-case (abc).
**
** ICU provides two types of case mapping, "general" case mapping and
** "language specific". Refer to ICU documentation for the differences
** between the two.
**
** To utilise "general" case mapping, the upper() or lower() scalar
** functions are invoked with one argument:
**
** upper('ABC') -> 'abc'
** lower('abc') -> 'ABC'
**
** To access ICU "language specific" case mapping, upper() or lower()
** should be invoked with two arguments. The second argument is the name
** of the locale to use. Passing an empty string ("") or SQL NULL value
** as the second argument is the same as invoking the 1 argument version
** of upper() or lower().
**
** lower('I', 'en_us') -> 'i'
** lower('I', 'tr_tr') -> 'ı' (small dotless i)
**
** http://www.icu-project.org/userguide/posix.html#case_mappings
*/
static void icuCaseFunc16(sqlite3_context *p, int nArg, sqlite3_value **apArg){
const UChar *zInput;
UChar *zOutput;
int nInput;
int nOutput;
UErrorCode status = U_ZERO_ERROR;
const char *zLocale = 0;
assert(nArg==1 || nArg==2);
if( nArg==2 ){
zLocale = (const char *)sqlite3_value_text(apArg[1]);
}
zInput = sqlite3_value_text16(apArg[0]);
if( !zInput ){
return;
}
nInput = sqlite3_value_bytes16(apArg[0]);
nOutput = nInput * 2 + 2;
zOutput = sqlite3_malloc(nOutput);
if( !zOutput ){
return;
}
if( sqlite3_user_data(p) ){
u_strToUpper(zOutput, nOutput/2, zInput, nInput/2, zLocale, &status);
}else{
u_strToLower(zOutput, nOutput/2, zInput, nInput/2, zLocale, &status);
}
if( !U_SUCCESS(status) ){
icuFunctionError(p, "u_strToLower()/u_strToUpper", status);
return;
}
sqlite3_result_text16(p, zOutput, -1, xFree);
}
static void test_destructor16(
sqlite3_context *pCtx,
int nArg,
sqlite3_value **argv
){
char *zVal;
int len;
test_destructor_count_var++;
assert( nArg==1 );
if( sqlite3_value_type(argv[0])==SQLITE_NULL ) return;
len = sqlite3_value_bytes16(argv[0]);
zVal = testContextMalloc(pCtx, len+3);
if( !zVal ){
return;
}
zVal[len+1] = 0;
zVal[len+2] = 0;
zVal++;
memcpy(zVal, sqlite3_value_text16(argv[0]), len);
sqlite3_result_text16(pCtx, zVal, -1, destructor);
}
const void *sqlite3_column_text16(sqlite3_stmt *pStmt, int i){
const void *val = sqlite3_value_text16( columnMem(pStmt,i) );
columnMallocFailure(pStmt);
return val;
}
/**
* This function is invoked as:
*
* _TOKENIZE('<token_table>', <data_row_id>, <data>, <delimiter>,
* <use_token_index>, <data_tag>)
*
* If <use_token_index> is omitted, it is treated as 0.
* If <data_tag> is omitted, it is treated as NULL.
*
* It will split <data> on each instance of <delimiter> and insert each token
* into <token_table>. The following columns in <token_table> are used:
* token TEXT, source INTEGER, token_index INTEGER, tag (any type)
* The token_index column is not required if <use_token_index> is 0.
* The tag column is not required if <data_tag> is NULL.
*
* One row is inserted for each token in <data>.
* In each inserted row, 'source' is <data_row_id>.
* In the first inserted row, 'token' is the hex collation key of
* the entire <data> string, and 'token_index' is 0.
* In each row I (where 1 <= I < N, and N is the number of tokens in <data>)
* 'token' will be set to the hex collation key of the I:th token (0-based).
* If <use_token_index> != 0, 'token_index' is set to I.
* If <data_tag> is not NULL, 'tag' is set to <data_tag>.
*
* In other words, there will be one row for the entire string,
* and one row for each token except the first one.
*
* The function returns the number of tokens generated.
*/
static void tokenize(sqlite3_context * context, int argc, sqlite3_value ** argv)
{
//ALOGD("enter tokenize");
int err;
int useTokenIndex = 0;
int useDataTag = 0;
if (!(argc >= 4 || argc <= 6)) {
ALOGE("Tokenize requires 4 to 6 arguments");
sqlite3_result_null(context);
return;
}
if (argc > 4) {
useTokenIndex = sqlite3_value_int(argv[4]);
}
if (argc > 5) {
useDataTag = (sqlite3_value_type(argv[5]) != SQLITE_NULL);
}
sqlite3 * handle = sqlite3_context_db_handle(context);
UCollator* collator = (UCollator*)sqlite3_user_data(context);
char const * tokenTable = (char const *)sqlite3_value_text(argv[0]);
if (tokenTable == NULL) {
ALOGE("tokenTable null");
sqlite3_result_null(context);
return;
}
// Get or create the prepared statement for the insertions
sqlite3_stmt * statement = (sqlite3_stmt *)sqlite3_get_auxdata(context, 0);
if (!statement) {
char const * tokenIndexCol = useTokenIndex ? ", token_index" : "";
char const * tokenIndexParam = useTokenIndex ? ", ?" : "";
char const * dataTagCol = useDataTag ? ", tag" : "";
char const * dataTagParam = useDataTag ? ", ?" : "";
char * sql = sqlite3_mprintf("INSERT INTO %s (token, source%s%s) VALUES (?, ?%s%s);",
tokenTable, tokenIndexCol, dataTagCol, tokenIndexParam, dataTagParam);
err = sqlite3_prepare_v2(handle, sql, -1, &statement, NULL);
sqlite3_free(sql);
if (err) {
ALOGE("prepare failed");
sqlite3_result_null(context);
return;
}
// This binds the statement to the table it was compiled against, which is argv[0].
// If this function is ever called with a different table the finalizer will be called
// and sqlite3_get_auxdata() will return null above, forcing a recompile for the new table.
sqlite3_set_auxdata(context, 0, statement, tokenize_auxdata_delete);
} else {
// Reset the cached statement so that binding the row ID will work properly
sqlite3_reset(statement);
}
// Bind the row ID of the source row
int64_t rowID = sqlite3_value_int64(argv[1]);
err = sqlite3_bind_int64(statement, 2, rowID);
if (err != SQLITE_OK) {
ALOGE("bind failed");
sqlite3_result_null(context);
return;
}
// Bind <data_tag> to the tag column
if (useDataTag) {
int dataTagParamIndex = useTokenIndex ? 4 : 3;
err = sqlite3_bind_value(statement, dataTagParamIndex, argv[5]);
if (err != SQLITE_OK) {
ALOGE("bind failed");
sqlite3_result_null(context);
return;
}
}
// Get the raw bytes for the string to tokenize
// the string will be modified by following code
// however, sqlite did not reuse the string, so it is safe to not dup it
UChar * origData = (UChar *)sqlite3_value_text16(argv[2]);
if (origData == NULL) {
sqlite3_result_null(context);
return;
}
// Get the raw bytes for the delimiter
const UChar * delim = (const UChar *)sqlite3_value_text16(argv[3]);
if (delim == NULL) {
ALOGE("can't get delimiter");
sqlite3_result_null(context);
return;
}
UChar * token = NULL;
UChar *state;
int numTokens = 0;
do {
if (numTokens == 0) {
token = origData;
}
// Reset the program so we can use it to perform the insert
sqlite3_reset(statement);
UErrorCode status = U_ZERO_ERROR;
char keybuf[1024];
uint32_t result = ucol_getSortKey(collator, token, -1, (uint8_t*)keybuf, sizeof(keybuf)-1);
if (result > sizeof(keybuf)) {
// TODO allocate memory for this super big string
ALOGE("ucol_getSortKey needs bigger buffer %d", result);
break;
}
uint32_t keysize = result-1;
uint32_t base16Size = keysize*2;
char *base16buf = (char*)malloc(base16Size);
base16Encode(base16buf, keybuf, keysize);
err = sqlite3_bind_text(statement, 1, base16buf, base16Size, SQLITE_STATIC);
if (err != SQLITE_OK) {
ALOGE(" sqlite3_bind_text16 error %d", err);
free(base16buf);
break;
}
if (useTokenIndex) {
err = sqlite3_bind_int(statement, 3, numTokens);
if (err != SQLITE_OK) {
ALOGE(" sqlite3_bind_int error %d", err);
free(base16buf);
break;
}
}
err = sqlite3_step(statement);
free(base16buf);
if (err != SQLITE_DONE) {
ALOGE(" sqlite3_step error %d", err);
break;
}
numTokens++;
if (numTokens == 1) {
// first call
u_strtok_r(origData, delim, &state);
}
} while ((token = u_strtok_r(NULL, delim, &state)) != NULL);
sqlite3_result_int(context, numTokens);
}
__declspec(dllexport) const void * WINAPI sqlite3_value_text16_interop(sqlite3_value *val, int *plen)
{
const void *pval = sqlite3_value_text16(val);
*plen = (pval != 0) ? wcslen((wchar_t *)pval) * sizeof(wchar_t) : 0;
return pval;
}
const void *sqlite3_column_text16(sqlite3_stmt *pStmt, int i){
return sqlite3_value_text16( columnMem(pStmt,i) );
}
static void mmenc_func(sqlite3_context *db, int argc, sqlite3_value **argv)
{
mm_cipher_context_t *ctx;
const UChar *src;
int32_t src_len;
char buf[1024];
char *dst = buf;
int32_t dst_len;
UErrorCode status = U_ZERO_ERROR;
int arg_type;
// only accept 1 argument.
if (argc != 1)
goto error_misuse;
// encoding BLOB data type is not supported.
arg_type = sqlite3_value_type(argv[0]);
if (arg_type == SQLITE_BLOB)
goto error_misuse;
// for data types other than TEXT, just return them.
if (arg_type != SQLITE_TEXT) {
sqlite3_result_value(db, argv[0]);
return;
}
ctx = (mm_cipher_context_t *) sqlite3_user_data(db);
src_len = sqlite3_value_bytes16(argv[0]) / 2;
src = (const UChar *) sqlite3_value_text16(argv[0]);
// transform input string to BOCU-1 encoding.
// try stack buffer first, if it doesn't fit, malloc a new buffer.
dst_len =
ucnv_fromUChars(ctx->cnv, dst, sizeof(buf), src, src_len, &status);
if (status == U_BUFFER_OVERFLOW_ERROR) {
status = U_ZERO_ERROR;
dst = (char *) sqlite3_malloc(dst_len);
dst_len =
ucnv_fromUChars(ctx->cnv, dst, dst_len, src, src_len, &status);
}
if (U_FAILURE(status) && status != U_STRING_NOT_TERMINATED_WARNING) {
sqlite3_mm_set_last_error(
"Failed transforming text to internal encoding.");
goto error_error;
}
// encrypt transformed BOCU-1 string.
do_rc4(ctx, dst, dst_len);
// return
sqlite3_result_blob(db, dst, dst_len, SQLITE_TRANSIENT);
if (dst != buf)
sqlite3_free(dst);
return;
error_error:
if (dst != buf)
sqlite3_free(dst);
sqlite3_result_error_code(db, SQLITE_ERROR);
return;
error_misuse:
if (dst != buf)
sqlite3_free(dst);
sqlite3_result_error_code(db, SQLITE_MISUSE);
return;
}
