1
0
mirror of https://github.com/sqlite/sqlite.git synced 2025-07-29 08:01:23 +03:00

Prototype implementation of the unistr() SQL function.

FossilOrigin-Name: 7cc302de05ed2a973372c05f55b048bf99af3d2590dd29f6fd0f379fb451aa0e
This commit is contained in:
drh
2025-02-22 23:18:38 +00:00
parent c071c47b51
commit 4d70dbad57
3 changed files with 137 additions and 7 deletions

View File

@ -1149,6 +1149,132 @@ void sqlite3QuoteValue(StrAccum *pStr, sqlite3_value *pValue){
}
}
/*
** Write a single UTF8 character whose value if v into the
** buffer starting at zOut. Return the number of bytes needed
** to encode that character.
*/
static int appendOneUtf8Char(char *zOut, u32 v){
if( v<0x00080 ){
zOut[0] = (u8)(v & 0xff);
return 1;
}
if( v<0x00800 ){
zOut[0] = 0xc0 + (u8)((v>>6) & 0x1f);
zOut[1] = 0x80 + (u8)(v & 0x3f);
return 2;
}
if( v<0x10000 ){
zOut[0] = 0xe0 + (u8)((v>>12) & 0x0f);
zOut[1] = 0x80 + (u8)((v>>6) & 0x3f);
zOut[2] = 0x80 + (u8)(v & 0x3f);
return 3;
}
zOut[0] = 0xf0 + (u8)((v>>18) & 0x07);
zOut[1] = 0x80 + (u8)((v>>12) & 0x3f);
zOut[2] = 0x80 + (u8)((v>>6) & 0x3f);
zOut[3] = 0x80 + (u8)(v & 0x3f);
return 4;
}
/*
** Return true if z[] begins with N hexadecimal digits, and write
** a decoding of those digits into *pVal. Or return false if any
** one of the first N characters in z[] is not a hexadecimal digit.
*/
static int isNHex(const char *z, int N, u32 *pVal){
int i;
int v = 0;
for(i=0; i<N; i++){
if( !sqlite3Isxdigit(z[i]) ) return 0;
v = (v<<4) + sqlite3HexToInt(z[i]);
}
*pVal = v;
return 1;
}
/*
** Implementation of the UNISTR() function.
**
** This is intended to be a work-alike of the UNISTR() function in
** PostgreSQL. Quoting from the PG documentation (PostgreSQL 17 -
** scraped on 2025-02-22):
**
** Evaluate escaped Unicode characters in the argument. Unicode
** characters can be specified as \XXXX (4 hexadecimal digits),
** \+XXXXXX (6 hexadecimal digits), \uXXXX (4 hexadecimal digits),
** or \UXXXXXXXX (8 hexadecimal digits). To specify a backslash,
** write two backslashes. All other characters are taken literally.
*/
static void unistrFunc(
sqlite3_context *context,
int argc,
sqlite3_value **argv
){
char *zOut;
const char *zIn;
int nIn;
int i, j, n;
u32 v;
assert( argc==1 );
zIn = (const char*)sqlite3_value_text(argv[0]);
if( zIn==0 ) return;
nIn = sqlite3_value_bytes(argv[0]);
zOut = sqlite3_malloc64(nIn+1);
if( zOut==0 ){
sqlite3_result_error_nomem(context);
return;
}
i = j = 0;
while( i<nIn ){
char *z = strchr(&zIn[i],'\\');
if( z==0 ){
n = nIn - i;
memmove(&zOut[j], &zIn[i], n);
j += n;
break;
}
n = z - &zIn[i];
if( n>0 ){
memmove(&zOut[j], &zIn[i], n);
j += n;
i += n;
}
if( zIn[i+1]=='\\' ){
i += 2;
zOut[j++] = '\\';
}else if( sqlite3Isxdigit(zIn[i+1]) ){
if( !isNHex(&zIn[i+1], 4, &v) ) goto unistr_error;
i += 5;
j += appendOneUtf8Char(&zOut[j], v);
}else if( zIn[i+1]=='+' ){
if( !isNHex(&zIn[i+2], 6, &v) ) goto unistr_error;
i += 8;
j += appendOneUtf8Char(&zOut[j], v);
}else if( zIn[i+1]=='u' ){
if( !isNHex(&zIn[i+2], 4, &v) ) goto unistr_error;
i += 6;
j += appendOneUtf8Char(&zOut[j], v);
}else if( zIn[i+1]=='U' ){
if( !isNHex(&zIn[i+2], 8, &v) ) goto unistr_error;
i += 10;
j += appendOneUtf8Char(&zOut[j], v);
}else{
goto unistr_error;
}
}
zOut[j] = 0;
sqlite3_result_text64(context, zOut, j, sqlite3_free, SQLITE_UTF8);
return;
unistr_error:
sqlite3_free(zOut);
sqlite3_result_error(context, "invalid Unicode escape", -1);
return;
}
/*
** Implementation of the QUOTE() function.
**
@ -2736,6 +2862,7 @@ void sqlite3RegisterBuiltinFunctions(void){
DFUNCTION(sqlite_version, 0, 0, 0, versionFunc ),
DFUNCTION(sqlite_source_id, 0, 0, 0, sourceidFunc ),
FUNCTION(sqlite_log, 2, 0, 0, errlogFunc ),
FUNCTION(unistr, 1, 0, 0, unistrFunc ),
FUNCTION(quote, 1, 0, 0, quoteFunc ),
VFUNCTION(last_insert_rowid, 0, 0, 0, last_insert_rowid),
VFUNCTION(changes, 0, 0, 0, changes ),