Use x'hex' syntax for null bytes in pdo::quote

Implements x'hex' encoding in pdo::quote for handling strings with null bytes,
 providing a reliable workaround for issue GH-13952.
An alternative fix is discussed in PR #13956

PR #13956 does something interesting, it avoids the overhead of copying to/from sqlite3_snprintf,
probably speeding up PDO::quote,
but right now I just want to keep things simple.

Co-authored-by: Niels Dossche <[email protected]>

Author: divinity76

ext/pdo_sqlite/sqlite_driver.c

@@ -227,20 +227,21 @@ static zend_string* sqlite_handle_quoter(pdo_dbh_t *dbh, const zend_string *unqu
 		return NULL;
 	}
 	if(ZSTR_LEN(unquoted) != 0 && memchr(ZSTR_VAL(unquoted), '\0', ZSTR_LEN(unquoted))) {
-		// x'hex'
-		zend_string *quoted = zend_string_safe_alloc(3 + (2 * ZSTR_LEN(unquoted)), 1, 0, 0);
+		// (''||x'hex')
+		// the odd (''||) thing is to make sure quote produce a sqlite datatype "string" rather than "blob" ...
+		// https://github.com/php/php-src/pull/13962/files#r1565485792
+		zend_string *quoted = zend_string_safe_alloc(9 + (2 * ZSTR_LEN(unquoted)), 1, 0, 0);
 		char *outptr = ZSTR_VAL(quoted);
 		const char *inptr = ZSTR_VAL(unquoted);
 		const char *const inendptr = inptr + ZSTR_LEN(unquoted);
-		*outptr++ = 'x';
-		*outptr++ = '\'';
+		memcpy(outptr, "(''||x'", 7);
+		outptr += 7;
 		while(inptr != inendptr) {
 			const unsigned char c = *inptr++;
 			*outptr++ = "0123456789ABCDEF"[c >> 4];
 			*outptr++ = "0123456789ABCDEF"[c & 0x0F];
 		}
-		*outptr++ = '\'';
-		*outptr = '\0'; // todo: does zend_string_safe_alloc do this for us? i don't know
+		memcpy(outptr, "')", 3); // todo: does zend_string_safe_alloc write the null terminator? if it does, reduce this to 2
 		return quoted;
 	}
 	quoted = safe_emalloc(2, ZSTR_LEN(unquoted), 3);

ext/pdo_sqlite/tests/gh13952.phpt

@@ -25,6 +25,8 @@ $test_cases = [
     "\x00foo\x00\x00\x00bar\x00",
     "\x00\x00\x00foo",
     "foo\x00\x00\x00",
+    "\x80", // << invalid UTF8
+    "\x00\x80\x00", // << invalid UTF8
 ];
 
 foreach($test_cases as $test){
@@ -38,7 +40,7 @@ $db->exec('CREATE TABLE test (name TEXT)');
 
 foreach ($test_cases as $test_case) {
     $quoted = $db->quote($test_case);
-    echo trim(json_encode($test_case), '"'), " -> $quoted\n";
+    echo trim(json_encode($test_case,  JSON_PARTIAL_OUTPUT_ON_ERROR), '"'), " -> $quoted\n";
     $db->exec("INSERT INTO test (name) VALUES (" . $quoted . ")");
 }
 
@@ -51,17 +53,19 @@ foreach ($stmt->fetchAll() as $result) {
 --EXPECTF--
 -> ''
 x -> 'x'
-\u0000 -> x'00'
-a\u0000b -> x'610062'
-\u0000\u0000\u0000 -> x'000000'
+\u0000 -> (''||x'00')
+a\u0000b -> (''||x'610062')
+\u0000\u0000\u0000 -> (''||x'000000')
 foobar -> 'foobar'
 foo'''bar -> 'foo''''''bar'
 'foo'''bar' -> '''foo''''''bar'''
-'foo'\u0000'bar' -> x'27666F6F27002762617227'
-foo\u0000\u0000\u0000bar -> x'666F6F000000626172'
-\u0000foo\u0000\u0000\u0000bar\u0000 -> x'00666F6F00000062617200'
-\u0000\u0000\u0000foo -> x'000000666F6F'
-foo\u0000\u0000\u0000 -> x'666F6F000000'
+'foo'\u0000'bar' -> (''||x'27666F6F27002762617227')
+foo\u0000\u0000\u0000bar -> (''||x'666F6F000000626172')
+\u0000foo\u0000\u0000\u0000bar\u0000 -> (''||x'00666F6F00000062617200')
+\u0000\u0000\u0000foo -> (''||x'000000666F6F')
+foo\u0000\u0000\u0000 -> (''||x'666F6F000000')
+null -> '€'
+null -> (''||x'008000')
 string(0) ""
 string(1) "x"
 string(1) "%0"
@@ -75,3 +79,5 @@ string(9) "foo%0%0%0bar"
 string(11) "%0foo%0%0%0bar%0"
 string(6) "%0%0%0foo"
 string(6) "foo%0%0%0"
+string(1) "€"
+string(3) "%0€%0"