ext/tidy: Refactor php_tidy_parse_string()

Girgias · Girgias · commit 4433986e406b · 2025-07-18T13:21:56.000+01:00
diff --git a/ext/tidy/tidy.c b/ext/tidy/tidy.c
@@ -775,11 +775,13 @@ static zend_result _php_tidy_apply_config_array(TidyDoc doc, const HashTable *ht
 	}
 }
 
-static zend_result php_tidy_parse_string(PHPTidyObj *obj, const char *string, uint32_t len, const char *enc)
+static zend_result php_tidy_parse_string(PHPTidyObj *obj, const zend_string *string, const char *enc)
 {
 	TidyBuffer buf;
 
-	if(enc) {
+	ZEND_ASSERT(!ZEND_SIZE_T_UINT_OVFL(ZSTR_LEN(string)));
+
+	if (enc) {
 		if (tidySetCharEncoding(obj->ptdoc->doc, enc) < 0) {
 			php_error_docref(NULL, E_WARNING, "Could not set encoding \"%s\"", enc);
 			return FAILURE;
@@ -789,9 +791,9 @@ static zend_result php_tidy_parse_string(PHPTidyObj *obj, const char *string, ui
 	obj->ptdoc->initialized = true;
 
 	tidyBufInit(&buf);
-	tidyBufAttach(&buf, (byte *) string, len);
+	tidyBufAttach(&buf, (byte *) ZSTR_VAL(string), (unsigned int) ZSTR_LEN(string));
 	if (tidyParseBuffer(obj->ptdoc->doc, &buf) < 0) {
-		php_error_docref(NULL, E_WARNING, "%s", obj->ptdoc->errbuf->bp);
+		php_error_docref(NULL, E_WARNING, "%s", (const char*) obj->ptdoc->errbuf->bp);
 		return FAILURE;
 	}
 	tidy_doc_update_properties(obj);
@@ -995,7 +997,7 @@ PHP_FUNCTION(tidy_parse_string)
 	obj = Z_TIDY_P(return_value);
 
 	if (php_tidy_apply_config(obj->ptdoc->doc, options_str, options_ht, 2) != SUCCESS
-	 || php_tidy_parse_string(obj, ZSTR_VAL(input), (uint32_t)ZSTR_LEN(input), enc) != SUCCESS) {
+	 || php_tidy_parse_string(obj, input, enc) != SUCCESS) {
 		zval_ptr_dtor(return_value);
 		RETURN_FALSE;
 	}
@@ -1063,7 +1065,7 @@ PHP_FUNCTION(tidy_parse_file)
 	obj = Z_TIDY_P(return_value);
 
 	if (php_tidy_apply_config(obj->ptdoc->doc, options_str, options_ht, 2) != SUCCESS
-	 || php_tidy_parse_string(obj, ZSTR_VAL(contents), (uint32_t)ZSTR_LEN(contents), enc) != SUCCESS) {
+	 || php_tidy_parse_string(obj, contents, enc) != SUCCESS) {
 		zval_ptr_dtor(return_value);
 		RETVAL_FALSE;
 	}
@@ -1348,7 +1350,7 @@ PHP_METHOD(tidy, __construct)
 		}
 		zend_restore_error_handling(&error_handling);
 
-		php_tidy_parse_string(obj, ZSTR_VAL(contents), (uint32_t)ZSTR_LEN(contents), enc);
+		php_tidy_parse_string(obj, contents, enc);
 
 		zend_string_release_ex(contents, 0);
 	}
@@ -1385,7 +1387,7 @@ PHP_METHOD(tidy, parseFile)
 	}
 
 	RETVAL_BOOL(php_tidy_apply_config(obj->ptdoc->doc, options_str, options_ht, 2) == SUCCESS
-				&& php_tidy_parse_string(obj, ZSTR_VAL(contents), (uint32_t)ZSTR_LEN(contents), enc) == SUCCESS);
+				&& php_tidy_parse_string(obj, contents, enc) == SUCCESS);
 
 	zend_string_release_ex(contents, 0);
 }
@@ -1413,7 +1415,7 @@ PHP_METHOD(tidy, parseString)
 	obj = Z_TIDY_P(ZEND_THIS);
 
 	RETURN_BOOL(php_tidy_apply_config(obj->ptdoc->doc, options_str, options_ht, 2) == SUCCESS
-				&& php_tidy_parse_string(obj, ZSTR_VAL(input), (uint32_t)ZSTR_LEN(input), enc) == SUCCESS);
+				&& php_tidy_parse_string(obj, input, enc) == SUCCESS);
 }
 
 

Original file line number	Diff line number	Diff line change
`@@ -775,11 +775,13 @@ static zend_result _php_tidy_apply_config_array(TidyDoc doc, const HashTable *ht`
`775`	`775`	`}`
`776`	`776`	`}`
`777`	`777`
`778`		`-static zend_result php_tidy_parse_string(PHPTidyObj obj, const char string, uint32_t len, const char *enc)`
	`778`	`+static zend_result php_tidy_parse_string(PHPTidyObj obj, const zend_string string, const char *enc)`
`779`	`779`	`{`
`780`	`780`	`TidyBuffer buf;`
`781`	`781`
`782`		`- if(enc) {`
	`782`	`+ ZEND_ASSERT(!ZEND_SIZE_T_UINT_OVFL(ZSTR_LEN(string)));`
	`783`	`+`
	`784`	`+ if (enc) {`
`783`	`785`	`if (tidySetCharEncoding(obj->ptdoc->doc, enc) < 0) {`
`784`	`786`	`php_error_docref(NULL, E_WARNING, "Could not set encoding \"%s\"", enc);`
`785`	`787`	`return FAILURE;`
`@@ -789,9 +791,9 @@ static zend_result php_tidy_parse_string(PHPTidyObj obj, const char string, ui`
`789`	`791`	`obj->ptdoc->initialized = true;`
`790`	`792`
`791`	`793`	`tidyBufInit(&buf);`
`792`		`- tidyBufAttach(&buf, (byte *) string, len);`
	`794`	`+ tidyBufAttach(&buf, (byte *) ZSTR_VAL(string), (unsigned int) ZSTR_LEN(string));`
`793`	`795`	`if (tidyParseBuffer(obj->ptdoc->doc, &buf) < 0) {`
`794`		`- php_error_docref(NULL, E_WARNING, "%s", obj->ptdoc->errbuf->bp);`
	`796`	`+ php_error_docref(NULL, E_WARNING, "%s", (const char*) obj->ptdoc->errbuf->bp);`
`795`	`797`	`return FAILURE;`
`796`	`798`	`}`
`797`	`799`	`tidy_doc_update_properties(obj);`
`@@ -995,7 +997,7 @@ PHP_FUNCTION(tidy_parse_string)`
`995`	`997`	`obj = Z_TIDY_P(return_value);`
`996`	`998`
`997`	`999`	`if (php_tidy_apply_config(obj->ptdoc->doc, options_str, options_ht, 2) != SUCCESS`
`998`		`- \|\| php_tidy_parse_string(obj, ZSTR_VAL(input), (uint32_t)ZSTR_LEN(input), enc) != SUCCESS) {`
	`1000`	`+ \|\| php_tidy_parse_string(obj, input, enc) != SUCCESS) {`
`999`	`1001`	`zval_ptr_dtor(return_value);`
`1000`	`1002`	`RETURN_FALSE;`
`1001`	`1003`	`}`
`@@ -1063,7 +1065,7 @@ PHP_FUNCTION(tidy_parse_file)`
`1063`	`1065`	`obj = Z_TIDY_P(return_value);`
`1064`	`1066`
`1065`	`1067`	`if (php_tidy_apply_config(obj->ptdoc->doc, options_str, options_ht, 2) != SUCCESS`
`1066`		`- \|\| php_tidy_parse_string(obj, ZSTR_VAL(contents), (uint32_t)ZSTR_LEN(contents), enc) != SUCCESS) {`
	`1068`	`+ \|\| php_tidy_parse_string(obj, contents, enc) != SUCCESS) {`
`1067`	`1069`	`zval_ptr_dtor(return_value);`
`1068`	`1070`	`RETVAL_FALSE;`
`1069`	`1071`	`}`
`@@ -1348,7 +1350,7 @@ PHP_METHOD(tidy, __construct)`
`1348`	`1350`	`}`
`1349`	`1351`	`zend_restore_error_handling(&error_handling);`
`1350`	`1352`
`1351`		`- php_tidy_parse_string(obj, ZSTR_VAL(contents), (uint32_t)ZSTR_LEN(contents), enc);`
	`1353`	`+ php_tidy_parse_string(obj, contents, enc);`
`1352`	`1354`
`1353`	`1355`	`zend_string_release_ex(contents, 0);`
`1354`	`1356`	`}`
`@@ -1385,7 +1387,7 @@ PHP_METHOD(tidy, parseFile)`
`1385`	`1387`	`}`
`1386`	`1388`
`1387`	`1389`	`RETVAL_BOOL(php_tidy_apply_config(obj->ptdoc->doc, options_str, options_ht, 2) == SUCCESS`
`1388`		`- && php_tidy_parse_string(obj, ZSTR_VAL(contents), (uint32_t)ZSTR_LEN(contents), enc) == SUCCESS);`
	`1390`	`+ && php_tidy_parse_string(obj, contents, enc) == SUCCESS);`
`1389`	`1391`
`1390`	`1392`	`zend_string_release_ex(contents, 0);`
`1391`	`1393`	`}`
`@@ -1413,7 +1415,7 @@ PHP_METHOD(tidy, parseString)`
`1413`	`1415`	`obj = Z_TIDY_P(ZEND_THIS);`
`1414`	`1416`
`1415`	`1417`	`RETURN_BOOL(php_tidy_apply_config(obj->ptdoc->doc, options_str, options_ht, 2) == SUCCESS`
`1416`		`- && php_tidy_parse_string(obj, ZSTR_VAL(input), (uint32_t)ZSTR_LEN(input), enc) == SUCCESS);`
	`1418`	`+ && php_tidy_parse_string(obj, input, enc) == SUCCESS);`
`1417`	`1419`	`}`
`1418`	`1420`
`1419`	`1421`