better position calculation

juan-morales · juan-morales · commit f4f5e08b61e7 · 2025-06-16T18:25:31.000-03:00
diff --git a/ext/json/json_parser.y b/ext/json/json_parser.y
@@ -35,7 +35,7 @@ int json_yydebug = 1;
 #define PHP_JSON_DEPTH_INC \
 	if (parser->max_depth && parser->depth >= parser->max_depth) { \
 		parser->scanner.errcode = PHP_JSON_ERROR_DEPTH; \
-		parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start); \
+		parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count); \
 		YYERROR; \
 	} \
 	++parser->depth
@@ -109,7 +109,7 @@ object_end:
 	|	']'
 			{
 				parser->scanner.errcode = PHP_JSON_ERROR_STATE_MISMATCH;
-				parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start); \
+				parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count); \
 				YYERROR;
 			}
 ;
@@ -166,7 +166,7 @@ array_end:
 	|	'}'
 			{
 				parser->scanner.errcode = PHP_JSON_ERROR_STATE_MISMATCH;
-				parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start); \
+				parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count); \
 				YYERROR;
 			}
 ;
@@ -245,7 +245,7 @@ static int php_json_parser_object_update(php_json_parser *parser, zval *object,
 	} else {
 		if (ZSTR_LEN(key) > 0 && ZSTR_VAL(key)[0] == '\0') {
 			parser->scanner.errcode = PHP_JSON_ERROR_INVALID_PROPERTY_NAME;
-			parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start); \
+			parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count); \
 			zend_string_release_ex(key, 0);
 			zval_ptr_dtor_nogc(zvalue);
 			zval_ptr_dtor_nogc(object);
@@ -304,7 +304,7 @@ static void php_json_yyerror(php_json_parser *parser, char const *msg)
 {
 	if (!parser->scanner.errcode) {
 		parser->scanner.errcode = PHP_JSON_ERROR_SYNTAX;
-		parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start);
+		parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count);
 	}
 }
 
diff --git a/ext/json/json_scanner.re b/ext/json/json_scanner.re
@@ -216,7 +216,7 @@ std:
 			return PHP_JSON_T_EOI;
 		} else {
 			s->errcode = PHP_JSON_ERROR_CTRL_CHAR;
-			s->errpos = (size_t)(s->str_start - s->input_start);
+			s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 			return PHP_JSON_T_ERROR;
 		}
 	}
@@ -229,24 +229,23 @@ std:
 	}
 	<JS>CTRL                 {
 		s->errcode = PHP_JSON_ERROR_CTRL_CHAR;
-		s->errpos = (size_t)(s->str_start - s->input_start);
+		s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 		return PHP_JSON_T_ERROR;
 	}
 	<JS>UTF8                 {
 		s->errcode = PHP_JSON_ERROR_SYNTAX;
-		s->errpos = (size_t)(s->str_start - s->input_start);
-		
+		s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 		return PHP_JSON_T_ERROR;
 	}
 	<JS>ANY                  {
 		s->errcode = PHP_JSON_ERROR_UTF8;
-		s->errpos = (size_t)(s->str_start - s->input_start);
+		s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 		return PHP_JSON_T_ERROR;
 	}
 
 	<STR_P1>CTRL             {
 		s->errcode = PHP_JSON_ERROR_CTRL_CHAR;
-		s->errpos = (size_t)(s->str_start - s->input_start);
+		s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 		return PHP_JSON_T_ERROR;
 	}
 	<STR_P1>UTF16_1          {
@@ -267,7 +266,7 @@ std:
 	}
 	<STR_P1>UCS2             {
 		s->errcode = PHP_JSON_ERROR_UTF16;
-		s->errpos = (size_t)(s->str_start - s->input_start);
+		s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 		return PHP_JSON_T_ERROR;
 	}
 	<STR_P1>ESC              {
@@ -276,7 +275,7 @@ std:
 	}
 	<STR_P1>ESCPREF           {
 		s->errcode = PHP_JSON_ERROR_SYNTAX;
-		s->errpos = (size_t)(s->str_start - s->input_start);
+		s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 		
 		return PHP_JSON_T_ERROR;
 	}
@@ -307,7 +306,7 @@ std:
 			if (s->options & PHP_JSON_INVALID_UTF8_SUBSTITUTE) {
 				if (s->utf8_invalid_count > INT_MAX - 2) {
 					s->errcode = PHP_JSON_ERROR_UTF8;
-					s->errpos = (size_t)(s->str_start - s->input_start);
+					s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 					return PHP_JSON_T_ERROR;
 				}
 				s->utf8_invalid_count += 2;
@@ -318,7 +317,7 @@ std:
 			PHP_JSON_CONDITION_GOTO(STR_P1);
 		}
 		s->errcode = PHP_JSON_ERROR_UTF8;
-		s->errpos = (size_t)(s->str_start - s->input_start);
+		s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 		return PHP_JSON_T_ERROR;
 	}
 
@@ -384,7 +383,7 @@ std:
 				break;
 			default:
 				s->errcode = PHP_JSON_ERROR_SYNTAX;
-				s->errpos = (size_t)(s->str_start - s->input_start);
+				s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 				
 				return PHP_JSON_T_ERROR;
 		}
@@ -414,7 +413,7 @@ std:
 
 	<*>ANY                   {
 		s->errcode = PHP_JSON_ERROR_SYNTAX;
-		s->errpos = (size_t)(s->str_start - s->input_start);
+		s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);
 		
 		return PHP_JSON_T_ERROR;
 	}
diff --git a/ext/json/tests/bug68546.phpt b/ext/json/tests/bug68546.phpt
@@ -16,5 +16,5 @@ NULL
 bool(true)
 NULL
 bool(true)
-string(54) "The decoded property name is invalid near character 20"
+string(54) "The decoded property name is invalid near character 15"
 Done
diff --git a/ext/json/tests/json_validate_002.phpt b/ext/json/tests/json_validate_002.phpt
@@ -9,6 +9,8 @@ json_validate_trycatchdump("");
 json_validate_trycatchdump("-");
 json_validate_trycatchdump("", -1);
 json_validate_trycatchdump('{"key1":"value1", "value2"}', 2);
+json_validate_trycatchdump('{"text":"Hello \u20AC World", "key1":"Hello \xE2\x28 World"}', 2);
+json_validate_trycatchdump('{"text":"Hello \u20AC World"}', 2);
 json_validate_trycatchdump('{"key1":"value1", "key2":"value2"}', 1);
 json_validate_trycatchdump('{"key1":"value1", "key2":"value2"}', 2);
 json_validate_trycatchdump("-", 0);
@@ -32,6 +34,12 @@ bool(false)
 int(4)
 string(30) "Syntax error near character 19"
 bool(false)
+int(4)
+string(30) "Syntax error near character 38"
+bool(true)
+int(0)
+string(8) "No error"
+bool(false)
 int(1)
 string(28) "Maximum stack depth exceeded"
 bool(true)

Original file line number	Diff line number	Diff line change
`@@ -35,7 +35,7 @@ int json_yydebug = 1;`
`35`	`35`	`#define PHP_JSON_DEPTH_INC \`
`36`	`36`	`if (parser->max_depth && parser->depth >= parser->max_depth) { \`
`37`	`37`	`parser->scanner.errcode = PHP_JSON_ERROR_DEPTH; \`
`38`		`- parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start); \`
	`38`	`+ parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count); \`
`39`	`39`	`YYERROR; \`
`40`	`40`	`} \`
`41`	`41`	`++parser->depth`
`@@ -109,7 +109,7 @@ object_end:`
`109`	`109`	`\| ']'`
`110`	`110`	`{`
`111`	`111`	`parser->scanner.errcode = PHP_JSON_ERROR_STATE_MISMATCH;`
`112`		`- parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start); \`
	`112`	`+ parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count); \`
`113`	`113`	`YYERROR;`
`114`	`114`	`}`
`115`	`115`	`;`
`@@ -166,7 +166,7 @@ array_end:`
`166`	`166`	`\| '}'`
`167`	`167`	`{`
`168`	`168`	`parser->scanner.errcode = PHP_JSON_ERROR_STATE_MISMATCH;`
`169`		`- parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start); \`
	`169`	`+ parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count); \`
`170`	`170`	`YYERROR;`
`171`	`171`	`}`
`172`	`172`	`;`
`@@ -245,7 +245,7 @@ static int php_json_parser_object_update(php_json_parser parser, zval object,`
`245`	`245`	`} else {`
`246`	`246`	`if (ZSTR_LEN(key) > 0 && ZSTR_VAL(key)[0] == '\0') {`
`247`	`247`	`parser->scanner.errcode = PHP_JSON_ERROR_INVALID_PROPERTY_NAME;`
`248`		`- parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start); \`
	`248`	`+ parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count); \`
`249`	`249`	`zend_string_release_ex(key, 0);`
`250`	`250`	`zval_ptr_dtor_nogc(zvalue);`
`251`	`251`	`zval_ptr_dtor_nogc(object);`
`@@ -304,7 +304,7 @@ static void php_json_yyerror(php_json_parser parser, char const msg)`
`304`	`304`	`{`
`305`	`305`	`if (!parser->scanner.errcode) {`
`306`	`306`	`parser->scanner.errcode = PHP_JSON_ERROR_SYNTAX;`
`307`		`- parser->scanner.errpos = (size_t)(parser->scanner.str_start - parser->scanner.input_start);`
	`307`	`+ parser->scanner.errpos = (size_t)((parser->scanner.str_start - parser->scanner.input_start) - parser->scanner.str_esc - parser->scanner.utf8_invalid_count);`
`308`	`308`	`}`
`309`	`309`	`}`
`310`	`310`
Original file line number	Diff line number	Diff line change
`@@ -216,7 +216,7 @@ std:`
`216`	`216`	`return PHP_JSON_T_EOI;`
`217`	`217`	`} else {`
`218`	`218`	`s->errcode = PHP_JSON_ERROR_CTRL_CHAR;`
`219`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`219`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`220`	`220`	`return PHP_JSON_T_ERROR;`
`221`	`221`	`}`
`222`	`222`	`}`
`@@ -229,24 +229,23 @@ std:`
`229`	`229`	`}`
`230`	`230`	`<JS>CTRL {`
`231`	`231`	`s->errcode = PHP_JSON_ERROR_CTRL_CHAR;`
`232`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`232`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`233`	`233`	`return PHP_JSON_T_ERROR;`
`234`	`234`	`}`
`235`	`235`	`<JS>UTF8 {`
`236`	`236`	`s->errcode = PHP_JSON_ERROR_SYNTAX;`
`237`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
`238`		`-`
	`237`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`239`	`238`	`return PHP_JSON_T_ERROR;`
`240`	`239`	`}`
`241`	`240`	`<JS>ANY {`
`242`	`241`	`s->errcode = PHP_JSON_ERROR_UTF8;`
`243`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`242`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`244`	`243`	`return PHP_JSON_T_ERROR;`
`245`	`244`	`}`
`246`	`245`
`247`	`246`	`<STR_P1>CTRL {`
`248`	`247`	`s->errcode = PHP_JSON_ERROR_CTRL_CHAR;`
`249`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`248`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`250`	`249`	`return PHP_JSON_T_ERROR;`
`251`	`250`	`}`
`252`	`251`	`<STR_P1>UTF16_1 {`
`@@ -267,7 +266,7 @@ std:`
`267`	`266`	`}`
`268`	`267`	`<STR_P1>UCS2 {`
`269`	`268`	`s->errcode = PHP_JSON_ERROR_UTF16;`
`270`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`269`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`271`	`270`	`return PHP_JSON_T_ERROR;`
`272`	`271`	`}`
`273`	`272`	`<STR_P1>ESC {`
`@@ -276,7 +275,7 @@ std:`
`276`	`275`	`}`
`277`	`276`	`<STR_P1>ESCPREF {`
`278`	`277`	`s->errcode = PHP_JSON_ERROR_SYNTAX;`
`279`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`278`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`280`	`279`
`281`	`280`	`return PHP_JSON_T_ERROR;`
`282`	`281`	`}`
`@@ -307,7 +306,7 @@ std:`
`307`	`306`	`if (s->options & PHP_JSON_INVALID_UTF8_SUBSTITUTE) {`
`308`	`307`	`if (s->utf8_invalid_count > INT_MAX - 2) {`
`309`	`308`	`s->errcode = PHP_JSON_ERROR_UTF8;`
`310`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`309`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`311`	`310`	`return PHP_JSON_T_ERROR;`
`312`	`311`	`}`
`313`	`312`	`s->utf8_invalid_count += 2;`
`@@ -318,7 +317,7 @@ std:`
`318`	`317`	`PHP_JSON_CONDITION_GOTO(STR_P1);`
`319`	`318`	`}`
`320`	`319`	`s->errcode = PHP_JSON_ERROR_UTF8;`
`321`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`320`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`322`	`321`	`return PHP_JSON_T_ERROR;`
`323`	`322`	`}`
`324`	`323`
`@@ -384,7 +383,7 @@ std:`
`384`	`383`	`break;`
`385`	`384`	`default:`
`386`	`385`	`s->errcode = PHP_JSON_ERROR_SYNTAX;`
`387`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`386`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`388`	`387`
`389`	`388`	`return PHP_JSON_T_ERROR;`
`390`	`389`	`}`
`@@ -414,7 +413,7 @@ std:`
`414`	`413`
`415`	`414`	`<*>ANY {`
`416`	`415`	`s->errcode = PHP_JSON_ERROR_SYNTAX;`
`417`		`- s->errpos = (size_t)(s->str_start - s->input_start);`
	`416`	`+ s->errpos = (size_t)(s->str_start - s->input_start - s->str_esc - s->utf8_invalid_count);`
`418`	`417`
`419`	`418`	`return PHP_JSON_T_ERROR;`
`420`	`419`	`}`