Syntax errors caused by unclosed {, [, ( mention specific location

Aside from a few very specific syntax errors for which detailed exceptions are
thrown, generally PHP just emits the default error messages generated by bison on syntax
error. These messages are very uninformative; they just say "Unexpected ... at line ...".

This is most problematic with constructs which can span an arbitrary number of lines, such
as blocks of code delimited by { }, 'if' conditions delimited by ( ), and so on. If a closing
delimiter is missed, the block will run for the entire remainder of the source file (which
could be thousands of lines), and then at the end, a parse error will be thrown with the
dreaded words: "Unexpected end of file".

Therefore, track the positions of opening and closing delimiters and ensure that they match
up correctly. If any mismatch or missing delimiter is detected, immediately throw a parse
error which points the user to the offending line. This is best done in the *lexer* and not
in the parser.

Thanks to Nikita Popov and George Peter Banyard for suggesting improvements.

Author: alexdowad

Zend/tests/bug60099.phpt

@@ -7,4 +7,4 @@ namespace foo {
 
 ?>
 --EXPECTF--
-Parse error: syntax error, unexpected end of file in %s on line %d
+Parse error: Unclosed '{' on line 2 in %s on line %d

Zend/tests/eval_parse_error_with_doc_comment.phpt

@@ -12,4 +12,4 @@ EOC
 
 ?>
 --EXPECTF--
-Parse error: syntax error, unexpected end of file in %s(%d) : eval()'d code on line %d
+Parse error: Unclosed '{' in %s(%d) : eval()'d code on line %d

Zend/tests/require_parse_exception.phpt

@@ -43,8 +43,8 @@ var_dump("\u{ffffff}");');
 ?>
 --EXPECT--
 Deprecated: Directive 'allow_url_include' is deprecated in Unknown on line 0
-syntax error, unexpected end of file on line 2
-syntax error, unexpected end of file on line 3
+Unclosed '{' on line 2
+Unclosed '{' on line 3
 syntax error, unexpected end of file, expecting '(' on line 2
 Invalid numeric literal on line 2
 Invalid UTF-8 codepoint escape sequence on line 2

Zend/zend_globals.h

@@ -285,6 +285,7 @@ struct _zend_php_scanner_globals {
 	int yy_state;
 	zend_stack state_stack;
 	zend_ptr_stack heredoc_label_stack;
+	zend_stack nest_location_stack; /* for syntax error reporting */
 	zend_bool heredoc_scan_ahead;
 	int heredoc_indentation;
 	zend_bool heredoc_indentation_uses_spaces;

Zend/zend_language_scanner.h

@@ -30,6 +30,7 @@ typedef struct _zend_lex_state {
 	int yy_state;
 	zend_stack state_stack;
 	zend_ptr_stack heredoc_label_stack;
+	zend_stack nest_location_stack; /* for syntax error reporting */
 
 	zend_file_handle *in;
 	uint32_t lineno;
@@ -63,6 +64,12 @@ typedef struct _zend_heredoc_label {
 	zend_bool indentation_uses_spaces;
 } zend_heredoc_label;
 
+/* Track locations of unclosed {, [, (, etc. for better syntax error reporting */
+typedef struct _zend_nest_location {
+	char text;
+	int  lineno;
+} zend_nest_location;
+
 BEGIN_EXTERN_C()
 ZEND_API void zend_save_lexical_state(zend_lex_state *lex_state);
 ZEND_API void zend_restore_lexical_state(zend_lex_state *lex_state);

Zend/zend_language_scanner.l

@@ -192,6 +192,7 @@ void startup_scanner(void)
 	CG(doc_comment) = NULL;
 	CG(extra_fn_flags) = 0;
 	zend_stack_init(&SCNG(state_stack), sizeof(int));
+	zend_stack_init(&SCNG(nest_location_stack), sizeof(zend_nest_location));
 	zend_ptr_stack_init(&SCNG(heredoc_label_stack));
 	SCNG(heredoc_scan_ahead) = 0;
 }
@@ -205,6 +206,7 @@ void shutdown_scanner(void)
 	CG(parse_error) = 0;
 	RESET_DOC_COMMENT();
 	zend_stack_destroy(&SCNG(state_stack));
+	zend_stack_destroy(&SCNG(nest_location_stack));
 	zend_ptr_stack_clean(&SCNG(heredoc_label_stack), (void (*)(void *)) &heredoc_label_dtor, 1);
 	zend_ptr_stack_destroy(&SCNG(heredoc_label_stack));
 	SCNG(heredoc_scan_ahead) = 0;
@@ -223,6 +225,9 @@ ZEND_API void zend_save_lexical_state(zend_lex_state *lex_state)
 	lex_state->state_stack = SCNG(state_stack);
 	zend_stack_init(&SCNG(state_stack), sizeof(int));
 
+	lex_state->nest_location_stack = SCNG(nest_location_stack);
+	zend_stack_init(&SCNG(nest_location_stack), sizeof(zend_nest_location));
+
 	lex_state->heredoc_label_stack = SCNG(heredoc_label_stack);
 	zend_ptr_stack_init(&SCNG(heredoc_label_stack));
 
@@ -258,6 +263,9 @@ ZEND_API void zend_restore_lexical_state(zend_lex_state *lex_state)
 	zend_stack_destroy(&SCNG(state_stack));
 	SCNG(state_stack) = lex_state->state_stack;
 
+	zend_stack_destroy(&SCNG(nest_location_stack));
+	SCNG(nest_location_stack) = lex_state->nest_location_stack;
+
 	zend_ptr_stack_clean(&SCNG(heredoc_label_stack), (void (*)(void *)) &heredoc_label_dtor, 1);
 	zend_ptr_stack_destroy(&SCNG(heredoc_label_stack));
 	SCNG(heredoc_label_stack) = lex_state->heredoc_label_stack;
@@ -1250,6 +1258,64 @@ static void copy_heredoc_label_stack(void *void_heredoc_label)
 	zend_ptr_stack_push(&SCNG(heredoc_label_stack), (void *) new_heredoc_label);
 }
 
+/* Check that { }, [ ], ( ) are nested correctly */
+static void report_bad_nesting(char opening, int opening_lineno, char closing)
+{
+	char   buf[256];
+	size_t used = 0;
+
+	used = snprintf(buf, sizeof(buf), "Unclosed '%c'", opening);
+
+	if (opening_lineno != CG(zend_lineno)) {
+		used += snprintf(buf + used, sizeof(buf) - used, " on line %d", opening_lineno);
+	}
+
+	if (closing) { 	/* 'closing' will be 0 if at end of file */
+		used += snprintf(buf + used, sizeof(buf) - used, " does not match '%c'", closing);
+	}
+
+	zend_throw_exception(zend_ce_parse_error, buf, 0);
+}
+
+static void enter_nesting(char opening)
+{
+	zend_nest_location nest_loc = {opening, CG(zend_lineno)};
+	zend_stack_push(&SCNG(nest_location_stack), &nest_loc);
+}
+
+static int exit_nesting(char closing)
+{
+	if (zend_stack_is_empty(&SCNG(nest_location_stack))) {
+		zend_throw_exception_ex(zend_ce_parse_error, 0, "Unmatched '%c'", closing);
+		return -1;
+	}
+
+	zend_nest_location *nest_loc = zend_stack_top(&SCNG(nest_location_stack));
+	char opening = nest_loc->text;
+	char closing = *location;
+
+	if ((opening == '{' && closing != '}') ||
+	    (opening == '[' && closing != ']') ||
+	    (opening == '(' && closing != ')')) {
+		report_bad_nesting(opening, nest_loc->lineno, closing);
+		return -1;
+	}
+
+	zend_stack_del_top(&SCNG(nest_location_stack));
+	return 0;
+}
+
+static int check_nesting_at_end()
+{
+	if (!zend_stack_is_empty(&SCNG(nest_location_stack))) {
+		zend_nest_location *nest_loc = zend_stack_top(&SCNG(nest_location_stack));
+		report_bad_nesting(nest_loc->text, nest_loc->lineno, 0);
+		return -1;
+	}
+
+	return 0;
+}
+
 #define PARSER_MODE() \
 	EXPECTED(elem != NULL)
 
@@ -1277,6 +1343,22 @@ static void copy_heredoc_label_stack(void *void_heredoc_label)
 		goto emit_token; \
 	} while (0)
 
+#define RETURN_EXIT_NESTING_TOKEN(_token) do { \
+		if (exit_nesting(_token) && PARSER_MODE()) { \
+			RETURN_TOKEN(T_ERROR); \
+		} else { \
+			RETURN_TOKEN(_token); \
+		} \
+	} while(0)
+
+#define RETURN_END_TOKEN do { \
+		if (check_nesting_at_end() && PARSER_MODE()) { \
+			RETURN_TOKEN(T_ERROR); \
+		} else { \
+			RETURN_TOKEN(END); \
+		} \
+	} while (0)
+
 int ZEND_FASTCALL lex_scan(zval *zendlval, zend_parser_stack_elem *elem)
 {
 int token;
@@ -1297,7 +1379,7 @@ BNUM	"0b"[01]+(_[01]+)*
 LABEL	[a-zA-Z_\x80-\xff][a-zA-Z0-9_\x80-\xff]*
 WHITESPACE [ \n\r\t]+
 TABS_AND_SPACES [ \t]*
-TOKENS [;:,.\[\]()|^&+-/*=%!~$<>?@]
+TOKENS [;:,.|^&+-/*=%!~$<>?@]
 ANY_CHAR [^]
 NEWLINE ("\r"|"\n"|"\r\n")
 
@@ -1770,29 +1852,40 @@ NEWLINE ("\r"|"\n"|"\r\n")
 	RETURN_TOKEN(T_SR);
 }
 
+<ST_IN_SCRIPTING>"]"|")" {
+	/* Check that ] and ) match up properly with a preceding [ or ( */
+	RETURN_EXIT_NESTING_TOKEN(yytext[0]);
+}
+
+<ST_IN_SCRIPTING>"["|"(" {
+	enter_nesting(yytext[0]);
+	RETURN_TOKEN(yytext[0]);
+}
+
 <ST_IN_SCRIPTING>{TOKENS} {
 	RETURN_TOKEN(yytext[0]);
 }
 
 
 <ST_IN_SCRIPTING>"{" {
 	yy_push_state(ST_IN_SCRIPTING);
+	enter_nesting('{');
 	RETURN_TOKEN('{');
 }
 
 
 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"${" {
 	yy_push_state(ST_LOOKING_FOR_VARNAME);
+	enter_nesting('{');
 	RETURN_TOKEN(T_DOLLAR_OPEN_CURLY_BRACES);
 }
 
-
 <ST_IN_SCRIPTING>"}" {
 	RESET_DOC_COMMENT();
 	if (!zend_stack_is_empty(&SCNG(state_stack))) {
 		yy_pop_state();
 	}
-	RETURN_TOKEN('}');
+	RETURN_EXIT_NESTING_TOKEN('}');
 }
 
 
@@ -2088,7 +2181,7 @@ string:
 
 <INITIAL>{ANY_CHAR} {
 	if (YYCURSOR > YYLIMIT) {
-		RETURN_TOKEN(END);
+		RETURN_END_TOKEN;
 	}
 
 inline_char_handler:
@@ -2165,7 +2258,7 @@ inline_char_handler:
 	RETURN_TOKEN(']');
 }
 
-<ST_VAR_OFFSET>{TOKENS}|[{}"`] {
+<ST_VAR_OFFSET>{TOKENS}|[[(){}"`] {
 	/* Only '[' or '-' can be valid, but returning other tokens will allow a more explicit parse error */
 	RETURN_TOKEN(yytext[0]);
 }
@@ -2569,6 +2662,7 @@ skip_escape_conversion:
 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"{$" {
 	yy_push_state(ST_IN_SCRIPTING);
 	yyless(1);
+	enter_nesting('{');
 	RETURN_TOKEN(T_CURLY_OPEN);
 }
 
@@ -2593,7 +2687,7 @@ skip_escape_conversion:
 	}
 
 	if (YYCURSOR > YYLIMIT) {
-		RETURN_TOKEN(END);
+		RETURN_END_TOKEN;
 	}
 	if (yytext[0] == '\\' && YYCURSOR < YYLIMIT) {
 		YYCURSOR++;
@@ -2640,7 +2734,7 @@ double_quotes_scan_done:
 
 <ST_BACKQUOTE>{ANY_CHAR} {
 	if (YYCURSOR > YYLIMIT) {
-		RETURN_TOKEN(END);
+		RETURN_END_TOKEN;
 	}
 	if (yytext[0] == '\\' && YYCURSOR < YYLIMIT) {
 		YYCURSOR++;
@@ -2689,7 +2783,7 @@ double_quotes_scan_done:
 	int newline = 0, indentation = 0, spacing = 0;
 
 	if (YYCURSOR > YYLIMIT) {
-		RETURN_TOKEN(END);
+		RETURN_END_TOKEN;
 	}
 
 	YYCURSOR--;
@@ -2813,7 +2907,7 @@ heredoc_scan_done:
 	int newline = 0, indentation = 0, spacing = -1;
 
 	if (YYCURSOR > YYLIMIT) {
-		RETURN_TOKEN(END);
+		RETURN_END_TOKEN;
 	}
 
 	YYCURSOR--;
@@ -2901,7 +2995,7 @@ nowdoc_scan_done:
 
 <ST_IN_SCRIPTING,ST_VAR_OFFSET>{ANY_CHAR} {
 	if (YYCURSOR > YYLIMIT) {
-		RETURN_TOKEN(END);
+		RETURN_END_TOKEN;
 	}
 
 	RETURN_TOKEN(T_BAD_CHARACTER);

ext/tokenizer/tests/token_get_all_variation17.phpt

@@ -32,6 +32,7 @@ try {
 } catch(Exception $e) {
     echo "caught exception:";
 }
+}
 ?>';
 $tokens =  token_get_all($source);
 var_dump($tokens);
@@ -40,7 +41,7 @@ echo "Done"
 ?>
 --EXPECTF--
 *** Testing token_get_all() : with exception keywords ***
-array(81) {
+array(83) {
   [0]=>
   array(3) {
     [0]=>
@@ -601,13 +602,25 @@ array(81) {
     int(14)
   }
   [80]=>
+  string(1) "}"
+  [81]=>
   array(3) {
     [0]=>
     int(%d)
     [1]=>
-    string(2) "?>"
+    string(1) "
+"
     [2]=>
     int(15)
   }
+  [82]=>
+  array(3) {
+    [0]=>
+    int(%d)
+    [1]=>
+    string(2) "?>"
+    [2]=>
+    int(16)
+  }
 }
 Done

ext/tokenizer/tokenizer.c

@@ -392,6 +392,8 @@ static zend_bool tokenize(zval *return_value, zend_string *source, zend_class_en
 	array_init(return_value);
 
 	while ((token_type = lex_scan(&token, NULL))) {
+		ZEND_ASSERT(token_type != T_ERROR);
+
 		add_token(
 			return_value, token_type, zendtext, zendleng, token_line,
 			token_class, &interned_strings);
@@ -408,7 +410,7 @@ static zend_bool tokenize(zval *return_value, zend_string *source, zend_class_en
 				&& --need_tokens == 0
 			) {
 				/* fetch the rest into a T_INLINE_HTML */
-				if (zendcursor != zendlimit) {
+				if (zendcursor < zendlimit) {
 					add_token(
 						return_value, T_INLINE_HTML, zendcursor, zendlimit - zendcursor,
 						token_line, token_class, &interned_strings);