Re: [ZEND-ENGINE-CVS] cvs: ZendEngine2 / zend_language_parser.y zend_language_scanner.l

Hi Dmitry,

I had looked at this Bug and wondering whether it should be fixed (I guess 
the answer is yes :-)) and how to do so.  I was only thinking along the 
lines of adding a stack to keep track of multiple heredocs in the scanner. 
Then I saw your fix and thought it was a cool method and better than 
whatever came to my mind...

However, it seems there's nothing to handle nested heredocs in 
zend_highlight/strip, tokenizer, etc., so they will break there (I think 
only the "inner" heredoc's ending label will be found).  Plus memory leak 
too I guess -- though that would have happened before as well if someone 
tried to nest heredocs...


- Matt


----- Original Message -----
From: "Dmitry Stogov"
Sent: Thursday, March 26, 2009

> dmitry Thu Mar 26 12:38:10 2009 UTC
>
>  Modified files:
>    /ZendEngine2 zend_language_parser.y zend_language_scanner.l
>  Log:
>  Fixed bug #47516 (nowdoc can not be embed in heredoc but can be embed in 
> double quote)
>
>
> http://cvs.php.net/viewvc.cgi/ZendEngine2/zend_language_parser.y?r1=1.214&r2=1.215&diff_format=u
> Index: ZendEngine2/zend_language_parser.y
> diff -u ZendEngine2/zend_language_parser.y:1.214 
> ZendEngine2/zend_language_parser.y:1.215
> --- ZendEngine2/zend_language_parser.y:1.214 Wed Dec 31 11:09:46 2008
> +++ ZendEngine2/zend_language_parser.y Thu Mar 26 12:38:10 2009
> @@ -18,7 +18,7 @@
> 
> +----------------------------------------------------------------------+
> */
>
> -/* $Id: zend_language_parser.y,v 1.214 2008/12/31 11:09:46 sebastian Exp 
> $ */
> +/* $Id: zend_language_parser.y,v 1.215 2009/03/26 12:38:10 dmitry Exp $ 
> */
>
> /*
>  * LALR shift/reduce conflicts and how they are resolved:
> @@ -768,13 +768,13 @@
>  | T_METHOD_C { $$ = $1; }
>  | T_FUNC_C { $$ = $1; }
>  | T_NS_C { $$ = $1; }
> - | start_heredoc T_ENCAPSED_AND_WHITESPACE T_END_HEREDOC { $$ = $2; }
> - | start_heredoc T_END_HEREDOC { if (CG(literal_type) == IS_UNICODE) { 
> ZVAL_EMPTY_UNICODE(&$$.u.constant); } else { 
> ZVAL_EMPTY_STRING(&$$.u.constant); } INIT_PZVAL(&$$.u.constant); 
> $$.op_type = IS_CONST; }
> + | start_heredoc T_ENCAPSED_AND_WHITESPACE T_END_HEREDOC { $$ = $2; 
> CG(heredoc) = Z_STRVAL($1.u.constant); CG(heredoc_len) = 
> Z_STRLEN($1.u.constant); }
> + | start_heredoc T_END_HEREDOC { if (CG(literal_type) == IS_UNICODE) { 
> ZVAL_EMPTY_UNICODE(&$$.u.constant); } else { 
> ZVAL_EMPTY_STRING(&$$.u.constant); } INIT_PZVAL(&$$.u.constant); 
> $$.op_type = IS_CONST; CG(heredoc) = Z_STRVAL($1.u.constant); 
> CG(heredoc_len) = Z_STRLEN($1.u.constant); }
> ;
>
> start_heredoc:
> - T_START_HEREDOC  { CG(literal_type) = 
> UG(unicode)?IS_UNICODE:IS_STRING; }
> - | T_BINARY_HEREDOC { CG(literal_type) = IS_STRING; }
> + T_START_HEREDOC  { CG(literal_type) = UG(unicode)?IS_UNICODE:IS_STRING; 
> $$ = $1; }
> + | T_BINARY_HEREDOC { CG(literal_type) = IS_STRING; $$ = $1; }
> ;
>
>
> @@ -802,7 +802,7 @@
>  | common_scalar { $$ = $1; }
>  | '"' { CG(literal_type) = UG(unicode)?IS_UNICODE:IS_STRING; } 
> encaps_list '"' { $$ = $3; }
>  | T_BINARY_DOUBLE { CG(literal_type) = IS_STRING; } encaps_list '"' { $$ 
> = $3; }
> - | start_heredoc encaps_list T_END_HEREDOC { $$ = $2; }
> + | start_heredoc encaps_list T_END_HEREDOC { $$ = $2; CG(heredoc) = 
> Z_STRVAL($1.u.constant); CG(heredoc_len) = Z_STRLEN($1.u.constant); }
> ;
>
>
> http://cvs.php.net/viewvc.cgi/ZendEngine2/zend_language_scanner.l?r1=1.203&r2=1.204&diff_format=u
> Index: ZendEngine2/zend_language_scanner.l
> diff -u ZendEngine2/zend_language_scanner.l:1.203 
> ZendEngine2/zend_language_scanner.l:1.204
> --- ZendEngine2/zend_language_scanner.l:1.203 Wed Mar 25 15:24:27 2009
> +++ ZendEngine2/zend_language_scanner.l Thu Mar 26 12:38:10 2009
> @@ -21,7 +21,7 @@
> 
> +----------------------------------------------------------------------+
> */
>
> -/* $Id: zend_language_scanner.l,v 1.203 2009/03/25 15:24:27 dmitry Exp $ 
> */
> +/* $Id: zend_language_scanner.l,v 1.204 2009/03/26 12:38:10 dmitry Exp $ 
> */
>
> #if 0
> # define YYDEBUG(s, c) printf("state: %d char: %c\n", s, c)
> @@ -2402,6 +2402,10 @@
>  char *s;
>  int bprefix = (yytext[0] != '<') ? 1 : 0;
>
> + /* save old heredoc label */
> + Z_STRVAL_P(zendlval) = CG(heredoc);
> + Z_STRLEN_P(zendlval) = CG(heredoc_len);
> +
>  CG(zend_lineno)++;
>  CG(heredoc_len) = yyleng-bprefix-3-1-(yytext[yyleng-2]=='\r'?1:0);
>  s = yytext+bprefix+3;
>
>