[PHP-DEV] PHP 4.0 Bug #8339: new preg_replace back reference handling not working From: padraic <email protected>
Date: 12/20/00

From: padraic <email protected>
Operating system: RedHat Linux 7
PHP version: 4.0.4
PHP Bug Type: PCRE related
Bug description: new preg_replace back reference handling not working

phpinfo() reports:
  PHP Version 4.0.4
  './configure' '--with-mysql' '--with-apxs=/usr/local/apache/bin/apxs'
  '--enable-rule=EAPI' '--with-imap' '--with-curl=/usr'
  '--with-config-file-path=/etc' '--enable-track-vars=yes'
  '--disable-debug'
  
I'm suddenly having problems with my PHPLIB template based application.
The problem seems to be due to the new 4.0.4 back reference handling in
preg_replace.

Test code:
  error_reporting(E_ALL);
  print(preg_replace("/{FOO}/","$7.50","Amount is {FOO}"));

Under PHP 4.0.2 the result is:
  Amount is $7.50

Under PHP 4.0.4 the result is:
  Amount is .50

No errors or warnings reported in either case.

I reviewed the Changelog for 4.0.4 and found:
  Modified preg_replace() to accept Perl-style $n backreferences
  in addition to \\n ones. (Andrei)
  Modified preg_replace() to ignore backreferences that refer
  to non-existing subpatterns. (Andrei)

The second Changelog entry seems to disagree with a statement in the
PCRE Pattern Syntax Manual section which states:
  However, if the decimal number following the backslash is
  less than 10, it is always taken as a back reference, and
  causes an error only if there are not that many capturing
  left parentheses in the entire pattern. In other words, the
  parentheses that are referenced need not be to the left of
  the reference for numbers less than 10. See the section
  entitled "Backslash" above for further details of the han-
  dling of digits following a backslash.
  
The manual seems to say that back references without corresponding
sub-patterns should generate an error/warning while the Changelog
seems to say that a back reference without a corresponding
sub-pattern is ignored.

The behavior of 4.0.4 (using my example code) seems to be that
the a back reference without a corresponding sub-pattern is
not ignored but treated as a back reference without reporting
an error warning. Note that change the replacement to "$17.50"
which makes the back reference greater than 10 still results
in a result of "Amount is .50".

The above indicates to me that there is either a problem
in the 4.0.4 code or the text of the manual.

I attempted to workaround this in 4.0.4 with the following code:
  error_reporting(E_ALL);
  print(preg_replace("/{FOO}/","\$7.50","Amount is {FOO}"));

Reading the PCRE Syntax Manual section made me think that I
could escape the "$" with a "\". That didn't work. The result
of the above code in 4.0.4 is:
  Amount is .50
  
Adding a space between the "$" and the "7" does stop the back
reference issue, but leaves me with a space in my output
which I really don't want.

Am I missing something here or have I found a problem.

Thanks for your help. 

-- 
Edit Bug report at: http://bugs.php.net/?id=8339&edit=1

-- 
PHP Development Mailing List <http://www.php.net/>
To unsubscribe, e-mail: php-dev-unsubscribe <email protected>
For additional commands, e-mail: php-dev-help <email protected>
To contact the list administrators, e-mail: php-list-admin <email protected>