RE: [PHP] Header

Well, standards are a Good Thing(tm). BTW, if anybody's concerned, I've written a function that can take any form of relative or absolute, full or partial path to redirect to, and redirects to a correct URL. It relies on another useful function I'm including as well. Here it is, quite commented (comments refer to the code BELOW them). Supposing I'd use it at http://www.wad.cz/foo/bar/script.php, redirect( '../otherscript.php' ) will result in header( 'Location: http://www.wad.cz/foo/otherscript.php' ), redirect( './otherscript.php' ) and redirect( 'otherscript.php' ) in header( 'Location: http://www.wad.cz/foo/bar/otherscript.php' ) ... Etc. function redirect( $input ) { $buffer = '' ; # won't work with CGI (I guess force-cgi-redirect might help, not sure $self = $GLOBALS['HTTP_SERVER_VARS']['PHP_SELF'] ; # ready to go, currently no additional checking is done if( "http://" == substr( $input , 0 , 7 ) ) { header( 'Location: ' . $input ) ; # absolute path without a hostname (i. e. just prepend the hostname) } elseif( '/' == $input[ 0 ] ) { $buffer = $input ; # parent. } elseif( '../' == substr( $input , 0 , 3 ) ) { # get the length of an initial segment composed solely of '.'s and '/'s # this, divided by 3, is the number of directory levels we'll climb up # this isn't the most robust code I can think of, but will do for now $up = strspn( $input , './' ) ; # get the current directory level $cur = substr_count( $self , '/' ) ; # echo "\$cur: $cur<BR>\n" ; # how deep from the documentroot will it end up? # is the value legal? if not, trigger an error $go = $cur - $up / 3 ; # echo "\$go: $go<BR>\n" ; if( $go <= 0 ) { trigger_error( "Cannot go above the document root, clipping\n\t" , E_USER_WARNING ) ; $go = 1 ; } ; # find the position of $go-th occurence of '/' in $self. that is: find # the length of the initial segment of $self we'll use and fetch the segment $pre = substr( $self , 0 , str_npos( $self , '/' , $go ) ) ; # substr( $input , $up ) returns everything BUT the initial '../'s $buffer = $pre . '/' . substr( $input , $up ) ; # current directory ( './otherscript.php' ) # faster than substr( $input , 0 , 2 ) ; } elseif( './' == $input[ 0 ] . $input[ 1 ] ) { $buffer = dirname( $self ) . substr( $input , 1 ) ; # current directory ( 'otherscript.php' ) } else { $buffer = dirname( $self ) . '/' . $input ; } header( 'Location: ' . 'http://' . $GLOBALS['HTTP_SERVER_VARS']['SERVER_NAME'] . $buffer ) ; } # find position of n-th occurence of $needle in $haystack function str_npos( $haystack , $needle , $occurrence ) { if( $occurrence > substr_count( $haystack , $needle ) ) { return false ; } for( $i = 0 , $offset = -1 ; $i < $occurrence ; $i++ ) { $offset = strpos( $haystack , $needle , $offset + 1 ) ; } return $offset ; } At 21:55 20.12. 2000, Dave VanAuken wrote the following: -------------------------------------------------------------- >I agree, and this is the primary reason why Javascript is barely a >tool in my opinion - the fact that half of its implementations don't >work on half the browsers... > >We also need to remember though that all these standards are simply >RFC's (Request For Comments) meaning that they are driven by feedback >and general concensus... The reason that we are still not limited to >HTML 0 is because of people pushing standards through the RFC >process(good) and browser monoliths from pushing THEIR own standards >(maybe not so good) due to market pull. > >Thus my comment, not judging the right or wrong or correctness of >nature, as its not my place to determine if someone should code their >site with code that only works on MSIE 7.0a codename "never work" or >elects to be HTML 1, 1.1 and so forth complient. > >Dave > >-----Original Message----- >From: Cynic [mailto:cynic@mail.cz] >Sent: Wednesday, December 20, 2000 2:44 PM >To: Dave VanAuken; php-general@lists.php.net >Subject: RE: [PHP] Header > > >Ever heard of WaSP? It's a group of people pushing HTML UA >vendors towards standards compliance. Because noncompliance >is what created the web hell where HTML people must have >a different version of a HTML page for every browser their >audience uses. >Noncompliance with HTTP 1.1 is what brought us coders problems >with proxies. > >header( "Location: ../blah/index.html") ; >might work with all current browsers, because they're not >compliant with the HTTP protocol. Next time you scream in >anger because of lack of support for standard X in browser Y, >rest assured it's not that the programmers played Minesweeper. >They spent the time they should've been implementing proper >web standards support coding around your laziness instead. > >No offense intended, I'm just... > > >At 20:05 20.12. 2000, Dave VanAuken wrote the following: >-------------------------------------------------------------- >>No, but it isn't "required" for the function to perform as intended. >> >>Stating what is proper to be compliant and stating what is required >>are two different things. >> >>Dave >> >>-----Original Message----- >>From: Cynic [mailto:cynic@mail.cz] >>Sent: Wednesday, December 20, 2000 12:49 PM >>To: mathias rockel; Dan Lowe; php-general@lists.php.net >>Subject: Re: [PHP] Header >> >> >>Is being standards-compliant such a PITA? >> >> >>At 18:41 20.12. 2000, mathias rockel wrote the following: >>-------------------------------------------------------------- >>>Hi! >>> >>>Although it is it seemed to me every browser I tried could use >>relative urls >>>as a location header. anyone made different experiences ? >>> >>>mats >>>----- Original Message ----- >>>From: "Dan Lowe" <dan@tangledhelix.com> >>>To: <php-general@lists.php.net> >>>Cc: <php-general@lists.php.net> >>>Sent: Wednesday, December 20, 2000 1:08 AM >>>Subject: Re: [PHP] Header >>> >>> >>>> Previously, Diogo Zulli said: >>>> > On [15 December 2000 18:32] (Cynic), wrote: >>>> > >>>> > } According to the HTTP protocol, the URL must be absolute. >>>> > >>>> > Does this means that >>>> > "http://hostname/" should be on "Location: >" >>>> > or just the /absolute/pathname must be there? >>>> >>>> Must be fully qualified with protocol, host, and anything extra, >so >>>> yes, http://hostname/extra/ is needed. >>>> >>>> See section 14.30 of RFC 2616 (HTTP 1.1) or section 10.11 of RFC >>1945 >>>> (HTTP 1.0). >>>> >>>> The Location response-header field defines the exact location >>of the >>>> resource that was identified by the Request-URI. For 3xx >>responses, >>>> the location must indicate the server's preferred URL for >>automatic >>>> redirection to the resource. Only one absolute URL is allowed. >>>> >>>> Location = "Location" ":" absoluteURI >>>> >>>> An example is >>>> >>>> Location: >>>> http://www.w3.org/hypertext/WWW/NewLocation.html >>>> >>>> -dan >>>> >>>> -- >>>> Dan Lowe <dan@tangledhelix.com> >>>> http://tangledhelix.com/