RE: [PECL-DEV] RFC for Post Parse Extension

IMO, this should be done in a generic way: 1/ Have an .ini option to skip the built-in mime parser. 2/ Expose mime parser functions so that the content can be parsed from any stream. So, a slightly modified version of your example: $fp = fopen('php://input', 'r'); // open a magical parser stream/resource $parser = postparser_init($fp); // get the content-type, length and encoding headers $headers = postparser_get_headers($parser); While ($token = postparser_get_token($parser)) { switch ($token['type']) { case POSTPARSER_VARIABLE: $_POST[$token['name']] = stream_get_contents($parser); break; case POSTPARSER_FILE: // $token contains content-type and filename etc. $fp = fopen($tmpfile, 'w'); stream_copy_to_stream($tmpfile, $parser); break; } } $parser = null; The important points here are: $parser is custom stream type that references the input file pointer; whenever you call postparser_get_token() it is rewound to reference the start of the post data item. Subsequent read calls will pull data out until the end of that item, where the $parser stream will register EOF. I've used stream_copy_to_stream() and stream_get_contents() to show how few lines are needed to process the data in bulk; if you are updating a progress bar of some kind, you will want to use fread() yourself in a loop. --Wez. > 3.0 Example > > /* > * Set up the environment and checks to see if multipart/form > * exists. > */ > $pph = postparse_init(); > > if ( $pph) { > /* > * POST data has been sent from client > */ > > while (! $rc = postparse_eof($pph) ) { > switch ($rc) { > > case POSTPARSE_VARIABLE: > > /* > * if no limit is desired for size of value > * a loop would be more appropriate > */ > postparse_read($pph, 1024, $value); > break; > > case POSTPARSE_FILE: > > $tmpname = tempnam('/tmp'); > $fp = fopen($tmpname, 'w'); > while ($bread = postparse_read($pph, 1024, $buf) ) { > fwrite($fp, $buf); > } > flcose($fp); > > break; > } > } > postparse_close($pph); > }