Re: Idea for a new extensions: checksum_from_file()

Hi, if the functions I mentioned in the last mail - especially in combination with a stream_wrapper/stream_filter, which might give you the offsets - aren't enough for your needs feel free to propose a patch. johannes On Sun, 2007-06-24 at 04:25 -0700, ecc wrote: > Ah, never seen hash extension. But there is no function, handeling offsets > internaly. > > hash_update could use offsets... but this isnt performant. Why handeling > this filedata within the php script? Thats the way, i allready handel this > without "hash". > > Can you give me an example to made this command performant with the hash > functions for an 600 MB file? > > In my example/idea, you dont have to read all into an variable to create the > checksum, only the php_extension do this transparent. Maybe its an idea to > add such a feature to the hash-extension. > > // My idea: > // all filedata is handeled transparent, only the checksums are returned. > checksum_from_file($file, array('crc32', 'md5'), 512, -128, array($this, > 'updateProcess', 1024)); > > I dont need the raw filedata in my script... i only need the checksum. I > think, done in c this is mutch faster, isnt it? > > With "hash_update" you have read the whole file using > > while (!feof($fp)) { > $buffer = fgets($fp, 65536); > hash_update($ctx, $buffer); > } > > Not really performant, isnt it? > > Regards, > Andreas > > > Johannes Schlueter wrote: > > > > Hi Andreas, > > > > since PHP 5.2.1 PHP bundles the hash extension, see php.net/hash which > > includes way more hashing algorithms (see hash_algos()) and a > > hash_file() function and should full-fill all your needs. > > > > johannes > > > > On Sun, 2007-06-24 at 02:16 -0700, ecc wrote: > >> Hi, > >> > >> there are 3 checksums supported in php. (md5, crc32 and sha1, Mabye there > >> are more...) > >> > >> md5 and sha1 allready have an *_file function available, crc32 dont have > >> them. crc32 dont have this function in the latest php version. > >> > >> These *_file functions are usefull in the most cases, but have a very big > >> disadvantage. They could not handle header offsets. > >> > >> Theses header offsets arent trivial, because e.g. MP3 files also contains > >> an > >> header. (In this case, its not really an header, because its an offset at > >> the end of the MP3 file containing the meta-informations ID3). So if you > >> parse the checksum of an MP3, you allway get the wrong checksum, because > >> the > >> song is the same in the most cases, but only the meta-infos are changed. > >> This isnt the best example, but it shows you the problem. > >> > >> This is my implementation idea (pseudocode). > >> > >> // get crc32 and md5 checksums from $file, begin at byte 512 and stop 128 > >> bytes before fileend. > >> // call method "updateProcess" everytime if 1024 bytes for file are > >> processed! (eg. for php-gtk2) > >> checksum_from_file($file, array('crc32', 'md5'), 512, -128, > >> array($this, > >> 'updateProcess', 1024)); > >> > >> public function checksum_from_file($file= false, $checksum_type = false, > >> $start_offset=false, $end_offset=false, $callback = false) { > >> $checksums = array(); > >> // get data > >> return $checksums; > >> } > >> > >> $file should be contain a filename, a filehandle or and something like > >> this > >> z:/test.zip#filename.php > >> So the creation of temp-copys of e.g. zip content could be transparent > >> done > >> without managing this externaly. > >> > >> This extension have to parse files bigger than 600 MB. So, the funktion > >> have > >> to be ready for this. (file_get_contents crash in this cases :-)) > >> > >> Now, i have to read the complete filecontent to the memory using a php > >> buildin function and create an checksum by this string, because i´ve had > >> to > >> manage the header offsets. Not really performant, and very very slow for > >> large files > 64 MB > >> > >> Regards, > >> Andreas > >> http://www.php-gtk.eu/apps/emucontrolcenter > >> > >> Here an overview of the allready implemented checksum function in php. > >> > >> string md5_file ( string $filename [, bool $raw_output] ) > >> http://de2.php.net/manual/de/function.md5-file.php > >> > >> string sha1_file ( string $filename [, bool $raw_output] ) > >> http://de2.php.net/manual/de/function.sha1-file.php > >> > >> int crc32 ( string $str ) > >> http://de2.php.net/manual/de/function.crc32.php > > > > -- > > PECL development discussion Mailing List (http://pecl.php.net/) > > To unsubscribe, visit: http://www.php.net/unsub.php > > > > > > > > -- > View this message in context: > http://www.nabble.com/Idea-for-a-new-extensions%3A-checksum_from_file%28%29-tf3971680.html#a11274130¤$ñßz†>¯‰ØÛ‹ƒã > > Sent from the Pecl - Dev mailing list archive at Nabble.com.