Changing from patch to improvement as there was no patch added

patch added (Currently only tested on php 5.3.1)

No to 1... no difference to object interface except that you must additionally unset the resource on a failure.

No to 2 (see 1)

No to 3... a false from finfo can also mean any other error and is not limited to a not recognised mimetype

No to 4... an automatic found magicfile should not throw an exception

No to 5... this is actually done afterwards to reduce payload

6: This is already done

to 1: I set the resource only if finfo_open retunred a resource and on destructor i close it

to 3: yes a false from finfo_file means there was an error - on error i throw an exception else it returns application/octed-stream (on my tests) -> I reset this to null

to 4: the automatic file detection is done on constructor and will catch the exception

to 5: within trunk the detection was on getMagicFile ?

to 6: I couldn't find it - where is it located ?

patch for changes in tests

No for the patch... it adds not allowed behaviour and also errors

Can you tell me which not allowed behavior do you mean and where errors are reported, please?

You were adding an exception within isValid() which is not allowed.

Additionally you added some other problems by erasing the check for save_mode which integrates already solved issues.

??You were adding an exception within isValid() which is not allowed.?? yes, but if finfo_file returns false I'm sure it's an internal error of finfo because it returns "application/octet-stream" if mimetype could not detect. (finfo_file: Returns a textual description of the contents of the filename argument, or FALSE if an error occurred.) -> On such cases than we need to add a new validate message and return false.

??Additionally you added some other problems by erasing the check for save_mode which integrates already solved issues.?? I will do some tests with it - don't tested it now

I see you closed ZF-9320, but: 1. if new finfo failes PHP-Notices are reported - This is also not allowed (if possible) on ZF 2. $mime isn't false if mime database is invalid. -> this is why I implemented it as open_finfo in my patch 3. You open an finfo instance twice -> on setMagicFile and on isValid /:

$const = defined('FILEINFO_MIME_TYPE') ? FILEINFO_MIME_TYPE : FILEINFO_MIME;
$mime = new finfo($const, __FILE__);
// PHP Notice:  finfo::finfo(): Warning: offset `<?php' invalid in .../finfo_test.php on line 4
// PHP Notice:  finfo::finfo(): Warning: type `<?php' invalid in .../finfo_test.php on line 4
// PHP Notice:  finfo::finfo(): Warning: offset `$const = defined('FILEINFO_MIME_TYPE') ? FILEINFO_MIME_TYPE : FILEINFO_MIME;' invalid in .../finfo_test.php on line 4
// PHP Notice:  finfo::finfo(): Warning: type `$const = defined('FILEINFO_MIME_TYPE') ? FILEINFO_MIME_TYPE : FILEINFO_MIME;' invalid in .../finfo_test.php on line 4
// PHP Notice:  finfo::finfo(): Warning: offset `$mime = new finfo($const, __FILE__);' invalid in .../finfo_test.php on line 4
// PHP Notice:  finfo::finfo(): Warning: type `$mime = new finfo($const, __FILE__);' invalid in .../finfo_test.php on line 4
// PHP Notice:  finfo::finfo(): Warning: offset `var_dump($mime);' invalid in .../finfo_test.php on line 4
// PHP Notice:  finfo::finfo(): Warning: type `var_dump($mime);' invalid in .../finfo_test.php on line 4
// PHP Warning:  finfo::finfo(): Failed to load magic database at '.../finfo_test.php'. in .../finfo_test.php on line 4

var_dump($mime);
// object(finfo)#1 (0) {
// }

Additionally: * getMagicFile don't return mime_magic.magicfile if finfo isn't available but mime_content_type is * setMagicFile don't throw an exception if finfo isn't available

Fixed as discussed within IRC