#31986 [NEW]: exif_read_data incorrectly calculates nesting_level, throwing warnings

[dpark at mit dot edu]

From:             dpark at mit dot edu
Operating system: Linux
PHP version:      4CVS-2005-02-15 (stable)
PHP Bug Type:     EXIF related
Bug description:  exif_read_data incorrectly calculates nesting_level, throwing 
warnings

Description:
------------
Related to #31797?  exif_read_data is still throwing the same warnings as
in #31797 on every camera-created image I give it (for cameras both new
and old, cheap and expensive).  

The Debian package in question is actually based off the 2005-02-06 4CVS
snapshot, which is later than the time at which #31797 was reported to be
fixed in 4CVS.  Adam Conrad, the package maintainer, confirms the bug is
in PHP, and can reproduce the problem with his own camera-created images.


Quoting an email from Adam Conrad:

Danny Park said:
> Anyway, when you say the nesting limit was increased from 5 to 25, are
> you saying 4.3.10-3 reflects that increase?

Indeed, it does.  However, it seems that the way the code's written, the
nesting_level increases quite rapidly, and obviously takes on a different
meaning than the original developer thought it had.

As an example your 5 images all work fine with jhead(1), which has a
hardcoded directory nesting limit of *5*.  However, with some debug
statements thrown into PHP's exif.c, we're reaching nesting levels of:

img_0949.jpg: 53
P1000415.JPG: 54
IMG_0669.JPG: 65
DSC00804.JPG: 48
dcp00884.jpg: 38

I don't currently have the time to hunt down what upstream's doing wrong
here, so I would encourage you to (re)file this bug upstream, and you're
welcome to quote this whole message.

I recommend they have a look at what jhead(1) is considering "directory
nesting" and mimic it, since PHP's obviously doing something a bit
differently.

If it doesn't get fixed properly upstream in a relatively timely fashion,
I'll make sure the next Debian packages uploaded either revert this
change, or hack MAX_IFD_NESTING_LEVEL to be something ridiculously high,
like 250.  I'll test with some pro cameras lying around the house here to
see just how much higher we can get before I do that.


A second email from Adam Conrad:

Woo.  An image from my EOS 300D gets the counter up to 75.  I'm just going
to set it at 250 in the next upload. :)

You should still talk to upsteam about unbreaking this, as it LOOKS like
they're incrementing the counter at altogether the wrong bounday.

Reproduce code:
---------------
$url  =
"http://dpark.mitacf.org/pix/pics/mit04/danny/03-sly_foliage/img_0949.jpg";;
print("\n\ntest img 1 - canon powershot S45\n");
$data = exif_read_data($url);
print_r($data);
$url  =
"http://dpark.mitacf.org/pix/pics/mit05/janice/01-christmas/P1000415.JPG";;
print("\n\ntest img 2 - panasonic DMC-FZ20\n");
$data = exif_read_data($url);
print_r($data);
$url  =
"http://dpark.mitacf.org/pix/pics/mit04/fslee/08-usgl_beach_sunrise/IMG_0669.JPG";;
print("\n\ntest img 3 - powershot S1 IS\n");
$data = exif_read_data($url);
print_r($data);
$url  =
"http://dpark.mitacf.org/pix/pics/mit04/chking/01-ivcf_sound_closet/DSC00804.JPG";;
print("\n\ntest img 4 - sony cybershot\n");
$data = exif_read_data($url);
print_r($data);
$url  =
"http://dpark.mitacf.org/pix/pics/mit00/dannys_camera/08-burton5-asbestos-evac/dcp00884.jpg";;
print("\n\ntest img 5 - kodak DC210\n");
$data = exif_read_data($url);
print_r($data);

Expected result:
----------------
Output of the print and print_r statements.

Actual result:
--------------
Output of the print and print_r statements interspersed with
"<b>Warning</b>:  exif_read_data(imagefilename.jpg): corrupt EXIF header:
maximum directory nesting level reached in <b>/path/to/script</b> on line
<b>line in script</b>"

Each image produces two or three such warnings.

-- 
Edit bug report at http://bugs.php.net/?id=31986&edit=1
-- 
Try a CVS snapshot (php4):   http://bugs.php.net/fix.php?id=31986&r=trysnapshot4
Try a CVS snapshot (php5.0): 
http://bugs.php.net/fix.php?id=31986&r=trysnapshot50
Try a CVS snapshot (php5.1): 
http://bugs.php.net/fix.php?id=31986&r=trysnapshot51
Fixed in CVS:                http://bugs.php.net/fix.php?id=31986&r=fixedcvs
Fixed in release:            http://bugs.php.net/fix.php?id=31986&r=alreadyfixed
Need backtrace:              http://bugs.php.net/fix.php?id=31986&r=needtrace
Need Reproduce Script:       http://bugs.php.net/fix.php?id=31986&r=needscript
Try newer version:           http://bugs.php.net/fix.php?id=31986&r=oldversion
Not developer issue:         http://bugs.php.net/fix.php?id=31986&r=support
Expected behavior:           http://bugs.php.net/fix.php?id=31986&r=notwrong
Not enough info:             
http://bugs.php.net/fix.php?id=31986&r=notenoughinfo
Submitted twice:             
http://bugs.php.net/fix.php?id=31986&r=submittedtwice
register_globals:            http://bugs.php.net/fix.php?id=31986&r=globals
PHP 3 support discontinued:  http://bugs.php.net/fix.php?id=31986&r=php3
Daylight Savings:            http://bugs.php.net/fix.php?id=31986&r=dst
IIS Stability:               http://bugs.php.net/fix.php?id=31986&r=isapi
Install GNU Sed:             http://bugs.php.net/fix.php?id=31986&r=gnused
Floating point limitations:  http://bugs.php.net/fix.php?id=31986&r=float
No Zend Extensions:          http://bugs.php.net/fix.php?id=31986&r=nozend
MySQL Configuration Error:   http://bugs.php.net/fix.php?id=31986&r=mysqlcfg