Preserve file paths on windows [#506526]

Running Apache/2.2.11 (Win32) PHP/5.2.9-1 on Server 2003 SP2.

When cron runs, the Recent log entries window is filled with "Could not extract any indexable text from" followed by the file name & path to the attached file.

It would seem that "escapeshellcmd" is misbehaving, though I'm not sure how to fix it (or describe it). If, at line 323 (or thereabouts) I remove that function call, the files get read and indexed.

Capturing the value of $cmd, with the function call in place I see:

	java -cp D: websites drupal6 sites all modules apachesolr_attachments tika -jar D: websites drupal6 sites all modules apachesolr_attachments tika/tika-0.3.jar -t D: websites drupal6 sites wiki.cfertech.com files CS4 Design Premium Read Me.pdf

With the function call removed:

	java -cp D:\websites\drupal6\sites\all\modules\apachesolr_attachments\tika -jar D:\websites\drupal6\sites\all\modules\apachesolr_attachments\tika/tika-0.3.jar -t "D:\websites\drupal6\sites\wiki.cfertech.com\files\CS4 Design Premium Read Me.pdf"

Comment	File	Size	Author
#10	escape-506526-10.patch	2.27 KB	pwolanin
#9	escape-506526-9.patch	2.39 KB	pwolanin
#7	escape-506526-7.patch	2.6 KB	pwolanin
#6	escape-506526-6.patch	984 bytes	pwolanin

Comments

Comment #1

dgarciad commented 9 August 2009 at 08:23

I think the problem is related to using "\" instead of "/" as it happens in windows.

I have also commented the escapeshellcmd:

// $cmd = escapeshellcmd($java .' -cp '. $tika_path .' -jar '. $tika .' -t "'. $filepath .'"');
$cmd = $java .' -cp '. $tika_path .' -jar '. $tika .' -t "'. $filepath .'"';

Maybe relacing every "\" with "/" will avoid troubles:

$cmd_tmp = $java .' -cp '. $tika_path .' -jar '. $tika .' -t "'. $filepath .'"';
$cmd = str_replace('\\', '/', $cmd_tmp);

Regards

Comment #2

cyberguyen commented 15 November 2009 at 12:48

Version:

6.x-2.x-dev

» 6.x-2.0-alpha1

As dgarciad suggest:

// $cmd = escapeshellcmd($java .' -cp '. $tika_path .' -jar '. $tika .' -t "'. $filepath .'"');

$cmd_tmp = $java .' -cp '. $tika_path .' -jar '. $tika .' -t "'. $filepath .'"';
$cmd = str_replace('\\', '/', $cmd_tmp);

This works very well on my windows installation of drupal so I would like the developer to implement something like it in the next release of Apache Solr Attachments.