Invalid xref table - error

[IanMcP]

Hi all,

I am using exiftool in an Applescript workflow which processes PDFs from a hotfolder, adding information from our MIS database. The files are delivered by Esko Backstage workflows.
I am periodically getting errors from exiftool as follows:

Code Select Expand

Error: Invalid xref table - /path/to/file.pdf
Error: Can't find Main object - /path/to/file.pdf

This has only occurred for files from 1 client account - all the error PDFs are version 1.4, although PDFs of various versions go through the workflow.
The files work eventually if they are put back through the hotfolder.

Has anyone come across this before? I have found very little information about xref tables and none of it involved exiftool.

Let me know if you need any more info
Thanks for any help

Ian

Exiftool is current (8.29) and is running on an Xserve (10.6.4 server, 2x2.66 quad, 12gb ram)

[Phil Harvey]

Hi Ian,

If you can send me a sample file that is causing problems I will take a look.  My email is philharvey66 at gmail.com

The xref table is the main directory in PDF files which tells where to find all other information.  If the file is not corrupt, then maybe there is a variation of the PDF format which exiftool isn't properly supporting.  But I need to see the file to be able to analyze this.

- Phil

[IanMcP]

Thanks Phil. Email sent

[Phil Harvey]

Hi Ian,

Thanks for the samples.  I tried reading and writing these files and didn't have any problems.

Could you give me a specific command line which demonstrates the problem?

Thanks.

- Phil

[Phil Harvey]

I just though of another possiblity:

Could it be that you are running exiftool before the file has been fully transfered (ie. a slow client pushing the file onto your server)?  You will get the "Invalid xref table" error if you run exiftool on a truncated PDF since the xref table comes at the end of the PDF.

Also, if I understand you correctly you can re-run exiftool on these files and the results are not consistent.  It is hard to imagine that exiftool could be the source of this inconsistency -- more likely that the source file changed between the two invocations.

- Phil

[IanMcP]

Phil,
Thanks for taking a look and the suggestions.
I have got a section of the script that should check that the file has finished copying, but I'll check it is doing it's job.
It seems a bit strange that it is only affecting 1 type of file (1.4) from 1 client account - I'll try and get them to output the same version with the same options as the other clients.

I'm not in work for a few days now, so I haven't got access to the code, but it is something like:

Code Select Expand

do shell script "exiftool -config path/to/file.txt -overwrite_original -v2 -barcode='123456' etc
The script stores values from the database in variables such as barcode, job title, etc and loads them all into exiftool in that one shell command. A custom namespace is setup, and all the database fields we require are stated in the custom config.

I'll take another look next week and try your suggestions

Thanks

Ian