Unrecognized File Type on PDF files

Hey there. Fileframework should be using mime types to "recognize" files of different types.

Whether or not it can perform any actions on the file(s) would have to do with the ability/availability of tools to work on them - for instance it can only convert a pdf to plain text if you have the pdftotext utility installed on your system.

I've had issues with mime type recognition and webdav, but if you are uploading via web browser - the default settings are that Fileframework will get mime type info from you (your browser).

Look at admin/settings/file/mime - is pdf listed there?

If you are using webdav - and even if not, I've got an issue open #406182: Is it possible to rebuild file types (mime.types)? which is somewhat related...I'm trying to figure out how to get Fileframework to take another look at files which have already been added.

I am able to upload pdfs and dwf's without issue, it properly recognizes .dwf as model/vnd.dwf although fileframework doesn't do anything special with dwf's....you can download them :)

Hope that helps.

You *might* try setting 'Server-side MIME type autodetection after upload' to 'Enabled' at admin/settings/file, but I haven't had much luck with that setting myself.

How this module handles MIME data isn't something I understand very well, so hopefully miglius will chime in on this soon.

Here is how it works in the following steps:

1. File arrives with the MIME type sent by the browser which has uploaded the file.
2. If the MIME is missing or if it is unrecognized ('application/octet-stream'), then the MIME is guessed by the file extension using the mime.types file for the mapping.
3. If Server-side MIME type autodetection is enabled or set to conditional, the system does not trust the mime type the file has so far and MIME is being redetected on the server using php fileinfo extension or shell 'file' command.

The server side detection is used to avoid situations when a browser sends a wrong MIME type and a user has renamed the file by adding a false extension. This situation means that a wrong file conversion utilities would be used to build a derivative files. From my experience though, the fileinfo extension fails to correctly detect some mime types and therefore should be used with caution.

I'm planning to step away from the mime.types file and use more flexible and configurable mapping, see #282955: Use other file MIME detection instead of the mime.types file.

The changes will affect only new files, but I see your point. If the file is already uploaded and them MIME is wrong, after changing the MIME table one has to reupload the file to set the proper MIME.

The redetection of MIME for currently uploaded files could be implemented, but I would need to think more what would be better - to do it per individual file, or should the script redetect MIMEs for all uploaded files in the system, which can be process intensive, since if on server autodetection is enabled, the file should be pulled from the bitcache, the MIME redetected and all new file conversions run again.

The way a MIME type is recognized has been changed drastically recently as of #282955: Use other file MIME detection instead of the mime.types file. Can you test the latest code to see if it solves the issue?

I had a test on this: it works only with visualnotion's modification on the current dev version.

In addition I managed to get ff to convert pdf to txt (and zip) as well:

add this to the same file (file_document.module) to the end of the function file_document_mime_converters() around line 200:

    'application/pdf' => array(
      'text/plain' => array(
        'pipeline' => '{pdftotext} "[in_file]" "[out_file]"',
        'handlers' => array('file_text_text'),
      ),
      'application/zip' => array(
        'pipeline' => 'pwd="`{pwd}`"; {mkdir} "[in_file].tmp"; {mv} "[in_file]" "[in_file].tmp/[filename]"; cd "[in_file].tm$
      ),
    ),

One flaw: this does NOT ocr any image that may be in your pdfs, but just extracts text.

Not sure what this issue is about.
pdf are well recognized and converted. See for example
http://web2.see.asso.fr/en/node/1922

hi ZyanKLee,
this lines of code already in

/formats/slideshow/file_slideshow.module

near line 100:

'application/pdf' => array(
      'text/plain'.....

and in

/formats/archive/file_archive.module

near line 113:

 'application/zip' => array(
      'text/plain' => array(
        'pipe

i think the problem was in older fileframework 6.x-1.0-alpha3
u only need to install the file_slideshow and file_archiv module
then the application/pdf appears in mime list (/admin/settings/file/mime) and as
converter (/admin/settings/file/convert/converter)