Anonymous Monk has asked for the wisdom of the Perl Monks concerning the following question:

I've got an application that accepts PDF file uploads and stores them in a BLOB column in MySQL. Another application pulls data from the database and sends it to the browser. When the database is queried the BLOB column is written to a file in a public HTML directory and a link is sent to the browser.

My problem is that when users click on the link the Acrobat plug-in loads and then errors out with "Object Label Badly Formatted". If I move the PDF to the directory myself then it loads correctly in the browsers. ls -Fla on the directory shows that the two files (uploaded through the application and copied in with Samba) have sizes that differ by 8 bytes (the uploaded file is larger).

Any ideas of where to look? Here's the code that is accepting the upload:
    my $pdf_name = $cgi_query->param('event_pdf');
    my $pdf = 'EventPDF';
    if ($pdf_name){    #A file (supposedly a PDF) has been sent in the
+ post data.  Load it into $pdf for storage in events table
        binmode($pdf_name);
        while (<$pdf_name>){
            $pdf .= $_;
        }
        $pdf = $db->quote($pdf);
    }
    my $mime_type;
    #(Not very robust) type checking for uploaded file.  This is depen
+dent on the browser behaving
    $mime_type = $cgi_query->uploadInfo($pdf_name)->{'Content-Type'} i
+f $pdf_name;
    if ($mime_type){
        return $self->display_error("Attached file is not a PDF\n") un
+less $mime_type eq 'application/pdf';
    }

[download]


Thanks!
--TWH

Replies are listed 'Best First'.
Re: Object Label Badly Formatted
by digger (Friar) on Mar 05, 2005 at 02:46 UTC

    You are telling perl to read your pdf line by line, and then putting the line into a variable. With binary data, a byte sequence that looks like a line ending (0A on unix, 0D0A on Windows) could appear anywhere in your file. What you end up with is binary data that is, for lack of a better term, mangled.

    You can get around the problem by doing this 

    my $readbuf;

  while (
    read (INF, $readbuf, 32767)    # read in (up to) 32k chunks
)
{$pdf .= $readbuf}

    [download]

    Now, instead of reading to a newline, which is probably a random byte somewhere inside the pdf, you are reading in 32k bytes at a time. Your pdf stays in tact, and you shouldn't have any trouble.

    HTH, digger

[reply]
[d/l]
Re: Object Label Badly Formatted
by Anonymous Monk on Mar 05, 2005 at 01:00 UTC
    I fixed the problem by changing:
    while (<$pdf_name>){
            $pdf .= $_;
        }

    [download]

    to:
    $pdf = do {local($/) ; <$pdf_name> };

    [download]

    but I don't quite understand why that fixes the problem. Any explanation would really be appreciated.
    Thanks!
    --TWH
[reply]
[d/l]
[select]

      I'm not sure that while (<$pdf_name>) works properly in binmode. You should be using read or sysread. In other words, don't read line-by-line when you've just told perl that it's a binary file. The second (new) code snippet doesn't read by line - it slurps the entire file as-is, with no translation. To get your while to work, just set local $/ = \1024 to read in 1024-byte chunks.

[reply]
[d/l]
[select]

Back to Seekers of Perl Wisdom