Wrong characters parsed with getText()

[karolisbarkauskas]

I have a code:

$parser = new Parser();
$pdf = $parser->parseFile(storage_path($file->full_path));

It's a simple code to parse file content, after doing ->getText() output is:

1\t\fü\x00\x03\x00\x1C\x00\x18\x00\x0F\x00\x16\x007\t0\x00\b\x00 \t\fü\x00\x03\x00\x1C\x00\x18\x00\x0F\x00\x16\x007\t\fü\x00\x03\x00\x1C\x00\x18\x00\x0F\x00\x16\x007\t\n
Costi di spedizione\t\fü\x00\x03\x00\x14\x00\x17\x00\x0F\x00\x1C\x009\t\fü\x00\x03\x00\x14\x00\x17\x00\x0F\x00\x1C\x009\t\fü\x00\x03\x00\x14\x00\x17\x00\x0F\x00\x1C\x009

Bus some part of pdf is being read normally.

Can you, please, help?

[machekhin]

I`ve make a little fix here:
#269

You can change this file in your project while this pull request will be merged

[Connum]

@OneSoftIO / @machekhin could you please provide a test file to reproduce this with? Because the fix for this breaks other PDF parsing processes and resulted in at least three new issues so far...

[machekhin]

@Connum , sorry, I can't show You this file even I find it, 'cause it's a commercial document.
Anyway, it was trouble with parse 'tab'-char \t. The previous version had a manual replace method and some special chars (like \f or \t) broke parsing.

[Connum]

That's unfortunate... I'm still unsuccessful in trying to produce a PDF with a tab or form feed character that will crash or fail with pdfparser prior to the fix...

[machekhin]

Maybe it was really \f. It is absent in the previous version:

4f4fd10

[Connum]

Maybe it was really \f. It is absent in the previous version:

4f4fd10

Yeah, probably, as \t was already present in the replacement code before.
I think the issue might be that stripcslashes() converts some but not all of the cases that the previous "manual" replacement did, therefore causing the issues with PDFs that worked fine before. So a combination of both could be the key to success. But we'll definitely need a PDF that shows the error when used with a version before the fix to get to the bottom of this...

I also fear that stripcslashes() might erroneously try to convert some cases that are not normally escaped in PDFs, but I'd need to find out more details about what that function does escape. The docs simply state

"Recognizes C-like \n, \r ..., octal and hexadecimal representation.

If that would indeed be the case, it could do more harm than it helps keeping the code simpler...

[Connum]

@machekhin If you do find the time and the PDF that failed, could you test if it is parsed fine when adding \f to the manual replacement code instead of using stripcslashes()?

            $text = str_replace(
                array('\\\\', '\(', '\)', '\n', '\r', '\t', '\f', '\ '),
                array('\\', '(', ')', "\n", "\r", "\t", "\f", ' '),
                $text
            );

That way, we could make sure that this issue can be closed again while resolving the other three issues as well.

[k00ni]

That way, we could make sure that this issue can be closed again while resolving the other three issues as well.

Can you list these issues here please (again)? Thanks.

EDIT: My bad, were they: #319, #322 and #334?