Hebrew text displayed in reverse

[bredisvictor]

Hello, for first thank you for the great parser.
The issue that when i parsing documents in Hebrew, all the text displayed reverse.

Thank you.

[k00ni]

Can you provide a PDF (for our test suite) which causes this error? It must be free of charge and without any obligations.

[bredisvictor]

Hello Konrad,

Yes, sure, file attached.
test_hebrew.pdf

It will help me very much, because my solution that fix it, not so good.

Thank you.

[k00ni]

Thank you for the PDF.

It will help me very much, because my solution that fix it, not so good.

We had similar issues in the past. Can you post your solution here please? It might help to create a fix.

[bredisvictor]

My fix is very simple and not so good (need to think over all cases), I didn't have for this issue so much time (did it in an hour). I'm just filter output text and make reverse for every word, except for the text in english and numbers. The problem in this solution that some words stick together after parsing, and it not easy to separate it correctly. I don't think that it will help you.

[k00ni]

@bredisvictor can you please test again with #402?

I don't think that it will help you.

In my opinion it is better to have a half solution than nothing. Your code showed a way to deal with the problem. I am not sure but it might helped @smalot to create a patch faster.

[bredisvictor]

@bredisvictor can you please test again with #402?

I don't think that it will help you.

In my opinion it is better to have a half solution than nothing. Your code showed a way to deal with the problem. I am not sure but it might helped @smalot to create a patch faster.

I tested it and received the same result as before fix.
After debugging i found the reason. It never pass this case and condition.

As a result, it is not enters to the part of the code that does reverse.

And i did some temporary solution that work

This condition checks the hebrew characters in the string
mb_ereg('[\x{0590}-\x{05FF}]', $text)

[smalot]

Hi @bredisvictor
Did you try using the test_hebrew.pdf you provided ?
If so, can you provide screenshots about what you expect to obtain and what you really obtain using the issue-398 branch ?
Many thanks

[bredisvictor]

Hi @bredisvictor
Did you try using the test_hebrew.pdf you provided ?
If so, can you provide screenshots about what you expect to obtain and what you really obtain using the issue-398 branch ?
Many thanks

Hello Sebastien,

Yes, you are right, i am test it with other one, test_hebrew.pdf parsed fine. But some files not parsed.

Try this file:
resumes_sample_Servicerepresentative.pdf
Also please look on numbers, looks like they are randomly scattered.

In additional, i add one more example. In this case reverse is works, but words break into several parts.
hebrew_test_2.pdf

Thank you

[k00ni]

Can we use these PDFs in our test environment? They must be free of charge and without any obligations.

[bredisvictor]

These are resume examples in hebrew downloaded from free, open source.

[smalot]

Adding support for right to left language makes me crazy.
When I handle parts of text, they seems to be good, but when I'm echoing on terminal.
And the mix of hebrew and ascii chars is really awful

[bredisvictor]

Adding support for right to left language makes me crazy.
When I handle parts of text, they seems to be good, but when I'm echoing on terminal.
And the mix of hebrew and ascii chars is really awful

Can i help some how? Your parser very important for my business.

[GreyWyvern]

The Hebrew text in the sample documents are displaying in the correct direction. The only remaining issue left here is that they are spaced out unnaturally. This is a result of decodeText() not taking into account the scale factor from the text matrix Tm when deciding to add whitespace.