I still have one concept. 

I have to first do the tests. 

Then if this will have any useful information. then I will share them.

It is because the text extraction routines recursively parses the XObjects and takes care of all the scaling and translation that may occur.  Proper text extraction is a complex process.  Our text extraction is actually using the page rendering code to extract the text.  There is no simple way to extract text for all PDF files.  There are many different ways to draw text in a PDF. 

Andrew.

mLipok,

That is a perfectly valid conten stream.

This is all documented in the PDF Specifications.

it is:

http://www.autoitscript.com/autoit3/docs/intro/lang_operators.htm

So how yous see it: adds a new string without deleting the previous content

But now I still think the problem is here: 

There you can read "...This function selects one of the selected page's content stream parts....".

If I understand you that you mean something like this:

you have to select the relevant content stream you wanna have first:

Cheers, Ingo

I have a document from the bank - the confirmation of the order.

When I retrieves text using ExtractFilePageText, I receive 1701 bytes of text.

But when I use:

BinaryToString ($oQP.GetContentStreamToVariant())

I receive only a short piece:

Below string represtnation of this binary data:

When I use ContentStreamCount then I get in results = 1