ExtractText yields nothing for apparently good PDF

[chrisinmtown]

PyPDF2 version 1.23 fails to extract any text from the first 3 pages of this PDF file:
http://emma.msrb.org/EP295293-EP10300-EP632440.pdf

The file seems well-formed to me; both Acrobat and evince display it nicely. The linux utility pdftotext converts it to text and I see the expected content just fine.

Here's the relevant bit of my little script:

    with open(filename, "rb") as pdf_file:
        try:
            pdf_obj = PdfFileReader(pdf_file)
            # gather properties
            prop_en = pdf_obj.getIsEncrypted()
            err = ""
            if not prop_en:
                # Look for any text on the first N pages
                prop_img = True
                prop_pg = pdf_obj.getNumPages()
                for i in xrange(min(prop_pg, 3)):
                    pagei = pdf_obj.getPage(i)
                    pageitext = pagei.extractText()
                    # Set property and stop searching at first text found
                    if len(pageitext) > 0:
                        prop_img = False
                        break

Is there a gotcha here that I'm missing? Pls advise, thanks in advance for help.

[chrisinmtown]

I would like to mention that I have many unprotected, machine-searchable (i.e., non-image) PDF files like this - I just posted one link. Unlike the last issue I opened about a freak PDF with a botched header, in this case PyPDF2 fails to get text from annoyingly many of the files I'm trying to process. Thanks for listening.

[zevaverbach]

@chrisinmtown I ran into a similar issue today; my PDF and yours are "page extraction: not allowed" according to Adobe Reader. :(

[chrisinmtown]

@zevav thanks for the comment but please let's not confuse issues.

Protected files are a whole different ball of wax and I don't expect PyPDF2 to extract anything from such files given no password.

The link I provided above yields a PDF that is not password protected. On this document Adobe Acrobat makes no complaint about extracting text, it happily saves-as plain text and the result is totally usable.

[zevaverbach]

@chrisinmtown hey, sorry to send you in the wrong direction; in Acrobat on my machine that document does show (document info) as "page extraction not allowed."

[chrisinmtown]

Thanks for clarifying. Now I'm concerned, I don't want to waste anyone's time here on non-issues!

I am using Adobe Acrobat XI on Win7_x64. With this document open in Acrobat I pick File -> Properties, switch to the Security tab of the Document Properties dialog, and there I read "Security Method: No Security", and under the restrictions everything is allowed (Printing, Changing, Copying ...).

Could there possibly be a difference in behavior between Reader and Acrobat on this document?

[zevaverbach]

Okay, my bad: I wrote "Acrobat" in my second comment, but I meant "Reader." Here's a screenshot of your file's info in that, on OS X.10, Reader 11.0.10.

[chrisinmtown]

I see the exact same thing in the Win7 version of Acrobat Reader XI: Document Assembly and Page Extract Not Allowed; all the rest (Content Copying ..) are Allowed. FWIW, PyPDF2 declares this document unprotected.

I'm starting to think the Properties window is reflecting features of Acrobat Reader rather than the document, do you agree? In my tests of Reader on other PDF documents, it invariably declares "Page Extract Not Allowed". Reader by definition cannot extract pages, right?

Just to be clear, I am sticking to my position :) that the original document is a valid PDF, unprotected, with text content, and I really would like PyPDF2 to be extended so it can handle this doc.

[zevaverbach]

Dang, you're right! I didn't think to check a PDF that I know PyPDF2 can extract the text of; Reader does indeed show that property for all PDFs. :(

What method in PyPDF2 tells you whether or not a document is protected?

[chrisinmtown]

The relevant method on PdfFileReader is getIsEncrypted()

[Rob1080]

I realise this is an old post, did you ever find the reason for text not being extracted?

[droid-surbhi]

Facing same problem. PyPDF2 version 1.26

[MartinThoma]

Sadly, the PDF mentioned above is no longer reachable. I think that #924 fixed the issue and hence I close this PR.

It might also be a duplicate of the underlying cause of #242 .

If you face the same issue, please open a new bug ticket and upload a PDF with the issue (to which you must have the copyright)