Problem in Page Binding of "Right to Left"

[mlove4u]

pypdfium2/src/pypdfium2/_helpers/text_extractor.py

Line 83 in 8ffcafc

def get_index(self, x, y, x_tol, y_tol):

When pdf page binding was set to "right to left", the text will be reversed after extracted.
I reported this problem to the pdfium team last year, but it doesn't seem to be solved so far. Although I don't know c++, I guess this can't be solved only inside pypdfium, right?
Test code:

import pypdfium2 as pdfium


def mytest(filename):
    doc = pdfium.PdfDocument(filename)
    page = doc.get_page(0)
    textpage = pdfium.PdfTextPage(page)
    text = textpage.get_text(45, 515, 375, 550) # pink rectangle
    return text


file1 = "left_to_right.pdf"
file2 = "right_to_left.pdf"
print(mytest(file1))  # 'To Every Thing There Is A Season'
# words are reversed
print(mytest(file2))  # 'Season A Is There Thing Every To'

left_to_right.pdf
right_to_left.pdf

[mara004]

I suppose you are referring to ViewerPreferences.Direction (L2R or R2L) as outlined in the PDF 1.7 Reference on page 371.
Perhaps you could use FPDF_VIEWERREF_GetName() to obtain the reading order and reverse the words if it's R2L?
That said, it would of course be better if the issue could be fixed in PDFium.

[mara004]

I've tried to create a patch for PDFium that would just remove the inversion: 0001-Remove-R2L-Inversion.patch.txt
Here's the output with the patch: output.txt
However, the change causes at least one failing test case, namely FPDFTextEmbedderTest.TextHebrewMirrored.
Apparently, the inversion is necessary in some cases to compensate flip, so disabling it completely as I did can't be the solution...

[mlove4u]

@mara004 Awesome.
Yes, it is "ViewerPreferences.Direction". I often export pdf from InDesign document, and it is called "page binding" in InDesign.
I used to use pymupdf and xpdf to extract text from pdf documents(these two works fine in R2L pdf), but because of their license, and then switched to use pdfium, and happened to stumble upon this problem.
Thanks.

vertical text layout sample pdf:
vertical_text_layout.pdf