Extracting text from highlighted annotations in a PDF file

Question

Since yesterday I'm trying to extract the text from some highlighted annotations in one pdf, using python-poppler-qt4.

According to this documentation, looks like I have to get the text using the Page.text() method, passing a Rectangle argument from the higlighted annotation, which I get using Annotation.boundary(). But I get only blank text. Can someone help me? I copied my code below and added a link for the PDF I am using. Thanks for any help!

import popplerqt4
import sys
import PyQt4


def main():

    doc = popplerqt4.Poppler.Document.load(sys.argv[1])
    total_annotations = 0
    for i in range(doc.numPages()):
        page = doc.page(i)
        annotations = page.annotations()
        if len(annotations) > 0:
            for annotation in annotations:
                if  isinstance(annotation, popplerqt4.Poppler.Annotation):
                    total_annotations += 1
                    if(isinstance(annotation, popplerqt4.Poppler.HighlightAnnotation)):
                        print str(page.text(annotation.boundary()))
    if total_annotations > 0:
        print str(total_annotations) + " annotation(s) found"
    else:
        print "no annotations found"

if __name__ == "__main__":
    main()

Test pdf: https://www.dropbox.com/s/10plnj67k9xd1ot/test.pdf

Seale

Answer 1

Looking at the documentation for Annotations it seems that the boundary property Returns this annotation's boundary rectangle in normalized coordinates. Although this seems a strange decision we can simply scale the coordinates by the page.pageSize().width() and .height() values.

import popplerqt4
import sys
import PyQt4


def main():

    doc = popplerqt4.Poppler.Document.load(sys.argv[1])
    total_annotations = 0
    for i in range(doc.numPages()):
        #print("========= PAGE {} =========".format(i+1))
        page = doc.page(i)
        annotations = page.annotations()
        (pwidth, pheight) = (page.pageSize().width(), page.pageSize().height())
        if len(annotations) > 0:
            for annotation in annotations:
                if  isinstance(annotation, popplerqt4.Poppler.Annotation):
                    total_annotations += 1
                    if(isinstance(annotation, popplerqt4.Poppler.HighlightAnnotation)):
                        quads = annotation.highlightQuads()
                        txt = ""
                        for quad in quads:
                            rect = (quad.points[0].x() * pwidth,
                                    quad.points[0].y() * pheight,
                                    quad.points[2].x() * pwidth,
                                    quad.points[2].y() * pheight)
                            bdy = PyQt4.QtCore.QRectF()
                            bdy.setCoords(*rect)
                            txt = txt + unicode(page.text(bdy)) + ' '

                        #print("========= ANNOTATION =========")
                        print(unicode(txt))

    if total_annotations > 0:
        print str(total_annotations) + " annotation(s) found"
    else:
        print "no annotations found"

if __name__ == "__main__":
    main()

Additionally, I decided to concatenate the .highlightQuads() to get a better representation of what was actually highlighted.

Please be aware of the explicit <space> I have appended to each quad region of text.

In the example document the returned QString could not be passed directly to print() or str(), the solution to this was to use unicode() instead.

I hope this helps someone as it helped me.

Note: Page rotation may affect the scaling values, I have not been able to test this.

Part

Thanks, I struggled to install popplerqt4 but this worked like a charm! – Sachsse

Any good tutorial for installing poppler qt. I was having issues. #51665600 – Regen

This updated to extract the highlighted images gist.github.com/v-sukt/a59f549864d8369162a237e608251b07 – Giles

If someone has issues with dependencies use docker images or Vagrant at github.com/v-sukt/extract-pdf-notes – Giles