Function bbox_count

tests/test_general.py:529–561

Make a page and insert identical text via different methods. Desired page rotation is a parameter. MediaBox and CropBox are chosen to be "awkward": MediaBox does not start at (0,0) and CropBox is a true subset of MediaBox.

(rot)

    (which is assumed correct) and TextWriter.write_text().
    """
    def bbox_count(rot):
        """Make a page and insert identical text via different methods.

        Desired page rotation is a parameter. MediaBox and CropBox are chosen
        to be "awkward": MediaBox does not start at (0,0) and CropBox is a
        true subset of MediaBox.
        """
        # bboxes of spans on page: same text positions are represented by ONE bbox
        bboxes = set()
        doc = pymupdf.open()
        # prepare a page with desired MediaBox / CropBox peculiarities
        mediabox = pymupdf.paper_rect("letter")
        page = doc.new_page(width=mediabox.width, height=mediabox.height)
        xref = page.xref
        newmbox = list(map(float, doc.xref_get_key(xref, "MediaBox")[1][1:-1].split()))
        newmbox = pymupdf.Rect(newmbox)
        mbox = newmbox + (10, 20, 10, 20)
        cbox = mbox + (10, 10, -10, -10)
        doc.xref_set_key(xref, "MediaBox", "[%g %g %g %g]" % tuple(mbox))
        doc.xref_set_key(xref, "CrobBox", "[%g %g %g %g]" % tuple(cbox))
        # set page to desired rotation
        page.set_rotation(rot)
        page.insert_text((50, 50), "Text inserted at (50,50)")
        tw = pymupdf.TextWriter(page.rect)
        tw.append((50, 50), "Text inserted at (50,50)")
        tw.write_text(page)
        blocks = page.get_text("dict")["blocks"]
        for b in blocks:
            for l in b["lines"]:
                for s in l["spans"]:
                    # store bbox rounded to 3 decimal places
                    bboxes.add(pymupdf.Rect(pymupdf.JM_TUPLE3(s["bbox"])))
        return len(bboxes)  # should be 1!

    # the following tests must all pass
    assert bbox_count(0) == 1