Text Positioning on Scanned JPG to PDF files

[arjunpaudyal]

Text positioning
Text positioning is not proper for PDF files created from a JPG image that has unusual DPI as 600 DPI or 1200 DPI

Describe the bug (mandatory)

For PDF files as one Page as one Image scanned at a higher resolution and DPI (1200 / 600 ), the text position (Fitz.Point @ X0, Y0) does not happen. They are rotated (can be fixed with PageRotation set, but position is out of clue.

To Reproduce (mandatory)

• Scan any image from NAPS (Not Another Portable Scanner 2) and make a PDF from there.
• (PDF can be made from PymuPDF Image to PDF feature)
• Use SumatraPDF for figuring out the coordinate of point for text insertion. (Press M/m until you get in Points)
• Use PyMuPDF for text insertion. It does not happen at expected coordinate, and text is rotated.

Code

import fitz
doc = fitz.open('APPLICATION.pdf')  # new or existing PDF
import sqlite3

dbfile = "FormData.db"
conn = sqlite3.connect(dbfile)
curr = conn.cursor()

query = "SELECT * FROM data"
md = {}
curr.execute(query)
for row in curr.fetchall():
    rid, page, itemtype, location, radius, fontname, fontsize, fontcolor, text = row
    if not fontname:
        fontname = "helv"
    if not fontcolor:
        fontcolor='(0,0,1)'
        
    if not fontsize:
        fontsize=10
    
    if not page in md.keys():
        md[page] = []
    md[page].append((itemtype, eval(location), radius, fontname, fontsize, eval(fontcolor), text))
        
        
for pno,dt in md.items():
    page = doc[int(pno)]
    shape = page.newShape()
    
    #     old_rotation = page.rotation
    #     if old_rotation:
    #         page.setRotation(old_rotation)
    #     
    for data in dt:
    
        itemtype, loc, radius, fontname, fontsize, color, text = data
        
        rotate = -90
        
        if itemtype.upper() == "TEXT":
            location = fitz.Point(*loc)
            # insert text
            page.insertText(location, text, fontname=fontname, fontsize=fontsize, color=color, rotate=rotate)
            # rotate = 90
        elif itemtype.upper() == "TEXTBOX":
            # textbox
            location = fitz.Rect(*loc)
            page.insertTextbox(location, text, fontname=fontname, fontsize=fontsize, color=color, rotate=rotate)
        elif itemtype.upper() == "CIRCLE":
            location = fitz.Point(*loc)
            shape.drawCircle(location, radius)
        elif itemtype.upper() == "ELLIPSE":
            location = fitz.Rect(*loc)
            shape.drawOval(location)
        else:
            pass

    shape.finish()





doc.save("filled.pdf")
doc.close()

Expected behavior (optional)

Properly aligned text with the co-ordinated, that co-ordinates with page shape

Screenshots (optional)

Database

** Generated PDF **

(Text expected at the top or page, appears on the bottom of page. X seems to be okay, Y seems to be off. Text rotation is corrected by forcefully rotating the text by -90)

Your configuration (mandatory)

• Windows 10
• PyMuPDF 1.18, Installed using PIP/3
• Python 3.9
• Python version, bitness

Additional context (optional)

Never failed with Text based PDF, never succeeded with the Image to pdf kind of PDF files.

[arjunpaudyal]

It might not be a bug as well. I have faced similar problem in older versions (hard to remember) but it still exists without sufficient information on manual.

[JorjMcKie]

I am unable to reproduce the problem like that.
Please provide me a PDF and demonstrate that text does not land at the specified position with some script.
Before you do that: please confirm, that you have checked proper page geometry handling via page.is_wrapped and reacted via page.wrap_contents() where applicable.

[JorjMcKie]

Any more data on this?
Please provide the requested information - otherwise I am going to close this issue within the next few days.

[arjunpaudyal]

Current document I am working is confidential to a personal level. (it does not work for sure). I can email you, but not post herein public space.

I am trying to generate the problem at a re-producible level, and I failed with couple of attempts.

Give me some time. I will keep updated.

But I am sure - this is recurring issue for me.

Edit: I am relying on SumatraPDF for co-ordinate / points measure. Hope that is matured and reliable too.

[JorjMcKie]

Your problem comes from page rotations. You seem to be entering text as if rotation were zero - whch is of course bound to fail.
We continue the discussion offline via e-mail ...

[JorjMcKie]

I am going to transfer this issue to Discussions category "Looking for Help".