OCR pdf file in python on the fly

With   PyMuPDF and tesserocr you can OCR image pdf easily

Requirements:

pip install PyMuPDF tesserocr

some pdf with (scanned images ;-) e.g. Tolstoy - GOD SEES THE TRUTH, BUT WAITS.pdf

Code

import fitz
import tesserocr
from PIL import Image


def get_PIL_image(pix):
    """Convert fitz pix to PIL.Image."""
    img = Image.frombytes("RGB",
                          [pix.width, pix.height],
                          pix.samples)
    return img


def binarize_image(image, thresh=100):
    """Custom image binarization."""
    binarized = image.convert("L").convert("1")
    return binarized

input_pdf = "Tolstoy - GOD SEES THE TRUTH, BUT WAITS.pdf"
output_name = "tolstoy.txt"

zoom = 2  # to increase the resolution
mat = fitz.Matrix(zoom, zoom)

doc = fitz.open(input_pdf)
ocr_api=tesserocr.PyTessBaseAPI(
    path=r"tessdata_best\tessdata"
    )
ocred_images = []

for page_idx, page in enumerate(doc):
    pix = page.getPixmap(matrix=mat)
    pil_image = get_PIL_image(pix)
    binarized = binarize_image(pil_image)
    print(f"OCRing page {page_idx+1} of {doc.page_count}...")
    ocr_api.SetImage(pil_image)
    ocred_images.append(ocr_api.GetUTF8Text())
doc.close()
ocr_api.End()

with open(output_name, mode='w', encoding='utf-8') as outfile:
    for page in ocred_images:
        outfile.write(page)