how to convert scanned pdf to searchable pdf

[Laxmi530]

Hai,
Actually, I am trying to convert a scanned pdf or non-searchable pdf to searchable pdf by using pytesseract. I am able to do if the pdf is one page pdf when it becomes multiple page it is doing only the last page.
Can someone please help me is there any method available in PyMuPDF. Below is the sample code and I am using.

file = r"C:\Users\name\Downloads\Scanned_pdf.pdf"
file_pdf = open('Converted_searchable.pdf', 'ab')
text_data = []
with fitz.open(file) as pdf:
    for page in pdf:
        _, _, img = pdf_text_box_image(page)
        text_data.append(pytesseract.image_to_pdf_or_hocr(img))
file_pdf.write(bytearray(item for item in text_data))
file_pdf.close()

Thanking you advance.

[JorjMcKie]

There is more than one option:

1. You can use the imprtable version of OCRmyPDF
2. You can use PyMuPDF's built-in Tesseract support as follows:

import fitz

DPI = 150  # desired resolution

src = fitz.open("input.pdf")
doc = fitz.open()  # output PDF with text layer

for page in src:
    pix = page.get_pixmap(dpi=DPI)
    imgpdf = fitz.open("pdf", pix.pdfocr_tobytes())  # make 1-page temp PDF with text layer
    doc.insert_pdf(imgpdf)  # append page
    imgpdf.close()

doc.save("input-ocr.pdf")

[Laxmi530]

Thanks @JorjMcKie that will be a great work. Please let me know once you update the library.

Thanking you in advance.

[SkaarFacee]

Adding to this I wanted to know if we have scope to use a OCR other than tesseract say maybe easyOCR or paddleOCR ?

[JorjMcKie]

No - only Tesseract has an in-built API in (py-) MuPDF.

[SkaarFacee]

Hey @JorjMcKie 👋🏻, thanks for the reply. Are there any plans to make it possible to input an raw ocr response and then be able to continue the task ?

[JorjMcKie]

There are no such plans at the time being.