run docling extraction for a particular page number

[ipalvikas]

Hey folks!
Is it possible to ask docling to extract text/table from a particular page number of a pdf document?

In my case, I have multiple pdfs of 60+ pages, but I want to deal with only couple of pages lets say 49,50. Can anyone please help me with this. Here is my code below:

def extract_text_docling_tesseractcli(source):
    
    from docling.backend.docling_parse_backend import DoclingParseDocumentBackend
    from docling.datamodel.base_models import InputFormat
    from docling.datamodel.pipeline_options import (
        AcceleratorDevice,
        AcceleratorOptions,
        EasyOcrOptions,
        OcrMacOptions,
        PdfPipelineOptions,
        RapidOcrOptions,
        TesseractCliOcrOptions,
        TesseractOcrOptions,
    )
    from docling.backend.pypdfium2_backend import PyPdfiumDocumentBackend
    from docling.document_converter import DocumentConverter, PdfFormatOption
    from docling.models.tesseract_ocr_cli_model import TesseractCliOcrOptions
    from docling.models.tesseract_ocr_model import TesseractOcrOptions
    
    # Docling Parse with Tesseract CLI
    # ----------------------
    pipeline_options = PdfPipelineOptions()
    pipeline_options.do_ocr = True
    pipeline_options.do_table_structure = True
    pipeline_options.table_structure_options.do_cell_matching = True
    pipeline_options.ocr_options = TesseractCliOcrOptions()
    pipeline_options.ocr_options.tesseract_cmd = '/filepath/tesseract/5.3.4/bin/tesseract'       

    doc_converter = DocumentConverter(
        format_options={
            InputFormat.PDF: PdfFormatOption(pipeline_options=pipeline_options)
        }
    )
    
    conv_result = doc_converter.convert(source)
    return conv_result.document.export_to_text()

text_from_pdf1 = extract_text_docling_tesseractcli(pdf_path)
text_from_pdf1

[Convl]

you can do that with the page_range parameter, which takes a tuple (start_page, end_page), e.g.

conv_result = doc_converter.convert(source, page_range=(5, 7))

[ipalvikas]

Thank you for sharing this information. I tried this, but it gives me validation error.
Am I doing it right?

from docling.document_converter import DocumentConverter

source = r'filepath/filename.pdf'
converter = DocumentConverter()
conv_result = converter.convert(source, page_range=(1,2))
conv_result.document.export_to_text()

Error message:
ValidationError: 1 validation error for DocumentConverter.convert
page_range

[ipalvikas]

Hello community friends,
Has anyone worked with page numbers in this context. Thanks for your help.

[ipalvikas]

Do you have any working example which you can share for my reference?
the above code for me did not work.

[kolhesamiksha]

Yes, with below docling version it's working for me

Docling version: 2.36.1
Docling Core version: 2.34.0
Docling IBM Models version: 3.4.4

working code:

try:
pipeline_options = PdfPipelineOptions()
pipeline_options.do_ocr = True
pipeline_options.do_table_structure = True
pipeline_options.table_structure_options.do_cell_matching = True
doc_converter = DocumentConverter(format_options={
InputFormat.PDF: PdfFormatOption(pipeline_options=pipeline_options)
})
result = doc_converter.convert(file_path, page_range=(1,1))
md_text = result.document.export_to_markdown()
return md_text
except Exception as e:
raise Exception(f"Error processing PDF: {e}")

[ipalvikas]

Thanks for sharing this!

Few things to note while using page_rank.

• Update docling to version. (current version 2.36.1)
• page_range only takes two inputs: start_page and end_page. (input should be a tuple)
• It will not work, if the your requirement is for multiple pages. It only works for start page to end page.

[JOSHMT0744]

When setting page_range with a powerpoint inputted, my output does not seem to be in a range. Instead it still processes the whole document. Is there a way to fix this?

[ipalvikas]

When setting page_range with a powerpoint inputted, my output does not seem to be in a range. Instead it still processes the whole document. Is there a way to fix this?

Do you have reference code which you have used? What you can do is, check if you did pass it in the form of tuple. it worked for me.