pdf_text_extraction_json_section_subsection

This repo contains codes for extraction of PDF text to JSON to show section number, section title, section body content, footnote

Document type

Legal

Types

3 Types

Thank you for providing the detailed sample of the Type 2 PDF. Based on your input, it's clear that the initial Type 2 parser needs enhancements to accurately capture the hierarchical structure of sections, chapters, and articles within the PDF. Below is the revised approach to handle Type 2 PDFs effectively.

Revised Approach for Type 2 PDF Parsing

Type 1

Certainly! Based on the provided sample of the Type 1 PDF, here's a clear understanding of its structure and key components:

Type 1 PDF Structure Overview

1. Document Hierarchy

The Type 1 PDF follows a hierarchical structure that organizes the legal document into Chapters, Articles, and Footnotes. Here's how each level is structured:

• Chapters

  • Designation: Each chapter is introduced with the keyword "Chapter" followed by a Roman numeral (e.g., "Chapter I").
  • Title: The chapter title follows the chapter number (e.g., "General provisions").

• Articles

  • Designation: Within each chapter, individual sections are designated as "Article" followed by an Arabic numeral (e.g., "Article 1").
  • Title: Each article has a specific title that summarizes its content.
  • Body Content: The main text of the article provides detailed provisions, regulations, or descriptions relevant to the title.

• Footnotes

  • Designation: Footnotes are introduced with the keyword "Footnote." and provide additional information, amendments, or references related to the preceding article.
  • Content: They often include references to laws, dates, or specific amendments that affect the article's provisions.

2. Detailed Structure Breakdown

Here's a more granular look at the structure based on the sample:

a. Chapter Level

• Example:

  Chapter I General provisions Article
  

• Components:

  • Chapter Number: "Chapter I"
  • Chapter Title: "General provisions"

b. Article Level

• Example:

  Article 1. Civil legal proceedings legislation of the Republic of Kazakhstan
  1. The procedure of legal proceedings in civil cases...
  2. International contractual and other obligations...
  

• Components:

  • Article Number: "Article 1"
  • Article Title: "Civil legal proceedings legislation of the Republic of Kazakhstan"
  • Article Body: Numbered paragraphs detailing specific provisions.

c. Footnote Level

• Example:

  Footnote. Article 1 as amended by the Law of the Republic of Kazakhstan dated...
  

• Components:

  • Footnote Identifier: "Footnote."
  • Footnote Content: Details about amendments, enforcement dates, or references to other laws.

Type 2

From your sample, the Type 2 PDF follows this hierarchical structure:

• Section: Denoted by SECTION n. TITLE
  • Chapter: Denoted by Chapter n. TITLE
    • Article: Denoted by Article n. TITLE
      • Content: Numbered paragraphs (e.g., 1. Some text)
      • Footnotes: Denoted by Footnote. ...

4. Integrating the Revised Parser

Update the process_pdf function to use the revised Type 2 parser.

def process_pdf_revised(pdf_path, pdf_type):
    text = extract_text_from_pdf(pdf_path)
    if pdf_type == 1:
        parsed_data = parse_type1(text)
    elif pdf_type == 2:
        parsed_data = parse_type2_revised(text)
    elif pdf_type == 3:
        parsed_data = parse_type3(text)
    else:
        raise ValueError("Unsupported PDF type")
    
    json_output = convert_to_json(parsed_data)
    return json_output

Example Usage

User must provide the pdf as below

pdf_path = "example-2.pdf"  # Chane the PDF as per the type 1 or 2 or 3

then the type of pdf such as 1, 2, 3

pdf_type = 2

Result

The result dirctory contains the json files for three example pdf 1, 2, 3 for correspodning output.

Licensing

This project is dual-licensed:

1. Non-Commercial Use:

   • Free for personal, educational, and non-commercial use under the terms outlined in the LICENSE file.

2. Commercial Use:

   • Requires a paid license. Contact Partha Pratim Ray at parthapratimray1986@gmail.com for more information.

Contact

For any licensing inquiries, please email parthapratimray1986@gmail.com.