PDF Book to Markdown Converter

A Windows-ready Python workflow for converting a directory of PDF books into simple Markdown with OCR fallback, table extraction, figure extraction, metadata detection, and master index files.

---

What this does

This project scans a folder of PDF books and attempts to:

• detect title, author, and ISBN from the first 10 pages
• apply a consistent naming convention
• convert each book into simple Markdown
• use OCR on low-text or scanned pages
• extract tables into CSV files
• extract figures/images into PNG files
• guess figure and table captions from page text
• create a per-book manifest.json
• create a master index.csv
• create a master index.md
• log failures into failures.json

---

Best use case

This works best for:

• digital PDFs with selectable text
• mixed PDFs where some pages are scanned
• building a searchable working archive of books

This is less reliable for:

• heavily scanned books with poor image quality
• complex multi-column layouts
• highly formatted textbooks with complicated tables
• pages where captions are far away from figures

---

Naming convention

Each book gets a stable folder name.

If ISBN is found

isbn_<isbn>

Example:

isbn_9780323445481

If ISBN is not found

<author_slug>__<title_slug>

Example:

wyatt_crosby__crazy_busy_doctor

---

Output structure

output/
  index.csv
  index.md
  failures.json
  isbn_9781234567890/
    isbn_9781234567890.md
    manifest.json
    assets/
      tables/
        page_0005_table_01.csv
      figures/
        page_0007_img_01.png

---

FULL SETUP GUIDE (BEGINNER)

Follow these steps exactly. You only need to do this once.

Step 1 — Install Python

1. Go to: https://www.python.org/downloads/

2. Download Python 3.11 or newer

3. During installation, check this box:

Add Python to PATH

4. After installing, open PowerShell and verify:

python --version

Expected output:

Python 3.11.x

Step 2 — Install Tesseract OCR (Required for scanned PDFs)

1. Download Tesseract OCR for Windows.
2. Install it using default settings.

Default install location:

C:\Program Files\Tesseract-OCR\tesseract.exe

You may need this path later.

Step 3 — Create your project folder

Open PowerShell:

mkdir pdf-book-md-converter
cd pdf-book-md-converter

Step 4 — Create the script file

Create a file named:

convert_books.py

Paste the full Python script into this file.

Step 5 — Create virtual environment

python -m venv .venv

Step 6 — Activate environment

.venv\Scripts\Activate.ps1

If you get a permission error, run:

Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass
.venv\Scripts\Activate.ps1

Step 7 — Install dependencies

pip install pymupdf pdfplumber pandas openpyxl pillow pytesseract

Step 8 — Run the script

Basic run:

python convert_books.py --input "D:\books" --output "D:\books_md"

If Tesseract is not detected:

python convert_books.py --input "D:\books" --output "D:\books_md" --tesseract-cmd "C:\Program Files\Tesseract-OCR\tesseract.exe"

---

What the script is doing (simple)

For each PDF:

1. scans first 10 pages
2. detects:
   • title
   • author
   • ISBN
3. creates a consistent folder name
4. extracts text
5. uses OCR if needed
6. removes repeated headers/footers
7. converts text to Markdown
8. extracts:
   • tables to CSV
   • images to PNG
9. detects captions
10. saves:

• markdown file
• manifest.json

Then creates:

• index.csv
• index.md
• failures.json

---

Tables and Figures

Tables

• saved as CSV
• linked in Markdown
• not flattened into messy text

Figures

• saved as PNG
• embedded in Markdown
• captions included when detected

---

Example run

python convert_books.py --input "D:\medical_books" --output "D:\md_books" --tesseract-cmd "C:\Program Files\Tesseract-OCR\tesseract.exe"

---

After setup (future runs)

Only run:

.venv\Scripts\Activate.ps1
python convert_books.py --input "D:\books" --output "D:\books_md"

---

Limitations

Expect weaker results for:

• scanned low-quality PDFs
• complex tables
• multi-column layouts
• distant captions
• poor metadata

---

Practical recommendation

Use this for:

• building a clean archive
• creating searchable datasets
• preprocessing for AI

Not for:

• perfect formatting reproduction

---

Suggested repo name

pdf-book-md-converter