PDF

Working with PDFs (.pdf)

You read and build PDFs by writing Python code, not by filling in a fixed schema. The docgen tool's run_python runs your code in the app's bundled interpreter, which has these libraries pre-installed:

• fitz (PyMuPDF) — the one tool for almost everything: extract text and tables, merge/split/reorder/rotate/delete pages, watermark, extract images, create new PDFs, render pages to PNG, and OCR scanned pages.
• PIL (Pillow) — image ops (resize, convert, compose) for images you pull out of or drop into a PDF.
• pptx, docx, openpyxl, pandas — if the source or target drifts to Office formats. To turn one of those into a PDF, export via LibreOffice (see the "Office → PDF" note below).

run_python is real execution with full filesystem access. print(...) is captured and returned to you. Iterate: write code → run → inspect → fix.

There is no pypdf / reportlab / pdfplumber in the bundle — don't import them. PyMuPDF covers all of those use cases; use fitz.

Reading / extracting text

To dump text, use the files read tool on the .pdf, or write a script:

import fitz
doc = fitz.open("/path/to/document.pdf")
for i, page in enumerate(doc, 1):
    print(f"## Page {i}")
    print(page.get_text())

page.get_text("text") gives plain reading order; "words" / "dict" / "blocks" give positions when you need layout. For a scanned (image-only) PDF get_text() returns empty — see OCR below.

Extracting tables

import fitz
doc = fitz.open("/path/to/document.pdf")
for i, page in enumerate(doc, 1):
    for t, table in enumerate(page.find_tables().tables, 1):
        print(f"## Page {i} table {t}")
        for row in table.extract():
            print(row)
        # table.to_pandas() gives a DataFrame if you want to write .csv/.xlsx

Merge / split / reorder / delete pages

import fitz

# Merge several PDFs into one
out = fitz.open()
for path in ["a.pdf", "b.pdf", "c.pdf"]:
    with fitz.open(path) as src:
        out.insert_pdf(src)
out.save("/abs/path/merged.pdf")

# Split: one file per page
src = fitz.open("input.pdf")
for i in range(src.page_count):
    one = fitz.open()
    one.insert_pdf(src, from_page=i, to_page=i)
    one.save(f"/abs/path/page_{i+1}.pdf")

# Reorder / subset: select() takes the new page order (0-based)
src.select([2, 0, 1])          # keep+reorder these pages
src.delete_page(0)             # or delete_pages(from_page=, to_page=)
src.save("/abs/path/reordered.pdf")