PDF-Helper

A simple python package that helps with doing simple stuff with PDFs.

Features

• Bundle: Bundle multiple files into one PDF
  • PDF inputs
  • Image inputs (e.g. PNG, JPG, etc.)
  • Markdown inputs
• Merge PDFs: Merge multiple PDFs into one PDF
• Split PDFs: Split a PDF into multiple PDFs, each containing a range of pages
• Export as image: Export designated pages from a PDF as image files
• Remove pages: Remove designated pages from a PDF
• Extract text: Export text from a PDF file and optionally save it to a text file
• Recipe system: Chain multiple operations together using YAML recipe files
• Add watermark: Overlay selectable vector text or an image on PDF pages
• Encrypt a PDF (user/owner passwords, AES-256/128, RC4)
• Decrypt a PDF (user or owner password)
• Extract images from a PDF
• Extract links from a PDF
• Set PDF metadata (title, author, etc.)

If you want any other feature to be added, feel free to open an issue or fork the repo and make a merge request after adding your contribution.

Usage

Installation

You can install PDF-Helper via pip:

pip install pdf-helper

# Or use uv to install the tool
uv tool install pdf-helper

Document operations (metadata editing, encryption, and the automatic metadata stamping) need the optional pikepdf dependency:

pip install "pdf-helper[pikepdf]"

Image operations (rendering pages to images, image watermarks, bundling or converting image files) need the optional Pillow dependency:

pip install "pdf-helper[pillow]"

Or everything at once:

pip install "pdf-helper[all]"

Everything else — bundling PDFs, splitting, text watermarking, text extraction, and recipes not using those operations — works without them. If a feature needs the missing extra, the command fails with the install instruction above instead of a traceback.

And run it using the command line:

pdf-helper <command> [options]

Or you can use uvx to run the package without installing it in a specific python environment:

uvx pdf-helper <command> [options]

You can also clone the repository and use uv run:

git clone https://gitlab.com/CodeWriter21/pdf-helper.git
cd pdf-helper
uv run pdf-helper <command> [options]

Bundle PDFs

Bundle multiple files into one PDF:

pdf-helper bundle <input_file_1> <input_file_2>... <input_file_n> <output_file>

# E.g. Bundle PDFs 1, 2 and 3 into a new PDF
pdf-helper bundle 1.pdf 2.pdf 3.pdf new.pdf

# E.g. Take 1.png, 2.jpg, and 3.png and create a PDF named 123.pdf and override
# if already exists
pdf-helper bundle 1.png 2.jpg 3.png 123.pdf -f

# E.g. Take part1.pdf, image1.png, ending.pdf and bundle them into a PDF named final.pdf
pdf-helper bundle part1.pdf image1.png ending.pdf final.pdf -v

Split PDFs

Split a PDF into multiple PDFs, each containing a range of pages:

pdf-helper split <input_file> <output_folder> -s <split_point_1>,<split_point_2>

# E.g. Split a PDF into three PDFs, one with pages 1-10, the second with pages 11-20 and
# the third with pages 21-end
pdf-helper split my-pdf.pdf my-split-pdfs -s 10,20

# E.g. Split a PDF into PDFs each containing one page
pdf-helper split my-pdf.pdf my-split-pdfs  # No need to specify split points

Export PDF pages as image files

Export PDF pages as image files:

pdf-helper to-image <input_file> <output_folder> \
        -p <page_number_1>,<page_number_2>,...,<page_number_n> -s <scale_factor>

# E.g. Export pages 1, 2, 3 and 6 from a PDF with scale factor 1
pdf-helper to-image 1.pdf images -p 1:3,6 -s 1

# E.g. Export the last three pages with scale factor 1
pdf-helper to-image 1.pdf images -p -3:-1 -s 1

# E.g. Export all pages from a PDF with scale 2
pdf-helper to-image my-pdf.pdf my-images

Remove pages from a PDF

Remove pages from a PDF:

pdf-helper remove-pages <input_file> <output_file> <page_number_1>,<page_number_2>,...,<page_number_n>

# E.g. Remove pages 1, 2, 3 and 6 from a PDF
pdf-helper remove-pages 1.pdf new.pdf 1:3,6

Add watermark to a PDF

Overlay selectable vector text or an image file on PDF pages:

# E.g. Stamp diagonal DRAFT text across all pages
pdf-helper watermark add my-pdf.pdf watermarked.pdf DRAFT

# E.g. Overlay a logo in the bottom-right corner of pages 1:3
pdf-helper watermark add my-pdf.pdf watermarked.pdf --watermark-image logo.png \
    --position bottom-right --image-scale 0.5 --pages 1:3

# E.g. Stamp 5% above the bottom edge, at the left edge
pdf-helper watermark add my-pdf.pdf watermarked.pdf "DRAFT" --position "0 -5%"

# E.g. Stamp text with a custom font (file path or installed name like Arial)
pdf-helper watermark add my-pdf.pdf watermarked.pdf "DRAFT" --font ./fonts/MyFont.ttf

# E.g. Stamp red text (CSS name, hex, or RGB tuple)
pdf-helper watermark add my-pdf.pdf watermarked.pdf "DRAFT" --color red

# E.g. Scale the text with the page: '50%' spans half the page width
pdf-helper watermark add my-pdf.pdf watermarked.pdf "DRAFT" --font-size "50%" --rotation 0

# E.g. Remove watermarks added by pdf-helper
pdf-helper watermark clear watermarked.pdf clean.pdf

# E.g. Swap one watermark for another in a single run
pdf-helper watermark replace watermarked.pdf new.pdf NEWMARK --remove-text DRAFT

Watermarks placed by pdf-helper carry an invisible mark, so clear removes exactly those; with text, matching text objects are stripped too (best effort).

--font-size accepts points (48) or a percent ("50%", quoted so the shell passes it through): a percent scales the text so its width spans that fraction of the page extent along the watermark direction — '50%' with --rotation 0 is a horizontal text half as wide as the page. Percents resolve per page, keeping mixed portrait/landscape documents proportional. Recipes accept the same values in font_size.

Text watermarks use Arial when it is installed (covering Latin, Greek, Cyrillic, Arabic and more), falling back to built-in Helvetica (Latin only) otherwise. RTL text (Persian/Arabic) is shaped into visual order automatically:

# E.g. Stamp a Persian watermark with an installed font
pdf-helper watermark add my-pdf.pdf watermarked.pdf "سلام" --font "Vazirmatn Regular"

Not sure which fonts are available? Query them (name or path both work with --font):

pdf-helper list-fonts
pdf-helper list-fonts arial

See examples/recipes/watermark-showcase.yaml for a recipe that chains a text watermark and an image watermark in one run.

Manage PDF metadata

Read, edit, or clear the metadata of a PDF file:

# E.g. Show all set metadata fields (accepts multiple files)
pdf-helper metadata show my-pdf.pdf other.pdf

# E.g. Set the title and author
pdf-helper metadata edit my-pdf.pdf tagged.pdf --title "Report" --author "Me"

# E.g. Remove all metadata (Info dictionary and XMP packet)
pdf-helper metadata clear my-pdf.pdf clean.pdf

Settable fields: title, author, subject, keywords, creator, producer, creationdate, moddate. Dates accept PDF date strings, ISO dates (2026-09-04, 2026-09-04 12:30), or datetimes, and show prints them in ISO format. Writes keep the XMP packet synchronized with the Info dictionary (and show prefers XMP values when present). Every file pdf-helper writes gets Producer (PDF-Helper <version>, unless already set), a CreationDate if missing, and a fresh ModDate automatically.

Encrypt and decrypt PDFs

Protect a PDF with a password, or remove the protection again:

# E.g. Require a password to open the file
pdf-helper encrypt my-pdf.pdf secure.pdf --password s3cret

# E.g. Acrobat-style lock: anyone may open, only the owner may edit
pdf-helper encrypt my-pdf.pdf secure.pdf --password "" --owner-password admin

# E.g. Decrypt with either password
pdf-helper decrypt secure.pdf open.pdf --password s3cret

The user password opens the file; the owner password (defaults to the user password) additionally grants full control. Passwords can also come from the PDF_HELPER_PASSWORD / PDF_HELPER_OWNER_PASSWORD environment variables, and are never echoed back. Algorithms: AES-256 (default), AES-128, RC4 (recipes additionally support per-operation permissions).

Every command that reads PDFs accepts --password (or PDF_HELPER_PASSWORD) for encrypted inputs — including metadata edit/clear, watermark, extract-text, and friends. In recipes, add a password: key (plain value or {input: name} reference) to any step that opens an encrypted file.

Export text from a PDF

To extract text from a PDF file and export them to text files you can do as follows:

pdf-helper extract-text <input_file> -o <output_file_name>

# E.g. Extract text from a PDF named my-pdf.pdf and save it to my-text.txt
pdf-helper extract-text my-pdf.pdf -o my-text.txt

Run Recipes

The recipe system lets you chain multiple PDF operations together in a single run using a YAML file. This unlocks features not available through individual CLI commands (e.g. selecting specific pages per file when bundling).

pdf-helper recipe run <recipe_file.yaml>

# E.g. Run a simple recipe
pdf-helper recipe run remove-pages.yaml

# E.g. Run with force overwrite and verbose logging
pdf-helper recipe run bundle-workflow.yaml --force --verbose

# E.g. Run a built-in recipe template instead of a file
pdf-helper recipe run --builtin remove-pages --input report.pdf --output cleaned.pdf

# E.g. Run a custom recipe from ~/.local/pdf-helper/recipes by name
pdf-helper recipe run --custom my-clean --input report.pdf --output cleaned.pdf

Custom recipes are plain YAML recipe files stored in ~/.local/pdf-helper/recipes (override with the PDF_HELPER_RECIPES_DIR environment variable) and run with the same --input/--output/--defines bindings. recipe generate lists them alongside the built-in templates.

Parameterized Recipes

Recipes can take their main input/output paths (and any extra values) from the command line instead of hardcoding them, so one recipe works on many files. Write {input}, {output}, or any {name} placeholder in the YAML and bind it at runtime:

name: "Remove specific pages"
version: "1.0"
parameters: [input, output]

steps:
  - id: clean
    operation: remove_pages
    input: "{input}"
    pages_to_remove: [2, 4, 6]
    output: "{output}"

pdf-helper recipe run clean.yaml --input report.pdf --output cleaned.pdf
pdf-helper recipe run clean.yaml --input a.pdf --output a-clean.pdf \
    --defines pages=1:3 --defines mode=strict

--input/--output bind {input}/{output}; repeatable --defines KEY=VALUE binds any other {name}. The optional parameters: list declares what a recipe expects — bare names are required, while name: default mappings provide defaults that CLI flags override:

parameters:
  - input
  - output
  - label: PDF-Helper
  - pages: "1,-1"

Path parts are available as {input.stem}, {input.name}, {input.suffix}, and {input.parent}, so outputs can be named after the input (e.g. output: "{input.stem}-pages"). Defaults may reference other placeholders, so - output: "{input.stem}-stamped-pages" works as an overridable default.

Repeat --input to pass multiple files as one array input: a bare {input} item inside a step list (e.g. bundle inputs: ["{input}"]) expands to every file. Array inputs need scalar outputs (path attributes like {input.stem} and embedding in longer strings are rejected with a clear error):

# E.g. Bundle two images, then stamp a label on every page
pdf-helper recipe run stamp-on-images.yaml --input a.png --input b.png \
    --output stamped.pdf --defines label=DRAFT

See examples/recipes/stamp-on-images.yaml for the full recipe.

Page selections use comma-separated items where start:end selects an inclusive range: 1:3,6 means pages 1, 2, 3 and 6. Numbers are 1-based (0 is rejected), and negatives count back from the last page, so 5:-2 means from page 5 to one page before the last. Values starting with - must use the --option=value form (--pages="-2:-1") so the CLI does not mistake them for flags. Built-in templates (recipe generate lists them) accept bindings the same way via --builtin <name>.

Recipe File Format

A recipe is a YAML file with a steps list. Each step has an id, an operation, input/output paths, and operation-specific options. Steps can reference each other's outputs using { step: step_id }.

name: "Remove specific pages"
description: "Removes pages 2, 4, 6 from a PDF."
version: "1.0"

steps:
  - id: clean
    operation: remove_pages
    input: document.pdf
    pages_to_remove: [2, 4, 6]
    output: cleaned.pdf

Supported Operations

Operations marked Planned are not yet implemented — the recipe runner logs a warning and copies the input file through, so pipelines don't break.

Watermark text is inserted as selectable vector objects; image watermarks are raster overlays. Both support position, opacity, rotation, and pages.

Advanced Example: Multi-step Pipeline

# yaml-language-server: $schema=https://gitlab.com/CodeWriter21/pdf-helper/-/raw/master/schemas/recipe-schema.json
name: "Split, Convert, and Extract Pipeline"
version: "1.0"

settings:
  temp_dir: "./.recipe-tmp"

steps:
  # Step 1: Split the PDF at pages 5 and 10
  - id: split
    operation: split_pdf
    input: report.pdf
    split_points: [5, 10]
    output_dir: .
    output_prefix: "report_part_"

  # Step 2: Convert the second chunk to images
  - id: to_images
    operation: pdf_to_image
    input:
      step: split
      file: report_part_2.pdf
    pages: "1:3"
    scale: 3
    output: ./output/images

  # Step 3: Extract text from the first chunk
  - id: extract
    operation: extract_text
    input:
      step: split
      file: report_part_1.pdf
    pages: "1:4"
    max_characters: 5000
    reverse_lines: true
    output: ./output/chapter-1-text.txt

Recipe Settings

Input values (e.g. passwords) can be sourced from environment variables or prompted at runtime:

inputs:
  password:
    env: PDF_PASSWORD
    prompt: "Enter output PDF password"

See examples/recipes/ for more example recipe files.

About

Author: CodeWriter21

GitLab: CodeWriter21/pdf-helper

Donations

Your donations are very welcome: nowpayments.io

You can also consider donating a Star to the repo.

License

This project is licensed under the MIT License.

See the LICENSE

References

• pypdfium2
• PILlow
• log21