pymd2pdf 0.2.0

Automated programmatic Markdown-to-PDF typesetting engine

Automated Programmatic Markdown-to-PDF Typesetting Engine

md2pdf converts structured Markdown documents into beautiful, print-ready PDFs. Unlike other conversion tools, it does not rely on heavy system dependencies like Pandoc, Node.js, or headless Chrome/Chromium browsers. It is written in pure Python and powered by ReportLab and mistletoe.

---

Architecture Overview

md2pdf is designed as a pipeline that runs in four distinct stages: preprocessing, parsing/validation, element rendering, and layout composition.

graph TD
    Input[Markdown File] --> Pre[Preprocessors]
    Pre --> Parser[MarkdownParser]
    Parser --> Tokens[Token Stream]
    Tokens --> Val[DocumentValidator]
    Tokens --> Reg[Plugin/Handler Registry]
    Reg --> Handlers[Element Handlers]
    Handlers --> Kroki[Kroki API / Cache]
    Handlers --> Flowables[ReportLab Flowables]
    Flowables --> Layout[LayoutComposer]
    Layout --> Post[Postprocessors]
    Post --> Renderer[ReportLab PDF Engine]
    Renderer --> Output[Output PDF]

---

Key Features

• Standard Elements: Headings (H1–H6), paragraphs, lists, blockquotes, horizontal rules, and hyperlinks.
• Multi-page Tables: Tables split cleanly across page boundaries. Headers repeat at the top of every page.
• Diagrams & Math Blocks: Renders Mermaid diagrams and LaTeX math blocks via the Kroki API, with automatic transparent/white margin cropping, offline fallbacks, and SHA-256 disk caching.
• PDF Bookmarks & Outline Panel: Every heading (H1–H6) is registered as a clickable, correctly nested entry in the PDF viewer's bookmarks/navigation panel.
• Broad Unicode Support: Bundled DejaVu Sans fonts cover Latin Extended, Greek, Cyrillic, mathematical operators, arrows, box-drawing characters, and more — no system font dependency required.
• Extensible Plugin System: Load custom element handlers, text-level preprocessors, post-processors, and stylesheet/theme layers.
• Typesetting Safeguards: Implements strict "anti-fail" layout rules including orphaned heading protection, ghost page elimination, and widow/orphan line settings.
• DX-First Validation: Pre-render validation runs to identify nested tables, empty diagrams, or unsupported elements before rendering.

---

Tech Stack

Component	Library/Tool	Description
Core Language	Python >= 3.11	Modern Python with strict type-hinting
PDF Generation	ReportLab >= 4.0	Low-level document layout and flowable engine
Markdown Parsing	mistletoe >= 1.3	Fast and extensible Markdown AST parser
HTTP Requests	requests >= 2.31	Handles communication with Kroki API
CLI Framework	typer >= 0.12	CLI builder for options and validation errors
Image Processing	pillow >= 10.0	Auto-cropping and dimension detection for diagrams

---

Project Structure

md2pdf/
├── docs/                   # Developer documentation
│   ├── plugin-authoring.md # Instructions for writing plugins
│   └── themes.md           # Themes and stylesheet reference
├── md2pdf/                 # Core source directory
│   ├── assets/             # Kroki client, caching, and fallback elements
│   ├── core/               # Engine pipeline, parser, validator, layout, registry
│   ├── handlers/           # Element-specific flowable generators (headings, tables, etc.)
│   ├── styles/             # Default stylesheet and theme configs
│   └── cli.py              # CLI entry point
├── tests/                  # Automated test suite
│   ├── fixtures/           # Markdown and configuration test files
│   ├── test_cli.py         # CLI integration tests
│   └── test_pipeline.py    # Pipeline validation tests
├── pyproject.toml          # Build system and dependency declaration
└── README.md               # Project overview

---

Logic Flows

The diagram below details the sequence of execution inside the Pipeline class:

sequenceDiagram
    autonumber
    actor CLI as CLI / User
    participant PL as Pipeline
    participant PR as Preprocessors
    participant MP as MarkdownParser
    participant DV as DocumentValidator
    participant REG as PluginRegistry
    participant LC as LayoutComposer
    participant PO as Postprocessors
    participant RL as ReportLab Engine

    CLI->>PL: run(raw_md)
    PL->>PR: run_all(raw_md)
    PR-->>PL: preprocessed_md
    PL->>MP: parse(preprocessed_md)
    MP-->>PL: token_stream
    PL->>DV: validate(token_stream)
    DV-->>PL: validation_issues (warnings/errors)
    PL->>REG: render(token_stream)
    REG-->>PL: raw_flowables
    PL->>LC: compose(raw_flowables)
    LC-->>PL: layout_guarded_flowables
    PL->>PO: run_all(layout_guarded_flowables)
    PO-->>PL: finalized_flowables
    PL->>RL: build(finalized_flowables)
    RL-->>CLI: output.pdf

---

Installation & Setup

Using uv (recommended):

uv tool install pymd2pdf

Or via standard pip:

pip install pymd2pdf

To initialize the project for local development:

# Clone the repository
git clone https://github.com/user/md2pdf.git
cd md2pdf

# Create virtual environment and install all dependencies (including dev)
uv sync
source .venv/bin/activate

---

Usage Examples

Command Line Interface

Convert a Markdown file:

md2pdf input.md -o output.pdf

Execute pre-render validation checks without producing a PDF:

md2pdf input.md --validate-only

Run in offline mode to avoid calling the Kroki API (places image boxes with source code in the PDF instead):

md2pdf input.md -o output.pdf --offline

CLI Options

Flag	Shortcut	Description
--output	-o	Path to save the output PDF file (default: output.pdf).
--config	-c	Path to a custom md2pdf.toml config file.
--theme	-t	Name of the theme to apply (default: default).
--offline		Skip external API requests (e.g. Kroki diagram rendering) and use local placeholders.
--validate-only		Execute pre-render validation checks and exit without building a PDF.
--verbose	-v	Output debug-level logging to stderr.

Programmatic Python Usage

from md2pdf import convert, Config, Pipeline

# Option 1: Simple conversion
convert("input.md", "output.pdf")

# Option 2: Advanced programmatic pipeline usage
config = Config(
    offline=False, 
    cache_dir=".md2pdf_cache", 
    output_file="my_document.pdf"
)
pipeline = Pipeline(config)

# Validate markdown document
issues = pipeline.validate("# Hello World")
for issue in issues:
    print(f"[{issue.severity}] {issue.code}: {issue.message}")

# Render markdown
pipeline.run(raw_md="# Document Title\n\nSome body text.")