tagz

A lightweight, type-safe Python library for building and parsing HTML documents programmatically without templates.

📚 Full documentation: https://mosquito.github.io/tagz/

Overview

tagz lets you construct HTML using pure Python code with a clean, intuitive API. No template engines, no DSLs — just Python functions and objects that map directly to HTML elements.

Key Features:

• Programmatic HTML construction — build documents using Python objects and methods
• HTML parser — parse existing HTML strings back into manipulable Tag objects
• Type-safe — full mypy support with comprehensive type annotations
• Streaming support — memory-efficient rendering with iter_lines(), iter_chunk(), and iter_string()
• Automatic escaping — XSS protection enabled by default
• CSS helpers — built-in Style and StyleSheet objects
• Fragments and raw — group elements without wrapper tags, or splice in pre-rendered HTML
• Page objects — high-level API for complete HTML documents with DOCTYPE support

Installation

pip install tagz

or with uv:

uv add tagz

Requires Python 3.10+.

Quick start

from tagz import Page, StyleSheet, Style, html


page = Page(
    lang="en",
    body_element=html.body(
        html.h1("Hello"),
        html.div(
            html.strong("world"),
        ),
        html.a(
            "example link",
            html.i("with italic text"),
            href="https://example.com/"
        ),
    ),
    head_elements=(
        html.meta(charset="utf-8"),
        html.meta(name="viewport", content="width=device-width, initial-scale=1"),
        html.title("tagz example page"),
        html.link(href="/static/css/bootstrap.min.css", rel="stylesheet"),
        html.script(src="/static/js/bootstrap.bundle.min.js"),
        html.style(
            StyleSheet({
                "body": Style(padding="0", margin="0"),
                (".container", ".container-fluid"): Style(transition="opacity 600ms ease-in"),
            })
        )
    ),
)

# `pretty=False` is the fast path; pretty=True produces human-readable output.
output = page.to_html5(pretty=True)
assert output.startswith("<!doctype html>")
assert "<strong>" in output

The pretty-printed output looks like:

<!doctype html>
<html lang="en">
	<head>
		<meta charset="utf-8"/>
		<meta content="width=device-width, initial-scale=1" name="viewport"/>
		<title>
			tagz example page
		</title>
		<link href="/static/css/bootstrap.min.css" rel="stylesheet"/>
		<script src="/static/js/bootstrap.bundle.min.js">
		</script>
		<style>
			body {margin: 0; padding: 0;}
			.container, .container-fluid {transition: opacity 600ms ease-in;}
		</style>
	</head>
	<body>
		<h1>
			Hello
		</h1>
		<div>
			<strong>
				world
			</strong>
		</div>
		<a href="https://example.com/">
			example link
			<i>
				with italic text
			</i>
		</a>
	</body>
</html>

Where to next?

The full documentation is organised in the Diátaxis style:

• Tutorials — guided lessons that take you from zero to a working page, parser, or streamed document.
• How-to guides — recipe-style answers to specific problems (streaming, callables, async data, data: URIs, CSV → table, …).
• Reference — the full API surface with type signatures.
• Explanation — design rationale: escaping model, callables and laziness, why there's no async render path.

Feature highlights

License

MIT — see LICENSE.