innertext 0.2.3

Pure Rust WHATWG-compliant innerText, outerText, and textContent extraction

innertext (Python)

Pure Rust WHATWG-compliant innerText, outerText, and textContent extraction from HTML.

Installation

pip install innertext

Quick Start

import innertext

html = """
<div id="root">
    Hello <span style="display:none">hidden</span> World
    <script>console.log("not shown")</script>
</div>
"""

# Get rendered text (CSS-aware)
print(innertext.inner_text(html))  # "Hello World"

# Get structural text (CSS-blind)
print(innertext.text_content(html))  # "Hello hidden console.log("not shown")"

Packaging Notes

The published Python package uses a two-layer layout:

• innertext is a regular Python package with a runtime __init__.py
• the compiled Rust extension is loaded as innertext._innertext

This avoids namespace-package import issues and ensures import innertext always exposes inner_text, outer_text, and text_content.

Type Hints

This package ships with PEP 561-compatible typing metadata:

• innertext/__init__.pyi defines the public API types
• innertext/py.typed marks the package as typed

Type checkers such as mypy and pyright can therefore infer:

• inner_text(html: str) -> str
• outer_text(html: str) -> str
• text_content(html: str) -> str

API

Functions

inner_text(html: str) -> str

Extract innerText from HTML string. Implements the WHATWG innerText algorithm:

• Respects display CSS property (skips display:none)
• Respects visibility CSS property
• Respects white-space CSS property (normal, pre, pre-line, pre-wrap)
• Respects text-transform CSS property
• Handles replaced elements (textarea, input, img)
• Converts <br> tags to newlines

outer_text(html: str) -> str

Extract outerText from HTML string. Per WHATWG spec, outerText getter is identical to innerText getter.

text_content(html: str) -> str

Extract textContent from HTML string. Performs CSS-blind structural walk:

• Ignores all CSS properties
• Includes display:none content
• Includes <script> and <style> content

Accuracy

100% Chromium parity on 36+ test cases covering:

• Display and visibility handling
• Whitespace normalization (normal, pre, pre-line)
• Block element newlines
• Table cell/row separators
• Replaced elements (textarea, input, img, button)
• Metadata elements (script, style)
• Unicode and entity handling

Performance

• Pure Rust implementation
• Zero external runtime dependencies
• <1ms per document (parsing + extraction)
• Minimal memory overhead (O(n) DOM tree)

License

MIT