html.elements

import html

html exposes this as html.elements, so import html is enough and the names are called as html.elements.*. import html.elements reaches the same definitions directly.

The element tables the tree construction algorithm consults on almost every token: which elements are “special”, which are formatting elements, which close which, and how SVG and MathML tag and attribute names have to be repaired after the tokenizer has lowercased them.

These are data, lifted from the WHATWG HTML Living Standard. They live apart from html.parser so the algorithm there reads as the algorithm and not as a wall of lists.

Constants

SPECIAL_HTML

html.elements.SPECIAL_HTML = [...]

The HTML elements the standard calls “special”: the ones a list item or definition term stops searching past, and the ones the adoption agency algorithm treats as a block boundary when it looks for the furthest block.

Nothing to do with CSS. This is purely the tree construction algorithm’s own category.

SPECIAL_MATHML

html.elements.SPECIAL_MATHML = [...]

The MathML elements that count as special.

SPECIAL_SVG

html.elements.SPECIAL_SVG = [...]

The SVG elements that count as special.

FORMATTING_ELEMENTS

html.elements.FORMATTING_ELEMENTS = [...]

The formatting elements: the ones the adoption agency algorithm reopens across a badly nested boundary, so that <b>a<p>b</b>c puts a fresh <b> inside the paragraph.

SCOPE_HTML

html.elements.SCOPE_HTML = [...]

The HTML elements that make up the “in scope” barrier every scope check shares. The narrower checks (list item scope, button scope) add to this list rather than replacing it.

IMPLIED_END_TAGS

html.elements.IMPLIED_END_TAGS = [...]

Elements whose end tag is implied by the start of a sibling, so that <li>a<li>b produces two list items rather than nesting.

THOROUGH_IMPLIED_END_TAGS

html.elements.THOROUGH_IMPLIED_END_TAGS = [...]

Everything in IMPLIED_END_TAGS plus the elements only closed when the standard says to generate implied end tags thoroughly, which is what </template> and the table sections do.

MATHML_TEXT_INTEGRATION_POINTS

html.elements.MATHML_TEXT_INTEGRATION_POINTS = [...]

The MathML elements whose children are parsed as HTML rather than as MathML.

SVG_HTML_INTEGRATION_POINTS

html.elements.SVG_HTML_INTEGRATION_POINTS = [...]

The SVG elements whose children are parsed as HTML rather than as SVG.

FOREIGN_BREAKOUT_TAGS

html.elements.FOREIGN_BREAKOUT_TAGS = [...]

Start tags that are always a mistake inside foreign content and that break out of it, closing SVG or MathML elements until an HTML insertion point is reached.

SVG_TAG_NAMES

html.elements.SVG_TAG_NAMES = {...}

SVG tag names the tokenizer lowercased and that have to be put back, keyed by the lowercase form. SVG is case-sensitive and <foreignobject> is not the same element as <foreignObject>.

SVG_ATTRIBUTES

html.elements.SVG_ATTRIBUTES = {...}

SVG attribute names the tokenizer lowercased and that have to be put back, keyed by the lowercase form.

MATHML_ATTRIBUTES

html.elements.MATHML_ATTRIBUTES = {...}

The one MathML attribute whose case the tokenizer destroys.

FOREIGN_ATTRIBUTES

html.elements.FOREIGN_ATTRIBUTES = {...}

Attributes in foreign content that belong to a namespace, mapping the lowercase name the tokenizer produced to the qualified name and the namespace URI it should carry.

QUIRKS_PUBLIC_PREFIXES

html.elements.QUIRKS_PUBLIC_PREFIXES = [...]

Public identifier prefixes that put a document in quirks mode, in lowercase for case-insensitive comparison.

QUIRKS_PUBLIC_EXACT

html.elements.QUIRKS_PUBLIC_EXACT = [...]

The two public identifiers that put a document in quirks mode on an exact match rather than a prefix match, in lowercase.

LIMITED_QUIRKS_PUBLIC_PREFIXES

html.elements.LIMITED_QUIRKS_PUBLIC_PREFIXES = [...]

Public identifier prefixes that always mean limited-quirks mode.

HTML4_TRANSITIONAL_PREFIXES

html.elements.HTML4_TRANSITIONAL_PREFIXES = [...]

Public identifier prefixes whose mode depends on whether the doctype also carries a system identifier: quirks without one, limited-quirks with one.

QUIRKS_SYSTEM_ID

html.elements.QUIRKS_SYSTEM_ID

The system identifier that alone puts a document in quirks mode, in lowercase.

Functions

is_special()

html.elements.is_special(namespace: string, name: string) -> bool

True when the element name in namespace namespace is one of the standard’s special elements.

Parameters

• namespace (string)
• name (string)

Returns bool

is_mathml_text_integration_point()

html.elements.is_mathml_text_integration_point(element) -> bool

True when element is a MathML text integration point, meaning its children are parsed as HTML.

Parameters

• element (any)

Returns bool

is_html_integration_point()

html.elements.is_html_integration_point(element) -> bool

True when element is an HTML integration point.

An <annotation-xml> qualifies only when its encoding attribute names an HTML flavour, which is how MathML embeds real markup.

Parameters

• element (any)

Returns bool

adjust_svg_tag_name()

html.elements.adjust_svg_tag_name(name: string) -> string

The correctly cased SVG tag name for the lowercase name the tokenizer produced, or name itself when it needs no repair.

Parameters

• name (string)

Returns string

adjust_svg_attributes()

html.elements.adjust_svg_attributes(attributes: dict) -> dict

Repairs the case of SVG attribute names in attributes and returns a new dictionary. Attributes that need no repair are copied across untouched, in their original order.

Parameters

• attributes (dict)

Returns dict

adjust_mathml_attributes()

html.elements.adjust_mathml_attributes(attributes: dict) -> dict

Repairs the case of MathML attribute names in attributes and returns a new dictionary.

Parameters

• attributes (dict)

Returns dict

doctype_mode()

html.elements.doctype_mode(name: ?string, public_id: ?string, system_id: ?string, force_quirks: bool) -> string

Which quirks mode a doctype puts a document in.

name may be nil for a doctype with no name at all, and the two identifiers may be nil when the doctype omits them; all three cases mean something different to the algorithm, which is why they are not folded into empty strings.

Returns 'no-quirks', 'limited-quirks' or 'quirks'. Nothing in this module behaves differently between them, but knowing which one a browser would have picked is exactly the question you are asking when you audit a legacy document.

Parameters

• name (?string)
• public_id (?string)
• system_id (?string)
• force_quirks (bool)

Returns string

2026, Richard Ore and The Zuri Contributors