html: tree construction for documents, first slice: `HtmlNode`, `HtmlDocument.parse` (step 2a of #79)

orion commented

2026-10-30 09:04:08 +00:00

Collaborator

Part of #79 (step 2, first slice). Step 1 (#439) is on main: HtmlTokenizer, HtmlToken, HtmlParseError.

Goal

std/html reads a document into a tree. This slice is the tree type and the construction for the document skeleton and plain content; the adoption agency algorithm, tables and foster parenting follow as the next slice(s) of step 2 (just in time, once this lands). Tree construction in full is more than one PR, so only this first slice is filed.

Acceptance criteria

HtmlNode (document, doctype, element with name and attributes, text, comment; children in order) and HtmlDocument.parse(text) answering the tree, the quirks mode and the HtmlParseError list (tokenizer and tree errors). Reading never fails, no Result (design record of #79).
The insertion modes initial, before html, before head, in head, after head, in body (without formatting elements and tables), text, after body, after after body; implied html, head, body; quirks mode from the doctype; the tokenizer switches for <script>, <style>, <textarea>, <title>, <plaintext> and similar raw-text elements. scripting stays off (<noscript> is markup).
The html5lib-tests tree-construction suite, vendored like tests/html5lib (same commit, NOTICE.md updated), runs in std/html/tests; every case this slice covers passes, the rest sit in a named skip list with the reason (formatting, table, foreign content, template, fragment) that the next slices shrink.
A docs page section and lib.trb exports for the new names; tools/gates.sh a green in CI.

Design pointers

WHATWG HTML section 13.2.6 (tree construction); layout models std/xml (node.trb, parser.trb) and the existing std/html tokenizer (tokenizer.trb, state.trb).
Names fixed by #79: HtmlNode, HtmlDocument, no type called Html (that is #93's).
Input is a String; no encoding detection.

Seed

No. Only std/html changes; the compiler imports no std/html name. If a compiler change turns out to be needed, split it into its own commit and stop: a seed touch means a teaching commit and a migration commit, and the seed refresh is the owner's step.

Size

One PR, one or two Conventional Commits (feat(html): ... for the tree, test(html): ... for the vendored suite).

Part of #79 (step 2, first slice). Step 1 (#439) is on `main`: `HtmlTokenizer`, `HtmlToken`, `HtmlParseError`. ## Goal `std/html` reads a document into a tree. This slice is the tree type and the construction for the document skeleton and plain content; the adoption agency algorithm, tables and foster parenting follow as the next slice(s) of step 2 (just in time, once this lands). Tree construction in full is more than one PR, so only this first slice is filed. ## Acceptance criteria - `HtmlNode` (document, doctype, element with name and attributes, text, comment; children in order) and `HtmlDocument.parse(text)` answering the tree, the quirks mode and the `HtmlParseError` list (tokenizer and tree errors). Reading never fails, no `Result` (design record of #79). - The insertion modes `initial`, `before html`, `before head`, `in head`, `after head`, `in body` (without formatting elements and tables), `text`, `after body`, `after after body`; implied `html`, `head`, `body`; quirks mode from the doctype; the tokenizer switches for `<script>`, `<style>`, `<textarea>`, `<title>`, `<plaintext>` and similar raw-text elements. `scripting` stays off (`<noscript>` is markup). - The html5lib-tests `tree-construction` suite, vendored like `tests/html5lib` (same commit, `NOTICE.md` updated), runs in `std/html/tests`; every case this slice covers passes, the rest sit in a named skip list with the reason (formatting, table, foreign content, template, fragment) that the next slices shrink. - A docs page section and `lib.trb` exports for the new names; `tools/gates.sh a` green in CI. ## Design pointers - WHATWG HTML section 13.2.6 (tree construction); layout models `std/xml` (`node.trb`, `parser.trb`) and the existing `std/html` tokenizer (`tokenizer.trb`, `state.trb`). - Names fixed by #79: `HtmlNode`, `HtmlDocument`, no type called `Html` (that is #93's). - Input is a `String`; no encoding detection. ## Seed No. Only `std/html` changes; the compiler imports no `std/html` name. If a compiler change turns out to be needed, split it into its own commit and stop: a seed touch means a teaching commit and a migration commit, and the seed refresh is the owner's step. ## Size One PR, one or two Conventional Commits (`feat(html): ...` for the tree, `test(html): ...` for the vendored suite).

orion added the

loop

ready

labels

2026-10-30 09:04:08 +00:00

orion referenced this issue

2026-10-30 09:04:37 +00:00

formats: std/html documents (epic) #79

orion commented

2026-10-30 19:31:34 +00:00

Author

Collaborator

[loop:torbscript] claim implementation by session:loop-torbscript-20261030 @anthropic:tkoehn until 2026-10-31T01:31Z

orion commented

2026-10-30 20:20:18 +00:00

Author

Collaborator

[loop:torbscript] release implementation by session:loop-torbscript-20261030 @anthropic:tkoehn: the worker run ended without a result (the worker runtime refused its oversized output log); no branch, commit or PR was pushed, so nothing is left to pick up. The item is still ready; the next worker starts from main and keeps its command output small.
next: implementation

[loop:torbscript] release implementation by session:loop-torbscript-20261030 @anthropic:tkoehn: the worker run ended without a result (the worker runtime refused its oversized output log); no branch, commit or PR was pushed, so nothing is left to pick up. The item is still `ready`; the next worker starts from main and keeps its command output small. next: implementation

Rows
Columns

html: tree construction for documents, first slice: HtmlNode, HtmlDocument.parse (step 2a of #79) #688

Goal

Acceptance criteria

Design pointers

Seed

Size

html: tree construction for documents, first slice: `HtmlNode`, `HtmlDocument.parse` (step 2a of #79) #688