RecursiveAI.OneHtml 1.1.0

RecursiveAI.OneHtml

Parse HTML exactly as a browser does, query it, compute styles, and edit HTML and CSS without disturbing a byte the edit didn't touch. Pure C# on the .NET base class library, safe code only, NativeAOT- and trim-compatible, net10.0.

• Encoding and parser. The WHATWG Encoding Standard (BOM, transport Content-Type, <meta> prescan) and the HTML tokenizer and tree construction, including the adoption agency algorithm, foster parenting, <template>, foreign content and fragment parsing. Parse errors are recorded as the spec says, never thrown.
• DOM as snapshots. A Document is an immutable arena; Node is a readonly struct handle. Every node knows its byte range in the original, and parser-inserted nodes are marked.
• Selectors Level 4 compiled for repeated use, and locators that are checked to match exactly their target.
• CSS. CSS Syntax parsing with source maps, a property table generated from the CSS specifications, and a cascade for every known property (layers, !important, var(), shorthands, media and supports conditions) under a StyleContext that holds the viewport, media features and one focus or hover state.
• Lossless editing. doc.Apply(edits) splices bytes only where an edit landed, re-parses the output and compares it with the expected tree, and returns a new snapshot plus a change set. Edits can be saved as versioned JSON and replayed on another copy of the page; a replay that cannot find exactly one matching target reports stale or ambiguous and never guesses.

using RecursiveAI.OneHtml;
using RecursiveAI.OneHtml.Editing;
Document doc = Document.Parse("<p>Hello <a href=/x>world</a>");
Node link = doc.Root.QuerySelector("a[href]");
EditResult edit = doc.Apply(new[] { Edit.SetAttribute(link, "title", "A link") });
byte[] output = edit.Document!.ToLosslessBytes();

Styles under a context:

using RecursiveAI.OneHtml.Css;
Document doc = Document.Parse("<style>a { color: red } a:focus { color: blue }</style><a href=/x>x</a>");
Node link = doc.Root.QuerySelector("a");
DocumentStyles styles = DocumentStyles.Create(doc);
string normal = styles.GetStyle(link).GetPropertyValue("color");
string focused = styles.WithContext(StyleContext.Default.WithFocus(link)).GetStyle(link).GetPropertyValue("color");

The library never reads files or the network, runs no script and does no layout. Linked stylesheets are supplied by the caller. Hostile-input limits (depth, attribute count, input size, stylesheet and selector complexity) are checked before the work they guard.

More

• Family overview, conformance numbers and benchmarks: README.
• Behaviour documents: docs/.
• Third-party material and licences: THIRD-PARTY-NOTICES.md.

Licensed under either of MIT or Apache-2.0 at your option. Copyright Recursive AI, LLC.