OfficeIMO.Html.AngleSharp 3.6.0

OfficeIMO.Html.AngleSharp

AngleSharp-backed parsing, selectors and serialization for the owned OfficeIMO HTML document model. Parsing is inert: it does not execute scripts or fetch resources.

dotnet add package OfficeIMO.Html.AngleSharp

This includes OfficeIMO.Html.Core. Keep directly referenced OfficeIMO packages on the same version. Install OfficeIMO.Html for resource policy, conversion, styles and rendering; it includes this provider.

using OfficeIMO.Html.Dom;
using OfficeIMO.Html.Providers;

HtmlDocument document = AngleSharpHtmlParser.Instance.ParseDocument(
    "<h1>Hello</h1>", new HtmlParseOptions());
HtmlDocument edited = document.Edit(editor => {
    editor.QuerySelector("h1")!.TextContent = "Updated";
});
string html = edited.OuterHtml;

HtmlDocument table = AngleSharpHtmlParser.Instance.ParseDocument(
    "<table><tbody><tr id='items'></tr></tbody></table>", new HtmlParseOptions());
HtmlDocumentFragment cells = AngleSharpHtmlParser.Instance.ParseFragment(
    "<td>A</td><td>B</td>", table.QuerySelector("#items")!, new HtmlParseOptions());

The owned tree is the long-lived source of truth. The provider caches its native selector and serializer projection weakly. When a caller still holds the native document, the bridge preserves that document identity and can rebuild only its maps after collection. HtmlConversionDocument keeps conversion-only parsing on the native fast path until its owned Document is requested, then allows the duplicate source projection to be reclaimed. The existing CSS/layout engine still uses a structural native-DOM adapter internally. This package does not claim dependency independence or browser execution. Source and tree budgets are enforced before downstream conversion work; parsing remains subject to the underlying parser's cooperative cancellation behavior.

AngleSharpEncodingProvider implements the owned IHtmlEncodingProvider charset contract. It retains AngleSharp web aliases and System.Text.Encoding.CodePages; initialization preserves the existing process-wide code-page registration. The HTML conversion package owns byte-order-mark precedence, HTML prescanning, CSS decoding rules and stream lifetime. This provider supplies label resolution.

Native trees are constructed before node/depth budgets can be checked. Source length is checked first, and parser cancellation is cooperative. Query results use owned nodes, including detached nodes. Invalid selectors throw ArgumentException. Conversion exports the attached tree, including template content. A detached tree is materialized only when that tree is queried or serialized; unrelated detached trees do not affect conversion or document queries. ParseDocument handles complete HTML documents. ParseFragment accepts an owned context element and returns a fragment in an independent owned document. The adapter reconstructs foreign and ancestor-form context where the retained native fragment API does not preserve it. Table insertion modes use the native fragment path. Both paths apply the same source, node, depth and cooperative-cancellation options.