activitypub: parse MathML to MFM

Ref: FEP-dc88
Changelog: Changed

packages/backend/src/mfm/from-html.ts

@@ -176,6 +176,53 @@ export function fromHtml(html: string, quoteUri?: string | null): string {
 				break;
 			}
 
+			// inline or block KaTeX
+			case 'math': {
+				// This node should contain <semantics>[...]<annotation/>[...]</semantics> tag with the "source code".
+				if (node.childNodes.length !== 1 || node.childNodes[0].nodeName !== 'semantics')
+					break;
+				const semantics = node.childNodes[0];
+
+				// only select well formed annotations
+				const annotations = semantics.childNodes
+					.filter(node =>
+						node.nodeName === 'annotation'
+						&& node.childNodes.length === 1
+						&& node.childNodes[0].nodeName === '#text'
+					);
+				if (annotations.length === 0)
+					break;
+
+				let annotation = annotations[0];
+				// try to prefer a TeX annotation if there are multiple annotations
+				const filteredAnnotations = annotations.filter(node => node.attrs.some(attribute => attribute.name === 'encoding' && attribute.value === 'application/x-tex'));
+				if (filteredAnnotations.length > 0) {
+					annotation = filteredAnnotations[0];
+				}
+
+				const formula = annotation.childNodes[0].value;
+				if (annotation.attrs.some(attribute => attribute.name === 'encoding' && attribute.value === 'application/x-tex')) {
+					// can be rendered as KaTeX, now decide if it is possible to render as inline or not
+					if (/[\r\n]/.test(formula)) {
+						// line break, this must be rendered as a block
+						text += '\n\\[' + formula + '\\]\n';
+					} else {
+						// render as inline
+						text += '\\(' + formula + '\\)';
+					}
+				} else {
+					// not KaTeX, but if there is a plaintext annotation it can still be rendered as code
+					if (/[\r\n]/.test(formula)) {
+						// line break, this must be rendered as a block
+						text += '\n```\n' + formula + '\n```\n';
+					} else {
+						// render as inline
+						text += '`' + formula + '`';
+					}
+				}
+				break;
+			}
+
 			case 'blockquote': {
 				const t = getText(node);
 				if (t) {