From 01a5300be88665ae414c15e8e0667c93799d8f0e Mon Sep 17 00:00:00 2001
From: dakkar <dakkar@thenautilus.net>
Date: Sat, 18 Jan 2025 12:51:38 +0000
Subject: [PATCH] handle more complex ruby from/to html - fixes #605

this is not exactly great, but it should be "good enough"

note that the new `group` function should not escape in the wild, as
we don't document it and only use it internally

I tried using `$[scale foo bar]` instead of `$[group foo bar]`, but
that would be rendered as `<i>foo bar</i>` when sent over the network
to non-misskey instances, and we don't want that
---
 packages/backend/src/core/MfmService.ts       | 67 +++++++++++++++++++
 packages/backend/test/unit/MfmService.ts      | 13 ++++
 .../frontend/src/components/global/MkMfm.ts   |  4 ++
 3 files changed, 84 insertions(+)

diff --git a/packages/backend/src/core/MfmService.ts b/packages/backend/src/core/MfmService.ts
index 42676d6f98..48995672c5 100644
--- a/packages/backend/src/core/MfmService.ts
+++ b/packages/backend/src/core/MfmService.ts
@@ -230,6 +230,67 @@ export class MfmService {
 					break;
 				}
 
+				case 'rp': break
+				case 'rt': {
+					appendChildren(node.childNodes);
+					break;
+				}
+				case 'ruby': {
+					if (node.childNodes) {
+						/*
+							we get:
+							```
+							<ruby>
+							some text <rp>(</rp> <rt>annotation</rt> <rp>)</rp>
+							more text <rt>more annotation<rt>
+							</ruby>
+							```
+
+							and we want to produce:
+							```
+							$[ruby $[group some text] annotation]
+							$[ruby $[group more text] more annotation]
+							```
+
+							that `group` is a hack, because when the `ruby` render
+							sees just text inside the `$[ruby]`, it splits on
+							whitespace, considers the first "word" to be the main
+							content, and the rest the annotation
+
+							with that `group`, we force it to consider the whole
+							group as the main content
+
+							(note that the `rp` are to be ignored, they only exist
+							for browsers who don't understand ruby)
+						*/
+						let nonRtNodes=[];
+						// scan children, ignore `rp`, split on `rt`
+						for (const child of node.childNodes) {
+							if (treeAdapter.isTextNode(child)) {
+								nonRtNodes.push(child);
+								continue;
+							}
+							if (!treeAdapter.isElementNode(child)) {
+								continue;
+							}
+							if (child.nodeName == 'rp') {
+								continue;
+							}
+							if (child.nodeName == 'rt') {
+								text += '$[ruby $[group ';
+								appendChildren(nonRtNodes);
+								text += '] ';
+								analyze(child);
+								text += '] ';
+								nonRtNodes=[];
+								continue;
+							}
+							nonRtNodes.push(child);
+						}
+					}
+					break;
+				}
+
 				default:	// includes inline elements
 				{
 					appendChildren(node.childNodes);
@@ -348,6 +409,12 @@ export class MfmService {
 						}
 					}
 
+					case 'group': { // this is mostly a hack for `ruby`
+						const el = doc.createElement('span');
+						appendChildren(node.children, el);
+						return el;
+					}
+
 					default: {
 						return fnDefault(node);
 					}
diff --git a/packages/backend/test/unit/MfmService.ts b/packages/backend/test/unit/MfmService.ts
index 8d5683329f..93ce0672dc 100644
--- a/packages/backend/test/unit/MfmService.ts
+++ b/packages/backend/test/unit/MfmService.ts
@@ -45,6 +45,12 @@ describe('MfmService', () => {
 			const output = '<p><pre><code>&lt;p&gt;Hello, world!&lt;/p&gt;</code></pre></p>';
 			assert.equal(mfmService.toHtml(mfm.parse(input)), output);
 		});
+
+		test('ruby', () => {
+			const input = '$[ruby $[group *some* text] ignore me]';
+			const output = '<p><ruby><span><i>some</i> text</span><rp>(</rp><rt>ignore me</rt><rp>)</rp></ruby></p>';
+			assert.equal(mfmService.toHtml(mfm.parse(input)), output);
+		});
 	});
 
 	describe('fromHtml', () => {
@@ -115,5 +121,12 @@ describe('MfmService', () => {
 		test('hashtag', () => {
 			assert.deepStrictEqual(mfmService.fromHtml('<p>a <a href="https://example.com/tags/a">#a</a> d</p>', ['#a']), 'a #a d');
 		});
+
+		test('ruby', () => {
+			assert.deepStrictEqual(
+				mfmService.fromHtml('<ruby> <i>some</i> text <rp>(</rp><rt>ignore me</rt><rp>)</rp> and <rt>more</rt></ruby>'),
+				'$[ruby $[group  <i>some</i> text ] ignore me] $[ruby $[group  and ] more]'
+			);
+		});
 	});
 });
diff --git a/packages/frontend/src/components/global/MkMfm.ts b/packages/frontend/src/components/global/MkMfm.ts
index 1039572a06..aceed17189 100644
--- a/packages/frontend/src/components/global/MkMfm.ts
+++ b/packages/frontend/src/components/global/MkMfm.ts
@@ -358,6 +358,10 @@ export default function (props: MfmProps, { emit }: { emit: SetupContext<MfmEven
 							return h('ruby', {}, [...genEl(token.children.slice(0, token.children.length - 1), scale), h('rt', text.trim())]);
 						}
 					}
+					case 'group': { // this is mostly a hack for the insides of `ruby`
+						style = '';
+						break;
+					}
 					case 'unixtime': {
 						const child = token.children[0];
 						const unixtime = parseInt(child.type === 'text' ? child.props.text : '');