From 39fb4465f6861de3dc98d638c14c424616f5bf6a Mon Sep 17 00:00:00 2001
From: Cadence Ember <cadence@disroot.org>
Date: Sat, 26 Aug 2023 19:07:19 +1200
Subject: [PATCH] fix m->d formatting of quotes and code

---
 m2d/converters/event-to-message.js      | 38 ++++++++++++++++++
 m2d/converters/event-to-message.test.js | 52 +++++++++++++++++++++++++
 2 files changed, 90 insertions(+)

diff --git a/m2d/converters/event-to-message.js b/m2d/converters/event-to-message.js
index 7893799..f4382a3 100644
--- a/m2d/converters/event-to-message.js
+++ b/m2d/converters/event-to-message.js
@@ -33,6 +33,40 @@ turndownService.addRule("strikethrough", {
 	}
 })
 
+turndownService.addRule("blockquote", {
+	filter: "blockquote",
+	replacement: function (content) {
+		content = content.replace(/^\n+|\n+$/g, "")
+		content = content.replace(/^/gm, "> ")
+		return content
+	}
+})
+
+turndownService.addRule("fencedCodeBlock", {
+	filter: function (node, options) {
+		return (
+			options.codeBlockStyle === "fenced" &&
+			node.nodeName === "PRE" &&
+			node.firstChild &&
+			node.firstChild.nodeName === "CODE"
+		)
+	},
+	replacement: function (content, node, options) {
+		const className = node.firstChild.getAttribute("class") || ""
+		const language = (className.match(/language-(\S+)/) || [null, ""])[1]
+		const code = node.firstChild
+		const visibleCode = code.childNodes.map(c => c.nodeName === "BR" ? "\n" : c.textContent).join("").replace(/\n*$/g, "")
+
+		var fence = "```"
+
+		return (
+			fence + language + "\n" +
+			visibleCode +
+			"\n" + fence
+		)
+	}
+})
+
 /**
  * @param {Ty.Event.Outer<Ty.Event.M_Room_Message>} event
  */
@@ -61,9 +95,13 @@ function eventToMessage(event) {
 		// input = input.replace(/ /g, "&nbsp;")
 		// There is also a corresponding test to uncomment, named "event2message: whitespace is retained"
 
+		// Element adds a bunch of <br> before </blockquote> but doesn't render them. I can't figure out how this works, so let's just delete those.
+		input = input.replace(/(?:\n|<br ?\/?>\s*)*<\/blockquote>/g, "</blockquote>")
+
 		// The matrix spec hasn't decided whether \n counts as a newline or not, but I'm going to count it, because if it's in the data it's there for a reason.
 		// But I should not count it if it's between block elements.
 		input = input.replace(/(<\/?([^ >]+)[^>]*>)?\n(<\/?([^ >]+)[^>]*>)?/g, (whole, beforeContext, beforeTag, afterContext, afterTag) => {
+			// console.error(beforeContext, beforeTag, afterContext, afterTag)
 			if (typeof beforeTag !== "string" && typeof afterTag !== "string") {
 				return "<br>"
 			}
diff --git a/m2d/converters/event-to-message.test.js b/m2d/converters/event-to-message.test.js
index 7aefdb1..afa40de 100644
--- a/m2d/converters/event-to-message.test.js
+++ b/m2d/converters/event-to-message.test.js
@@ -271,6 +271,58 @@ test("event2message: code blocks work", t => {
 	)
 })
 
+test("event2message: code block contents are formatted correctly and not escaped", t => {
+	t.deepEqual(
+		eventToMessage({
+			"type": "m.room.message",
+			"sender": "@cadence:cadence.moe",
+			"content": {
+				"msgtype": "m.text",
+				"body": "wrong body",
+				"format": "org.matrix.custom.html",
+				"formatted_body": "<pre><code>input = input.replace(/(&lt;\\/?([^ &gt;]+)[^&gt;]*&gt;)?\\n(&lt;\\/?([^ &gt;]+)[^&gt;]*&gt;)?/g,\n_input_ = input = input.replace(/(&lt;\\/?([^ &gt;]+)[^&gt;]*&gt;)?\\n(&lt;\\/?([^ &gt;]+)[^&gt;]*&gt;)?/g,\n</code></pre>\n<p><code>input = input.replace(/(&lt;\\/?([^ &gt;]+)[^&gt;]*&gt;)?\\n(&lt;\\/?([^ &gt;]+)[^&gt;]*&gt;)?/g,</code></p>\n"
+			},
+			"origin_server_ts": 1693031482275,
+			"unsigned": {
+				"age": 99,
+				"transaction_id": "m1693031482146.511"
+			},
+			"event_id": "$pGkWQuGVmrPNByrFELxhzI6MCBgJecr5I2J3z88Gc2s",
+			"room_id": "!BpMdOUkWWhFxmTrENV:cadence.moe"
+		}),
+		[{
+			username: "cadence",
+			content: "```\ninput = input.replace(/(<\\/?([^ >]+)[^>]*>)?\\n(<\\/?([^ >]+)[^>]*>)?/g,\n_input_ = input = input.replace(/(<\\/?([^ >]+)[^>]*>)?\\n(<\\/?([^ >]+)[^>]*>)?/g,\n```\n\n`input = input.replace(/(<\\/?([^ >]+)[^>]*>)?\\n(<\\/?([^ >]+)[^>]*>)?/g,`",
+			avatar_url: undefined
+		}]
+	)
+})
+
+test("event2message: quotes have an appropriate amount of whitespace", t => {
+	t.deepEqual(
+		eventToMessage({
+			content: {
+				msgtype: "m.text",
+				body: "wrong body",
+				format: "org.matrix.custom.html",
+				formatted_body: "<blockquote>Chancellor of Germany Angela Merkel, on March 17, 2017: they did not shake hands<br><br><br></blockquote><br>🤨"
+			},
+			event_id: "$g07oYSZFWBkxohNEfywldwgcWj1hbhDzQ1sBAKvqOOU",
+			origin_server_ts: 1688301929913,
+			room_id: "!kLRqKKUQXcibIMtOpl:cadence.moe",
+			sender: "@cadence:cadence.moe",
+			type: "m.room.message",
+			unsigned: {
+				age: 405299
+			}
+		}),
+		[{
+			username: "cadence",
+			content: "> Chancellor of Germany Angela Merkel, on March 17, 2017: they did not shake hands\n🤨",
+			avatar_url: undefined
+		}]
+	)
+})
 
 test("event2message: m.emote markdown syntax is escaped", t => {
 	t.deepEqual(