From 6738290d99c7984b96e495e7870320d6761764f8 Mon Sep 17 00:00:00 2001
From: Cadence Ember <cadence@disroot.org>
Date: Wed, 21 Feb 2024 00:00:11 +1300
Subject: [PATCH] m->d: Reliably put < > around matrix.to links

This replaces the turndown brackets system with a regexp over body
and formatted_body.
---
 m2d/converters/event-to-message.js      | 12 ++---
 m2d/converters/event-to-message.test.js | 58 ++++++++++++++++++++++---
 2 files changed, 60 insertions(+), 10 deletions(-)

diff --git a/m2d/converters/event-to-message.js b/m2d/converters/event-to-message.js
index 4ffe26e..ae9be22 100644
--- a/m2d/converters/event-to-message.js
+++ b/m2d/converters/event-to-message.js
@@ -126,12 +126,10 @@ turndownService.addRule("inlineLink", {
 		if (node.getAttribute("data-message-id")) return `https://discord.com/channels/${node.getAttribute("data-guild-id")}/${node.getAttribute("data-channel-id")}/${node.getAttribute("data-message-id")}`
 		if (node.getAttribute("data-channel-id")) return `<#${node.getAttribute("data-channel-id")}>`
 		const href = node.getAttribute("href")
-		let brackets = ["", ""]
 		content = content.replace(/ @.*/, "")
-		if (href.startsWith("https://matrix.to")) brackets = ["<", ">"]
-		if (href === content) return brackets[0] + href + brackets[1]
+		if (href === content) return href
 		if (href.startsWith("https://matrix.to/#/@") && content[0] !== "@") content = "@" + content
-		return "[" + content + "](" + brackets[0] + href + brackets[1] + ")"
+		return "[" + content + "](" + href + ")"
 	}
 })
 
@@ -709,6 +707,9 @@ async function eventToMessage(event, guild, di) {
 			// @ts-ignore bad type from turndown
 			content = turndownService.turndown(root)
 
+			// Put < > around any surviving matrix.to links to hide the URL previews
+			content = content.replace(/\bhttps?:\/\/matrix\.to\/[^ )]*/, "<$&>")
+
 			// It's designed for commonmark, we need to replace the space-space-newline with just newline
 			content = content.replace(/  \n/g, "\n")
 
@@ -725,7 +726,8 @@ async function eventToMessage(event, guild, di) {
 				content = `* ${displayName} ${content}`
 			}
 
-			content = await handleRoomOrMessageLinks(content, di)
+			content = await handleRoomOrMessageLinks(content, di) // Replace matrix.to links with discord.com equivalents where possible
+			content = content.replace(/\bhttps?:\/\/matrix\.to\/[^ )]*/, "<$&>") // Put < > around any surviving matrix.to links to hide the URL previews
 
 			const result = await checkWrittenMentions(content, guild, di)
 			if (result) {
diff --git a/m2d/converters/event-to-message.test.js b/m2d/converters/event-to-message.test.js
index d8926b6..3865999 100644
--- a/m2d/converters/event-to-message.test.js
+++ b/m2d/converters/event-to-message.test.js
@@ -2731,7 +2731,7 @@ test("event2message: mentioning events falls back to original link when the chan
 	)
 })
 
-test("event2message: link to event in an unknown room", async t => {
+test("event2message: link to event in an unknown room (href link)", async t => {
 	t.deepEqual(
 		await eventToMessage({
 			content: {
@@ -2741,13 +2741,61 @@ test("event2message: link to event in an unknown room", async t => {
 				formatted_body: 'ah yeah, here\'s where the bug was reported: <a href="https://matrix.to/#/!QtykxKocfZaZOUrTwp:matrix.org/$1542477546853947KGhZL:matrix.org">https://matrix.to/#/!QtykxKocfZaZOUrTwp:matrix.org/$1542477546853947KGhZL:matrix.org</a>'
 			},
 			event_id: "$g07oYSZFWBkxohNEfywldwgcWj1hbhDzQ1sBAKvqOOU",
-			origin_server_ts: 1688301929913,
 			room_id: "!kLRqKKUQXcibIMtOpl:cadence.moe",
 			sender: "@cadence:cadence.moe",
 			type: "m.room.message",
-			unsigned: {
-				age: 405299
-			}
+		}),
+		{
+			ensureJoined: [],
+			messagesToDelete: [],
+			messagesToEdit: [],
+			messagesToSend: [{
+				username: "cadence [they]",
+				content: "ah yeah, here's where the bug was reported: <https://matrix.to/#/!QtykxKocfZaZOUrTwp:matrix.org/$1542477546853947KGhZL:matrix.org>",
+				avatar_url: undefined
+			}]
+		}
+	)
+})
+
+test("event2message: link to event in an unknown room (bare link)", async t => {
+	t.deepEqual(
+		await eventToMessage({
+			content: {
+				body: "wrong body",
+				format: "org.matrix.custom.html",
+				formatted_body: "PK API failure, tho idk how you'd handle that https://matrix.to/#/!SeYQChwXBnZaQLoZfI:sleeping.town/$AAPZ56B2P7TfROYPTtuoJjgvXmaBM11NoNceM8GCJ7s",
+				msgtype: "m.text"
+			},
+			sender: "@cadence:cadence.moe",
+			type: "m.room.message",
+			event_id: "$-UwntiHseGfch1GMjTROIgDbgLGIOwMx0vJdTi-dmok",
+			room_id: "!kLRqKKUQXcibIMtOpl:cadence.moe"
+		}),
+		{
+			ensureJoined: [],
+			messagesToDelete: [],
+			messagesToEdit: [],
+			messagesToSend: [{
+				username: "cadence [they]",
+				content: "PK API failure, tho idk how you'd handle that <https://matrix.to/#/!SeYQChwXBnZaQLoZfI:sleeping.town/$AAPZ56B2P7TfROYPTtuoJjgvXmaBM11NoNceM8GCJ7s>",
+				avatar_url: undefined
+			}]
+		}
+	)
+})
+
+test("event2message: link to event in an unknown room (plaintext)", async t => {
+	t.deepEqual(
+		await eventToMessage({
+			content: {
+				msgtype: "m.text",
+				body: "ah yeah, here's where the bug was reported: https://matrix.to/#/!QtykxKocfZaZOUrTwp:matrix.org/$1542477546853947KGhZL:matrix.org"
+			},
+			event_id: "$g07oYSZFWBkxohNEfywldwgcWj1hbhDzQ1sBAKvqOOU",
+			room_id: "!kLRqKKUQXcibIMtOpl:cadence.moe",
+			sender: "@cadence:cadence.moe",
+			type: "m.room.message",
 		}),
 		{
 			ensureJoined: [],