Merge attachments with message when possible

2026-01-21 13:25:30 +13:00
parent ddc7387fa0
commit 345b7d6135
3 changed files with 87 additions and 33 deletions
--- a/src/d2m/converters/message-to-event.js
+++ b/src/d2m/converters/message-to-event.js
@@ -542,16 +542,9 @@ async function messageToEvent(message, guild, options = {}, di) {
 			$type: "m.room.message",
 			"m.mentions": mentions,
 			msgtype,
-			body: body
-		}
-
-		const isPlaintext = body === html
-
-		if (!isPlaintext || options.alwaysReturnFormattedBody) {
-			Object.assign(newTextMessageEvent, {
-				format: "org.matrix.custom.html",
-				formatted_body: html
-			})
+			body: body,
+			format: "org.matrix.custom.html",
+			formatted_body: html
 		}

 		events.push(newTextMessageEvent)
@@ -695,7 +688,18 @@ async function messageToEvent(message, guild, options = {}, di) {
 	// Then attachments
 	if (message.attachments) {
 		const attachmentEvents = await Promise.all(message.attachments.map(attachmentToEvent.bind(null, mentions)))
-		events.push(...attachmentEvents)
+
+		// Try to merge attachment events with the previous event
+		// This means that if the attachments ended up as a text link, and especially if there were many of them, the events will be joined together.
+		let prev = events.at(-1)
+		for (const atch of attachmentEvents) {
+			if (atch.msgtype === "m.text" && prev?.body && prev?.formatted_body && ["m.text", "m.notice"].includes(prev?.msgtype)) {
+				prev.body = prev.body + "\n" + atch.body
+				prev.formatted_body = prev.formatted_body + "<br>" + atch.formatted_body
+			} else {
+				events.push(atch)
+			}
+		}
 	}

 	// Then embeds
@@ -829,6 +833,16 @@ async function messageToEvent(message, guild, options = {}, di) {
 		})
 	}

+	// Strip formatted_body where equivalent to body
+	if (!options.alwaysReturnFormattedBody) {
+		for (const event of events) {
+			if (["m.text", "m.notice"].includes(event.msgtype) && event.body === event.formatted_body) {
+				delete event.format
+				delete event.formatted_body
+			}
+		}
+	}
+
 	return events
 }

--- a/src/d2m/converters/message-to-event.test.js
+++ b/src/d2m/converters/message-to-event.test.js
@@ -869,14 +869,62 @@ test("message2event: very large attachment is linked instead of being uploaded",
 		$type: "m.room.message",
 		"m.mentions": {},
 		msgtype: "m.text",
-		body: "hey"
-	}, {
+		body: "hey\n📄 Uploaded file: https://bridge.example.org/download/discordcdn/123/456/789.mega (100 MB)",
+		format: "org.matrix.custom.html",
+		formatted_body: 'hey<br>📄 Uploaded file: <a href="https://bridge.example.org/download/discordcdn/123/456/789.mega">hey.jpg</a> (100 MB)'
+	}])
+})
+
+test("message2event: multiple attachments are combined into the same event where possible", async t => {
+	const events = await messageToEvent({
+		content: "hey",
+		attachments: [{
+			filename: "hey.jpg",
+			url: "https://cdn.discordapp.com/attachments/123/456/789.mega",
+			content_type: "application/i-made-it-up",
+			size: 100e6
+		}, {
+			filename: "SPOILER_secret.jpg",
+			url: "https://cdn.discordapp.com/attachments/123/456/SPOILER_secret.jpg",
+			content_type: "image/jpeg",
+			size: 38291
+		}, {
+			filename: "my enemies.txt",
+			url: "https://cdn.discordapp.com/attachments/123/456/my_enemies.txt",
+			content_type: "text/plain",
+			size: 8911
+		}, {
+			filename: "hey.jpg",
+			url: "https://cdn.discordapp.com/attachments/123/456/789.mega",
+			content_type: "application/i-made-it-up",
+			size: 100e6
+		}]
+	})
+	t.deepEqual(events, [{
 		$type: "m.room.message",
 		"m.mentions": {},
 		msgtype: "m.text",
-		body: "📄 Uploaded file: https://bridge.example.org/download/discordcdn/123/456/789.mega (100 MB)",
+		body: "hey"
+			+ "\n📄 Uploaded file: https://bridge.example.org/download/discordcdn/123/456/789.mega (100 MB)"
+			+ "\n📸 Uploaded SPOILER file: https://bridge.example.org/download/discordcdn/123/456/SPOILER_secret.jpg (38 KB)"
+			+ "\n📄 Uploaded file: https://bridge.example.org/download/discordcdn/123/456/789.mega (100 MB)",
 		format: "org.matrix.custom.html",
-		formatted_body: '📄 Uploaded file: <a href="https://bridge.example.org/download/discordcdn/123/456/789.mega">hey.jpg</a> (100 MB)'
+		formatted_body: "hey"
+			+ `<br>📄 Uploaded file: <a href="https://bridge.example.org/download/discordcdn/123/456/789.mega">hey.jpg</a> (100 MB)`
+			+ `<br><blockquote>📸 Uploaded SPOILER file: <a href="https://bridge.example.org/download/discordcdn/123/456/SPOILER_secret.jpg">https://bridge.example.org/download/discordcdn/123/456/SPOILER_secret.jpg</a> (38 KB)</blockquote>`
+			+ `<br>📄 Uploaded file: <a href="https://bridge.example.org/download/discordcdn/123/456/789.mega">hey.jpg</a> (100 MB)`
+	}, {
+		$type: "m.room.message",
+		"m.mentions": {},
+		msgtype: "m.file",
+		body: "my enemies.txt",
+		filename: "my enemies.txt",
+		external_url: "https://bridge.example.org/download/discordcdn/123/456/my_enemies.txt",
+		url: "mxc://cadence.moe/y89EOTRp2lbeOkgdsEleGOge",
+		info: {
+			mimetype: "text/plain",
+			size: 8911
+		}
 	}])
 })

@@ -1494,21 +1542,12 @@ test("message2event: forwarded message with unreferenced mention", async t => {
 			}
 		]
 	})
-	t.deepEqual(events, [
-		{
-			$type: "m.room.message",
-			msgtype: "m.text",
-			body: "[🔀 Forwarded message]\n» @unknown-user:",
-			format: "org.matrix.custom.html",
-			formatted_body: `🔀 <em>Forwarded message</em><br><blockquote>@unknown-user:</blockquote>`,
-			"m.mentions": {}
-		}, {
-			$type: "m.room.message",
-			msgtype: "m.text",
-			body: "» 🎞️ Uploaded file: https://bridge.example.org/download/discordcdn/893634327722721290/1463174815119704114/2022-10-18_16-49-46.mp4 (51 MB)",
-			format: "org.matrix.custom.html",
-			formatted_body: "<blockquote>🎞️ Uploaded file: <a href=\"https://bridge.example.org/download/discordcdn/893634327722721290/1463174815119704114/2022-10-18_16-49-46.mp4\">2022-10-18_16-49-46.mp4</a> (51 MB)</blockquote>",
-			"m.mentions": {}
-		}
-	])
+	t.deepEqual(events, [{
+		$type: "m.room.message",
+		msgtype: "m.text",
+		body: "[🔀 Forwarded message]\n» @unknown-user:\n» 🎞️ Uploaded file: https://bridge.example.org/download/discordcdn/893634327722721290/1463174815119704114/2022-10-18_16-49-46.mp4 (51 MB)",
+		format: "org.matrix.custom.html",
+		formatted_body: "🔀 <em>Forwarded message</em><br><blockquote>@unknown-user:<br>🎞️ Uploaded file: <a href=\"https://bridge.example.org/download/discordcdn/893634327722721290/1463174815119704114/2022-10-18_16-49-46.mp4\">2022-10-18_16-49-46.mp4</a> (51 MB)</blockquote>",
+		"m.mentions": {}
+	}])
 })
--- a/test/ooye-test-data.sql
+++ b/test/ooye-test-data.sql
@@ -151,7 +151,8 @@ INSERT INTO file (discord_url, mxc_url) VALUES
 ('https://cdn.discordapp.com/attachments/112760669178241024/1197621094786531358/Ins_1960637570.mp4', 'mxc://cadence.moe/kMqLycqMURhVpwleWkmASpnU'),
 ('https://cdn.discordapp.com/attachments/1099031887500034088/1112476845502365786/voice-message.ogg', 'mxc://cadence.moe/MRRPDggXQMYkrUjTpxQbmcxB'),
 ('https://cdn.discordapp.com/attachments/122155380120748034/1174514575220158545/the.yml', 'mxc://cadence.moe/HnQIYQmmlIKwOQsbFsIGpzPP'),
-('https://cdn.discordapp.com/attachments/112760669178241024/1296237494987133070/100km.gif', 'mxc://cadence.moe/qDAotmebTfEIfsAIVCEZptLh');
+('https://cdn.discordapp.com/attachments/112760669178241024/1296237494987133070/100km.gif', 'mxc://cadence.moe/qDAotmebTfEIfsAIVCEZptLh'),
+('https://cdn.discordapp.com/attachments/123/456/my_enemies.txt', 'mxc://cadence.moe/y89EOTRp2lbeOkgdsEleGOge');

 INSERT INTO emoji (emoji_id, name, animated, mxc_url) VALUES
 ('230201364309868544', 'hippo', 0, 'mxc://cadence.moe/qWmbXeRspZRLPcjseyLmeyXC'),