m->d: fix image captions spec, fix upload issues

src/m2d/converters/event-to-message.js

@@ -315,7 +315,7 @@ function getUserOrProxyOwnerID(mxid) {
  * At the time of this executing, we know what the end of message emojis are, and we know that at least one of them is unknown.
  * This function will strip them from the content and generate the correct pending file of the sprite sheet.
  * @param {string} content
- * @param {{id: string, name: string}[]} attachments
+ * @param {{id: string, filename: string}[]} attachments
  * @param {({name: string, mxc: string} | {name: string, mxc: string, key: string, iv: string} | {name: string, buffer: Buffer})[]} pendingFiles
  * @param {(mxc: string) => Promise<Buffer | undefined>} mxcDownloader function that will download the mxc URLs and convert to uncompressed PNG data. use `getAndConvertEmoji` or a mock.
  */
@@ -329,9 +329,9 @@ async function uploadEndOfMessageSpriteSheet(content, attachments, pendingFiles,
 	// Create a sprite sheet of known and unknown emojis from the end of the message
 	const buffer = await emojiSheet.compositeMatrixEmojis(endOfMessageEmojis, mxcDownloader)
 	// Attach it
-	const name = "emojis.png"
-	attachments.push({id: String(attachments.length), name})
-	pendingFiles.push({name, buffer})
+	const filename = "emojis.png"
+	attachments.push({id: String(attachments.length), filename})
+	pendingFiles.push({name: filename, buffer})
 	return content
 }
 
@@ -486,6 +486,7 @@ async function eventToMessage(event, guild, di) {
 	}
 
 	let content = event.content.body // ultimate fallback
+	/** @type {{id: string, filename: string}[]} */
 	const attachments = []
 	/** @type {({name: string, mxc: string} | {name: string, mxc: string, key: string, iv: string} | {name: string, buffer: Buffer})[]} */
 	const pendingFiles = []
@@ -493,7 +494,45 @@ async function eventToMessage(event, guild, di) {
 	const ensureJoined = []
 
 	// Convert content depending on what the message is
-	if (event.type === "m.room.message" && (event.content.msgtype === "m.text" || event.content.msgtype === "m.emote")) {
+	// Handle images first - might need to handle their `body`/`formatted_body` as well, which will fall through to the text processor
+	let shouldProcessTextEvent = event.type === "m.room.message" && (event.content.msgtype === "m.text" || event.content.msgtype === "m.emote")
+	if (event.type === "m.room.message" && (event.content.msgtype === "m.file" || event.content.msgtype === "m.video" || event.content.msgtype === "m.audio" || event.content.msgtype === "m.image")) {
+		content = ""
+		const filename = event.content.filename || event.content.body
+		if ("url" in event.content) {
+			// Unencrypted
+			attachments.push({id: "0", filename})
+			pendingFiles.push({name: filename, mxc: event.content.url})
+		} else {
+			// Encrypted
+			assert.equal(event.content.file.key.alg, "A256CTR")
+			attachments.push({id: "0", filename})
+			pendingFiles.push({name: filename, mxc: event.content.file.url, key: event.content.file.key.k, iv: event.content.file.iv})
+		}
+		// Check if we also need to process a text event for this image - if it has a caption that's different from its filename
+		if ((event.content.body && event.content.filename && event.content.body !== event.content.filename) || event.content.formatted_body) {
+			shouldProcessTextEvent = true
+		}
+	}
+	if (event.type === "m.sticker") {
+		content = ""
+		let filename = event.content.body
+		if (event.type === "m.sticker") {
+			let mimetype
+			if (event.content.info?.mimetype?.includes("/")) {
+				mimetype = event.content.info.mimetype
+			} else {
+				const res = await di.api.getMedia(event.content.url, {method: "HEAD"})
+				if (res.status === 200) {
+					mimetype = res.headers.get("content-type")
+				}
+				if (!mimetype) throw new Error(`Server error ${res.status} or missing content-type while detecting sticker mimetype`)
+			}
+			filename += "." + mimetype.split("/")[1]
+		}
+		attachments.push({id: "0", filename})
+		pendingFiles.push({name: filename, mxc: event.content.url})
+	} else if (shouldProcessTextEvent) {
 		// Handling edits. If the edit was an edit of a reply, edits do not include the reply reference, so we need to fetch up to 2 more events.
 		// this event ---is an edit of--> original event ---is a reply to--> past event
 		await (async () => {
@@ -780,40 +819,6 @@ async function eventToMessage(event, guild, di) {
 			// @ts-ignore bad type from turndown
 			content = turndownService.escape(content)
 		}
-	} else if (event.type === "m.room.message" && (event.content.msgtype === "m.file" || event.content.msgtype === "m.video" || event.content.msgtype === "m.audio" || event.content.msgtype === "m.image")) {
-		content = ""
-		const filename = event.content.filename || event.content.body
-		// A written `event.content.body` will be bridged to Discord's image `description` which is like alt text.
-		// Bridging as description rather than message content in order to match Matrix clients (Element, Neochat) which treat this as alt text or title text.
-		const description = (event.content.body !== event.content.filename && event.content.filename && event.content.body) || undefined
-		if ("url" in event.content) {
-			// Unencrypted
-			attachments.push({id: "0", description, filename})
-			pendingFiles.push({name: filename, mxc: event.content.url})
-		} else {
-			// Encrypted
-			assert.equal(event.content.file.key.alg, "A256CTR")
-			attachments.push({id: "0", description, filename})
-			pendingFiles.push({name: filename, mxc: event.content.file.url, key: event.content.file.key.k, iv: event.content.file.iv})
-		}
-	} else if (event.type === "m.sticker") {
-		content = ""
-		let filename = event.content.body
-		if (event.type === "m.sticker") {
-			let mimetype
-			if (event.content.info?.mimetype?.includes("/")) {
-				mimetype = event.content.info.mimetype
-			} else {
-				const res = await di.api.getMedia(event.content.url, {method: "HEAD"})
-				if (res.status === 200) {
-					mimetype = res.headers.get("content-type")
-				}
-				if (!mimetype) throw new Error(`Server error ${res.status} or missing content-type while detecting sticker mimetype`)
-			}
-			filename += "." + mimetype.split("/")[1]
-		}
-		attachments.push({id: "0", filename})
-		pendingFiles.push({name: filename, mxc: event.content.url})
 	}
 
 	content = displayNameRunoff + replyLine + content

src/m2d/converters/event-to-message.test.js

@@ -3770,7 +3770,7 @@ test("event2message: text attachments work", async t => {
 				username: "cadence [they]",
 				content: "",
 				avatar_url: "https://bridge.example.org/download/matrix/cadence.moe/azCAhThKTojXSZJRoWwZmhvU",
-				attachments: [{id: "0", description: undefined, filename: "chiki-powerups.txt"}],
+				attachments: [{id: "0", filename: "chiki-powerups.txt"}],
 				pendingFiles: [{name: "chiki-powerups.txt", mxc: "mxc://cadence.moe/zyThGlYQxvlvBVbVgKDDbiHH"}]
 			}]
 		}
@@ -3806,14 +3806,14 @@ test("event2message: image attachments work", async t => {
 				username: "cadence [they]",
 				content: "",
 				avatar_url: "https://bridge.example.org/download/matrix/cadence.moe/azCAhThKTojXSZJRoWwZmhvU",
-				attachments: [{id: "0", description: undefined, filename: "cool cat.png"}],
+				attachments: [{id: "0", filename: "cool cat.png"}],
 				pendingFiles: [{name: "cool cat.png", mxc: "mxc://cadence.moe/IvxVJFLEuksCNnbojdSIeEvn"}]
 			}]
 		}
 	)
 })
 
-test("event2message: image attachments can have a custom description", async t => {
+test("event2message: image attachments can have a plaintext caption", async t => {
 	t.deepEqual(
 		await eventToMessage({
 			type: "m.room.message",
@@ -3840,10 +3840,62 @@ test("event2message: image attachments can have a custom description", async t =
 			messagesToEdit: [],
 			messagesToSend: [{
 				username: "cadence [they]",
-				content: "",
+				content: "Cat emoji surrounded by pink hearts",
 				avatar_url: "https://bridge.example.org/download/matrix/cadence.moe/azCAhThKTojXSZJRoWwZmhvU",
-				attachments: [{id: "0", description: "Cat emoji surrounded by pink hearts", filename: "cool cat.png"}],
-				pendingFiles: [{name: "cool cat.png", mxc: "mxc://cadence.moe/IvxVJFLEuksCNnbojdSIeEvn"}]
+				attachments: [{id: "0", filename: "cool cat.png"}],
+				pendingFiles: [{name: "cool cat.png", mxc: "mxc://cadence.moe/IvxVJFLEuksCNnbojdSIeEvn"}],
+				allowed_mentions: {
+					parse: ["users", "roles"]
+				}
+			}]
+		}
+	)
+})
+
+test("event2message: image attachments can have a formatted caption", async t => {
+	t.deepEqual(
+		await eventToMessage({
+			content: {
+				body: "this event has `formatting`",
+				filename: "5740.jpg",
+				format: "org.matrix.custom.html",
+				formatted_body: "this event has <code>formatting</code>",
+				info: {
+					h: 1340,
+					mimetype: "image/jpeg",
+					size: 226689,
+					thumbnail_info: {
+						h: 670,
+						mimetype: "image/jpeg",
+						size: 80157,
+						w: 540
+					},
+					thumbnail_url: "mxc://thomcat.rocks/XhLsOCDBYyearsLQgUUrbAvw",
+					w: 1080,
+					"xyz.amorgan.blurhash": "KHJQG*55ic-.}?0M58J.9v"
+				},
+				msgtype: "m.image",
+				url: "mxc://thomcat.rocks/RTHsXmcMPXmuHqVNsnbKtRbh"
+			},
+			origin_server_ts: 1740607766895,
+			sender: "@cadence:cadence.moe",
+			type: "m.room.message",
+			event_id: "$NqNqVgukiQm1nynm9vIr9FIq31hZpQ3udOd7cBIW46U",
+			room_id: "!BnKuBPCvyfOkhcUjEu:cadence.moe"
+		}),
+		{
+			ensureJoined: [],
+			messagesToDelete: [],
+			messagesToEdit: [],
+			messagesToSend: [{
+				username: "cadence [they]",
+				content: "this event has `formatting`",
+				avatar_url: "https://bridge.example.org/download/matrix/cadence.moe/azCAhThKTojXSZJRoWwZmhvU",
+				attachments: [{id: "0", filename: "5740.jpg"}],
+				pendingFiles: [{name: "5740.jpg", mxc: "mxc://thomcat.rocks/RTHsXmcMPXmuHqVNsnbKtRbh"}],
+				allowed_mentions: {
+					parse: ["users", "roles"]
+				}
 			}]
 		}
 	)
@@ -3892,7 +3944,7 @@ test("event2message: encrypted image attachments work", async t => {
 				username: "cadence [they]",
 				content: "",
 				avatar_url: "https://bridge.example.org/download/matrix/cadence.moe/azCAhThKTojXSZJRoWwZmhvU",
-				attachments: [{id: "0", description: undefined, filename: "image.png"}],
+				attachments: [{id: "0", filename: "image.png"}],
 				pendingFiles: [{
 					name: "image.png",
 					mxc: "mxc://heyquark.com/LOGkUTlVFrqfiExlGZNgCJJX",

src/matrix/mreq.js

@@ -1,9 +1,9 @@
 // @ts-check
 
-const mixin = require("@cloudrac3r/mixin-deep")
 const stream = require("stream")
 const streamWeb = require("stream/web")
-const getStream = require("get-stream")
+const {buffer} = require("stream/consumers")
+const mixin = require("@cloudrac3r/mixin-deep")
 
 const {reg, writeRegistration} = require("./read-registration.js")
 
@@ -19,20 +19,33 @@ class MatrixServerError extends Error {
 	}
 }
 
+/**
+ * @param {undefined | string | object | streamWeb.ReadableStream | stream.Readable} body
+ * @returns {Promise<string | streamWeb.ReadableStream | stream.Readable | Buffer>}
+ */
+async function _convertBody(body) {
+	if (body == undefined || Object.is(body.constructor, Object)) {
+		return JSON.stringify(body) // almost every POST request is going to follow this one
+	} else if (body instanceof stream.Readable && reg.ooye.content_length_workaround) {
+		return await buffer(body) // content length workaround is set, so convert to buffer. the buffer consumer accepts node streams.
+	} else if (body instanceof stream.Readable) {
+		return stream.Readable.toWeb(body) // native fetch can only consume web streams
+	} else if (body instanceof streamWeb.ReadableStream && reg.ooye.content_length_workaround) {
+		return await buffer(body) // content lenght workaround is set, so convert to buffer. the buffer consumer accepts async iterables, which web streams are.
+	}
+	return body
+}
+
+/* c8 ignore start */
+
 /**
  * @param {string} method
  * @param {string} url
- * @param {string | object | streamWeb.ReadableStream | stream.Readable} [body]
+ * @param {string | object | streamWeb.ReadableStream | stream.Readable} [bodyIn]
  * @param {any} [extra]
  */
-async function mreq(method, url, body, extra = {}) {
-	if (body == undefined || Object.is(body.constructor, Object)) {
-		body = JSON.stringify(body)
-	} else if (body instanceof stream.Readable && reg.ooye.content_length_workaround) {
-		body = await getStream.buffer(body)
-	} else if (body instanceof streamWeb.ReadableStream && reg.ooye.content_length_workaround) {
-		body = await stream.consumers.buffer(stream.Readable.fromWeb(body))
-	}
+async function mreq(method, url, bodyIn, extra = {}) {
+	const body = await _convertBody(bodyIn)
 
 	/** @type {RequestInit} */
 	const opts = mixin({
@@ -86,3 +99,4 @@ module.exports.MatrixServerError = MatrixServerError
 module.exports.baseUrl = baseUrl
 module.exports.mreq = mreq
 module.exports.withAccessToken = withAccessToken
+module.exports._convertBody = _convertBody

src/matrix/mreq.test.js

@@ -0,0 +1,47 @@
+// @ts-check
+
+const assert = require("assert")
+const stream = require("stream")
+const streamWeb = require("stream/web")
+const {buffer} = require("stream/consumers")
+const {test} = require("supertape")
+const {_convertBody} = require("./mreq")
+const {reg} = require("./read-registration")
+
+async function *generator() {
+	yield "a"
+	yield "b"
+}
+
+reg.ooye.content_length_workaround = false
+
+test("convert body: converts object to string", async t => {
+	t.equal(await _convertBody({a: "1"}), `{"a":"1"}`)
+})
+
+test("convert body: leaves undefined as undefined", async t => {
+	t.equal(await _convertBody(undefined), undefined)
+})
+
+test("convert body: leaves web readable as web readable", async t => {
+	const webReadable = stream.Readable.toWeb(stream.Readable.from(generator()))
+	t.equal(await _convertBody(webReadable), webReadable)
+})
+
+test("convert body: converts node readable to web readable (for native fetch upload)", async t => {
+	const readable = stream.Readable.from(generator())
+	const webReadable = await _convertBody(readable)
+	assert(webReadable instanceof streamWeb.ReadableStream)
+	t.deepEqual(await buffer(webReadable), Buffer.from("ab"))
+})
+
+test("convert body: converts node readable to buffer", async t => {
+	reg.ooye.content_length_workaround = true
+	const readable = stream.Readable.from(generator())
+	t.deepEqual(await _convertBody(readable), Buffer.from("ab"))
+})
+
+test("convert body: converts web readable to buffer", async t => {
+	const webReadable = stream.Readable.toWeb(stream.Readable.from(generator()))
+	t.deepEqual(await _convertBody(webReadable), Buffer.from("ab"))
+})

src/types.d.ts

@@ -167,6 +167,8 @@ export namespace Event {
 	export type M_Room_Message_File = {
 		msgtype: "m.file" | "m.image" | "m.video" | "m.audio"
 		body: string
+		format?: "org.matrix.custom.html"
+		formatted_body?: string
 		filename?: string
 		url: string
 		info?: any
@@ -184,6 +186,8 @@ export namespace Event {
 	export type M_Room_Message_Encrypted_File = {
 		msgtype: "m.file" | "m.image" | "m.video" | "m.audio"
 		body: string
+		format?: "org.matrix.custom.html"
+		formatted_body?: string
 		filename?: string
 		file: {
 			url: string

src/web/server.test.js

@@ -1,14 +1,14 @@
 // @ts-check
 
 const streamWeb = require("stream/web")
-const {test} = require("supertape")
+const {test} = require("../../test/web")
 const {router} = require("../../test/web")
 const assert = require("assert").strict
 
 require("./server")
 
 test("web server: can get home", async t => {
-	t.match(await router.test("get", "/", {}), /Add the bot to your Discord server./)
+	t.has(await router.test("get", "/", {}), /a bridge between the Discord and Matrix chat apps/)
 })
 
 test("web server: can get htmx", async t => {