out-of-your-element/m2d/converters/event-to-message.js

286 lines
11 KiB
JavaScript
Raw Normal View History

2023-07-02 13:06:05 +00:00
// @ts-check
2023-07-04 05:19:17 +00:00
const Ty = require("../../types")
2023-07-02 13:06:05 +00:00
const DiscordTypes = require("discord-api-types/v10")
2023-08-25 13:43:17 +00:00
const chunk = require("chunk-text")
const TurndownService = require("turndown")
2023-07-02 13:06:05 +00:00
const passthrough = require("../../passthrough")
const { sync, db, discord } = passthrough
/** @type {import("../../matrix/file")} */
const file = sync.require("../../matrix/file")
/** @type {import("../converters/utils")} */
const utils = sync.require("../converters/utils")
2023-07-02 13:06:05 +00:00
2023-08-25 13:43:17 +00:00
const BLOCK_ELEMENTS = [
"ADDRESS", "ARTICLE", "ASIDE", "AUDIO", "BLOCKQUOTE", "BODY", "CANVAS",
"CENTER", "DD", "DETAILS", "DIR", "DIV", "DL", "DT", "FIELDSET", "FIGCAPTION", "FIGURE",
"FOOTER", "FORM", "FRAMESET", "H1", "H2", "H3", "H4", "H5", "H6", "HEADER",
"HGROUP", "HR", "HTML", "ISINDEX", "LI", "MAIN", "MENU", "NAV", "NOFRAMES",
"NOSCRIPT", "OL", "OUTPUT", "P", "PRE", "SECTION", "SUMMARY", "TABLE", "TBODY", "TD",
"TFOOT", "TH", "THEAD", "TR", "UL"
]
function cleanAttribute (attribute) {
return attribute ? attribute.replace(/(\n+\s*)+/g, "\n") : ""
}
2023-08-25 13:43:17 +00:00
const turndownService = new TurndownService({
2023-08-25 14:04:49 +00:00
hr: "----",
headingStyle: "atx",
preformattedCode: true,
codeBlockStyle: "fenced"
2023-08-25 13:43:17 +00:00
})
2023-08-26 08:30:22 +00:00
turndownService.remove("mx-reply")
2023-08-25 13:43:17 +00:00
turndownService.addRule("strikethrough", {
filter: ["del", "s", "strike"],
replacement: function (content) {
return "~~" + content + "~~"
}
})
2023-08-26 10:59:22 +00:00
turndownService.addRule("underline", {
filter: ["u"],
replacement: function (content) {
return "__" + content + "__"
}
})
2023-08-26 07:07:19 +00:00
turndownService.addRule("blockquote", {
filter: "blockquote",
replacement: function (content) {
content = content.replace(/^\n+|\n+$/g, "")
content = content.replace(/^/gm, "> ")
return content
}
})
turndownService.addRule("inlineLink", {
filter: function (node, options) {
return (
options.linkStyle === "inlined" &&
node.nodeName === "A" &&
node.getAttribute("href")
)
},
replacement: function (content, node) {
if (node.getAttribute("data-user-id")) return `<@${node.getAttribute("data-user-id")}>`
if (node.getAttribute("data-channel-id")) return `<#${node.getAttribute("data-channel-id")}>`
const href = node.getAttribute("href")
let title = cleanAttribute(node.getAttribute("title"))
if (title) title = ` "` + title + `"`
let brackets = ["", ""]
if (href.startsWith("https://matrix.to")) brackets = ["<", ">"]
return "[" + content + "](" + brackets[0] + href + title + brackets[1] + ")"
}
})
2023-08-26 07:07:19 +00:00
turndownService.addRule("fencedCodeBlock", {
filter: function (node, options) {
return (
options.codeBlockStyle === "fenced" &&
node.nodeName === "PRE" &&
node.firstChild &&
node.firstChild.nodeName === "CODE"
)
},
replacement: function (content, node, options) {
const className = node.firstChild.getAttribute("class") || ""
const language = (className.match(/language-(\S+)/) || [null, ""])[1]
const code = node.firstChild
const visibleCode = code.childNodes.map(c => c.nodeName === "BR" ? "\n" : c.textContent).join("").replace(/\n*$/g, "")
var fence = "```"
return (
fence + language + "\n" +
visibleCode +
"\n" + fence
)
}
})
/**
* @param {string} roomID
* @param {string} mxid
* @returns {Promise<{displayname?: string?, avatar_url?: string?}>}
*/
async function getMemberFromCacheOrHomeserver(roomID, mxid, api) {
const row = db.prepare("SELECT displayname, avatar_url FROM member_cache WHERE room_id = ? AND mxid = ?").get(roomID, mxid)
if (row) return row
return api.getStateEvent(roomID, "m.room.member", mxid).then(event => {
db.prepare("REPLACE INTO member_cache (room_id, mxid, displayname, avatar_url) VALUES (?, ?, ?, ?)").run(roomID, mxid, event?.displayname || null, event?.avatar_url || null)
return event
}).catch(() => {
return {displayname: null, avatar_url: null}
})
}
2023-07-02 13:06:05 +00:00
/**
2023-07-04 05:19:17 +00:00
* @param {Ty.Event.Outer<Ty.Event.M_Room_Message>} event
2023-08-26 08:30:22 +00:00
* @param {import("discord-api-types/v10").APIGuild} guild
* @param {{api: import("../../matrix/api")}} di simple-as-nails dependency injection for the matrix API
2023-07-02 13:06:05 +00:00
*/
2023-08-26 08:30:22 +00:00
async function eventToMessage(event, guild, di) {
2023-07-02 13:06:05 +00:00
/** @type {(DiscordTypes.RESTPostAPIWebhookWithTokenJSONBody & {files?: {name: string, file: Buffer}[]})[]} */
2023-08-25 13:43:17 +00:00
let messages = []
2023-07-02 13:06:05 +00:00
let displayName = event.sender
let avatarURL = undefined
2023-08-27 13:30:07 +00:00
/** @type {string[]} */
let messageIDsToEdit = []
2023-08-26 08:30:22 +00:00
let replyLine = ""
// Extract a basic display name from the sender
const match = event.sender.match(/^@(.*?):/)
if (match) displayName = match[1]
// Try to extract an accurate display name and avatar URL from the member event
const member = await getMemberFromCacheOrHomeserver(event.room_id, event.sender, di?.api)
if (member.displayname) displayName = member.displayname
if (member.avatar_url) avatarURL = utils.getPublicUrlForMxc(member.avatar_url)
2023-08-25 13:43:17 +00:00
// Convert content depending on what the message is
let content = event.content.body // ultimate fallback
if (event.content.format === "org.matrix.custom.html" && event.content.formatted_body) {
let input = event.content.formatted_body
if (event.content.msgtype === "m.emote") {
input = `* ${displayName} ${input}`
}
// Note: Element's renderers on Web and Android currently collapse whitespace, like the browser does. Turndown also collapses whitespace which is good for me.
// If later I'm using a client that doesn't collapse whitespace and I want turndown to follow suit, uncomment the following line of code, and it Just Works:
// input = input.replace(/ /g, "&nbsp;")
// There is also a corresponding test to uncomment, named "event2message: whitespace is retained"
2023-08-27 13:30:07 +00:00
// 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 () => {
if (!event.content["m.new_content"]) return
const relatesTo = event.content["m.relates_to"]
if (!relatesTo) return
// Check if we have a pointer to what was edited
const relType = relatesTo.rel_type
if (relType !== "m.replace") return
const originalEventId = relatesTo.event_id
if (!originalEventId) return
console.log("a", originalEventId)
messageIDsToEdit = db.prepare("SELECT message_id FROM event_message WHERE event_id = ? ORDER BY part").pluck().all(originalEventId)
if (!messageIDsToEdit.length) return
// Get the original event, then check if it was a reply
const originalEvent = await di.api.getEvent(event.room_id, originalEventId)
if (!originalEvent) return
const repliedToEventId = originalEvent.content["m.relates_to"]?.["m.in_reply_to"]?.event_id
if (!repliedToEventId) return
console.log("c")
// After all that, it's an edit of a reply.
// We'll be sneaky and prepare the message data so that everything else can handle it just like original messages.
Object.assign(event.content, event.content["m.new_content"])
input = event.content.formatted_body || event.content.body
relatesTo["m.in_reply_to"] = {event_id: repliedToEventId}
})()
2023-08-26 08:30:22 +00:00
// Handling replies. We'll look up the data of the replied-to event from the Matrix homeserver.
2023-08-27 13:30:07 +00:00
// Note that an <mx-reply> element is not guaranteed because this might be m.new_content.
2023-08-26 08:30:22 +00:00
await (async () => {
2023-08-27 13:30:07 +00:00
const repliedToEventId = event.content["m.relates_to"]?.["m.in_reply_to"]?.event_id
2023-08-26 08:30:22 +00:00
if (!repliedToEventId) return
const repliedToEvent = await di.api.getEvent(event.room_id, repliedToEventId)
if (!repliedToEvent) return
const row = db.prepare("SELECT channel_id, message_id FROM event_message INNER JOIN message_channel USING (message_id) WHERE event_id = ? ORDER BY part").get(repliedToEventId)
2023-08-26 08:30:22 +00:00
if (row) {
replyLine = `<:L1:1144820033948762203><:L2:1144820084079087647>https://discord.com/channels/${guild.id}/${row.channel_id}/${row.message_id} `
} else {
replyLine = `<:L1:1144820033948762203><:L2:1144820084079087647>`
}
const sender = repliedToEvent.sender
const senderName = sender.match(/@([^:]*)/)?.[1] || sender
const authorID = db.prepare("SELECT discord_id FROM sim WHERE mxid = ?").pluck().get(repliedToEvent.sender)
if (authorID) {
replyLine += `<@${authorID}>: `
} else {
replyLine += `Ⓜ️**${senderName}**: `
}
const repliedToContent = repliedToEvent.content.formatted_body || repliedToEvent.content.body
const contentPreviewChunks = chunk(repliedToContent.replace(/.*<\/mx-reply>/, "").replace(/(?:\n|<br>)+/g, " ").replace(/<[^>]+>/g, ""), 50)
2023-08-26 08:30:22 +00:00
const contentPreview = contentPreviewChunks.length > 1 ? contentPreviewChunks[0] + "..." : contentPreviewChunks[0]
replyLine = `> ${replyLine}\n> ${contentPreview}\n`
2023-08-26 08:30:22 +00:00
})()
// Handling mentions of Discord users
input = input.replace(/("https:\/\/matrix.to\/#\/(@[^"]+)")>/g, (whole, attributeValue, mxid) => {
if (!utils.eventSenderIsFromDiscord(mxid)) return whole
const userID = db.prepare("SELECT discord_id FROM sim WHERE mxid = ?").pluck().get(mxid)
if (!userID) return whole
return `${attributeValue} data-user-id="${userID}">`
})
// Handling mentions of Discord rooms
input = input.replace(/("https:\/\/matrix.to\/#\/(![^"]+)")>/g, (whole, attributeValue, roomID) => {
const channelID = db.prepare("SELECT channel_id FROM channel_room WHERE room_id = ?").pluck().get(roomID)
if (!channelID) return whole
return `${attributeValue} data-channel-id="${channelID}">`
})
2023-08-26 08:30:22 +00:00
// Element adds a bunch of <br> before </blockquote> but doesn't render them. I can't figure out how this even works in the browser, so let's just delete those.
2023-08-26 07:07:19 +00:00
input = input.replace(/(?:\n|<br ?\/?>\s*)*<\/blockquote>/g, "</blockquote>")
2023-08-25 13:43:17 +00:00
// 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) => {
2023-08-26 07:07:19 +00:00
// console.error(beforeContext, beforeTag, afterContext, afterTag)
2023-08-25 13:43:17 +00:00
if (typeof beforeTag !== "string" && typeof afterTag !== "string") {
return "<br>"
}
beforeContext = beforeContext || ""
beforeTag = beforeTag || ""
afterContext = afterContext || ""
afterTag = afterTag || ""
if (!BLOCK_ELEMENTS.includes(beforeTag.toUpperCase()) && !BLOCK_ELEMENTS.includes(afterTag.toUpperCase())) {
return beforeContext + "<br>" + afterContext
} else {
return whole
}
})
2023-08-25 13:43:17 +00:00
// @ts-ignore bad type from turndown
2023-08-25 13:43:17 +00:00
content = turndownService.turndown(input)
// It's optimised for commonmark, we need to replace the space-space-newline with just newline
content = content.replace(/ \n/g, "\n")
2023-08-25 14:04:49 +00:00
} else {
// Looks like we're using the plaintext body!
// Markdown needs to be escaped
content = content.replace(/([*_~`#])/g, `\\$1`)
2023-07-02 13:06:05 +00:00
}
2023-08-26 08:30:22 +00:00
content = replyLine + content
2023-08-25 13:43:17 +00:00
// Split into 2000 character chunks
const chunks = chunk(content, 2000)
messages = messages.concat(chunks.map(content => ({
content,
username: displayName,
avatar_url: avatarURL
})))
2023-08-27 13:30:07 +00:00
const messagesToEdit = []
const messagesToSend = []
for (let i = 0; i < messages.length; i++) {
if (messageIDsToEdit.length) {
messagesToEdit.push({id: messageIDsToEdit.shift(), message: messages[i]})
} else {
messagesToSend.push(messages[i])
}
}
return {
messagesToEdit,
messagesToSend,
messagesToDelete: messageIDsToEdit
}
2023-07-02 13:06:05 +00:00
}
module.exports.eventToMessage = eventToMessage