From 8478cf443720a0b0284f5be3a5bce9a59642c22a Mon Sep 17 00:00:00 2001 From: Il Harper Date: Tue, 5 Mar 2024 19:28:39 +0800 Subject: [PATCH] feat(event): add parser --- .../src/parser/index.ts | 546 ++++++++++++++++++ .../engine-chronocat-event/src/parser/msgt.ts | 141 +++++ 2 files changed, 687 insertions(+) create mode 100644 packages/engine-chronocat-event/src/parser/index.ts create mode 100644 packages/engine-chronocat-event/src/parser/msgt.ts diff --git a/packages/engine-chronocat-event/src/parser/index.ts b/packages/engine-chronocat-event/src/parser/index.ts new file mode 100644 index 0000000..2dd395a --- /dev/null +++ b/packages/engine-chronocat-event/src/parser/index.ts @@ -0,0 +1,546 @@ +import type { + Channel, + ChronocatContext, + ChronocatLogCurrentConfig, + ChronocatSatoriEventsConfig, + Event, + Guild, + GuildMember, + RedMessage, +} from '@chronocat/shell' +import { + ChannelType, + ChatType, + FaceType, + MsgType, + SendType, +} from '@chronocat/shell' +import { l } from '@chronocat/shell/lib/services/logger' +import { PLATFORM } from '@chronocat/shell/lib/utils/consts' +import h from '@satorijs/element' +import { Buffer } from 'node:buffer' +import type { O } from 'ts-toolbelt' +import { parseMsgTypes } from './msgt' + +export const buildParser = + ( + ctx: ChronocatContext, + config: O.Intersect, + ) => + (message: RedMessage) => + parseMessageRecv(ctx, config, message) + +export const parseMessageRecv = async ( + ctx: ChronocatContext, + config: O.Intersect, + message: RedMessage, +) => { + const parsed = await parseMessage(ctx, config, message) + + if (!parsed) return undefined + + const result: Event[] = [] + + for (const event of parsed) { + if (!event.message?.id && !event.user?.id) { + l.warn('satori: parser: 丢弃了一条消息', { code: 2127 }) + continue + } else if (!event.message?.id) + l.warn( + `satori: parser: 来自 ${event.user?.name} (${event.user?.id}) 的消息不带有 messageId，请注意。`, + { code: 2128 }, + ) + else if (!event.user?.id) + l.warn( + `satori: parser: 消息 ${event.message.id} 不带有 userId，请注意。`, + { code: 2129 }, + ) + else if (!event.user?.name && !event.member?.nick) + l.warn( + `satori: parser: 消息 ${event.message.id} 不带有 userName，请注意。`, + { + code: 2130, + }, + ) + + result.push(event) + } + + return result +} + +export const parseMessage = async ( + ctx: ChronocatContext, + config: O.Intersect, + message: RedMessage, +) => { + const event: Event = { + id: undefined as unknown as number, + type: undefined as unknown as string, + + platform: undefined as unknown as string, + self_id: undefined as unknown as string, + timestamp: Number(message.msgTime) * 1000, + } + + event.user = { + id: message.senderUin, + name: (message.sendNickName || undefined) as unknown as string, + avatar: `http://thirdqq.qlogo.cn/headimg_dl?dst_uin=${message.senderUin}&spec=640`, + } + + const ntMsgTypes = parseMsgTypes(message) + + // 无论哪种消息都有 Channel 和 User + event.channel = {} as Channel + + // 判断消息来源 + switch (ntMsgTypes.chatType) { + case ChatType.Private: + event.channel.type = ChannelType.DIRECT + event.channel.id = `private:${event.user.id}` + event.channel.name = event.user.name + break + + case ChatType.Group: + // Guild 和 Member 只有群聊有 + event.guild = {} as Guild + event.member = {} as GuildMember + + if (message.sendMemberName) event.member.nick = message.sendMemberName + + event.channel.type = ChannelType.TEXT + event.channel.id = event.guild.id = message.peerUid + event.channel.name = event.guild.name = message.peerName + event.guild.avatar = `https://p.qlogo.cn/gh/${message.peerUid}/${message.peerUid}/640` + break + } + + if ( + ntMsgTypes.msgType === MsgType.Ark && + message.subMsgType === 0 && + ntMsgTypes.sendType === SendType.Normal + ) + // ARK 卡片消息，elementType = 10 + // 不处理 + return undefined + else if ( + ntMsgTypes.msgType === MsgType.Normal || + ntMsgTypes.msgType === MsgType.Value3 || + ntMsgTypes.msgType === MsgType.Ptt || + ntMsgTypes.msgType === MsgType.Video || + ntMsgTypes.msgType === MsgType.WithRecords || + ntMsgTypes.msgType === MsgType.Vaule17 + ) + return parseChatMessage(ctx, config, event, message).then((x) => [ + x[0], + ...x[1], + ]) + // else if (event.__CHRONO_UNSAFE_NTMSGTYPES__.subMsgType.multiForward) + // // 合并转发消息 + // // multiForwardMsgElement（elementType = 16）内并不带有合并转发的全部内容， + // // 需要后续通过 API 再请求 + // // 考虑到合并转发消息解析需求较少，不调度此 session + // return undefined + else if ( + ntMsgTypes.msgType === MsgType.System && // 5 + message.subMsgType === 8 && // 8 + ntMsgTypes.sendType === SendType.System && // 3 + message.elements[0]!.elementType === 8 && // 8 + message.elements[0]!.grayTipElement!.subElementType === 4 && // 4 + message.elements[0]!.grayTipElement!.groupElement!.type === 1 // 1 + ) + // 新人自行入群 + return await parseGuildMemberAddedMessage(ctx, config, event, message) + else if ( + ntMsgTypes.msgType === MsgType.System && // 5 + message.subMsgType === 8 && // 8 + ntMsgTypes.sendType === SendType.System && // 3 + message.elements[0]!.elementType === 8 && // 8 + message.elements[0]!.grayTipElement!.subElementType === 4 && // 4 + message.elements[0]!.grayTipElement!.groupElement!.type === 8 // 8 + ) + // 他人被禁言 + return await parseGuildMemberMuteMessage(ctx, config, event, message) + else if ( + ntMsgTypes.msgType === MsgType.System && // 5 + message.subMsgType === 8 && // 8 + ntMsgTypes.sendType === SendType.System && // 3 + message.elements[0]!.elementType === 8 && // 8 + message.elements[0]!.grayTipElement!.subElementType === 4 && // 4 + message.elements[0]!.grayTipElement!.groupElement!.type === 5 // 5 + ) + // 群名称变更 + return undefined + else if ( + ntMsgTypes.msgType === MsgType.System && // 5 + message.subMsgType === 12 && // 12 + ntMsgTypes.sendType === SendType.System && // 3 + message.elements[0]!.elementType === 8 && // 8 + message.elements[0]!.grayTipElement!.subElementType === 12 && // 12 + message.elements[0]!.grayTipElement!.xmlElement!.busiType === '1' && // 1 + message.elements[0]!.grayTipElement!.xmlElement!.busiId === '10145' // 10145 + ) + // 旧版群成员邀请新人入群 + return await parseGuildMemberAddedLegacyInviteMessage( + ctx, + config, + event, + message, + ) + else if ( + ntMsgTypes.msgType === MsgType.System && + message.subMsgType === 17 && + ntMsgTypes.sendType === SendType.System + ) + // 群主禁止群内临时通话 + // 群主禁止群内发起新的群聊 + return undefined + + return undefined +} + +/** + * 解析聊天消息。 + * + * @remarks + * 在消息没有除了消息元素以外的其他属性需要处理的情况下，直接使用此方法。 + */ +async function parseChatMessage( + ctx: ChronocatContext, + config: O.Intersect, + event: Event, + message: RedMessage, +) { + const [elements, extraEvents] = await parseElements(ctx, config, message) + event.type = 'message-created' + event.message = { + id: message.msgId, + content: elements.join(''), + } + return [event, extraEvents] as const +} + +/** + * 解析新人自行入群消息。 + * + * @remarks + * 通过解析 `grayTipElement`（elementType = 8）中的 + * `groupElement`（subElementType = 4）即可直接提取 QQ 号。 + */ +async function parseGuildMemberAddedMessage( + ctx: ChronocatContext, + config: O.Intersect, + event: Event, + message: RedMessage, +) { + const [event2, extraEvents] = await parseChatMessage( + ctx, + config, + event, + message, + ) + event2.type = 'guild-member-added' + + event2.operator = { + id: message.elements[0]!.grayTipElement!.groupElement!.adminUin!, + name: undefined as unknown as string, + } + + event2.user = { + id: message.elements[0]!.grayTipElement!.groupElement!.memberUin!, + name: message.elements[0]!.grayTipElement!.groupElement!.memberNick!, + avatar: `http://thirdqq.qlogo.cn/headimg_dl?dst_uin=${ + message.elements[0]!.grayTipElement!.groupElement!.memberUin + }&spec=640`, + } + + if (event2.member) delete event2.member + + return [event2, ...extraEvents] +} + +/** + * 解析他人被禁言消息。 + */ +async function parseGuildMemberMuteMessage( + ctx: ChronocatContext, + config: O.Intersect, + event: Event, + message: RedMessage, +) { + const [event2, extraEvents] = await parseChatMessage( + ctx, + config, + event, + message, + ) + if ( + Number(message.elements[0]!.grayTipElement!.groupElement!.shutUp!.duration) + ) + event2.type = 'unsafe-guild-mute' + else event2.type = 'unsafe-guild-unmute' + + event2.operator = { + id: message.elements[0]!.grayTipElement!.groupElement!.shutUp!.admin.uin, + name: undefined as unknown as string, + } + + event2.user = { + id: message.elements[0]!.grayTipElement!.groupElement!.shutUp!.member.uin, + name: message.elements[0]!.grayTipElement!.groupElement!.shutUp!.member + .name, + avatar: `http://thirdqq.qlogo.cn/headimg_dl?dst_uin=${ + message.elements[0]!.grayTipElement!.groupElement!.shutUp!.member.uin + }&spec=640`, + } + + if (event2.member) delete event2.member + + return [event2, ...extraEvents] +} + +const regexGuildMemberAddedLegacyInviteMessage = /jp="(\d+)".*jp="(\d+)"/gim + +/** + * 解析旧版群成员邀请新人入群消息。使用 NT + * 以前的客户端邀请他人加群会收到此消息。 + * + * @remarks + * 遗憾地，旧版群成员邀请新人入群消息不存在能够直接获取成员 QQ + * 号的方法。需要通过解析 `grayTipElement`（elementType = 8）中的 + * `xmlElement`（subElementType = 12）中的 HTML 来提取 QQ 号。 + * + * 一个 HTML 的示例： + * + * ```html + * + * + * + * + * + * + * ``` + * + * 目前使用正则进行 QQ 号的提取。如果未来正则失效，则应考虑使用 cheerio + * 进行提取。 + */ +async function parseGuildMemberAddedLegacyInviteMessage( + ctx: ChronocatContext, + config: O.Intersect, + event: Event, + message: RedMessage, +) { + const [event2, extraEvents] = await parseChatMessage( + ctx, + config, + event, + message, + ) + event2.type = 'guild-member-added' + + const execArr = regexGuildMemberAddedLegacyInviteMessage.exec( + message.elements[0]!.grayTipElement!.xmlElement!.content!, + ) + if (!Array.isArray(execArr) || execArr.length < 3) return undefined + const [_, operatorId, userId] = execArr + + event2.operator = { + id: operatorId!, + name: undefined as unknown as string, + } + + event2.user = { + id: userId!, + name: undefined as unknown as string, + avatar: `http://thirdqq.qlogo.cn/headimg_dl?dst_uin=${userId}&spec=640`, + } + + if (event2.member) delete event2.member + + return [event2, ...extraEvents] +} + +/** + * 解析消息元素。 + */ +async function parseElements( + ctx: ChronocatContext, + config: O.Intersect, + message: RedMessage, +) { + const elements: h[] = [] + const extraEvents: Event[] = [] + + for (const m of message.elements) { + switch (m.elementType) { + case 1: { + // 文本消息 + switch (m.textElement!.atType) { + case 0: { + // 纯文本消息 + elements.push(h.text(m.textElement?.content)) + break + } + + case 2: { + // at 消息 + const id = m.textElement!.atNtUin + const name = m.textElement!.content.slice(1) + if (!id) { + l.warn( + `satori: parser: at 目标 ${name} 不带有 id，将跳过该元素。`, + { code: 2131 }, + ) + break + } + elements.push( + h('at', { + id, + name, + }), + ) + break + } + } + break + } + + case 2: { + // 图片消息 + elements.push( + h('img', { + src: `${config.self_url}/v1/assets/${Buffer.from( + JSON.stringify({ + msgId: message.msgId, + chatType: message.chatType, + peerUid: message.peerUid, + elementId: m.elementId, + thumbSize: m.picElement!.thumbFileSize, + }), + ).toString('base64url')}`, + }), + ) + break + } + + case 3: { + // 文件消息 + elements.push( + h('file', { + src: `${config.self_url}/v1/assets/${Buffer.from( + JSON.stringify({ + msgId: message.msgId, + chatType: message.chatType, + peerUid: message.peerUid, + elementId: m.elementId, + thumbSize: m.fileElement!.thumbFileSize, + }), + ).toString('base64url')}`, + }), + ) + break + } + + case 4: { + // 语音消息 + elements.push( + h('audio', { + src: `${config.self_url}/v1/assets/${Buffer.from( + JSON.stringify({ + msgId: message.msgId, + chatType: message.chatType, + peerUid: message.peerUid, + elementId: m.elementId, + thumbSize: 0, + }), + ).toString('base64url')}`, + }), + ) + break + } + + case 5: { + // 视频消息 + elements.push( + h('video', { + src: `${config.self_url}/v1/assets/${Buffer.from( + JSON.stringify({ + msgId: message.msgId, + chatType: message.chatType, + peerUid: message.peerUid, + elementId: m.elementId, + thumbSize: m.videoElement!.thumbSize, + }), + ).toString('base64url')}`, + }), + ) + break + } + + case 6: { + // 表情 + switch (m.faceElement!.faceType) { + case FaceType.PCPoke: { + elements.push( + h(`${PLATFORM}:pcpoke`, { + id: m.faceElement!.pokeType, + }), + ) + break + } + + case FaceType.Normal: + case FaceType.Super: { + elements.push( + h(`${PLATFORM}:face`, { + id: m.faceElement!.faceIndex, + name: `[${(await ctx.chronocat.api['chronocat.internal.qface.get'](`${m.faceElement!.faceIndex}`))!.QDes.slice(1)}]`, + platform: PLATFORM, + 'unsafe-super': m.faceElement!.faceType === FaceType.Super, + 'unsafe-result-id': m.faceElement!.resultId, + 'unsafe-chain-count': m.faceElement!.chainCount, + }), + ) + break + } + } + break + } + + case 7: { + // 引用消息 + const source = message.records.find( + (x) => x.msgId === m.replyElement!.sourceMsgIdInRecords, + )! + + elements.push( + h( + 'quote', + { + 'chronocat:seq': m.replyElement!.replayMsgSeq, + }, + [ + await parseAuthor(source), + ...(await parseElements(ctx, config, source))[0], + ], + ), + ) + break + } + + default: + break + } + } + + return [elements, extraEvents] as const +} + +async function parseAuthor(message: RedMessage) { + return h('author', { + id: message.senderUin, + name: message.sendMemberName || message.sendNickName, + avatar: `http://thirdqq.qlogo.cn/headimg_dl?dst_uin=${message.senderUin}&spec=640`, + }) +} diff --git a/packages/engine-chronocat-event/src/parser/msgt.ts b/packages/engine-chronocat-event/src/parser/msgt.ts new file mode 100644 index 0000000..3cfd04d --- /dev/null +++ b/packages/engine-chronocat-event/src/parser/msgt.ts @@ -0,0 +1,141 @@ +import type { ChatType, RedMessage, SendType } from '@chronocat/shell' +import { MsgType } from '@chronocat/shell' + +/** + * QQNT 消息类型。 + * + * @remarks + * # 消息类型 + * + * QQNT 的消息类型共有四种，它们分别是 {@link RedMessage.chatType}、{@link RedMessage.msgType}、{@link RedMessage.subMsgType} + * 和 {@link RedMessage.sendType}。下面分别介绍这四种类型。 + * + * ## {@link RedMessage.chatType} + * + * 消息的归属。目前只有 {@link ChatType.Private} 和 {@link ChatType.Group}。 + * + * ## {@link RedMessage.msgType} + * + * 消息的主要类型。 + * + * ## {@link RedMessage.subMsgType} + * + * 消息内容中包含的元素类型。普通消息的情况下以 Bitset 提供，可使用 {@link adaptMsgTypes} + * 解析为一组 {@link Boolean} 字段。 + * + * ## {@link RedMessage.sendType} + * + * 消息的呈现类型。 + */ +export interface MsgTypes { + chatType: ChatType + + msgType: MsgType + + subMsgType: { + /** + * 文本消息，如纯文本消息、at 消息等。 + * + * @remarks + * - Bit: `0000 0010 | 0000 0000 0000 0001` = `2-1` + * - Bit: `0000 1001 | 0000 0000 0000 0001` = `9-1` + * - `elementType`: `1` + */ + text: boolean + + /** + * 图片。 + * + * @remarks + * - Bit: `0000 0010 | 0000 0000 0000 0010` = `2-2` + * - `elementType`: `2` + */ + pic: boolean + + /** + * 普通表情。 + * + * @remarks + * - Bit: `0000 0010 | 0000 0000 0001 0000` = `2-16` + * - Bit: `0000 1001 | 0000 0000 0001 0000` = `9-16` + * - `elementType`: `6` + */ + face: boolean + + /** + * 链接。 + * + * @remarks + * - Bit: `0000 0010 | 0000 0000 1000 0000` = `2-128` + * - `elementType`: `1` + */ + link: boolean + + /** + * 合并转发。 + * + * @remarks + * - Bit: `0000 0010 | 0000 0000 0000 1000` = `2-8` (NT 合并转发) + * - Bit: `0000 1000 | 0000 0000 0000 0000` = `8-0` (手 Q 合并转发) + * - `elementType`: `16` + */ + multiForward: boolean + + /** + * 回复（Quote）。 + * + * @remarks + * - Bit: `0000 1001 | 0000 0000 0010 0000` = `9-32` + * - `elementType`: `7` + */ + reply: boolean + + /** + * 原创表情。 + * + * @remarks + * - Bit: `0001 0001 | 0000 0000 0000 1000` = `17-8` + * - `elementType`: `11` + */ + marketFace: boolean + + /** + * 文件。 + * + * @remarks + * - Bit: `0000 0011 | 0000 0010 0000 0000` = `3-512` + * - `elementType`: `3` + */ + file: boolean + } + + sendType: SendType +} + +/** + * 解析 QQNT 消息类型。 + */ +export const parseMsgTypes = (message: RedMessage): MsgTypes => ({ + chatType: message.chatType, + + msgType: message.msgType, + + subMsgType: { + text: Boolean(message.subMsgType & (1 << 0)), + pic: Boolean(message.subMsgType & (1 << 1)), + face: Boolean(message.subMsgType & (1 << 4)), + link: Boolean(message.subMsgType & (1 << 7)), + multiForward: Boolean(message.subMsgType & (1 << 3)), + reply: + message.msgType === MsgType.WithRecords && + Boolean(message.subMsgType & (1 << 5)), + marketFace: + message.msgType == MsgType.Vaule17 && + Boolean(message.subMsgType & (1 << 3)), + file: + message.msgType === MsgType.Value3 && + Boolean(message.subMsgType & (1 << 9)), + }, + + sendType: message.sendType, +})