update

packages/migrate-vijos/README.md

@@ -0,0 +1,17 @@
+# migrate-vijos
+
+这个模块用于从Vijos4.0的数据库导入数据。
+
+## 迁移须知
+
+迁移前，您需要指定数据来源。  
+**请不要**将 Hydro 正在使用的数据库设置为数据来源的数据库。  
+迁移过程中，Hydro 的以下内容会被**清空**：
+
+题目列表，提交记录，用户列表，比赛列表，比赛成绩表，训练列表，训练进度，站内消息，题解，讨论  
+
+以下内容将被迁移：  
+
+题目与测试数据、题解，讨论与回复，比赛、训练、作业相关数据，提交记录，用户列表，站内消息。
+
+迁移过程中请保持网络畅通。

packages/migrate-vijos/lib.ts

@@ -0,0 +1,17 @@
+import 'hydrooj';
+
+const { md5, sha1 } = global.Hydro.lib;
+
+interface Udoc {
+    uname: string,
+    unameLower: string,
+}
+
+function hash(password: string, salt: string, udoc: Udoc) {
+    const { uname, unameLower } = udoc;
+    const pmd5 = md5(password);
+    const mixedSha1 = sha1(md5(unameLower + pmd5) + salt + sha1(pmd5 + salt));
+    return `${Buffer.from(uname).toString('base64')}|${mixedSha1}`;
+}
+
+global.Hydro.lib['hash.vj2'] = hash;

packages/migrate-vijos/package.json

@@ -0,0 +1,24 @@
+{
+    "name": "@hydrooj/migrate-vijos",
+    "version": "1.3.3",
+    "main": "package.json",
+    "repository": "git@github.com:hydro-dev/Hydro.git",
+    "author": "undefined <masnn0@outlook.com>",
+    "license": "MIT",
+    "scripts": {
+        "lint": "eslint lib.ts script.ts --fix",
+        "build": "tsc"
+    },
+    "devDependencies": {
+        "@types/js-yaml": "^3.12.5",
+        "@types/mongodb": "^3.5.25",
+        "hydrooj": "2.11.5"
+    },
+    "peerDependencies": {
+        "hydrooj": "^2.11.3"
+    },
+    "dependencies": {
+        "js-yaml": "^3.14.0",
+        "mongodb": "^3.6.0"
+    }
+}

packages/migrate-vijos/script.ts

@@ -0,0 +1,524 @@
+/* eslint-disable no-await-in-loop */
+import {
+    Rdoc, TestCase, Mdoc, Drrdoc,
+} from 'hydrooj';
+import fs from 'fs';
+import os from 'os';
+import path from 'path';
+import yaml from 'js-yaml';
+import mongodb, { ObjectID, Db, Cursor } from 'mongodb';
+
+const dst = global.Hydro.service.db;
+const { file, discussion, document } = global.Hydro.model;
+const { testdataConfig } = global.Hydro.lib;
+
+// TODO output enhancement
+const map = {};
+
+const pid = (id) => {
+    if (map[id.toString()]) return map[id.toString()];
+    return id;
+};
+
+const tasks = {
+    user: async (doc) => ({
+        _id: doc._id,
+        uname: doc.uname,
+        unameLower: doc.uname_lower,
+        salt: doc.salt,
+        hash: doc.hash.split('|')[1],
+        hashType: doc.hash.split('|')[0] === 'vj4' ? 'hydro' : doc.hash.split('|')[0],
+        priv: doc.priv,
+        gravatar: doc.gravatar,
+        mail: doc.mail,
+        mailLower: doc.mail_lower,
+        regat: doc.regat,
+        regip: doc.regip,
+        loginat: doc.loginat,
+        loginip: doc.loginip,
+        codeLang: doc.code_lang,
+        codeTemplate: doc.code_template,
+        timezone: doc.timezone,
+        viewLang: doc.viewLang,
+        bio: doc.bio,
+        gender: doc.gender,
+        qq: doc.qq,
+    }),
+    document: {
+        _id: '_id',
+        doc_id: 'docId',
+        doc_type: 'docType',
+        num_submit: 'nSubmit',
+        num_accept: 'nAccept',
+        difficulty: 'difficulty',
+        difficulty_admin: null,
+        difficulty_algo: 'difficultyAlgo',
+        difficulty_setting: 'difficultySetting',
+        pname: 'pid',
+        title: 'title',
+        content: 'content',
+        owner_uid: 'owner',
+        category: 'category',
+        hidden: 'hidden',
+        data: 'data',
+        tag: 'tag',
+        vote: 'vote',
+        reply: {
+            field: 'reply',
+            processer: (reply) => {
+                const res = [];
+                for (const r of reply) {
+                    const drrdoc: Drrdoc = {
+                        _id: r._id,
+                        content: r.content,
+                        owner: r.owner_uid,
+                        ip: r.ip,
+                    };
+                    res.push(drrdoc);
+                }
+                return res;
+            },
+        },
+        ac_msg: 'acMsg',
+        parent_doc_id: {
+            field: 'parentId',
+            processer: (parentId, doc) => {
+                if (doc.parent_doc_type === global.Hydro.model.document.TYPE_PROBLEM) {
+                    return pid(parentId);
+                }
+                return parentId;
+            },
+        },
+        parent_doc_type: 'parentType',
+        num_replies: 'nReply',
+        views: 'views',
+        highlight: 'highlight',
+        ip: 'ip',
+        domain_id: 'domainId',
+        update_at: 'updateAt',
+        begin_at: 'beginAt',
+        end_at: 'endAt',
+        penalty_since: 'penaltySince',
+        penalty_rules: {
+            field: 'penaltyRules',
+            processer: (rule) => {
+                const n = {};
+                for (const key in rule) {
+                    n[parseInt(key, 10) / 3600] = rule;
+                }
+                return n;
+            },
+        },
+        rule: {
+            field: 'rule',
+            processer: (rule: number) => {
+                const rules = {
+                    2: 'oi',
+                    3: 'acm',
+                    11: 'homework',
+                };
+                return rules[rule];
+            },
+        },
+        pids: {
+            field: 'pids',
+            processer: (pids) => pids.map((p) => pid(p)),
+        },
+        attend: 'attend',
+        desc: 'description',
+        enroll: 'enroll',
+        rated: 'rated',
+        dag: {
+            field: 'dag',
+            processer: (dag) => {
+                const r = [];
+                for (const t of dag) {
+                    r.push({
+                        _id: t._id,
+                        title: t.title,
+                        requireNids: t.require_nids,
+                        pids: t.pids.map((id) => pid(id)),
+                    });
+                }
+                return r;
+            },
+        },
+    },
+    'document.status': {
+        _id: '_id',
+        doc_id: 'docId',
+        doc_type: 'docType',
+        uid: 'uid',
+        domain_id: 'domainId',
+        num_accept: 'nAccept',
+        num_submit: 'nSubmit',
+        status: 'status',
+        accept: 'accept',
+        vote: 'vote',
+        star: 'star',
+        enroll: 'enroll',
+        rid: 'rid',
+        rev: 'rev',
+        attend: 'attend',
+        journal: {
+            field: 'journal',
+            processer: (journal) => {
+                const r = [];
+                for (const i of journal) {
+                    r.push({ ...i, pid: pid(i.pid) });
+                }
+                return r;
+            },
+        },
+        penalty_score: 'penaltyScore',
+        detail: {
+            field: 'detail',
+            processer: (detail) => {
+                const r = [];
+                for (const i of detail) {
+                    r.push({ ...i, pid: pid(i.pid) });
+                }
+                return r;
+            },
+        },
+        score: 'score',
+        time: 'time',
+        done: 'done',
+        done_nids: 'doneNids',
+        done_pids: {
+            field: 'donePids',
+            processer: (pids) => pids.map((id) => pid(id)),
+        },
+        rp: null,
+    },
+    'domain.user': {
+        _id: '_id',
+        domain_id: 'domainId',
+        uid: 'uid',
+        num_problems: 'nProblem',
+        num_submit: 'nSubmit',
+        num_accept: 'nAccept',
+        num_liked: 'nLike',
+        level: 'level',
+        role: 'role',
+        join_at: 'joinAt',
+        display_name: 'displayName',
+        rp: null,
+        rank: null,
+    },
+    record: async (doc) => {
+        const testCases: TestCase[] = [];
+        for (const c of doc.cases || []) {
+            testCases.push({
+                status: c.status,
+                time: c.time_ms || c.time,
+                memory: c.memory_kb || c.memory,
+                message: (c.judge_text || '') + (c.message || ''),
+            });
+        }
+        const rdoc: Rdoc = {
+            _id: doc._id,
+            status: doc.status,
+            score: doc.score,
+            time: doc.time_ms,
+            memory: doc.memory_kb,
+            code: doc.code,
+            lang: doc.lang,
+            uid: doc.uid,
+            pid: pid(doc.pid),
+            domainId: doc.domain_id,
+            judger: doc.judge_uid || 1,
+            judgeAt: doc.judge_at || new Date(),
+            judgeTexts: doc.judge_texts || [],
+            compilerTexts: doc.compiler_texts || [],
+            testCases,
+            hidden: !!doc.hidden,
+            rejudged: !!doc.rejudged,
+        };
+        if (doc.rejudged) rdoc.rejudged = true;
+        if (doc.tid) {
+            rdoc.contest = {
+                type: doc.ttype || document.TYPE_CONTEST,
+                tid: doc.tid,
+            };
+        }
+        return rdoc;
+    },
+    domain: {
+        _id: '_id',
+        doc_id: 'docId',
+        doc_type: 'docType',
+        owner_uid: 'owner',
+        name: 'name',
+        roles: {
+            field: 'roles',
+            processer: (roles) => {
+                const res = {};
+                for (const role in roles) {
+                    res[role] = roles[role].toString();
+                }
+                return res;
+            },
+        },
+        gravatar: 'gravatar',
+        bulletin: 'bulletin',
+        pid_counter: 'pidCounter',
+    },
+    'fs.files': async (doc) => doc,
+    'fs.chunks': async (doc) => doc,
+    file: async (doc) => ({
+        _id: doc._id,
+        count: 0,
+        secret: doc.metadata.secret,
+        size: doc.length,
+        md5: doc.md5,
+    }),
+};
+
+type CursorGetter = (s: Db) => Cursor<any>;
+
+const cursor: NodeJS.Dict<CursorGetter> = {
+    user: (s) => s.collection('user').find(),
+    document: (s) => s.collection('document').find({ doc_type: { $ne: 20 } }),
+    'document.status': (s) => s.collection('document.status').find(),
+    'domain.user': (s) => s.collection('domain.user').find(),
+    record: (s) => s.collection('record').find(),
+    domain: (s) => s.collection('domain').find(),
+    'fs.files': (s) => s.collection('fs.files').find(),
+    'fs.chunks': (s) => s.collection('fs.chunks').find(),
+    file: (s) => s.collection('fs.files').find(),
+};
+
+async function discussionNode(src: Db, report: Function) {
+    const count = await src.collection('document').find({ doc_type: 20 }).count();
+    await report({ progress: 1, message: `discussion.node: ${count}` });
+    const total = Math.floor(count / 5);
+    for (let i = 0; i <= total; i++) {
+        const docs = await src.collection('document')
+            .find({ doc_type: 20 }).skip(i * 5).limit(5)
+            .toArray();
+        for (const doc of docs) {
+            const t = [];
+            for (const item of doc.content || []) {
+                const category = item[0];
+                const nodes = item[1];
+                for (const node of nodes || []) {
+                    if (node.pic) {
+                        t.push(discussion.addNode(
+                            doc.domain_id, node.name, category, { pic: node.pic },
+                        ));
+                    } else {
+                        t.push(discussion.addNode(doc.domain_id, node.name, category, {}));
+                    }
+                }
+            }
+            await Promise.all(t).catch((e) => e);
+        }
+        await report({ progress: Math.round(100 * ((i + 1) / (total + 1))) });
+    }
+}
+
+function addSpace(content: string) {
+    const lines = content.split('\r\n');
+    for (let i = 0; i < lines.length; i++) {
+        // eslint-disable-next-line no-continue
+        if (lines[i].endsWith('|')) continue; // Markdown table;
+        if (!lines[i].endsWith('  ')) lines[i] = `${lines[i]}  `;
+    }
+    return lines.join('\n');
+}
+
+async function fix(doc) {
+    await dst.collection('document').updateOne(
+        { _id: doc._id },
+        { $set: { pid: doc.pid || doc.docId.toString(), content: addSpace(doc.content) } },
+    );
+    if (doc.data && doc.data instanceof ObjectID) {
+        const r = await file.get(doc.data);
+        const p = path.resolve(os.tmpdir(), 'hydro', `migrate.vijos.${doc._id}.zip`);
+        const w = fs.createWriteStream(p);
+        await new Promise((resolve, reject) => {
+            w.on('finish', resolve);
+            w.on('error', reject);
+            r.pipe(w);
+        });
+        const config = yaml.safeLoad(await testdataConfig.readConfig(p));
+        await dst.collection('document').updateOne(
+            { _id: doc._id },
+            { $set: { config } },
+        );
+        fs.unlinkSync(p);
+    }
+}
+
+async function fixProblem(report: Function) {
+    const count = await dst.collection('document').find({ docType: 10 }).count();
+    await report({ progress: 1, message: `Fix pid: ${count}` });
+    const total = Math.floor(count / 50);
+    for (let i = 0; i <= total; i++) {
+        const docs = await dst.collection('document')
+            .find({ docType: 10 }).skip(i * 50).limit(50)
+            .toArray();
+        for (const doc of docs) {
+            await fix(doc).catch((e) => report({ message: `${e.toString()}\n${e.stack}` }));
+        }
+        await report({ progress: Math.round(100 * ((i + 1) / (total + 1))) });
+    }
+}
+
+function objid(ts: Date) {
+    const p = Math.floor(ts.getTime() / 1000).toString(16);
+    const id = new mongodb.ObjectID();
+    return new mongodb.ObjectID(p + id.toHexString().slice(8, 8 + 6 + 4 + 6));
+}
+
+// FIXME this seems not working
+async function message(src: Db, report: Function) {
+    const count = await src.collection('message').find().count();
+    await report({ progress: 1, message: `Messages: ${count}` });
+    const total = Math.floor(count / 50);
+    for (let i = 0; i <= total; i++) {
+        const docs = await src.collection('message')
+            .find().skip(i * 50).limit(50)
+            .toArray();
+        for (const doc of docs) {
+            for (const msg of doc.reply) {
+                const mdoc: Mdoc = {
+                    _id: objid(msg.at),
+                    from: msg.sender_uid,
+                    to: msg.sender_uid === doc.sender_uid ? doc.sendee_uid : doc.sender_uid,
+                    content: msg.content,
+                    // Mark all as read
+                    flag: 0,
+                };
+                await dst.collection('message').insertOne(mdoc);
+            }
+        }
+        await report({ progress: Math.round(100 * ((i + 1) / (total + 1))) });
+    }
+}
+
+async function task(name: string, src: Db, report: Function) {
+    const count = await cursor[name](src).count();
+    await report({ progress: 1, message: `${name}: ${count}` });
+    const total = Math.floor(count / 50);
+    let lastProgress = -1;
+    for (let i = 0; i <= total; i++) {
+        const docs = await cursor[name](src).skip(i * 50).limit(50).toArray();
+        const res = [];
+        for (const doc of docs) {
+            let d: any = {};
+            if (typeof tasks[name].call === 'function') {
+                d = await tasks[name](doc);
+            } else {
+                const mapper = tasks[name];
+                for (const key in doc) {
+                    if (typeof mapper[key] === 'string') {
+                        d[mapper[key]] = doc[key];
+                    } else if (mapper[key] === null) {
+                        // Ignore this key
+                    } else if (typeof mapper[key] === 'object') {
+                        d[mapper[key].field] = mapper[key].processer(doc[key], doc);
+                    } else {
+                        await report({ message: `Unknown key ${key} in collection ${name}` });
+                    }
+                }
+            }
+            if (d) {
+                if (d.rule) d.rated = true; // Hack contest rated field
+                const docWithoutId = {};
+                const docWithoutDid = {};
+                for (const key in d) {
+                    if (key !== '_id') {
+                        if (key !== 'domainId' && key !== 'docId' && key !== 'docType' && key !== 'uid') {
+                            docWithoutDid[key] = d[key];
+                        }
+                        docWithoutId[key] = d[key];
+                    }
+                }
+                if (d.domainId && d.docId && d.docType) {
+                    const query: any = { domainId: d.domainId, docId: d.docId, docType: d.docType };
+                    if (d.uid) query.uid = d.uid;
+                    res.push((async () => {
+                        const data = await dst.collection(name).findOne(query);
+                        if (data) {
+                            await dst.collection(name).updateOne(query, { $set: docWithoutDid });
+                        } else if (d._id) {
+                            const dat = await dst.collection(name).findOne({ _id: d._id });
+                            if (dat) {
+                                await dst.collection(name).updateOne({
+                                    _id: d._id,
+                                }, { $set: docWithoutId });
+                            } else {
+                                await dst.collection(name).insertOne(d);
+                            }
+                        } else {
+                            await dst.collection(name).insertOne(d);
+                        }
+                    })());
+                } else if (d._id) {
+                    res.push(dst.collection(name).updateOne({
+                        _id: d._id,
+                    }, { $set: docWithoutId }, { upsert: true }));
+                } else res.push(dst.collection(name).insertOne(d));
+            }
+        }
+        await Promise.all(res).catch((e) => report({ message: `${e}\n${e.stack}` }));
+        const progress = Math.round(100 * ((i + 1) / (total + 1)));
+        if (progress > lastProgress) {
+            await report({ progress });
+            lastProgress = progress;
+        }
+    }
+}
+
+export async function run({
+    host = 'localhost', port = 27017, name = 'vijos4', username, password,
+}, report: Function) {
+    let mongourl = 'mongodb://';
+    if (username) mongourl += `${username}:${password}@`;
+    mongourl += `${host}:${port}/${name}`;
+    const Database = await mongodb.MongoClient.connect(mongourl, {
+        useNewUrlParser: true, useUnifiedTopology: true,
+    });
+    const src = Database.db(name);
+    await report({ progress: 0, message: 'Database connected.' });
+    const userCounter = await src.collection('system').findOne({ _id: 'user_counter' });
+    if (!userCounter) {
+        report({ message: 'No valid installation found' });
+        return false;
+    }
+    await dst.collection('system').updateOne(
+        { _id: 'user' },
+        {
+            $set: {
+                value: userCounter.value,
+            },
+        },
+        { upsert: true },
+    );
+    await report({ progress: 1, message: 'Collection:system done.' });
+    if (!await dst.collection('system').findOne({ _id: 'migrateVijosFs' })) {
+        const f = ['fs.files', 'fs.chunks'];
+        for (const i of f) {
+            await dst.collection(i).deleteMany({});
+            await task(i, src, report);
+        }
+        await dst.collection('system').insertOne({ _id: 'migrateVijosFs', value: true });
+    }
+    await dst.collection('user').deleteMany({ _id: { $nin: [0, 1] } });
+    await dst.collection('message').deleteMany({});
+    const d = ['domain', 'user', 'document', 'document.status', 'domain.user', 'record', 'file'];
+    for (const i of d) await task(i, src, report);
+    await fixProblem(report);
+    await discussionNode(src, report);
+    await message(src, report);
+    return true;
+}
+
+export const description = 'migrate from vijos';
+export const validate = {
+    host: 'string', port: 'number', name: 'string', username: 'string', password: 'string',
+};
+
+global.Hydro.script.migrateVijos = { run, description, validate };

packages/migrate-vijos/tsconfig.json

@@ -0,0 +1,11 @@
+{
+    "extends": "../../tsconfig.base",
+    "compilerOptions": {
+        "outDir": ".",
+        "rootDir": "."
+    },
+    "include": [
+        "*.ts"
+    ],
+    "exclude": []
+}