Merge pull request #859 from syuilo/improve/serializers

improve performance & use GridFS for files
This commit is contained in:
こぴなたみぽ 2017-11-06 18:56:14 +09:00 committed by GitHub
commit 00198a827c
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
21 changed files with 330 additions and 205 deletions

View file

@ -95,6 +95,7 @@
"webpack": "3.8.1"
},
"dependencies": {
"@prezzemolo/rap": "0.1.2",
"accesses": "2.5.0",
"animejs": "2.2.0",
"autwh": "0.0.1",

View file

@ -4,14 +4,27 @@ import * as gm from 'gm';
import * as debug from 'debug';
import fileType = require('file-type');
import prominence = require('prominence');
import DriveFile from '../models/drive-file';
import DriveFile, { getGridFSBucket } from '../models/drive-file';
import DriveFolder from '../models/drive-folder';
import serialize from '../serializers/drive-file';
import event from '../event';
import config from '../../conf';
import { Duplex } from 'stream';
const log = debug('misskey:register-drive-file');
const addToGridFS = (name, binary, metadata): Promise<any> => new Promise(async (resolve, reject) => {
const dataStream = new Duplex();
dataStream.push(binary);
dataStream.push(null);
const bucket = await getGridFSBucket();
const writeStream = bucket.openUploadStream(name, { metadata });
writeStream.once('finish', (doc) => { resolve(doc); });
writeStream.on('error', reject);
dataStream.pipe(writeStream);
});
/**
* Add file to drive
*
@ -58,7 +71,7 @@ export default (
// Generate hash
const hash = crypto
.createHash('sha256')
.createHash('md5')
.update(data)
.digest('hex') as string;
@ -67,8 +80,8 @@ export default (
if (!force) {
// Check if there is a file with the same hash
const much = await DriveFile.findOne({
user_id: user._id,
hash: hash
md5: hash,
'metadata.user_id': user._id
});
if (much !== null) {
@ -82,13 +95,13 @@ export default (
// Calculate drive usage
const usage = ((await DriveFile
.aggregate([
{ $match: { user_id: user._id } },
{ $match: { 'metadata.user_id': user._id } },
{ $project: {
datasize: true
length: true
}},
{ $group: {
_id: null,
usage: { $sum: '$datasize' }
usage: { $sum: '$length' }
}}
]))[0] || {
usage: 0
@ -131,21 +144,15 @@ export default (
}
// Create DriveFile document
const file = await DriveFile.insert({
created_at: new Date(),
const file = await addToGridFS(`${user._id}/${name}`, data, {
user_id: user._id,
folder_id: folder !== null ? folder._id : null,
data: data,
datasize: size,
type: mime,
name: name,
comment: comment,
hash: hash,
properties: properties
});
delete file.data;
log(`drive file has been created ${file._id}`);
resolve(file);

View file

@ -14,16 +14,16 @@ module.exports = (params, user) => new Promise(async (res, rej) => {
// Calculate drive usage
const usage = ((await DriveFile
.aggregate([
{ $match: { user_id: user._id } },
{ $match: { 'metadata.user_id': user._id } },
{
$project: {
datasize: true
length: true
}
},
{
$group: {
_id: null,
usage: { $sum: '$datasize' }
usage: { $sum: '$length' }
}
}
]))[0] || {

View file

@ -13,35 +13,35 @@ import serialize from '../../serializers/drive-file';
* @param {any} app
* @return {Promise<any>}
*/
module.exports = (params, user, app) => new Promise(async (res, rej) => {
module.exports = async (params, user, app) => {
// Get 'limit' parameter
const [limit = 10, limitErr] = $(params.limit).optional.number().range(1, 100).$;
if (limitErr) return rej('invalid limit param');
if (limitErr) throw 'invalid limit param';
// Get 'since_id' parameter
const [sinceId, sinceIdErr] = $(params.since_id).optional.id().$;
if (sinceIdErr) return rej('invalid since_id param');
if (sinceIdErr) throw 'invalid since_id param';
// Get 'max_id' parameter
const [maxId, maxIdErr] = $(params.max_id).optional.id().$;
if (maxIdErr) return rej('invalid max_id param');
if (maxIdErr) throw 'invalid max_id param';
// Check if both of since_id and max_id is specified
if (sinceId && maxId) {
return rej('cannot set since_id and max_id');
throw 'cannot set since_id and max_id';
}
// Get 'folder_id' parameter
const [folderId = null, folderIdErr] = $(params.folder_id).optional.nullable.id().$;
if (folderIdErr) return rej('invalid folder_id param');
if (folderIdErr) throw 'invalid folder_id param';
// Construct query
const sort = {
_id: -1
};
const query = {
user_id: user._id,
folder_id: folderId
'metadata.user_id': user._id,
'metadata.folder_id': folderId
} as any;
if (sinceId) {
sort._id = 1;
@ -57,14 +57,11 @@ module.exports = (params, user, app) => new Promise(async (res, rej) => {
// Issue query
const files = await DriveFile
.find(query, {
fields: {
data: false
},
limit: limit,
sort: sort
});
// Serialize
res(await Promise.all(files.map(async file =>
await serialize(file))));
});
const _files = await Promise.all(files.map(file => serialize(file)));
return _files;
};

View file

@ -24,13 +24,9 @@ module.exports = (params, user) => new Promise(async (res, rej) => {
// Issue query
const files = await DriveFile
.find({
name: name,
user_id: user._id,
folder_id: folderId
}, {
fields: {
data: false
}
'metadata.name': name,
'metadata.user_id': user._id,
'metadata.folder_id': folderId
});
// Serialize

View file

@ -12,28 +12,26 @@ import serialize from '../../../serializers/drive-file';
* @param {any} user
* @return {Promise<any>}
*/
module.exports = (params, user) => new Promise(async (res, rej) => {
module.exports = async (params, user) => {
// Get 'file_id' parameter
const [fileId, fileIdErr] = $(params.file_id).id().$;
if (fileIdErr) return rej('invalid file_id param');
if (fileIdErr) throw 'invalid file_id param';
// Fetch file
const file = await DriveFile
.findOne({
_id: fileId,
user_id: user._id
}, {
fields: {
data: false
}
'metadata.user_id': user._id
});
if (file === null) {
return rej('file-not-found');
throw 'file-not-found';
}
// Serialize
res(await serialize(file, {
const _file = await serialize(file, {
detail: true
}));
});
return _file;
};

View file

@ -24,11 +24,7 @@ module.exports = (params, user) => new Promise(async (res, rej) => {
const file = await DriveFile
.findOne({
_id: fileId,
user_id: user._id
}, {
fields: {
data: false
}
'metadata.user_id': user._id
});
if (file === null) {
@ -38,7 +34,7 @@ module.exports = (params, user) => new Promise(async (res, rej) => {
// Get 'name' parameter
const [name, nameErr] = $(params.name).optional.string().pipe(validateFileName).$;
if (nameErr) return rej('invalid name param');
if (name) file.name = name;
if (name) file.metadata.name = name;
// Get 'folder_id' parameter
const [folderId, folderIdErr] = $(params.folder_id).optional.nullable.id().$;
@ -46,7 +42,7 @@ module.exports = (params, user) => new Promise(async (res, rej) => {
if (folderId !== undefined) {
if (folderId === null) {
file.folder_id = null;
file.metadata.folder_id = null;
} else {
// Fetch folder
const folder = await DriveFolder
@ -59,14 +55,14 @@ module.exports = (params, user) => new Promise(async (res, rej) => {
return rej('folder-not-found');
}
file.folder_id = folder._id;
file.metadata.folder_id = folder._id;
}
}
DriveFile.update(file._id, {
await DriveFile.update(file._id, {
$set: {
name: file.name,
folder_id: file.folder_id
'metadata.name': file.metadata.name,
'metadata.folder_id': file.metadata.folder_id
}
});

View file

@ -30,6 +30,5 @@ module.exports = (params, user) => new Promise(async (res, rej) => {
});
// Serialize
res(await Promise.all(folders.map(async folder =>
await serialize(folder))));
res(await Promise.all(folders.map(folder => serialize(folder))));
});

View file

@ -4,7 +4,7 @@
import $ from 'cafy';
import DriveFolder from '../../../models/drive-folder';
import { isValidFolderName } from '../../../models/drive-folder';
import serialize from '../../../serializers/drive-file';
import serialize from '../../../serializers/drive-folder';
import event from '../../../event';
/**

View file

@ -54,9 +54,7 @@ module.exports = (params, user) => new Promise(async (res, rej) => {
if (fileId !== undefined) {
file = await DriveFile.findOne({
_id: fileId,
user_id: user._id
}, {
data: false
'metadata.user_id': user._id
});
if (file === null) {

View file

@ -44,9 +44,7 @@ module.exports = (params, user: IUser, app) => new Promise(async (res, rej) => {
// SELECT _id
const entity = await DriveFile.findOne({
_id: mediaId,
user_id: user._id
}, {
_id: true
'metadata.user_id': user._id
});
if (entity === null) {

View file

@ -2,6 +2,7 @@
* Module dependencies
*/
import $ from 'cafy';
import rap from '@prezzemolo/rap';
import Post from '../../models/post';
import ChannelWatching from '../../models/channel-watching';
import getFriends from '../../common/get-friends';
@ -15,32 +16,33 @@ import serialize from '../../serializers/post';
* @param {any} app
* @return {Promise<any>}
*/
module.exports = (params, user, app) => new Promise(async (res, rej) => {
module.exports = async (params, user, app) => {
// Get 'limit' parameter
const [limit = 10, limitErr] = $(params.limit).optional.number().range(1, 100).$;
if (limitErr) return rej('invalid limit param');
if (limitErr) throw 'invalid limit param';
// Get 'since_id' parameter
const [sinceId, sinceIdErr] = $(params.since_id).optional.id().$;
if (sinceIdErr) return rej('invalid since_id param');
if (sinceIdErr) throw 'invalid since_id param';
// Get 'max_id' parameter
const [maxId, maxIdErr] = $(params.max_id).optional.id().$;
if (maxIdErr) return rej('invalid max_id param');
if (maxIdErr) throw 'invalid max_id param';
// Check if both of since_id and max_id is specified
if (sinceId && maxId) {
return rej('cannot set since_id and max_id');
throw 'cannot set since_id and max_id';
}
const { followingIds, watchChannelIds } = await rap({
// ID list of the user itself and other users who the user follows
const followingIds = await getFriends(user._id);
followingIds: getFriends(user._id),
// Watchしているチャンネルを取得
const watches = await ChannelWatching.find({
watchChannelIds: ChannelWatching.find({
user_id: user._id,
// 削除されたドキュメントは除く
deleted_at: { $exists: false }
}).then(watches => watches.map(w => w.channel_id))
});
//#region Construct query
@ -65,7 +67,7 @@ module.exports = (params, user, app) => new Promise(async (res, rej) => {
}, {
// Watchしているチャンネルへの投稿
channel_id: {
$in: watches.map(w => w.channel_id)
$in: watchChannelIds
}
}]
} as any;
@ -90,7 +92,6 @@ module.exports = (params, user, app) => new Promise(async (res, rej) => {
});
// Serialize
res(await Promise.all(timeline.map(async post =>
await serialize(post, user)
)));
});
const _timeline = await Promise.all(timeline.map(post => serialize(post, user)));
return _timeline;
};

View file

@ -1,11 +1,22 @@
import db from '../../db/mongodb';
import * as mongodb from 'mongodb';
import monkDb, { nativeDbConn } from '../../db/mongodb';
const collection = db.get('drive_files');
const collection = monkDb.get('drive_files.files');
(collection as any).createIndex('hash'); // fuck type definition
export default collection as any; // fuck type definition
const getGridFSBucket = async (): Promise<mongodb.GridFSBucket> => {
const db = await nativeDbConn();
const bucket = new mongodb.GridFSBucket(db, {
bucketName: 'drive_files'
});
return bucket;
};
export { getGridFSBucket };
export function validateFileName(name: string): boolean {
return (
(name.trim().length > 0) &&

View file

@ -31,44 +31,40 @@ export default (
if (mongo.ObjectID.prototype.isPrototypeOf(file)) {
_file = await DriveFile.findOne({
_id: file
}, {
fields: {
data: false
}
});
} else if (typeof file === 'string') {
_file = await DriveFile.findOne({
_id: new mongo.ObjectID(file)
}, {
fields: {
data: false
}
});
} else {
_file = deepcopy(file);
}
// Rename _id to id
_file.id = _file._id;
delete _file._id;
if (!_file) return reject('invalid file arg.');
delete _file.data;
// rendered target
let _target: any = {};
_file.url = `${config.drive_url}/${_file.id}/${encodeURIComponent(_file.name)}`;
_target.id = _file._id;
_target.created_at = _file.uploadDate;
if (opts.detail && _file.folder_id) {
_target = Object.assign(_target, _file.metadata);
_target.url = `${config.drive_url}/${_target.id}/${encodeURIComponent(_target.name)}`;
if (opts.detail && _target.folder_id) {
// Populate folder
_file.folder = await serializeDriveFolder(_file.folder_id, {
_target.folder = await serializeDriveFolder(_target.folder_id, {
detail: true
});
}
if (opts.detail && _file.tags) {
if (opts.detail && _target.tags) {
// Populate tags
_file.tags = await _file.tags.map(async (tag: any) =>
_target.tags = await _target.tags.map(async (tag: any) =>
await serializeDriveTag(tag)
);
}
resolve(_file);
resolve(_target);
});

View file

@ -44,7 +44,7 @@ const self = (
});
const childFilesCount = await DriveFile.count({
folder_id: _folder.id
'metadata.folder_id': _folder.id
});
_folder.folders_count = childFoldersCount;

View file

@ -12,6 +12,7 @@ import serializeChannel from './channel';
import serializeUser from './user';
import serializeDriveFile from './drive-file';
import parse from '../common/text';
import rap from '@prezzemolo/rap';
/**
* Serialize a post
@ -21,13 +22,13 @@ import parse from '../common/text';
* @param options? serialize options
* @return response
*/
const self = (
const self = async (
post: string | mongo.ObjectID | IPost,
me?: string | mongo.ObjectID | IUser,
options?: {
detail: boolean
}
) => new Promise<any>(async (resolve, reject) => {
) => {
const opts = options || {
detail: true,
};
@ -56,6 +57,8 @@ const self = (
_post = deepcopy(post);
}
if (!_post) throw 'invalid post arg.';
const id = _post._id;
// Rename _id to id
@ -70,28 +73,29 @@ const self = (
}
// Populate user
_post.user = await serializeUser(_post.user_id, meId);
_post.user = serializeUser(_post.user_id, meId);
// Populate app
if (_post.app_id) {
_post.app = await serializeApp(_post.app_id);
_post.app = serializeApp(_post.app_id);
}
// Populate channel
if (_post.channel_id) {
_post.channel = await serializeChannel(_post.channel_id);
_post.channel = serializeChannel(_post.channel_id);
}
// Populate media
if (_post.media_ids) {
_post.media = await Promise.all(_post.media_ids.map(async fileId =>
await serializeDriveFile(fileId)
_post.media = Promise.all(_post.media_ids.map(fileId =>
serializeDriveFile(fileId)
));
}
// When requested a detailed post data
if (opts.detail) {
// Get previous post info
_post.prev = (async () => {
const prev = await Post.findOne({
user_id: _post.user_id,
_id: {
@ -105,9 +109,11 @@ const self = (
_id: -1
}
});
_post.prev = prev ? prev._id : null;
return prev ? prev._id : null;
})();
// Get next post info
_post.next = (async () => {
const next = await Post.findOne({
user_id: _post.user_id,
_id: {
@ -121,24 +127,26 @@ const self = (
_id: 1
}
});
_post.next = next ? next._id : null;
return next ? next._id : null;
})();
if (_post.reply_id) {
// Populate reply to post
_post.reply = await self(_post.reply_id, meId, {
_post.reply = self(_post.reply_id, meId, {
detail: false
});
}
if (_post.repost_id) {
// Populate repost
_post.repost = await self(_post.repost_id, meId, {
_post.repost = self(_post.repost_id, meId, {
detail: _post.text == null
});
}
// Poll
if (meId && _post.poll) {
_post.poll = (async (poll) => {
const vote = await Vote
.findOne({
user_id: meId,
@ -146,15 +154,19 @@ const self = (
});
if (vote != null) {
const myChoice = _post.poll.choices
const myChoice = poll.choices
.filter(c => c.id == vote.choice)[0];
myChoice.is_voted = true;
}
return poll;
})(_post.poll);
}
// Fetch my reaction
if (meId) {
_post.my_reaction = (async () => {
const reaction = await Reaction
.findOne({
user_id: meId,
@ -163,12 +175,18 @@ const self = (
});
if (reaction) {
_post.my_reaction = reaction.reaction;
return reaction.reaction;
}
return null;
})();
}
}
resolve(_post);
});
// resolve promises in _post object
_post = await rap(_post);
return _post;
};
export default self;

View file

@ -8,6 +8,7 @@ import serializePost from './post';
import Following from '../models/following';
import getFriends from '../common/get-friends';
import config from '../../conf';
import rap from '@prezzemolo/rap';
/**
* Serialize a user
@ -55,6 +56,8 @@ export default (
_user = deepcopy(user);
}
if (!_user) return reject('invalid user arg.');
// Me
const meId: mongo.ObjectID = me
? mongo.ObjectID.prototype.isPrototypeOf(me)
@ -104,26 +107,30 @@ export default (
if (meId && !meId.equals(_user.id)) {
// If the user is following
_user.is_following = (async () => {
const follow = await Following.findOne({
follower_id: meId,
followee_id: _user.id,
deleted_at: { $exists: false }
});
_user.is_following = follow !== null;
return follow !== null;
})();
// If the user is followed
_user.is_followed = (async () => {
const follow2 = await Following.findOne({
follower_id: _user.id,
followee_id: meId,
deleted_at: { $exists: false }
});
_user.is_followed = follow2 !== null;
return follow2 !== null;
})();
}
if (opts.detail) {
if (_user.pinned_post_id) {
// Populate pinned post
_user.pinned_post = await serializePost(_user.pinned_post_id, meId, {
_user.pinned_post = serializePost(_user.pinned_post_id, meId, {
detail: true
});
}
@ -132,23 +139,24 @@ export default (
const myFollowingIds = await getFriends(meId);
// Get following you know count
const followingYouKnowCount = await Following.count({
_user.following_you_know_count = Following.count({
followee_id: { $in: myFollowingIds },
follower_id: _user.id,
deleted_at: { $exists: false }
});
_user.following_you_know_count = followingYouKnowCount;
// Get followers you know count
const followersYouKnowCount = await Following.count({
_user.followers_you_know_count = Following.count({
followee_id: _user.id,
follower_id: { $in: myFollowingIds },
deleted_at: { $exists: false }
});
_user.followers_you_know_count = followersYouKnowCount;
}
}
// resolve promises in _user object
_user = await rap(_user);
resolve(_user);
});
/*

View file

@ -1,11 +1,38 @@
import * as mongo from 'monk';
import config from '../conf';
const uri = config.mongodb.user && config.mongodb.pass
? `mongodb://${config.mongodb.user}:${config.mongodb.pass}@${config.mongodb.host}:${config.mongodb.port}/${config.mongodb.db}`
: `mongodb://${config.mongodb.host}:${config.mongodb.port}/${config.mongodb.db}`;
/**
* monk
*/
import * as mongo from 'monk';
const db = mongo(uri);
export default db;
/**
* MongoDB native module (officialy)
*/
import * as mongodb from 'mongodb';
let mdb: mongodb.Db;
const nativeDbConn = async (): Promise<mongodb.Db> => {
if (mdb) return mdb;
const db = await ((): Promise<mongodb.Db> => new Promise((resolve, reject) => {
mongodb.MongoClient.connect(uri, (e, db) => {
if (e) return reject(e);
resolve(db);
});
}))();
mdb = db;
return db;
};
export { nativeDbConn };

View file

@ -9,7 +9,7 @@ import * as cors from 'cors';
import * as mongodb from 'mongodb';
import * as gm from 'gm';
import File from '../api/models/drive-file';
import DriveFile, { getGridFSBucket } from '../api/models/drive-file';
/**
* Init app
@ -97,17 +97,28 @@ app.get('/:id', async (req, res) => {
return;
}
const file = await File.findOne({ _id: new mongodb.ObjectID(req.params.id) });
const fileId = new mongodb.ObjectID(req.params.id);
const file = await DriveFile.findOne({ _id: fileId });
if (file == null) {
res.status(404).sendFile(`${__dirname}/assets/dummy.png`);
return;
} else if (file.data == null) {
res.sendStatus(400);
return;
}
send(file.data.buffer, file.type, req, res);
const bucket = await getGridFSBucket();
const buffer = await ((id): Promise<Buffer> => new Promise((resolve, reject) => {
const chunks = [];
const readableStream = bucket.openDownloadStream(id);
readableStream.on('data', chunk => {
chunks.push(chunk);
});
readableStream.on('end', () => {
resolve(Buffer.concat(chunks));
});
}))(fileId);
send(buffer, file.metadata.type, req, res);
});
app.get('/:id/:name', async (req, res) => {
@ -117,17 +128,28 @@ app.get('/:id/:name', async (req, res) => {
return;
}
const file = await File.findOne({ _id: new mongodb.ObjectID(req.params.id) });
const fileId = new mongodb.ObjectID(req.params.id);
const file = await DriveFile.findOne({ _id: fileId });
if (file == null) {
res.status(404).sendFile(`${__dirname}/assets/dummy.png`);
return;
} else if (file.data == null) {
res.sendStatus(400);
return;
}
send(file.data.buffer, file.type, req, res);
const bucket = await getGridFSBucket();
const buffer = await ((id): Promise<Buffer> => new Promise((resolve, reject) => {
const chunks = [];
const readableStream = bucket.openDownloadStream(id);
readableStream.on('data', chunk => {
chunks.push(chunk);
});
readableStream.on('end', () => {
resolve(Buffer.concat(chunks));
});
}))(fileId);
send(buffer, file.metadata.type, req, res);
});
module.exports = app;

View file

@ -1152,9 +1152,12 @@ async function insertHimawari(opts) {
}
async function insertDriveFile(opts) {
return await db.get('drive_files').insert(Object.assign({
return await db.get('drive_files.files').insert({
length: opts.datasize,
metadata: Object.assign({
name: 'strawberry-pasta.png'
}, opts));
}, opts)
});
}
async function insertDriveFolder(opts) {

View file

@ -0,0 +1,49 @@
// for Node.js interpret
const { default: db } = require('../../built/db/mongodb')
const { default: DriveFile, getGridFSBucket } = require('../../built/api/models/drive-file')
const { Duplex } = require('stream')
const writeToGridFS = (bucket, buffer, ...rest) => new Promise((resolve, reject) => {
const writeStream = bucket.openUploadStreamWithId(...rest)
const dataStream = new Duplex()
dataStream.push(buffer)
dataStream.push(null)
writeStream.once('finish', resolve)
writeStream.on('error', reject)
dataStream.pipe(writeStream)
})
const migrateToGridFS = async (doc) => {
const id = doc._id
const buffer = doc.data.buffer
const created_at = doc.created_at
delete doc._id
delete doc.created_at
delete doc.datasize
delete doc.hash
delete doc.data
const bucket = await getGridFSBucket()
const added = await writeToGridFS(bucket, buffer, id, `${id}/${doc.name}`, { metadata: doc })
const result = await DriveFile.update(id, {
$set: {
uploadDate: created_at
}
})
return added && result.ok === 1
}
const main = async () => {
const docs = await db.get('drive_files').find()
const all = await Promise.all(docs.map(migrateToGridFS))
return all
}
main().then(console.dir).catch(console.error)