ReactiveDB
一个 Reactive 风格的前端 ORM。基于 Lovefield 与 RxJS。
Fork from teambition/ReactiveDB
Features
-
响应式查询
支持以 Observable 的形式返回响应式数据
-
数据一致性
所有的执行过程都是事务性
的,在遇到环境异常时(indexDB 异常,浏览器限制,隐私模式导致的功能性缺失等) 也不会产生脏数据。
-
数据持久化
大量的数据场景下,极端如单页应用不间断运行几个月的情况下,不会造成内存占用量过多。所有的数据都可以持久化在本地存储而非内存中,并支持丰富的数据换页配置[WIP]。
-
debug tools
Lovefield debug tool for Chrome
Documents
Scenarios
在单页实时性应用的场景下,抽象出在前端维护数据以及其关联的数据的变更的逻辑
考虑下面的场景,在一个单页前端应用中,需要展示 A,B, C, D 四个列表:
其中列表 A 展示所有 ownerId 为 user1
的 Item :
[
{
"_id": 1,
"name": "item 1",
"ownerId": "user1",
"creatorId": "user2",
"created": "2016-01-31T16:00:00.000Z",
"owner": {
"_id": "user1",
"name": "user1 name"
},
"creator": {
"_id": "user2",
"name": "user2 name"
}
},
{
"_id": 3,
"name": "item 1",
"ownerId": "user1",
"creatorId": "user3",
"created": "2016-05-03T16:00:00.000Z",
"owner": {
"_id": "user1",
"name": "user1 name"
},
"creator": {
"_id": "user3",
"name": "user3 name"
}
}
...
]
列表 B 展示所有 creatorId 为 user2
的 Item:
[
{
"_id": 1,
"name": "item 1",
"ownerId": "user1",
"creatorId": "user2",
"created": "2016-01-31T16:00:00.000Z",
"owner": {
"_id": "user1",
"name": "user1 name"
},
"creator": {
"_id": "user2",
"name": "user2 name"
}
},
{
"_id": 2,
"name": "item 1",
"ownerId": "user2",
"creatorId": "user3",
"created": "2016-04-20T16:00:00.000Z",
"owner": {
"_id": "user2",
"name": "user2 name"
},
"creator": {
"_id": "user3",
"name": "user3 name"
}
}
...
]
列表 C 展示所有 created
时间为 2016年3月1日
以后的 Item:
[
{
"_id": 2,
"name": "item 1",
"ownerId": "user2",
"creatorId": "user3",
"created": "2016-04-20T16:00:00.000Z",
"owner": {
"_id": "user2",
"name": "user2 name"
},
"creator": {
"_id": "user3",
"name": "user3 name"
}
},
{
"_id": 3,
"name": "item 1",
"ownerId": "user1",
"creatorId": "user3",
"created": "2016-05-03T16:00:00.000Z",
"owner": {
"_id": "user1",
"name": "user1 name"
},
"creator": {
"_id": "user3",
"name": "user3 name"
}
}
]
列表 D 展示所有的用户信息:
[
{
"_id": "user1",
"name": "user1 name",
"avatarUrl": "user1 avatarUrl",
"birthday": "user1 birthday"
},
{
"_id": "user2",
"name": "user2 name",
"avatarUrl": "user2 avatarUrl",
"birthday": "user2 birthday"
},
{
"_id": "user3",
"name": "user3 name",
"avatarUrl": "user3 avatarUrl",
"birthday": "user3 birthday"
}
]
这四个列表的数据分别从四个 API 获取。在大多数单页应用的架构中,数据层会缓存这几个接口的数据,避免重复请求。而在实时性的单页应用中,这些数据的更新通常需要通过 WebSocket
等手段进行更新。根据缓存策略的不同(单例存储/同一 ID 存储多份数据),则有不同的更新方式。但这个过程一般是 业务化且难以抽象 的。
比如单一引用存储数据时, 上面场景中列举到的数据只会存储为:
{
item1, item2, item3,
user1, user2, user3
}
在这种缓存策略下,一个数据变更后,将变更后的结果通知到所属的集合是一件非常麻烦的事情。
假设现在我们的应用收到一条 socket 消息:
{
"change:item1": {
"ownerId": "user3"
}
}
按照业务需求我们应该将 item1
从 ListA
中移除。在这种缓存策略中,如果使用的 pub/sub
的模型进行通知(Backbone 之类),则会导致数据层外的代码不得不进行大量的计算,不停的 filter
一个变更是否满足某个列表的需求。这种重复的过程是非常难以维护,业务化,且难以抽象的。
而按照 ReactiveDB
的设计理念,所有的数据都有可选的响应模式,即任何与之相关的变动都会让数据自行
更新为最新的值:
伪代码如下:
database.get<ItemSchema>('Item', {
where: {
ownerId: 'user1'
},
fields: [
'_id', 'name', 'ownerId',
{
owner: ['_id', 'name']
}
]
})
.changes()
.subscribe(items => {
console.log(items)
})
使用 ReactiveDB 的情况下,无论是 Item
本身的变更还是与之关联的 User
变更,都会产生新的 items 值。
更复杂的比如 ListC
:
database.get<ItemSchema>('Item', {
where: {
created: {
'$gte': new Date(2016, 3, 1).valueOf()
}
},
fields: [
'_id', 'name', 'ownerId',
{
owner: ['_id', 'name']
}
]
})
.changes()
.subscribe(items => {
console.log(items)
})
Publish
On master
branch,
> npm version [version name]
> git push --follow-tags
The ./.circleci/config.yml script should take care of the following jobs:
- build ReactiveDB/core packages;
- verify that all tests are passing;
- given that the current commit is a release commit, publish the built packages to NPM repository.
Done :)
If a release is to be published from a branch other than master
, please make sure to version it as an alpha or a beta; v0.9.15-alpha.1-description
for example. And you will have to build (npm run build_all
) and publish (npm run publish_all
) from local.