A directive for Angular framework to provide unlimited bidirectional virtual scrolling over limited viewport. Built on top of vscroll native virtual scrolling engine. Demo is available at dhilt.github.io/ngx-ui-scroll.
can donate? go here 👉
make open-source world better
The ngx-ui-scroll library has no breaking changes in its API, but there are inevitable changes in how it is built and distributed to the host app depending on the version of the Angular.
| ngx-ui-scroll | Angular | compiled | support | notes |
|---|---|---|---|---|
| v1 | 5-12 | View Engine | no | no dependencies (vscroll is not extracted) |
| v2 | 5-12 | View Engine | maintenance | vscroll is a bundle-dependency |
| v3 | 12+ | Ivy | active | vscroll is a peer-dependency |
So if the consumer app is view-engine compatible, you should use ngx-ui-scroll v2 which is in maintenance mode and under v2-legacy branch.
Scrolling large datasets may cause performance issues. Many DOM elements, many data-bindings, many event listeners... The common way to improve the performance is to render only a small portion of the dataset visible to a user. Other dataset elements that are not visible to a user are virtualized with upward and downward empty padding elements which should provide a consistent viewport with consistent scrollbar parameters.
The ngx-ui-scroll library provides the *uiScroll structural directive that works like *ngFor and renders a templated element once per item from a collection. By requesting the external Datasource (the implementation of which is a developer responsibility) the *uiScroll directive fetches necessary portion of the dataset and renders corresponded elements until the visible part of the viewport is filled out. It starts to retrieve new data to render new elements again if a user scrolls to the edge of visible element list. It dynamically destroys elements as they become invisible and recreates them if they become visible again.
- unlimited bidirectional virtual scroll
- lots of virtualization settings
- super easy templating
- infinite mode, demo
- horizontal mode, demo
- entire window scrollable, demo
- items with non-constant heights, demo
- API Adapter object to manipulate and assess the Scroller, demos
The *uiScroll directive is a part of UiScrollModule which is available via npm –
npm install ngx-ui-scroll
The UiScrollModule has to be imported in the App/feature module where it is going to be used.
import { NgModule } from '@angular/core';
import { BrowserModule } from '@angular/platform-browser';
import { UiScrollModule } from 'ngx-ui-scroll';
import { AppComponent } from './app.component';
@NgModule({
declarations: [AppComponent],
imports: [
BrowserModule,
UiScrollModule
],
bootstrap: [AppComponent]
})
export class AppModule { }Basic usage template may look like
<div class="viewport">
<div *uiScroll="let item of datasource">
<b>{{item.text}}</b>
</div>
</div>where the viewport is a scrollable area of finite height:
.viewport {
height: 300px;
overflow-y: auto;
}If the height of the viewport is not constrained, it will pull the entire content of the datasource and no scrollbar will appear.
*uiScroll acts like *ngFor in its simplest form, where the datasource is an object of special type (IDatasource), which implements the get method used by the *uiScroll directive to access data by index and count parameters. The directive calls the Datasource.get method each time the user scrolls the list of visible elements to the edge.
import { IDatasource } from 'ngx-ui-scroll';
@Component({ ... })
export class AppComponent {
datasource: IDatasource = {
get: (index, count, success) => {
const data = [];
for (let i = index; i <= index + count - 1; i++) {
data.push({ text: 'item #' + i });
}
success(data);
}
};
}Datasource.get method must provide an array of count data-items started from index position. If there are no items within given range [index; index + count - 1], an empty array has to be passed. Empty result (or result which length is less than count) is being treated as the edge of the dataset (eof/bof), and no further requests for preceding/tailing items will be issued.
Datasource.get has 3 signatures: callback based, Promise based and Observable based. So, if we want some remote API to be a source of our data, basically it may look like
datasource: IDatasource = {
get: (index, count) =>
this.http.get(`${myApiUrl}?index=${index}&count=${count}`)
};More details could be found on the Datasource demo page.
Datasource implementation along with get method property may include settings object property:
datasource: IDatasource = {
get: ...,
settings: {
minIndex: 0,
startIndex: 0,
...
}
};Settings are being applied during the Scroller initialization and have an impact on how the Scroller behaves. Below is the list of available settings with descriptions, defaults, types and demos.
| Name | Type | Default | Description |
|---|---|---|---|
| bufferSize | number, integer |
5 | Fixes minimal size of the pack of the datasource items to be requested per single Datasource.get call. Can't be less than 1. |
| padding | number, float |
0.5 | Determines the viewport outlets containing real but not visible items. The value is relative to the viewport's size. For example, 0.25 means that there will be as many items at a moment as needed to fill out 100% of the visible part of the viewport, + 25% of the viewport size in the backward direction and + 25% in the forward direction. The value can't be less than 0.01. |
| startIndex | number, integer |
1 | Specifies item index to be requested/rendered first. Can be any, but the real datasource boundaries should be taken into account. |
| minIndex | number, integer |
-Infinity | Fixes absolute minimal index of the dataset. The datasource left boundary. |
| maxIndex | number, integer |
+Infinity | Fixes absolute maximal index of the dataset. The datasource right boundary. |
| infinite | boolean | false | Enables "infinite" mode, when items rendered once are never removed. |
| horizontal | boolean | false | Enables "horizontal" mode, when the viewport's orientation is horizontal. |
| sizeStrategy | string enum, 'average' | 'frequent' | 'constant' | 'average' | Defines how the default item size is calculated. If item has never been rendered, its size is assumed to be the default size: an average or most frequent among all items that have been rendered before, or constant. This has an impact on the process of virtualization. |
| windowViewport | boolean | false | Enables "entire window scrollable" mode, when the entire window becomes the scrollable viewport. |
The Scroller has API to assess its parameters and provide some manipulations at runtime. This API is available via special Adapter object. The datasource needs to be instantiated via operator "new" for the Adapter object to be added to it:
import { Datasource } from 'ngx-ui-scroll';
...
datasource = new Datasource({
get: ... ,
settings: { ... }
});Then this.datasource.adapter.packageInfo, this.datasource.adapter.reload() and other Adapter expressions become available. For better typing, it is recommended to specify the Datasource Item type as follows:
import { Datasource } from 'ngx-ui-scroll';
...
datasource = new Datasource<MyItem>({
get: ... ,
settings: { ... }
});MyItem should reflect the structure of items that the Datasource will deal with. It is "unknown" by default, and if not set, for example, this.datasource.adapter.firstVisible.data.id expression will produce typescript error: Object is of type 'unknown'. There are some Adapter props and methods dealing with MyItem, and if used, MyItem should be specified.
Below is the list of read-only properties of the Adapter API with descriptions and links to demos.
| Name | Type | Description |
|---|---|---|
| packageInfo | IPackages { consumer: IPackage; core: IPackage; } IPackage { name: string; version: string } |
Information about versions of the library ant its core. For example: "ngx-ui-scroll" v3.1.0 (consumer), "vscroll" v1.5.5 (core). |
| init | boolean | Indicates whether the Scroller is initialized ot not. |
| init$ | Subject<boolean> | An Observable version of "init" property. |
| isLoading | boolean | Indicates whether the Scroller is working ot not. |
| isLoading$ | Subject<boolean> | An Observable version of "isLoading" property. |
| itemsCount | number | A number of items that are rendered in the viewport at a moment. |
| bufferInfo | IBufferInfo { firstIndex: number; lastIndex: number; minIndex: number; maxIndex: number; absMinIndex: number; absMaxIndex: number; } |
|
| bof | boolean | Indicates whether the beginning of the dataset is reached or not. |
| bof$ | Subject<boolean> | An Observable version of "bof" property. |
| eof | boolean | Indicates whether the end of the dataset is reached or not. |
| eof$ | Subject<boolean> | An Observable version of "eof" property. |
| firstVisible | ItemAdapter<MyItem> { $index: number; data: MyItem; element?: HTMLElement; } |
Object of ItemAdapter type containing information about first visible item, where "$index" corresponds to the Datasource item index value, "data" is exactly the item's content passed via Datasource, "element" is a link to DOM element which is relevant to the item. |
| firstVisible$ | BehaviorSubject <ItemAdapter<MyItem>> |
An observable version of "firstVisible" property. |
| lastVisible | ItemAdapter<MyItem> { $index: number; data: MyItem; element?: HTMLElement; } |
Object of ItemAdapter type containing information about last visible item. |
| lastVisible$ | BehaviorSubject <ItemAdapter<MyItem>> |
An observable version of "lastVisible" property. |
| paused | boolean | Indicates whether the Scroller is paused ot not. |
| paused$ | Subject<boolean> | An Observable version of "paused" property. |
Below is the list of invocable methods of the Adapter API with description and links to demos.
| Name | Parameters | Description |
|---|---|---|
| relax | (callback?: Function) | Resolves asynchronously when there are no pending processes. If the callback is set, it will be executed right before resolving. Basically, it needs to protect with the relax every piece of the App logic, that might be sensitive to the internal processes of the Scroller, to avoid interference and race conditions. Also, every Adapter method returns the same promise as the relax method, so for example, explicitly waiting for the async result of the append method is equivalent to waiting for the async result of the relax method right after synchronous calling the append method. |
| reload | (startIndex?: number) | Resets the items buffer, resets the viewport params and starts fetching items from startIndex (if set). |
| reset | (datasource?: IDatasource) | Performs hard reset of the Scroller's internal state by re-instantiating all its entities (instead of reusing them when reload). If datasource argument is passed, it will be treated as new Datasource. All props of the datasource are optional and the result Datasource will be a combination (merge) of the original one and the one passed as an argument. |
| pause | Pauses the Scroller, so the scroll events are not processed by the engine. Also, when paused, the Adapter methods do nothing but resolve immediately without affecting UI (except Adapter.resume and Adapter.reset). | |
| resume | Resumes the Scroller if it was paused. | |
| check | Checks if any of current items changed it's size and runs a procedure to provide internal consistency and new items fetching if needed. | |
| clip | (options: { forwardOnly?: boolean, backwardOnly?: boolean }) |
Removes out-of-viewport items on demand. The direction in which invisible items should be clipped can be specified by passing an options object. If no options is passed (or both properties are set to true), clipping will occur in both directions. |
| append | (options: { items: MyItem[], eof?: boolean decrease?: boolean }) |
Adds items to the end of the Scroller's buffer/dataset. If eof flag is not set, items will be inserted right after the last buffered item and rendered immediately. If eof is true, rendering will occur only if the right border of the buffer matches the right border of the dataset (end-of-file is reached); otherwise, items will be virtualized as appended to the end of the dataset. Indexes increase by default. If decrease is set to true, indexes are decremented. See also bof/eof. |
| prepend | (options: { items: MyItem[], bof?: boolean increase?: boolean }) |
Adds items to the beginning of the Scroller's buffer/dataset. If bof flag is not set, items will be inserted right before the first buffered item and rendered immediately. If bof is true, rendering will occur only if the left border of the buffer matches the left border of the dataset (begin-of-file is reached); otherwise, items will be virtualized as prepended to the beginning of the dataset. Indexes decrease by default. If increase is set to true, indexes are incremented. Note, by historical reasons, items are being reversed during prepending, so if you need to have "initial" order, you may reverse the items array before prepend, or use Adapter.insert API instead. See also bof/eof. |
| insert | (options: { items: MyItem[], before?: ItemsPredicate, after?: ItemsPredicate, beforeIndex?: number, afterIndex?: number, decrease?: boolean }) |
Inserts items into the buffer or virtually. Only one of the before, after, beforeIndex and afterIndex options is allowed. If before or after option is used, the Scroller will try to insert items before or after the item that presents in the buffer and satisfies the predicate condition. If beforeIndex or afterIndex option is used, the Scroller will try to insert items by index. If the index to insert is out of the buffer but still belongs to the known datasource boundaries, then the items will be virtualized. Indexes increase by default. Decreasing strategy can be enabled via decrease option. |
| remove | (options: { predicate?: ItemsPredicate, indexes?: number[], increase?: boolean }) type ItemsPredicate = (item: ItemAdapter) => boolean |
Removes items form buffer and/or virtually. Predicate is a function to be applied to every item presently in the buffer. Predicate must return a boolean value. If predicate's return value is true, the item will be removed. Alternatively, if indexes array is passed, the items whose indexes match the list will be removed. Only one of the predicate and indexes options is allowed. In case of indexes, the deletion is performed also virtually. By default, indexes of the items following the deleted ones are decremented. Instead, if increase is set to true, the indexes of the items before the removed ones are incremented. |
| replace | (options: { predicate: ItemsPredicate, items: MyItem[], fixRight?: boolean }) |
Replaces items that continuously match the predicate with an array of new items. Indexes are maintained on the assumption that the left border of the dataset is fixed. To release the left border and fix the right one the fixRight option should be set to true. |
| update | (options: { predicate: BufferUpdater, fixRight?: boolean }) type BufferUpdater = (item: ItemAdapter) => unknown |
Updates existing items by running the predicate function over the Scroller's buffer. The return value of the predicate determines the operation: falsy or empty array to remove, truthy or array with only 1 current item to keep unchanged, non-empty array to replace/insert. Indexes are maintained on the assumption that the left border of the dataset is fixed. To release the left border and fix the right one the fixRight option should be set to true. |
Along with the documented API there are some undocumented features that can be treated as experimental. They are not tested enough and might change over time. Some of them can be found on the experimental tab of the demo app.
All of the Adapter methods are asynchronous, because they work with DOM and may take time to complete. For this purpose the Adapter methods return a Promise resolving at the moment when the Scroller terminates its internal processes, which were triggered by the invocation of the correspondent Adapter method. This is called the Adapter Return API. This promise has exactly the same nature as the promise of the Adapter.relax method. Both "Relax" and "Return API" are the instruments of the App-Scroller processes normalization. It might be quite important to run some logic after the Scroller finishes its job and relaxes. Below is an example of how an explicit sequence of the Adapter methods can be safely implemented:
const { adapter } = this.datasource;
const predicate = ({ $index }) => $index === indexToReplace;
await adapter.relax();
await adapter.remove({ predicate });
await adapter.insert({ items: [itemToReplace], before: predicate });
console.log('Two-phase replacement done');For more information, see Adapter demo page.
There are some npm scripts available from package.json:
npm startto run demo App on port 4200npm testto run Karma testsnpm run build:libto build the ngx-ui-scroll module into the ./dist/scroller foldernpm run build:demoto build the demo App into the ./dist/demo folder
Along with settings object the datasource implementation may include also devSettings object:
import { Datasource } from 'ngx-ui-scroll';
...
datasource = new Datasource({
get: ... ,
settings: { ... },
devSettings: {
debug: true,
immediateLog: true,
...
}
});The development settings are not documented. Information about it can be taken directly from the source code. The Scroller has "debug" mode with powerful logging which can be enabled via devSettings.debug = true, see Dev Log doc.
Below is the quick guide for vscroll/ngx-ui-scroll integrated development (this is relevant since ngx-ui-scroll v2 depends on vscroll):
- clone both ngx-ui-scroll and vscroll repositories into the same folder
- replace "vscroll" import with local sources here and here.
Also, there are some environment variables for additional customization of the dev process. In accordance with dotenv approach, the .env file should be placed in the root folder, and it may contain the following variables.
| Name | Value | Description |
|---|---|---|
| DEV_SERVER_PORT | 4200 | Port the dev server (webpack) will use. Need to run npm run start:env instead of npm run to make this setting work. |
| TEST_BROWSER | default | chrome | firefox | Platform for running tests. By default a headless chrome is used; "chrome" or "firefox" are for running tests in real (chrome/ff) browser |
| TEST_SERVER_PORT | 9876 | Port that will be used by non-default testing browser |
Any support and participation are welcome, so feel free to donate, submit new Issues and open Pull Requests.
2024 © dhilt