I have a text field represented as: field = {text: "", valid: false}, and an input with [(ngModel)]="field.text".
I want to make that field only accept a defined set of characters (for this issue, numbers), and doing (keypress) doesn't work on mobile, so I did: (ngModelChange)="fieldChanged(field)"
The method does the following:
fieldChanged(field) {
console.log(field.text);
field.text = Array.from(field.text).filter((char:string) => "0123456789".indexOf(char) != -1).join("");
console.log(field.text);
}
And it's behaviour is extremely weird.
Legend:
- input: what key was pressed
- before update: first console.log
- after update: second console.log
- output: what I see on screen in the input
| input | before update | after update | output |
|---------|---------------|--------------|--------|
| "" | "" | "" | "" | <- starting position, no event
| "a" | "a" | "" | "a" |
| "a" | "aa" | "" | "aa" |
| "4" | "aa4" | "4" | "4" |
| "a" | "4a" | "4" | "4a" |
| "a" | "4aa" | "4" | "4aa" |
| "4" | "4aa4" | "44" | "44" |
Why does it always update the output when I enter a legal character? It should be working for each event call.
Edit: Plunker
I think the cause is that modifying the value on ngModelChange breaks change detection, for example if you change the value back to the previous value, because an invalid character was added.
A workaround:
constructor(private cdRef:ChangeDetectorRef) {}
fieldChanged(field) {
console.log(field.text);
field.text = Array.from(field.text).filter((char:string) => "0123456789".indexOf(char) != -1).join("");
console.log(field.text);
var tmp = field.text;
field.text = null; // or some other value that normally won't ever be in `field.text`
this.cdRef.detectChanges();
field.text = tmp;
this.cdRef.detectChanges(); // I guess this 2nd call won't be necessary
}
If anyone having the issues in updating the value, use setTimeout function while updating
// setTimeout function
setTimeout(() => {
field.text = temp;
this.cdRef.detectChanges();
}, 1);