WebContents: executeJavaScript() method
The executeJavaScript() method runs code in the web page and brings the result back to the app.
Syntax
executeJavaScript(code)Parameters
code
The code to run in the web page, a string. It runs in the page's environment: it can use document and window, but not the app's variables.
The code comes in two forms, with different results:
- A single expression, such as
'document.title'or'fetch("/api").then((r) => r.status)'. A trailing semicolon doesn't change this;'foo();'also counts as an expression. - Multiple statements, such as
'var a = 1; a + 1'. Code starting with{(a block),function, orclassis also treated as multiple statements; names declared byfunctionandclassgo into the page's global scope.
Return value
Promise<any>.
| Case | Result |
|---|---|
| The code is a single expression | Fulfills with its value. If the value is a Promise, waits for it; if that is rejected, this one is rejected too |
| The code is multiple statements | Fulfills with the value of the last expression, without waiting for Promises. If the value is an object, member order isn't guaranteed |
| The page throws, or the code has a syntax error | Rejects with an Error whose message is the page's original text, such as Error: boom or SyntaxError: … |
| The result can't be passed back (a function, a DOM node, a Promise produced by multiple statements) | Rejects with TypeError: executeJavaScript: the result cannot be passed back; it is a function, a DOM node, or a Promise produced by multi-statement code |
| The page has been destroyed | Rejects with TypeError: Object has been destroyed |
The result is passed back as JSON: dates become strings, NaN becomes null, Map becomes {}. For all the rules, see How values are passed.
Exceptions
Doesn't throw directly. All errors reject the returned Promise.
Description
- It works even before any page has loaded; it then runs in a blank document.
- It works as usual when the page forbids scripts (Content Security Policy).
- The result comes from the page and is untrusted input; check its type and range before using it.
Examples
Reading content from the page
const contents = new WebView().webContents;
await contents.loadURL('https://example.com/');
const title = await contents.executeJavaScript('document.title');
const count = await contents.executeJavaScript('document.querySelectorAll("a").length');
console.log(title, count);Result: It prints the page's title and the number of links.
Waiting for a Promise in multiple statements
Multiple statements don't wait for Promises. To use await inside, wrap the code in an immediately invoked async function; the whole thing then becomes a single expression:
const contents = new WebView().webContents;
await contents.loadURL('https://example.com/');
const text = await contents.executeJavaScript(`(async () => {
const response = await fetch('/api/user');
return (await response.json()).name;
})()`);Result: text is the name the API returns.
The page throws an exception
const contents = new WebView().webContents;
contents.executeJavaScript('throw new Error("boom")').catch((err) => console.log(err.message));Result: It prints Error: boom.
Differences from Electron
Same as Electron's contents.executeJavaScript(code), with three differences: code with multiple statements doesn't wait for Promises; objects returned from multiple statements don't guarantee member order; results are passed as JSON (dates are strings, Map is {}). Electron's second argument, userGesture, isn't supported.
The native iOS equivalents are [webView evaluateJavaScript:completionHandler:] and callAsyncJavaScript:….