Using getWordUnderCursor to get words that contain apostrophes?

I'm using monaco-editor in an Electron application where users will be typing poetry. The language set for the editor is plaintext.

I'm running into an issue where if a user has typed a word with an ' (apostrophe) in it, getWordAtPosition won't return the full word. Depending on the cursor's position (which is what I pass into getWordAtPosition), it'll either return what's before the apostrophe, or what's after it.

For example, using a | (pipe) to represent the cursor's location, if the user has typed:

couldn't|

getWordAtPosition will return t

For |couldn't, it will return couldn.

What I want is for, regardless of cursor position, it will return couldn't.

What I've tried so far is supplying wordSeparators in the options when creating the editor that doesn't include the apostrophe (but is otherwise the default), but that hasn't helped.

Editor creation block:

editorInstance = monaco.editor.create(document.getElementById('editor'), {
            language: 'plaintext',
            wordSeparators: '`~!@#$%^&*()-=+[{]}\\|;:",.<>/?'
        });

Block using the getWordAtPosition:

fromEventPattern((handler: NodeEventHandler) => editorInstance.onDidChangeCursorPosition(handler))
            .pipe(
                map(() => {
                    const cursorPosition: IPosition = editorInstance.getPosition();

                    return editorInstance.getModel()
                        .getWordAtPosition(cursorPosition)
                        ?.word;
                }),
                filter((value: string) => !!value)
            );

Any help would be greatly appreciated! Thanks in advance!

You can set a language configuration (either for a new language, or an existing one) and provide a wordPattern regex that will be used to match against words.

monaco.languages.setLanguageConfiguration('plaintext', {
    wordPattern: /'?\w[\w'-.]*[?!,;:"]*/
});


monaco.editor.create(document.getElementById("container"), {
	language: 'plaintext',
});

Even though plaintext already exists as a language, this still works. I'm not sure if setLanguageConfiguration does any merging (I'd assume not?) so keep that in mind.

The regex in my code snippet isn't perfect, but it matches words like:

• Couldn't
• 'cause
• you?
• Hello

which was important for my use-case. Keep that in mind if you're a future searcher looking at this answer!