Small script to detect caret pixel position in contenteditable div

caret.js

    /**
     * Get the caret position in all cases
     *
     * @returns {object} left, top distance in pixels
     */
    getCaretTopPoint () {
      const sel = document.getSelection()
      const r = sel.getRangeAt(0)
      let rect
      let r2
      // supposed to be textNode in most cases
      // but div[contenteditable] when empty
      const node = r.startContainer
      const offset = r.startOffset
      if (offset > 0) {
        // new range, don't influence DOM state
        r2 = document.createRange()
        r2.setStart(node, (offset - 1))
        r2.setEnd(node, offset)
        // https://developer.mozilla.org/en-US/docs/Web/API/range.getBoundingClientRect
        // IE9, Safari?(but look good in Safari 8)
        rect = r2.getBoundingClientRect()
        return { left: rect.right, top: rect.top }
      } else if (offset < node.length) {
        r2 = document.createRange()
        // similar but select next on letter
        r2.setStart(node, offset)
        r2.setEnd(node, (offset + 1))
        rect = r2.getBoundingClientRect()
        return { left: rect.left, top: rect.top }
      } else { // textNode has length
        // https://developer.mozilla.org/en-US/docs/Web/API/Element.getBoundingClientRect
        rect = node.getBoundingClientRect()
        const styles = getComputedStyle(node)
        const lineHeight = parseInt(styles.lineHeight)
        const fontSize = parseInt(styles.fontSize)
        // roughly half the whitespace... but not exactly
        const delta = (lineHeight - fontSize) / 2
        return { left: rect.left, top: (rect.top + delta) }
      }
    }