Function whitespaceCleaner

modules/tinymce/src/core/main/ts/api/html/DomParser.ts:176–283

(root: AstNode, schema: Schema, settings: DomParserSettings, args: ParserArgs)

//
// Returns [ preprocess, postprocess ]
const whitespaceCleaner = (root: AstNode, schema: Schema, settings: DomParserSettings, args: ParserArgs): [WalkerCallback, WalkerCallback] => {
  const validate = settings.validate;
  const nonEmptyElements = schema.getNonEmptyElements();
  const whitespaceElements = schema.getWhitespaceElements();
  const blockElements: Record<string, string> = extend(makeMap(extraBlockLikeElements), schema.getBlockElements());
  const textRootBlockElements = getTextRootBlockElements(schema);
  const allWhiteSpaceRegExp = /[ \t\r\n]+/g;
  const startWhiteSpaceRegExp = /^[ \t\r\n]+/;
  const endWhiteSpaceRegExp = /[ \t\r\n]+$/;

  const hasWhitespaceParent = (node: AstNode) => {
    let tempNode = node.parent;
    while (Type.isNonNullable(tempNode)) {
      if (tempNode.name in whitespaceElements) {
        return true;
      } else {
        tempNode = tempNode.parent;
      }
    }
    return false;
  };

  const isTextRootBlockEmpty = (node: AstNode) => {
    let tempNode: AstNode | null | undefined = node;
    while (Type.isNonNullable(tempNode)) {
      if (tempNode.name in textRootBlockElements) {
        return isEmpty(schema, nonEmptyElements, whitespaceElements, tempNode);
      } else {
        tempNode = tempNode.parent;
      }
    }
    return false;
  };

  const isBlock = (node: AstNode) =>
    node.name in blockElements || TransparentElements.isTransparentAstBlock(schema, node) || (Namespace.isNonHtmlElementRootName(node.name) && node.parent === root);

  const isAtEdgeOfBlock = (node: AstNode, start: boolean): boolean => {
    const neighbour = start ? node.prev : node.next;
    if (Type.isNonNullable(neighbour) || Type.isNullable(node.parent)) {
      return false;
    }

    // Make sure our parent is actually a block, and also make sure it isn't a temporary "context" element
    // that we're probably going to unwrap as soon as we insert this content into the editor
    return isBlock(node.parent) && (node.parent !== root || args.isRootContent === true);
  };

  const preprocess = (node: AstNode) => {
    if (node.type === 3) {
      // Remove leading whitespace here, so that all whitespace in nodes to the left of us has already been fixed
      if (!hasWhitespaceParent(node)) {
        let text = node.value ?? '';
        text = text.replace(allWhiteSpaceRegExp, ' ');

        if (isLineBreakNode(node.prev, isBlock) || isAtEdgeOfBlock(node, true)) {
          text = text.replace(startWhiteSpaceRegExp, '');
        }