parse html

parse.js

var startTagReg = /^<([^\s!/>]+)[^>]*[^/]?>/;
var endTagReg = /^<\/([^>\s]+)>/;
var selfCloseTagReg = /^<([^\s>!/]+)[^>]*\/>/;
var textContentTagReg = /^[^<]+/;
var commentTagReg = /^<!--[\s\S]*?-->/;
var selfCloseTagsNameReg = /^img|link|meta|br|hr|base$/i;

function parseHTML(html) {
  var rootChildren = [];
  var stack = [{ children: rootChildren }];
  var pendingContent = html;
  var match = undefined;
  var cursorPosition = 0;

  while (pendingContent.length) {
    var closetTag = stack[stack.length - 1];
    var source;

    if ((match = pendingContent.match(startTagReg))) {
      appendElementNode(match[1], match[0], isSelfCloseTag(match[1]));
    } else if ((match = pendingContent.match(endTagReg))) {
      var source = match[0];

      if (closetTag.name === match[1]) {
        closetTag.endAt = closetTag.endAt + source.length;
        closetTag = stack.pop();
      } else {
        appendTextContentNode(source);
      }
    } else if ((match = pendingContent.match(selfCloseTagReg))) {
      appendElementNode(match[1], match[0], true);
    } else if ((match = pendingContent.match(commentTagReg))) {
      appendCommentNode(match[0]);
    } else if ((match = pendingContent.match(textContentTagReg))) {
      appendTextContentNode(match[0]);
    } else if (pendingContent.charAt(0) === "<") {
      match = ["<"];
      appendTextContentNode(match[0]);
    }

    if (match) {
      source = match[0];
      cursorPosition += source.length;
      pendingContent = pendingContent.slice(source.length);
      closetTag.endAt = cursorPosition + source.length;
    }
  }

  return rootChildren;

  function appendElementNode(name, source, closeSelf) {
    let node = {
      name: name,
      type: "element",
      source: source,
      startAt: cursorPosition,
      endAt: cursorPosition + source.length
    };

    if (closeSelf) {
      node.closeSelf = closeSelf;
    } else {
      node.children = [];
      stack.push(node);
    }

    closetTag.children.push(node);
  }

  function appendCommentNode(source) {
    closetTag.children.push({
      type: "comment",
      source: source,
      startAt: cursorPosition,
      endAt: cursorPosition + source.length
    });
  }

  function appendTextContentNode(source) {
    let node = closetTag.children[closetTag.children.length - 1];

    if (!node || node.type !== "textContent") {
      node = {
        type: "textContent",
        source: source,
        startAt: cursorPosition,
        endAt: cursorPosition + source.length
      };

      closetTag.children.push(node);
    } else {
      node.source += source;
      node.endAt += source.length;
    }
  }
}

function isSelfCloseTag(tagName) {
  return selfCloseTagsNameReg.test(tagName);
}

666