[Feature Request] Return All Nodes from XPath

[alixaxel]

Right now, the current page.xpath() implementation only returns the first matched element.

I find this implementation somewhat restrictive, specially considering it's tied to ElementHandle.

For instance, if we wish to find all ANY_* XPath result type constants on the MDN website:

id("wikiArticle")/table[@class="standard-table"]//td[1]/code[starts-with(., 'ANY_')]

The above XPath would do the job, but with the current implementation we are forced to either traverse the DOM using good old page.evaluate() or known how many elements would be matched beforehand and issue n separate XPath queries in the form of (expression)[1..n]. Or just run XPath in page.evaluate:

await page.evaluate(() => {
  function $x(expression) {
    let result = [];
    let element = null;
    let elements = document.evaluate(expression, document, null, XPathResult.ANY_TYPE, null);

    while ((element = elements.iterateNext()) !== null) {
      result.push(element);
    }

    return result;
  }

  return $x('id("wikiArticle")/table[@class="standard-table"]//td[1]/code[starts-with(., "ANY_")]').map(
    (element) => {
      return element.textContent;
    }
  );
});

// ["ANY_TYPE", "ANY_UNORDERED_NODE_TYPE"]

It would be great to have more capable XPath querying abilities before PPTR 1.0.0 is released.

[aslushnikov]

@JoelEinbinder how about returning array of all matched elements from page.xpath?

[yujiosaka]

FYI

selenium-webdriver has both getAttributeByXpath and getAttributesByXpath.
the former returns a single element and the latter returns all.