HtmlAgilityPack selecting childNodes not as expected

Question

I am attempting to use the HtmlAgilityPack library to parse some links in a page, but I am not seeing the results I would expect from the methods. In the following I have a HtmlNodeCollection of links. For each link I want to check if there is an image node and then parse its attribures but the SelectNodes and SelectSingleNode methods of linkNode seems to be searching the parent document not the childNodes of linkNode what gives?

HtmlDocument htmldoc = new HtmlDocument();
htmldoc.LoadHtml(content);
HtmlNodeCollection linkNodes = htmldoc.DocumentNode.SelectNodes("//a[@href]");

foreach(HtmlNode linkNode in linkNodes)
{
    string linkTitle = linkNode.GetAttributeValue("title", string.Empty);
    if (linkTitle == string.Empty)
    {
        HtmlNode imageNode = linkNode.SelectSingleNode("/img[@alt]");     
    }
}

Is there any other way I could get the alt attribute of the image childnode of linkNode if it exists?

Solution 1

You should remove the forwardslash prefix from "/img[@alt]" as it signifies that you want to start at the root of the document.

HtmlNode imageNode = linkNode.SelectSingleNode("img[@alt]");

Solution 2

With an xpath query you can also use "." to indicate the search should start at the current node.

HtmlNode imageNode = linkNode.SelectSingleNode(".//img[@alt]");

Solution 3

Also, Watch out for Null Check. SelectNodes returns null instead of blank collection.

HtmlNodeCollection linkNodes = htmldoc.DocumentNode.SelectNodes("//a[@href]");

**if(linkNodes!=null)**
{
   foreach(HtmlNode linkNode in linkNodes)
  {
     string linkTitle = linkNode.GetAttributeValue("title", string.Empty);
     if (linkTitle == string.Empty)
     {
       **HtmlNode imageNode = linkNode.SelectSingleNode("img[@alt]");**   
     }
  }
}

Comments

[Sheff]

Errrm OK. That was pretty daft of me. I thought I was missing something. Sorry for wasting question space Thanks.

[binarydreams]

There's always plenty of space :)

[mpen]

Which was a really stupid design decision IMO. There's no reason it shouldn't return an empty collection.

[binarydreams]

The default axis is children to a prefix is actually not required at all.

[Matt]

see also stackoverflow.com/questions/8619724/…

[Andranik Hovhannisyan]

You the man! A sec ago i was cursing at the HtmlAgility project, but turns out they just implemented xpath the right way :)

[wal]

This didnt work for me (HtmlAgilityPack 1.4.9) - I had to use the .// notation (answer below)

[binarydreams]

@wal The syntax above assumes the target img is a direct child of linkNode. If you had to use .//, I'm guessing the img was a descendant but not a direct child.