HtmlAgilityPack and selecting Nodes and Subnodes

问题:

Hope somebody can help me.

Let´s say I have a html document that contains multiple divs like this example:

<div class="search_hit">

    <span prop="name">Richard Winchester</span>
    <span prop="company">Kodak</span>
    <span prop="street">Arlington Road 1</span>

</div>
<div class="search_hit">

    <span prop="name">Ted Mosby</span>
    <span prop="company">HP</span>
    <span prop="street">Arlington Road 2</span>

</div>

I´m using HtmlAgilityPack to get the html document. What i need to know is how can i get the spans for each "search_hit"-div?

My first thought was something like this:

foreach (HtmlAgilityPack.HtmlNode node in doc.DocumentNode.SelectNodes("//div[@class='search_hit']"))
{
     foreach (HtmlAgilityPack.HtmlNode node2 in node.SelectNodes("//span[@prop]"))
     {

     }
}

Each div should be a object with the included spans as properties. I. e.

public class Record
    {
        public string Name { get; set; }
        public string company { get; set; }
        public string street { get; set; }
    }

And this List shall be filled then:

public List<Record> Results = new List<Record>();

But the XPATH i´m using is not doing a search in the subnode as it should do. It seams that it searches the whole document again and again.

I mean I already got it working in that way that i just get the the spans of the whole page. But then i have no relation between the spans and divs. Means: I don´t know anymore which span is related to which div.

Does somebody know a solution? I already played around that much that i´m totally confused now :)

Any help is appreciated!

回答1:

The following works for me. The important bit is just as BeniBela noted to add a dot in second call to 'SelectNodes'.

List<Record> lstRecords=new List<Record>();
foreach (HtmlNode node in doc.DocumentNode.SelectNodes("//div[@class='search_hit']"))
{
  Record record=new Record();
  foreach (HtmlNode node2 in node.SelectNodes(".//span[@prop]"))
  {
    string attributeValue = node2.GetAttributeValue("prop", "");
    if (attributeValue == "name")
    {
      record.Name = node2.InnerText;
    }
    else if (attributeValue == "company")
    {
      record.company = node2.InnerText;
    }
    else if (attributeValue == "street")
    {
      record.street = node2.InnerText;
    }
  }
  lstRecords.Add(record);
}

回答2:

If you use //, it searches from the document begin.

Use .// to search all from the current node

 foreach (HtmlAgilityPack.HtmlNode node2 in node.SelectNodes(".//span[@prop]"))

Or drop the prefix entirely to search just for direct children:

 foreach (HtmlAgilityPack.HtmlNode node2 in node.SelectNodes("span[@prop]"))

回答3:

First of all, take a look at this: Html Agility Pack - Problem selecting subnode

Here is a full working solution for your question:

IList<Record> results = new List<Record>();
foreach (var node in doc.DocumentNode.SelectNodes("//div[@class='search_hit']")) {
    var record = new Record();
    record.Name = node.SelectSingleNode(".//span[@prop='name']").InnerText;
    record.company = node.SelectSingleNode(".//span[@prop='company']").InnerText;
    record.street = node.SelectSingleNode(".//span[@prop='street']").InnerText;
    results.Add(record);
}

If you read the question I pointed you to, you will see that doing ./span[@prop='name'] is exactly the same, since those span nodes are (direct) children of the div node.

---

If the span nodes do not have those prop attributes, and you want to assign them depending on the order they appear, you can do:

foreach (var node in doc.DocumentNode.SelectNodes("//div[@class='search_hit']")) {
    var spanNodes = node.SelectNodes("./span");
    var record = new Record();
    record.Name = spanNodes[0].InnerText;
    record.company = spanNodes[1].InnerText;
    record.street = spanNodes[2].InnerText;
    results.Add(record);
}

回答4:

Shame on me :)

All of you were right.

I found the problem. This NullReferenceException kept nagging me so I spent more time to look at it in detail. In between all those divs there was one div with the same "class='search-hit'" attribute but without the spans inside. Thats why it throughs an error at the second loop.

foreach (HtmlAgilityPack.HtmlNode node in doc.DocumentNode.SelectNodes("//span[@prop]/ancestor::div[@class='search_hit']"))
   {
        Record rec = new Record();
        foreach (HtmlAgilityPack.HtmlNode node2 in node.SelectNodes(".//span[@prop]"))
           {
           }
           rList.Results.Add(rec);
   }

The code above is working.

Thank you guys for your time and help!

回答5:

I used that. class convert id

  HtmlNodeCollection nodes = dokuman.DocumentNode.SelectNodes("//div[@id='search_hit']//span[@prop]");


            for (int i = 0; i < nodes .Count; i++)
        {
            var record = new Record();


                record.Name = links[i].InnerText;   results.Add(record);  }