I am developing Hackernator - a Hacker News client for Windows Phone. One of the challenges I faced at the start of the project was the fact that I needed to somehow get Hacker News data. There are two ways to do it. First, I could use the RSS feed. A pretty good solution, but it only gave me the ability to read the titles of the links posted. I wanted to get the number of comments and points as well. Add to that the domain and the author of the link. The second potential way would be using the iHackerNews API. It would be really easy to handle, but the API itself sometimes experiences availability issues.

So I decided to go the hard way and actually read the page manually, by going through the HTML content. It ended up being a much more interesting task than I thought. This is what I had to work with:

Classic work with tables. Do you see anything some elements have in common? You get bonus points if you looked at td class="title" and td class="subtext". Those are the primary indicators for areas in HTML that I should pay attention for. The first sector is the link title, the second one is connected to metadata. For a single link, I came up with this model implementation:

namespace Hackernator.Models
{
    public class Link
    {
        public string Title { get; set; }
        public string ID { get; set; }
        public string Domain { get; set; }
        public string Url { get; set; }
        public int Points { get; set; }
        public string Author { get; set; }
        public int Comments { get; set; }
    }
}

Every single property here is self-explanatory, other than ID. That is the unique identifier assigned to every link on HN. For example, Gigabit Internet for $80 has the ID value of 3637599. By using the ID, I am able to read comment data for a given link directly on Hacker News instead of going to the URL.

Here is the class that parses the content:

using System;
using System.Collections.Generic;
using Hackernator.Models;

namespace Hackernator.HNAPI
{
    public class HackerNewsParser
    {
        public IEnumerable<Link> GetLinksFromRaw(string content)
        {
            List<Link> links = new List<Link>();
            string titleFlag = "<td class=\"title\">";
            string closingFlag = "</td>";
            string dataToManipulate = content;
            int titleLocation = dataToManipulate.IndexOf(titleFlag);
            while (titleLocation != -1)
            {
                Link hLink = new Link();
                dataToManipulate = dataToManipulate.Remove(0, titleLocation);
                int closingIndex = dataToManipulate.IndexOf(closingFlag);
                string subParsedItem = dataToManipulate.Substring(0, closingIndex + closingFlag.Length);
                dataToManipulate = dataToManipulate.Replace(subParsedItem, string.Empty);
                int index = subParsedItem.IndexOf("href");
                if (index != -1)
                {
                    subParsedItem = subParsedItem.Remove(0, index);
                    index = subParsedItem.IndexOf("\">");
                    hLink.Url = subParsedItem.Substring(6, index - 6);
                    subParsedItem = subParsedItem.Remove(0, index + 2);
                    index = subParsedItem.IndexOf("</");
                    hLink.Title = subParsedItem.Substring(0, index);
                    if (hLink.Title != "More")
                    {
                        subParsedItem = subParsedItem.Remove(0, index);
                        index = subParsedItem.IndexOf("(");
                        if (index != -1)
                            hLink.Domain = subParsedItem.Substring(index, subParsedItem.IndexOf(")") - index + 1);
                        else
                            hLink.Domain = "HackerNews";
                        index = dataToManipulate.IndexOf("<td class=\"subtext\">");
                        if (index != -1)
                        {
                            dataToManipulate = dataToManipulate.Remove(0, index);
                            index = dataToManipulate.IndexOf(closingFlag);
                            subParsedItem = dataToManipulate.Substring(0, index + closingFlag.Length);
                            dataToManipulate = dataToManipulate.Replace(subParsedItem, string.Empty);
                            index = subParsedItem.IndexOf("score_");
                            if (index != -1)
                            {
                                subParsedItem = subParsedItem.Remove(0, index);
                                index = subParsedItem.IndexOf(">");
                                hLink.ID = subParsedItem.Substring(6, index - 6);
                                subParsedItem = subParsedItem.Remove(0, index + 1);
                            }
                            index = subParsedItem.IndexOf("points");
                            if (index != -1)
                            {
                                hLink.Points = Convert.ToInt32(subParsedItem.Substring(0, index));
                                index = subParsedItem.IndexOf("\">");
                                subParsedItem = subParsedItem.Remove(0, index + 2);
                                index = subParsedItem.IndexOf("<");
                                hLink.Author = subParsedItem.Substring(0, index);
                                subParsedItem = subParsedItem.Remove(0, index);
                                index = subParsedItem.IndexOf("\">");
                                subParsedItem = subParsedItem.Remove(0, index + 2);
                            }
                            else
                                hLink.Points = 0;
                            index = subParsedItem.IndexOf("comments");
                            if (index != -1)
                                hLink.Comments = Convert.ToInt32(subParsedItem.Substring(0, index));
                            else
                                hLink.Comments = 0;
                            links.Add(hLink);
                        }
                    }
                }
                titleLocation = dataToManipulate.IndexOf(titleFlag);
            }
            return links;
        }
    }
}

To sum it all up - I am directly manipulating the string data by iterating through sections and removing content that I already worked with. My plans of using a XML serializer or even go as far as try using XDocument.Parse for specific fragments failed because of the way HttpWebRequest was returning data - in fact, I suspect that the guys running HN are in the middle of this problem too. Some attributes use quotes, and some don't. This throws the XML parser off and I get an exception, failing to get the content I need.

For some items I had to make sure that those exist in the first place. The following items can be non-existent:

• domain
• full URL
• number of points
• author
• number of comments

A good example of such an outlier can be found here. So far, I managed to successfully avoid exceptions connected to these links because of index verification. If you are curious to see a visible result of the work done by the parser shown above, look at this screenshot:

PS: This was a quick implementation. An alternative parser might be implemented via RegEx, and I will write about it once I complete it. If you have other suggestions on parsing the HN home page, let me know.