bookmark-manager/Models/BookmarkItemModel.cs

using bzit.bomg.Models;
using HtmlAgilityPack;
using RyzStudio.Net;
using System;
using System.Drawing;
using System.IO;
using System.Net;
using System.Text;
using HtmlDocument = HtmlAgilityPack.HtmlDocument;

namespace bzit.bomg
{
    public class BookmarkItemModel : BookmarkItemViewModel
    {
        protected HttpWeb webClient = null;

        public BookmarkItemModel()
        {
            this.Clear();
        }

        public void Clear()
        {
            this.SiteName = string.Empty;
            this.SiteAddress = string.Empty;
            this.SiteDescription = string.Empty;
            this.FaviconAddress = string.Empty;
            this.TreeviewPath = string.Empty;
        }

        public new string ToString()
        {
            StringBuilder sb = new StringBuilder();
            sb.Append("Name = ");
            sb.Append(this.SiteName ?? string.Empty);
            sb.Append(Environment.NewLine);

            sb.Append("Address = ");
            sb.Append(this.SiteAddress ?? string.Empty);
            sb.Append(Environment.NewLine);

            sb.Append("Description = ");
            sb.Append(this.SiteDescription ?? string.Empty);
            sb.Append(Environment.NewLine);

            return sb.ToString();
        }

        public BookmarkItemViewModel ToViewModel()
        {
            return new BookmarkItemViewModel()
            {
                SiteName = this.SiteName,
                SiteAddress = this.SiteAddress,
                SiteDescription = this.SiteDescription,
                FaviconAddress = this.FaviconAddress,
                TreeviewPath = this.TreeviewPath
            };
        }

        public bool Update()
        {
            string sourceCode = retrieveSourceCode();
            if (string.IsNullOrWhiteSpace(sourceCode))
            {
                return false;
            }

            HtmlDocument document = new HtmlDocument();
            document.LoadHtml(sourceCode);

            // title
            this.SiteName = parseSiteTitle(document);

            // description
            this.SiteDescription = parseSiteDescription(document);

            // favicon
            this.FaviconAddress = parseSiteIcon(document);

            // default infurred icon
            if (string.IsNullOrWhiteSpace(this.FaviconAddress))
            {
                this.FaviconAddress = "/favicon.ico";
            }

            if (!string.IsNullOrWhiteSpace(this.FaviconAddress))
            {
                Uri iconAddressURI;
                bool rv = Uri.TryCreate(new Uri(this.SiteAddress), this.FaviconAddress, out iconAddressURI);
                if (rv)
                {
                    this.FaviconAddress = iconAddressURI.ToString();
                }
            }

            return true;
        }

        public bool UpdateFavicon()
        {
            string sourceCode = retrieveSourceCode();
            if (string.IsNullOrWhiteSpace(sourceCode))
            {
                return false;
            }

            HtmlDocument document = new HtmlDocument();
            document.LoadHtml(sourceCode);

            // favicon
            this.FaviconAddress = parseSiteIcon(document);

            // default infurred icon
            if (string.IsNullOrWhiteSpace(this.FaviconAddress))
            {
                this.FaviconAddress = "/favicon.ico";
            }

            if (!string.IsNullOrWhiteSpace(this.FaviconAddress))
            {
                Uri iconAddressURI;
                bool rv = Uri.TryCreate(new Uri(this.SiteAddress), this.FaviconAddress, out iconAddressURI);
                if (rv)
                {
                    this.FaviconAddress = iconAddressURI.ToString();
                }
            }

            return true;
        }

        public Bitmap RetrieveFavicon()
        {
            if (string.IsNullOrWhiteSpace(this.FaviconAddress))
            {
                return null;
            }

            byte[] iconData = null;
            WebClient webClient = new WebClient();
            webClient.CachePolicy = new System.Net.Cache.RequestCachePolicy(System.Net.Cache.RequestCacheLevel.NoCacheNoStore);

            try
            {
                iconData = webClient.DownloadData(this.FaviconAddress);

                if (!RyzStudio.IO.FileType.IsImage(iconData))
                {
                    throw new Exception("Not a supported image");
                }

                Image img = Image.FromStream(new MemoryStream(iconData));
                return new Bitmap(img, 16, 16);
            }
            catch
            {
                iconData = null;
                this.FaviconAddress = null;

                return null;
            }
        }

        public Bitmap RetrieveFavicon(out byte[] rawData)
        {
            rawData = null;

            if (string.IsNullOrWhiteSpace(this.FaviconAddress))
            {
                return null;
            }

            byte[] iconData = null;
            WebClient webClient = new WebClient();
            webClient.CachePolicy = new System.Net.Cache.RequestCachePolicy(System.Net.Cache.RequestCacheLevel.NoCacheNoStore);

            try
            {
                iconData = webClient.DownloadData(this.FaviconAddress);

                if (!RyzStudio.IO.FileType.IsImage(iconData))
                {
                    throw new Exception("Not a supported image");
                }

                rawData = iconData;

                Image img = Image.FromStream(new MemoryStream(iconData));
                return new Bitmap(img, 16, 16);
            }
            catch
            {
                iconData = null;
                this.FaviconAddress = null;

                return null;
            }
        }

        protected string retrieveSourceCode()
        {
            if (webClient == null)
            {
                webClient = new HttpWeb();
            }

            string sourceCode;
            int statusCode = webClient.GetResponse(out sourceCode, this.SiteAddress);
            if ((statusCode == 200) || (statusCode == 301) || (statusCode == 302))
            {
                return sourceCode;
            }

            return null;

            //WebClient webClient = new WebClient();
            //webClient.CachePolicy = new System.Net.Cache.RequestCachePolicy(System.Net.Cache.RequestCacheLevel.NoCacheNoStore);

            //try
            //{
            //    return webClient.DownloadString(this.SiteAddress);
            //}
            //catch (Exception exc)
            //{
            //    return null;
            //}
        }

		protected string parseTagValue(HtmlDocument doc, string xpath, string defaultValue = "")
		{
			HtmlNodeCollection hnc = doc.DocumentNode.SelectNodes(xpath);
			if (hnc == null)
			{
				return defaultValue;
			}

			if (hnc.Count <= 0)
			{
				return defaultValue;
			}

			foreach (HtmlNode hn in hnc)
			{
				if (string.IsNullOrWhiteSpace(hn.InnerHtml))
				{
					continue;
				}

				string rs = WebUtility.HtmlDecode(hn.InnerHtml)?.Replace("\r", "")?.Replace("\n", " ")?.Trim();
				if (string.IsNullOrWhiteSpace(rs))
				{
					continue;
				}

				return rs;
			}

			return defaultValue;
		}

		protected string parseTagValue_Attr(HtmlDocument doc, string xpath, string attr, string defaultValue = "")
		{
			HtmlNodeCollection hnc = doc.DocumentNode.SelectNodes(xpath);
			if (hnc == null)
			{
				return defaultValue;
			}

			if (hnc.Count <= 0)
			{
				return defaultValue;
			}

			foreach (HtmlNode hn in hnc)
			{
				if (hn.Attributes[attr] == null)
				{
					continue;
				}

				if (string.IsNullOrWhiteSpace(hn.Attributes[attr].Value))
				{
					continue;
				}

				return System.Web.HttpUtility.HtmlDecode(hn.Attributes[attr].Value?.Trim());
			}

			return defaultValue;
		}

		protected string parseSiteTitle(HtmlDocument doc)
		{
			string rs = null;

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue(doc, "//title", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@property='og:title']", "content", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@name='twitter:title']", "content", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@property='og:site_name']", "content", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@itemprop='name']", "content", string.Empty);
			}

			return rs?.Trim();
		}

		protected string parseSiteDescription(HtmlDocument doc)
		{
			string rs = null;

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@name='description']", "content", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@property='og:description']", "content", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@name='twitter:description']", "content", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@property='og:description']", "content", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				rs = parseTagValue_Attr(doc, "//meta[@itemprop='description']", "content", string.Empty);
			}

			return rs;
		}

		protected string parseSiteIcon(HtmlDocument doc)
		{
			string rs = null;

			if (string.IsNullOrWhiteSpace(rs))
			{
				//rs = parseTagValue_Attr(doc, "//link[@rel='shortcut icon']", "href", string.Empty);
				rs = parseTagValue_Attr(doc, "//link[translate(@rel, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'shortcut icon']", "href", string.Empty);
			}

			if (string.IsNullOrWhiteSpace(rs))
			{
				//rs = parseTagValue_Attr(doc, "//link[@rel='icon']", "href", string.Empty);
                rs = parseTagValue_Attr(doc, "//link[translate(@rel, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'icon']", "href", string.Empty);
            }

            if (string.IsNullOrWhiteSpace(rs))
			{
				//rs = parseTagValue_Attr(doc, "//link[@rel='apple-touch-icon']", "href", string.Empty);
                rs = parseTagValue_Attr(doc, "//link[translate(@rel, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'apple-touch-icon']", "href", string.Empty);
            }

            if (string.IsNullOrWhiteSpace(rs))
			{
				//rs = parseTagValue_Attr(doc, "//link[@rel='apple-touch-icon-precomposed']", "href", string.Empty);
                rs = parseTagValue_Attr(doc, "//link[translate(@rel, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'apple-touch-icon-precomposed']", "href", string.Empty);
            }

            if (string.IsNullOrWhiteSpace(rs))
			{
				//rs = parseTagValue_Attr(doc, "//meta[@property='og:image']", "content", string.Empty);
                rs = parseTagValue_Attr(doc, "//meta[translate(@property, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'og:image']", "content", string.Empty);
            }

            if (string.IsNullOrWhiteSpace(rs))
			{
				//rs = parseTagValue_Attr(doc, "//meta[@name='twitter:image']", "content", string.Empty);
                rs = parseTagValue_Attr(doc, "//meta[translate(@name, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'twitter:image']", "content", string.Empty);
            }

            if (string.IsNullOrWhiteSpace(rs))
			{
				//rs = parseTagValue_Attr(doc, "//meta[@property='og:image']", "content", string.Empty);
                rs = parseTagValue_Attr(doc, "//meta[translate(@property, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'og:image']", "content", string.Empty);
            }

            if (string.IsNullOrWhiteSpace(rs))
			{
				//rs = parseTagValue_Attr(doc, "//meta[@itemprop='image']", "content", string.Empty);
                rs = parseTagValue_Attr(doc, "//meta[translate(@itemprop, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'image']", "content", string.Empty);
            }

            return rs;
		}

	}
}