bookmark-manager-r4/Services/WebProvider.cs

using System;
using System.Drawing;
using System.Net;
using System.Threading.Tasks;
using HtmlAgilityPack;
using RyzStudio.Net;

namespace BookmarkManager.Services
{
    public class WebProvider
    {
        private readonly WebClientProvider _webClientProvider;


        public WebProvider()
        {
            _webClientProvider = new WebClientProvider();
            _webClientProvider.Timeout = 4;
        }

        public async Task<HtmlAgilityPack.HtmlDocument> RetrieveHtmlDocument(string url)
        {
            var sourceCode = await this.RetrieveSourceCode(url);
            if (string.IsNullOrWhiteSpace(sourceCode))
            {
                return null;
            }

            var document = new HtmlAgilityPack.HtmlDocument();

            try
            {
                document.LoadHtml(sourceCode);
            }
            catch (Exception)
            {
                return null;
            }

            return document;
        }

        public async Task<string> RetrieveSourceCode(string url)
        {
            if (string.IsNullOrWhiteSpace(url))
            {
                return null;
            }

            if (!Uri.TryCreate(url, UriKind.Absolute, out Uri uri))
            {
                return null;
            }

            System.Net.Http.HttpResponseMessage response;

            try
            {
                response = await _webClientProvider.Get(url?.Trim());
            }
            catch (Exception)
            {
                return null;
            }

            if (response.StatusCode != HttpStatusCode.OK)
            {
                return null;
            }

            var sourceCode = await response?.Content?.ReadAsStringAsync();
            if (string.IsNullOrWhiteSpace(sourceCode))
            {
                return null;
            }

            return sourceCode;
        }

        public async Task<Image> RetrieveImage(string url)
        {
            if (string.IsNullOrWhiteSpace(url))
            {
                return null;
            }

            if (!Uri.TryCreate(url, UriKind.Absolute, out Uri uri))
            {
                return null;
            }

            System.Net.Http.HttpResponseMessage response;

            try
            {
                response = await _webClientProvider.Get(url?.Trim());
            }
            catch (Exception)
            {
                return null;
            }

            if (response.StatusCode != HttpStatusCode.OK)
            {
                return null;
            }

            var stream = await response?.Content?.ReadAsStreamAsync();

            Image result = null;

            try
            {
                result = Image.FromStream(stream);
            }
            catch (Exception)
            {
                return null;
            }

            return result;
        }

        public async Task<Image> RetrieveImage(HtmlAgilityPack.HtmlDocument document)
        {
            var iconUrl = this.ParseFavicon(document);
            if (string.IsNullOrWhiteSpace(iconUrl))
            {
                return null;
            }

            return await this.RetrieveImage(iconUrl);
        }

        public string ParseTitle(HtmlAgilityPack.HtmlDocument document)
        {
            string result = null;

            result = ParseTagValue(document, "//title", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[@property='og:title']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[@name='twitter:title']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[@property='og:site_name']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[@itemprop='name']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            return string.Empty;
        }

        public string ParseMetaDescription(HtmlAgilityPack.HtmlDocument document)
        {
            string result = null;

            result = ParseTagValue_Attr(document, "//meta[@name='description']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[@property='og:description']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[@name='twitter:description']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[@property='og:description']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[@itemprop='description']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            return result;
        }

        public string ParseFavicon(HtmlAgilityPack.HtmlDocument document)
        {
            string result = null;

            result = ParseTagValue_Attr(document, "//link[translate(@rel, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'shortcut icon']", "href", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//link[translate(@rel, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'icon']", "href", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//link[translate(@rel, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'apple-touch-icon']", "href", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//link[translate(@rel, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'apple-touch-icon-precomposed']", "href", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[translate(@property, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'og:image']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[translate(@name, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'twitter:image']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[translate(@property, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'og:image']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            result = ParseTagValue_Attr(document, "//meta[translate(@itemprop, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'image']", "content", string.Empty)?.Trim();
            if (!string.IsNullOrWhiteSpace(result))
            {
                return result;
            }

            return "/favicon.ico";
        }

        private string ParseTagValue(HtmlAgilityPack.HtmlDocument document, string xPath, string defaultValue = "")
        {
            var hnc = document.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;
                }

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

                return result;
            }

            return defaultValue;
        }

        private string ParseTagValue_Attr(HtmlAgilityPack.HtmlDocument document, string xPath, string attr, string defaultValue = "")
        {
            var hnc = document.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;
        }

    }
}