예제 #1
0
        public CalculatedSpecialty(SpecialtyInfo specialty)
        {
            var tb         = specialty.GetType();
            var properties = tb.GetProperties();
            var t          = GetType();

            properties.ToList().ForEach(property =>
            {
                var isPresent = t.GetProperty(property.Name);
                if (isPresent != null)
                {
                    var value = tb.GetProperty(property.Name).GetValue(specialty, null);
                    t.GetProperty(property.Name).SetValue(this, value, null);
                }
            });
        }
예제 #2
0
        private async void ParsePage(HttpContent httpContent, string link)
        {
#if DEBUG
            Console.WriteLine($"Parsing {link}");
#endif
            var page = await httpContent.ReadAsStringAsync();

            httpContent.Dispose();
            var doc = new HtmlDocument();
            doc.LoadHtml(page);
            var sp = new SpecialtyInfo
            {
                Content = new ContentVisualiser {
                    Content = new Dictionary <string, List <string> >()
                },
                Links = new LinksVisualiser {
                    Links = new Dictionary <string, List <LinkItem> >()
                },
                ChairsProvidesProg = new TupleVisualiser()
            };

            try
            {
                var f = doc.GetElementbyId("yoo-zoo").Element("div");
                foreach (var el in f.Elements("div"))
                {
                    switch (el.GetAttributeValue("class", "null"))
                    {
                    case "pos-top":
                    {
                        // List<string> tmpContent = new List<string>();
                        var er = el.Element("div");
                        sp.BranchName = new Item
                        {
                            Content = new List <string>()
                        };
                        sp.BranchName.Content.Add(er.ChildNodes[2].InnerText);
                        var urltmp = er.ChildNodes[2].GetAttributeValue("href", "");
                        sp.BranchName.Url   = string.IsNullOrEmpty(urltmp) ? "" : abitUrl + urltmp;
                        sp.BranchName.Title = er.ChildNodes[1].InnerText.TrimEnd(' ').TrimStart(' ');
                        break;
                    }

                    case "pos-content":
                    {
                        foreach (var i in el.Elements("div"))
                        {
                            var atr = i.GetAttributeValue("class", "null");
                            if (atr != "null")
                            {
                                switch (atr.Substring("element element-".Length))
                                {
                                /* case "relateditems":
                                 *   {
                                 *       try
                                 *       {
                                 *           if (i.Element("h3").InnerText.ToLowerInvariant().Contains("контракт"))
                                 *           {
                                 *               sp.Modulus = Specialty.ModulusList.GetModulusFromHtml(i.Element("ul").Elements("li"), sp.Modulus, false);
                                 *           }
                                 *           else if (i.Element("h3").InnerText.ToLowerInvariant().Contains("держ"))
                                 *           {
                                 *               sp.Modulus = Specialty.ModulusList.GetModulusFromHtml(i.Element("ul").Elements("li"), sp.Modulus, true);
                                 *           }
                                 *       }
                                 *       catch (Exception)
                                 *       {
                                 *           Console.SetCursorPosition(0, rx++);
                                 *           Console.Write("Check this url: " + sitesUrl + link as string);
                                 *       }
                                 *       break;
                                 *   }*/
                                case "text first":
                                {
                                    try
                                    {
                                        sp.Code = i.ChildNodes[2].InnerText;
                                    }
                                    catch (Exception)
                                    {
#if DEBUG
                                        Console.WriteLine("Check this url: " + abitUrl + link);
#endif
                                    }
                                    break;
                                }

                                case "text":
                                {
                                    List <string> tmpContent = new List <string>
                                    {
                                        i.ChildNodes[2].InnerText
                                    };
                                    sp.Content.Content.Add(i.ChildNodes[1].InnerText, tmpContent);
                                    break;
                                }

                                case "textarea":
                                {
                                    List <string> list = new List <string>();
                                    var           ds   = i.Element("h3").NextSibling;
                                    list.Add(ds.OuterHtml);
                                    while (true)
                                    {
                                        if (ds.NextSibling != null)
                                        {
                                            ds = ds.NextSibling;
                                            if (ds.OuterHtml.Length > 3)
                                            {
                                                list.Add(ds.OuterHtml);
                                            }
                                        }
                                        else
                                        {
                                            break;
                                        }
                                    }
                                    sp.Content.Content.Add(i.ChildNodes[1].InnerText, list);
                                    break;
                                }

                                case "link":
                                {
                                    var links = new List <LinkItem>();
                                    foreach (var op in i.Elements("a"))
                                    {
                                        links.Add(new LinkItem
                                                {
                                                    Url   = op.GetAttributeValue("href", ""),
                                                    Title = op.GetAttributeValue("title", "")
                                                });
                                    }

                                    sp.Links.Links.Add(i.ChildNodes[1].InnerText, links);

                                    break;
                                }

                                case "relatedcategories":
                                {
                                    var links = new List <LinkItem>();
                                    foreach (var r in i.ChildNodes[2].ChildNodes)
                                    {
                                        var urltmp = r.ChildNodes[0].GetAttributeValue("href", "");
                                        links.Add(new LinkItem
                                                {
                                                    Url   = string.IsNullOrEmpty(urltmp) ? "" : abitUrl + urltmp,
                                                    Title = r.ChildNodes[0].InnerText
                                                });
                                    }
                                    sp.ChairsProvidesProg.ChairsProvidesProg = new Tuple <string, List <LinkItem> >(i.ChildNodes[1].InnerText, links);
                                    break;
                                }
                                }
                            }
                        }
                        break;
                    }
                    }
                }
                var title = f.Element("h1").InnerText.Trim();
                if (title.Contains('('))
                {
                    sp.Title = title.Substring(0, title.IndexOf('('));
                    var sb = new Regex(@"(?<=\()\W.*?(?=\))", RegexOptions.ECMAScript).Match(title).Groups[0].Value;
                    if (sb != "Бакалавр")
                    {
                        sp.SubTitle = UppercaseFirst(sb);
                    }
                }
                else
                {
                    sp.Title = title;
                }
                sp.Title = sp.Title.Trim().Replace('\n', ' ').Replace("  ", " ").Replace('’', '\'').Replace('`', '\'');
                sp.URL   = abitUrl + link;

                SpecialtyList.Add(sp);
            }
            catch (Exception ex)
            {
                sp.Errors = new List <string>
                {
                    ex.Message + ' ' + Tools.AnonymizeStack(ex.StackTrace)
                };
                SpecialtyList.Add(sp);
            }

#if DEBUG
            Console.WriteLine($"Finished {link}");
#endif
        }