/// <summary>
/// Scrapes extra data about a course which is not given on first pass
/// </summary>
/// <param name="course">Course to scrape</param>
/// <returns>A list of courses</returns>
public static List <Course> AddData(this Course course)
{
var page = new HtmlWeb().Load(course.CourseUrl).DocumentNode;
/* UCAS CODE */
var ucasCodeSection = page.SelectSingleNode("//strong[text()='UCAS code']");
string ucasCode = ucasCodeSection.ParentNode.NextSibling.InnerText;
/* ADMISSION STATS */
var admissions = AdmissionStats(page);
if (admissions != null)
{
course.Admission = new Admission(admissions);
}
/* DURATION STATS */
course.Duration.AddRange(DurationStats(page));
/* ADMISSIONS TEST */
course.AdmissionTest = AdmissionTest(page);
/* WRITTEN WORK */
var writtenWorkSection = page.SelectSingleNode("//strong[contains(text(), 'Written work')]");
var writtenWork = writtenWorkSection.ParentNode.NextSibling.InnerText;
course.WrittenWork = (writtenWork == "None") ? null : writtenWork;
/* SUBJECT REQUIREMENTS */
course.ALevelCourse.AddRange(SubjectRequirements(page));
/* SPLIT COURSES */
//if only one ucas code, return as normal
if (ucasCode.Length == 4)
{
course.UcasCode = ucasCode;
return(new List <Course> {
course
});
}
//if has brackets (indicates there are 2 courses on one page)
if (ucasCode.Contains('('))
{
return(Fetcher.TwoCodes(course, ucasCode));
}
//if contains a hyperlink, indicates a link to a table elsewhere
if (ucasCodeSection.ParentNode.NextSibling.SelectSingleNode("a") != null)
{
var link = ucasCodeSection.ParentNode.NextSibling.SelectSingleNode("a");
return(Fetcher.HyperlinkTable(course, link.Attributes["href"].Value));
}
//if no URL in ucas code, can simply read table from same page
return(Fetcher.Table(course, Fetcher.FindTable(page)));
}
static void Main(string[] args)
{
string docurl = "https://ebiz.khnp.co.kr/getTmpView.do?type=spbidnoti&loadId=fff84246-b957-47fc-a249-e1d023131890";
var doc = new HtmlWeb().Load(docurl).DocumentNode;
string aprice1 = doc.SelectSingleNode("//th[text()='가격점수제외금액(A)']").NextSibling.NextSibling.InnerText;
string aprice2 = doc.SelectSingleNode("//th[text()[contains(., '가격점수제외금액(A)')]]").NextSibling.NextSibling.InnerText;
Console.WriteLine(aprice1);
Console.WriteLine(aprice2);
}
/// <summary>
/// Adds extra data to a college
/// </summary>
/// <param name="college">College to add data to</param>
public static void AddData(this College college)
{
var page = new HtmlWeb().Load(college.CollegeUrl).DocumentNode;
/* CAMPUS CODE */
var campusCodeSelection = page.SelectSingleNode("//h2[contains(text(), 'UCAS campus code:')]");
string campusCode = campusCodeSelection.InnerText.Split(' ').Last();
college.CampusCode = campusCode;
/* OFFERED COURSES */
var coursePage = new HtmlWeb().Load($"{college.CollegeUrl}#content-tab--3").DocumentNode;
var courseList = coursePage.SelectSingleNode("//div[@id='content-tab--3']/ul");
college.Courses = new List <string>();
//temp hardcoded solution, planning to fix
if (college.CampusCode == "P")
{
college.Courses.Add("Theology and Religion");
college.Courses.Add("Philosophy and Theology");
}
else
{
foreach (var a in courseList.SelectNodes("li/a"))
{
college.Courses.Add(a.InnerText);
}
}
/* NUMBER OF STUDENTS */
var studentNumSelection = page.SelectSingleNode("//h3[contains(text(), 'Student numbers')]").NextSibling;
var textSections = studentNumSelection.InnerHtml.Split("<br>");
//Mathces 1 or more digits (for both regexes)
college.UndergradStudents = Convert.ToInt32(Regex.Match(textSections[0], @"\d{1,}").Value);
college.GradStudents = Convert.ToInt32(Regex.Match(textSections[1], @"\d{1,}").Value);
college.TotalStudents = college.UndergradStudents + college.GradStudents;
/* FACILITIES */
var facilityPage = new HtmlWeb().Load($"{college.CollegeUrl}#content-tab--2").DocumentNode;
var facilities = facilityPage.SelectSingleNode("//h2[text()='College facilities']").ParentNode.SelectSingleNode("table/tbody");
college.Facilities = new CollegeFacilities(facilities);
}
/// <summary>
/// Get a list of type T from a url, by parsing the expected table format
/// </summary>
/// <typeparam name="T">Type for list to use</typeparam>
/// <param name="url">Url to parse</param>
/// <returns>A list of data parsed from url</returns>
public static List <T> GetList <T>(string url)
{
List <T> outputList = new List <T>();
var page = new HtmlWeb().Load(url).DocumentNode;
var tableNode = page.SelectSingleNode("//*[@class='table-reduced']");
var trNode = tableNode.SelectSingleNode("tbody/tr");
foreach (var td in trNode.ChildNodes)
{
foreach (var link in td.SelectNodes("p/a"))
{
string name = link.InnerText;
string linkUrl = "https:" + link.Attributes["href"].Value;
//create an instance of T with given parameters - hacky workaround, but only solution i can find
T data = (T)Activator.CreateInstance(typeof(T), name, linkUrl);
outputList.Add(data);
}
}
return(outputList);
}
public static string DownloadDeck(string url)
{
if (url.Contains("legends-decks.com"))
{
var page = new HtmlWeb().Load(url).DocumentNode;
return(page.SelectSingleNode("//div[@id='deckModal']//div[@class='well_full']").InnerHtml.Replace("<br>", "\r\n").Trim());
}
if (url.Contains("eternalwarcry.com"))
{
var page = new HtmlWeb().Load(url).DocumentNode;
var name = page.SelectSingleNode("//h1").InnerText.Trim();
var cards = page.SelectSingleNode("//textarea[@id='export-deck-text']").InnerHtml.Trim();
return($"### {name} ###\r\n{cards}");
}
return(null);
}
/// <summary>
/// Creates multiple courses based upon a table on another page
/// </summary>
/// <param name="course">Course to split</param>
/// <param name="url">Url of other page</param>
/// <returns>A list of correct courses</returns>
public static List <Course> HyperlinkTable(Course course, string url)
{
var page = new HtmlWeb().Load($"https:{url}").DocumentNode;
var ucasSection = page.SelectSingleNode("//h3[contains(text(), 'UCAS codes')]/following-sibling::div");
List <Course> newCourses = new List <Course>();
foreach (var table in ucasSection.SelectNodes("table").Take(3))
{
newCourses.AddRange(Table(course, table));
}
return(newCourses);
}
public static string DownloadDeck(string url)
{
if (url.Contains("legends-decks.com"))
{
var page = new HtmlWeb().Load(url).DocumentNode;
return(page.SelectSingleNode("//div[@id='deckModal']//div[@class='well_full']").InnerHtml.Replace("<br>", "\r\n").Trim());
}
if (url.Contains("eternalwarcry.com"))
{
var page = new HtmlWeb().Load(url).DocumentNode;
var name = page.SelectSingleNode("//h1").InnerText.Trim();
var cards = page.SelectSingleNode("//textarea[@id='export-deck-text']").InnerHtml.Trim();
return($"### {name} ###\r\n{cards}");
}
if (url.Contains("teslegends.pro"))
{
var slug = url.Substring(url.TrimEnd('/').LastIndexOf('/')).Trim('/');
var response = new WebClient().UploadValues("https://teslegends.pro/dc/do.php", "POST", new NameValueCollection {
{ "exportdeck", slug }
});
return(Encoding.UTF8.GetString(response));
}
return(null);
}
private void Update()
{
_html = "";
_lastStoredHrefBCPEA = ReadLastCarHref(_filenameBCPEA);
_lastStoredHrefNap = ReadLastCarHref(_filenameNap);
try
{
var docNap = new HtmlWeb().Load(_urlNap).DocumentNode;
var nDivs = docNap.SelectNodes("//div[@class]");
var carsNapDivs = nDivs.Where(at => at.GetAttributeValue("class", "").Trim() == "aoh-item").ToList();
carsNapDivs.ForEach(div => div.InnerHtml = div.InnerHtml.Replace("href=\"/targ/", "href=\"" + _websiteNapLink + "/targ/")
.Replace("src=\"/", "src=\"" + _websiteNapLink + "/"));
var lastestCarsNap = GetLastestCars(carsNapDivs, _lastStoredHrefNap);
UpdateLastStoredHref(lastestCarsNap, _filenameNap, ref _lastStoredHrefNap);
if (lastestCarsNap.Any())
{
string title = "Original link: <a href=\"" + _urlNap + "\">" + _urlNap + "</a><hr>";
_html += title + string.Join("<hr>", lastestCarsNap.Select(s => s.InnerHtml)) + "<br/>";
Logger.Write("lastestCarsNap: " + lastestCarsNap.Count);
}
var doc = new HtmlWeb().Load(_urlBcpea).DocumentNode;
var nodeBody = doc.SelectSingleNode("//body");
var n = nodeBody.SelectNodes("//ul[@class]");
var carsCsiLis = n.First(at => at.Attributes.AttributesWithName("class").Select(a => a.Value == "results_list").First()).SelectNodes("li").ToList();
carsCsiLis.ForEach(li => li.InnerHtml = li.InnerHtml.Replace("href=\"/bg/auto/", "href=\"" + _websiteBCPEALink + "/bg/auto/"));
var lastestCars = GetLastestCars(carsCsiLis, _lastStoredHrefBCPEA);
UpdateLastStoredHref(lastestCars, _filenameBCPEA, ref _lastStoredHrefBCPEA);
if (lastestCars.Any())
{
string title = "Original link: <a href=\"" + _urlBcpea + "\">" + _urlBcpea + "</a><hr>";
_html += title + string.Join("<hr>", lastestCars.Select(s => s.InnerHtml));
Logger.Write("lastestCars: " + lastestCars.Count);
}
}
catch (Exception ex)
{
Logger.Write("Error in Update(): ", ex);
}
}
public static List <(string, double, double)> GetNorrington()
{
List <(string, double, double)> outputList = new List <(string, double, double)>();
var page = new HtmlWeb().Load(@"https://en.wikipedia.org/wiki/Norrington_Table").DocumentNode;
var tableNode = page.SelectSingleNode("//table/tbody");
foreach (var tr in tableNode.SelectNodes("tr").Skip(1))
{
(string, double, double)value = new ValueTuple <string, double, double>
{
Item1 = tr.SelectSingleNode("td[1]/a").InnerText,
Item2 = Convert.ToDouble(tr.SelectSingleNode("td[2]").InnerText),
Item3 = Convert.ToDouble(tr.SelectSingleNode("td[3]").InnerText)
};
outputList.Add(value);
}
return(outputList);
}
public void ParseFrom(string url, int StartPage, int FinishPage)
{
try
{
var doc = new HtmlWeb().Load(url + "/goods.php");
// int pageCount = MarketItems.PageCount(url);
//проверка по пагинации
for (int page = StartPage; page <= FinishPage; page++)
{
doc = new HtmlWeb().Load(url + "/goods.php?cid=5&page=" + page); //текущая страница
var pageGoods = doc.DocumentNode.SelectNodes("//div[@class='ernr']/ul/li/div/h3/a").
Select(a => url + "/" + a.ChildAttributes("href").
FirstOrDefault().Value); //ссылки на все товары на странице
if (pageGoods.Count() == 0)
{
throw new Exception("Товары не найдены!\nВозможно, сайт поменял верстку. Требуется обновить ПО!");
}
else
{
foreach (var good in pageGoods.ToList())
{
MarketItem item = new MarketItem();//будем заполнять товар
try
{
var goodNode = new HtmlWeb().Load(good).DocumentNode.SelectSingleNode("//div[@class='cps']"); //html товара. отсюда и вытащим всё
var Options = MarketItems.ConvertList(goodNode.SelectNodes(".//div[@class='tabmen']/ul/li").Select(a => a.InnerText).ToList());
var Quantities = ConvertList(goodNode.SelectNodes(".//div[@id='tabconten']/ul/li").Select(a => a.InnerText.Replace("双", "")).ToList()); //
item.Model = Count != 0 ? this.Max(a => a.Model) + 1 : 1;
item.Name = goodNode.SelectSingleNode(".//h6").InnerText + " " + item.Model;
item.Description = "<p><br></p>";
item.SEO_url = item.Name.Replace(" ", "-");
item.Out_stock_status = "";
item.Option_type = "radio";
item.Price = "";//сам вводит
item.Main_image = url + "/" + goodNode.SelectSingleNode(".//img").ChildAttributes("src").FirstOrDefault().Value;
/*TODO*/
item.Manufacturer = Manuf(item.Name);
item.Option = MarketItems.Option(Options);
string resName = "";
if (item.Manufacturer.Contains("Timber"))
{
resName += "Ботинки";
}
else
{
resName += "Кроссовки";
}
if (Options.Min() <= 36)
{
resName += " Женские ";
}
else
{
resName += " Мужские ";
}
item.Name = resName + goodNode.SelectSingleNode(".//h6").InnerText + " " + item.Model;
item.Meta_title = item.Name;
//по парам
item.Quantity = Quantities.First().ToString();
item.Option_value = Options.First().ToString();
if (Options.Count == Quantities.Count)
{
if (Options.Count > 1)
{
for (int i = 1; i < Options.Count; i++)
{
MarketItem child = new MarketItem();
child.Model = item.Model;
child.Name = item.Name;
child.Description = "";
child.Meta_title = "";
child.SEO_url = "";
child.Option = item.Option;
child.Option_type = item.Option_type;
child.Main_image = "";
child.Quantity = Quantities[i].ToString();
child.Option_value = Options[i].ToString();
item.ChildrenItems.Add(child);
}
}
}
else
{
throw new Exception();
}
}
catch (Exception exception)
{
item.Error = exception.Message.ToString();
}
Add(item);//после парсинга страницы добавляем элемент
}
}
}
}
catch (Exception e)
{
throw new Exception("Произошла ошибка при попытке получить страницу по адресу: " + url + "\n\t\t" + e.Message + "\nПожалуйста, проверьте правильность ввода ресурса!");
}
}