public ArticleMark(string link, List <ArticleCriteria> articleList)
{
this.FacultyName = ArticleCriteria.GetFaculty(link);
this.ArticleNumber = (articleList.Count);
this.LastArticleDate = articleList.Max(a => a.Date);
}
public List <ArticleCriteria> CollectNewsInfo(string link, DateTime fromDate)
{
string faculty = GetFaculty(link);
List <ArticleCriteria> articleList = new List <ArticleCriteria>();
bool keepGoing = true;
int pageNumber = 1;
while (keepGoing)
{
string pageLink;
if (pageNumber == 1)
{
pageLink = link;
}
else
{
pageLink = link + "/page/" + pageNumber.ToString();
}
HtmlWeb web = new HtmlWeb();
var res = web.Load(pageLink);
if (res != null)
{
// list of raw html fragments <article>(.*?)</article>
var articlesRaw = res.DocumentNode.SelectNodes("//article");
foreach (var articleRaw in articlesRaw)
{
var article = new ArticleCriteria(articleRaw, pageNumber, faculty);
try
{
if (article.Date >= fromDate)
{
articleList.Add(article);
}
else
{
//stop
keepGoing = false;
break;
}
}
catch (Exception e)
{
articleList.Add(article);
}
}
pageNumber += 1;
}
else
{
keepGoing = false;
}
}
return(articleList);
}
public void MainFunction()
{
// посилання на факультети
string filenameRead = @"Data\faculty_links.txt";
List <string> links = ExtractFacultyLinks(filenameRead);
// посилання на списки кафедр, новин
List <string> departmentLinks = ConcatanateLinks(links, "/about/departments");
List <string> newsLinks = ConcatanateLinks(links, "/news");
string filenameWrite = ("report_department" + DateTime.Today.ToString("u") + ".txt").Replace(':', '.');
using (StreamWriter f = new StreamWriter(filenameWrite, false))
{
f.WriteLine(string.Format("ФАКУЛЬТЕТ \t Кафедра \t Оцінка на {0}", DateTime.Today.ToString("dd.MM.yyyy")));
for (int i = 0; i < departmentLinks.Count; i++)
{
HtmlWeb web = new HtmlWeb();
string departmentLink = departmentLinks[i],
newsLink = newsLinks[i];
var mainNews = ExtractArticles(web.Load(newsLink).DocumentNode, "");
var htmlDoc = web.Load(departmentLink);
var departments = ExtractAllDepartments(htmlDoc.DocumentNode);
htmlDoc = web.Load(newsLink);
var filters = ExtractDepartmentsFromFilter(htmlDoc.DocumentNode);
var departmentFilter = LinkDeparmentsToFilters(departments, filters);
double facultyMark = new double();
string info_line = string.Format("{0}\n", ArticleCriteria.GetFaculty(departmentLink));
f.WriteLine(info_line);
foreach (var department in departmentFilter.Keys)
{
var departmentArticles = ExtractArticles(web.Load(department.
Attributes["href"].Value).DocumentNode, "//body//div//div//section");
HtmlNodeCollection filterArticles;
try
{
filterArticles = ExtractArticles(web.Load(departmentFilter[department].
Attributes["href"].Value).DocumentNode, "//body//div[@class='content news']");
}
catch (Exception e)
{
filterArticles = null;
}
double departmentMark = GetDepartmentMark(departmentArticles, filterArticles, mainNews);
dump_department_mark(f, department.InnerText, departmentMark);
facultyMark += departmentMark;
}
facultyMark /= departments.Count;
f.WriteLine(string.Format("ФАКУЛЬТЕТ \t Середня оцінка {0}", facultyMark));
}
}
}
//public void MainFunction()
//{
// // посилання на персонал факультетів
// string filenameRead = @"Data\staff_links.txt";
// List<string> staffLinks = ExtractFacultyLinks(filenameRead);
// string filenameWrite = ("report_staff_" + DateTime.Now.ToString() + ".txt").Replace(':', '.');
// using (StreamWriter f = new StreamWriter(filenameWrite, false))
// {
// f.WriteLine(string.Format("Викладачі \t Оцінка на {0}", DateTime.Today.ToString("dd.MM.yyyy")));
// for (int i = 0; i < staffLinks.Count; i++)
// {
// HtmlWeb web = new HtmlWeb();
// string staffLink = staffLinks[i];
// var linksNodes = GetAllLecturerLinksNodes(web.Load(staffLink).DocumentNode);
// //отримання посилань на всіх працівників факультету
// var lecturerlinks = GetAllLecturerLinks(linksNodes);
// double facultyStaffMark = new double();
// string info_line = string.Format("Факультет {0}\n", ArticleCriteria.GetFaculty(staffLink));
// f.WriteLine(info_line);
// var lecturers = new List<Lecturer>();
// foreach (var lecturerLink in lecturerlinks)
// {
// var lecturer = new Lecturer();
// lecturer.SetLink(lecturerLink);
// try
// {
// var lecturerPage = web.Load(lecturerLink).DocumentNode;
// var infoRows = GetAllInfoRows(lecturerPage);
// lecturer.SetName(lecturerPage);
// lecturer.SetInfoFromRows(infoRows);
// }
// catch
// {
// }
// f.WriteLine(lecturer);
// //double departmentMark = GetDepartmentMark(departmentArticles, filterArticles, mainNews);
// //dump_department_mark(f, department.InnerText, departmentMark);
// //facultyMark += departmentMark;
// }
// //facultyMark /= departments.Count;
// //f.WriteLine(string.Format("ФАКУЛЬТЕТ \t Середня оцінка {0}", facultyMark));
// }
// f.WriteLine(string.Format("Час завершення {0}", DateTime.Now.ToString()));
// }
//}
public void MainFunction()
{
// посилання на персонал факультетів
string filenameRead = @"Data\staff_links.txt";
List <string> staffLinks = ExtractFacultyLinks(filenameRead);
string filenameWrite = ("report_staff_" + DateTime.Now.ToString() + ".txt").Replace(':', '.');
using (StreamWriter f = new StreamWriter(filenameWrite, false))
{
f.WriteLine(string.Format("Викладачі \t Оцінка на {0}", DateTime.Today.ToString("dd.MM.yyyy")));
}
for (int i = 0; i < staffLinks.Count; i++)
{
HtmlWeb web = new HtmlWeb();
string staffLink = staffLinks[i];
var linksNodes = GetAllLecturerLinksNodes(web.Load(staffLink).DocumentNode);
//отримання посилань на всіх працівників факультету
var lecturerlinks = GetAllLecturerLinks(linksNodes);
double facultyStaffMark = new double();
using (StreamWriter f = new StreamWriter(filenameWrite, true))
{
string info_line = string.Format("Факультет {0}\n", ArticleCriteria.GetFaculty(staffLink));
f.WriteLine(info_line);
}
var lecturers = new List <Lecturer>();
foreach (var lecturerLink in lecturerlinks)
{
var lecturer = new Lecturer();
lecturer.SetLink(lecturerLink);
try
{
var lecturerPage = web.Load(lecturerLink).DocumentNode;
var infoRows = GetAllInfoRows(lecturerPage);
lecturer.SetName(lecturerPage);
lecturer.SetInfoFromRows(infoRows);
var sections = GetAllSections(lecturerPage);
lecturer.SetInfoFromSections(sections);
}
catch
{
}
using (StreamWriter f = new StreamWriter(filenameWrite, true))
{
f.WriteLine(lecturer);
}
//double departmentMark = GetDepartmentMark(departmentArticles, filterArticles, mainNews);
//dump_department_mark(f, department.InnerText, departmentMark);
//facultyMark += departmentMark;
}
//facultyMark /= departments.Count;
//f.WriteLine(string.Format("ФАКУЛЬТЕТ \t Середня оцінка {0}", facultyMark));
}
using (StreamWriter f = new StreamWriter(filenameWrite, true))
{
f.WriteLine(string.Format("Час завершення {0}", DateTime.Now.ToString()));
}
}
public ArticleMark(DateTime LastArticleDate, int ArticleNumber, string FacultyName)
{
this.FacultyName = ArticleCriteria.GetFaculty(FacultyName);
this.ArticleNumber = (ArticleNumber);
this.LastArticleDate = (LastArticleDate);
}
