private string prepareBook() { bool val = sf.isSafePull(); if (!val) { while (!val) { val = sf.isSafePull(); } //break; } //string HTML = (wbBrowser.Document as mshtml.IHTMLDocument2).body.innerHTML; HtmlDocument doc = new HtmlDocument(); doc.LoadHtml(this.Document.DocumentNode.OuterHtml); this.Document = doc; WebScraper.StartScrape scrape = new WebScraper.StartScrape(this.CurrentBook.PageType, doc); //string myString = File.ReadAllText(Environment.GetFolderPath(Environment.SpecialFolder.Desktop) + "\\btth2.html"); //HtmlDocument d = new HtmlDocument(); //d.LoadHtml(myString); //WebScraper.StartScrape scrape = new WebScraper.StartScrape(this.CurrentBook.PageType, d); DatabaseWriter db = new DatabaseWriter(@"Server=localhost\SQLEXPRESS01; Database=Books; Trusted_Connection=true;"); db.createSource(this.CurrentBook.AssociatedWebsite.WebsiteName, Websites.WuxiaWorld.WebsiteAddress); db.createBook(this.CurrentBook.BookName, this.CurrentBook.AssociatedWebsite.WebsiteName); db.createChapter(scrape.CurrentChapterName.Trim(), this.CurrentBook.BookName, scrape.BookContent, ProgramStatics.ChapterAddress, scrape.HTML); BookSaver saver = new BookSaver(this.CurrentBook.AssociatedWebsite.WebsiteName + "\\" + this.CurrentBook.BookName, scrape.CurrentChapterName.Trim()); string s = scrape.BookContent; //s = replaceValues(s); //s.Trim(); saver.saveBook(scrape.BookContent); if (string.IsNullOrEmpty(scrape.NextChapter)) { ProgramStatics.NextBook = true; return(""); } if (scrape.NextChapter.Equals("Page Not Found")) { ProgramStatics.NextBook = true; return(""); } if (scrape.NextChapter.Equals(ProgramStatics.ChapterAddress) || scrape.NextChapter.Equals(ProgramStatics.PreviousChapter)) { ProgramStatics.NextBook = true; return(""); } ProgramStatics.PreviousChapter = ProgramStatics.ChapterAddress; ProgramStatics.ChapterAddress = scrape.NextChapter; System.Threading.Thread.Sleep(2000); return(scrape.NextChapter); }
private string prepareBook() { bool val = sf.isSafePull(); if (!val) { while (!val) { val = sf.isSafePull(); } //break; } //string HTML = (wbBrowser.Document as mshtml.IHTMLDocument2).body.innerHTML; HtmlDocument doc = new HtmlDocument(); doc.LoadHtml(this.Document.DocumentNode.OuterHtml); this.Document = doc; //WebScraper.PageTypes p = WebScraper.PageTypes[this.CurrentBook.SourceID]; WebScraper.StartScrape scrape = new WebScraper.StartScrape(this.CurrentBook.SourceID, doc, this.CurrentBook.BookID); //Thread.Sleep(15000); string nextChapter = scrape.NextChapter; string previousChapter = scrape.PreviousChapter; int bookID = this.CurrentBook.BookID; DatabaseWriter dbWrite = new DatabaseWriter(); dbWrite.BookID = bookID; //dbWrite.ChapterAddress = Chapter; dbWrite.ChapterAddress = ProgramStatics.ChapterAddress; dbWrite.ChapterHtml = scrape.HTML; dbWrite.ChapterName = scrape.CurrentChapterName; dbWrite.ChapterText = scrape.BookContent; dbWrite.PreviousChapterAddress = scrape.PreviousChapter; dbWrite.SaveChapter(); Logging.Logging.LogMessage("", Environment.GetFolderPath(Environment.SpecialFolder.Desktop) + "\\Logs", "BookScraper_DBSaved", " BookID" + this.CurrentBook.BookID + "\r\n\r\n Book Next Chapter Address" + this.CurrentBook.NextChapterAddress + "\r\n\r\n Book Previous Chapter Address: " + this.CurrentBook.PreviousChapterAddress + "\r\n\r\n Book LastChapterAddress: " + this.CurrentBook.BookLastChapterAddress + "\r\n\r\n Book SourceID: " + this.CurrentBook.SourceID + "\r\n\r\n Chapter: " + this.Chapter + "\r\n\r\n DB Next Chapter Address: " + scrape.NextChapter + "\r\n\r\n DB Previous Chapter Address: " + scrape.PreviousChapter + "\r\n\r\n DB Chapter Name: " + scrape.CurrentChapterName); //DatabaseWriter db = new DatabaseWriter(@"Server=localhost\SQLEXPRESS; Database=Book; Trusted_Connection=true;"); //db.createSource(this.CurrentBook.AssociatedWebsite.WebsiteName, Websites.WuxiaWorld.WebsiteAddress); //db.createBook(this.CurrentBook.BookName, this.CurrentBook.AssociatedWebsite.WebsiteName); //db.createChapter(scrape.CurrentChapterName, this.CurrentBook.BookName, scrape.BookContent, ProgramStatics.ChapterAddress, scrape.HTML); //BookSaver saver = new BookSaver(this.CurrentBook.AssociatedWebsite.WebsiteName + "\\" + this.CurrentBook.BookName, scrape.CurrentChapterName.Trim()); BookSaver saver = new BookSaver(this.websites[this.CurrentBook.SourceID] + "\\" + this.CurrentBook.BookName, scrape.CurrentChapterName.Trim()); string s = scrape.BookContent; //s = replaceValues(s); //s.Trim(); //saver.saveBook(scrape.BookContent); /* * if (string.IsNullOrEmpty(scrape.NextChapter)) * return ""; * if (scrape.NextChapter.Equals("Page Not Found")) * return ""; */ if (scrape.NextChapter.Equals("Page Not Found") || scrape.NextChapter.Equals("Page Not Found/") || string.IsNullOrEmpty(scrape.NextChapter) || scrape.NextChapter.Equals(ProgramStatics.ChapterAddress) || scrape.NextChapter.Equals(ProgramStatics.PreviousChapter) || ChapterAlreadyExists(scrape.NextChapter)) { ProgramStatics.NextBook = true; return(""); } ProgramStatics.PreviousChapter = ProgramStatics.ChapterAddress; ProgramStatics.ChapterAddress = scrape.NextChapter; //System.Threading.Thread.Sleep(15000); return(scrape.NextChapter); }