public bool RGP(int PageNum = 1)
{
string url = "http://rgporn.com/index.php?newsid=" + PageNum;
//tel.Update(txtURL, url);
string html = web.DownloadHTML(url);
if (html == "")
{
return(false);
}
resultLines = xml.Parse_HTML_XML(html, "//body/div/div/div/div/table/tbody/tr/td/div/div/div/p"); // extract sections of text from html xml
// extract tags
string display = "TAGS: ";
List <string> tags = new List <string>();
string Tags = "";
foreach (string line in resultLines)
{
if (line.Contains("=tags"))
{
string tag = line.Replace("<a", "");
tag = tag.Replace("</a>", "");
tag = ahk.StringSplit(tag, ">", 1);
tag = tag.Trim();
tags.Add(tag);
display = display + "\n" + tag;
if (Tags != "")
{
Tags = Tags + "|" + tag;
}
if (Tags == "")
{
Tags = tag;
}
}
}
// extract title
List <string> titleLines = xml.Parse_HTML_XML(html, "//body/div/div/div/div/table/tbody/tr/td/div/div/div"); // extract sections of text from html xml
string title = lst.Return_List_Value(titleLines, 2);
title = title.Replace("<h1 class=\"shead\">", "");
title = title.Replace("</h1>", "");
string imagehtml = "";
foreach (string ima in titleLines)
{
if (ima.Contains(".jpg"))
{
imagehtml = imagehtml + "\n" + ima;
}
}
//ahk.MsgBox(title);
List <string> links = rg.Regex_RGLinks(html);
bool AllOnline = true;
int onlineLinkCount = 0;
string linkText = "";
foreach (string link in links)
{
RGInfo info = rg.RapidGatorCheckStatus(link);
if (!info.FileOnline)
{
AllOnline = false;
}
else
{
onlineLinkCount++;
}
string wlink = link + "|" + info.FileSize;
if (linkText != "")
{
linkText = linkText + "\n" + wlink;
}
if (linkText == "")
{
linkText = wlink;
}
}
string linksMsg = onlineLinkCount + "/" + links.Count + " Links Online";
// Extract / Display Images From Page
int downloaded = 0;
if (AllOnline)
{
string SaveDir = ahk.AppDir() + "\\Posts\\RGP\\" + ahk.AddLeadingZeros(PageNum, 5).ToString();
ahk.FileCreateDir(SaveDir);
ahk.Sleep(500);
ahk.FileAppend(Tags, SaveDir + "\\Tags.txt");
ahk.FileAppend(title, SaveDir + "\\Title.txt");
ahk.FileAppend(linkText, SaveDir + "\\Links.txt");
downloaded = DownloadImages(imagehtml, SaveDir, ahk.AddLeadingZeros(PageNum, 5).ToString());
//tel.Update(txtResults, "Title: " + title + "\n\nTags: " + Tags + "\n\n" + linksMsg);
}
//ahk.MsgBox("Found " + links.Count + " Links\nDownloaded " + downloaded.ToString() + " + Images");
return(AllOnline);
}
public List <TCP> Parse_TCPPage(string URL, string imageRoot = @"H:\SiteParse\TheClassic")
{
List <TCP> Films = new List <TCP>();
string xpath = "/html/body/div[3]/div/div[2]/div[4]/ul";
string html = ahk.Download_HTML(URL);
List <string> resultLines = xml.Parse_HTML_XML(html, xpath); // extract sections of text from html xml
foreach (string line in resultLines) // pull segments of page (20 total)
{
TCP Film = new TCP();
List <string> ImageLinks = new List <string>();
//string FilmURL = ""; string CoverURL = ""; string Title = ""; List<string> ImageLinks = new List<string>();
//string VidID = "";
string seg = line.Replace("\">", "\">\n");
List <string> newLines = lst.Text_To_List(seg); bool TitleFound = false;
foreach (string newline in newLines)
{
//FILM URL
string videoURLCheck = "thumb-video-link\" href=\"/videos";
if (newline.Contains(videoURLCheck))
{
string videoURL = newline.Replace("<a class=\"thumb-video-link\" href=\"", "");
videoURL = videoURL.Replace("/\">", "");
Film.FilmURL = "https://theclassicporn.com" + videoURL; // + @"\";
//ahk.MsgBox("FilmURL: " + FilmURL);
}
// VIDEO ID
if (newline.Contains("covers.jpg"))
{
List <string> items = ahk.StringSplit_List(newline, "/");
foreach (string item in items)
{
if (!item.Contains("covers.jpg"))
{
Film.VideoID = item;
}
if (item.Contains("covers.jpg"))
{
break;
}
}
//ahk.MsgBox(vidID);
}
// VIDEO COVER URL
if (newline.Contains("covers.jpg"))
{
List <string> items = ahk.StringSplit_List(newline, "\"");
foreach (string item in items)
{
if (item.Contains("covers.jpg"))
{
Film.CoverURL = item; break;
}
}
//ahk.MsgBox(CoverURL);
}
// VIDEO TITLE
if (newline.Contains("class=\"link-blue link-no-border\""))
{
TitleFound = true; continue;
}
if (TitleFound)
{
Film.FilmName = newline.Replace("</a>", "");
//ahk.MsgBox(Title);
TitleFound = false;
}
if (newline.Contains(".jpg") && !newline.Contains("covers.jpg"))
{
string ImageLine = newline.Replace("\" alt=\"\">", "");
ImageLine = ImageLine.Replace("<img class=\"screen-thumb\" src=\"", "");
//ahk.MsgBox(ImageLine);
ImageLinks.Add(ImageLine);
}
}
string links = "";
if (imageRoot != "")
{
string saveDir = imageRoot + "\\" + Film.VideoID;
ahk.FileCreateDir(saveDir);
// download film images to dirs sorted by VidID
foreach (string image in ImageLinks)
{
string fileName = ahk.StringSplit(image, "/", 0, true);
ahk.Download_File(image, saveDir + "\\" + fileName);
if (links == "")
{
links = image;
}
else
{
links = links + "\n" + image;
}
}
// download film cover
if (Film.CoverURL != null)
{
string filename = ahk.StringSplit(Film.CoverURL, "/", 0, true);
ahk.Download_File(Film.CoverURL, saveDir + "\\" + filename);
}
}
else
{
// add image links to string list
foreach (string image in ImageLinks)
{
if (links == "")
{
links = image;
}
else
{
links = links + "\n" + image;
}
}
}
Film.ImageLinks = links;
Film.ImgLinks = ImageLinks;
Films.Add(Film);
bool added = TCP_UpdateInsert(Film);
}
return(Films);
}
public void Download_ReleaseBB_MovieIndex(int StartPage = 1, int LastPage = 1366)
{
_Parse.XML xml = new _Parse.XML();
_Lists lst = new _Lists();
_Sites.Imdb imdb = new Imdb();
string startURL = "http://rlsbb.ru/category/movies/page/";
int LastPageNum = LastPage;
int pageNum = StartPage;
do
{
string URL = startURL + pageNum + "/";
string HTML = ahk.Download_HTML(URL);
string xmlPath = "//*[@id=\"contentArea\"]";
List <string> segs = xml.Parse_HTML_XML(HTML, xmlPath); // extract sections of text from html xml
int segNum = 1;
foreach (string seg in segs)
{
ReleaseBBMovies obj = new ReleaseBBMovies();
// skip segs on page that aren't posts
if (segNum == 1)
{
segNum++; continue;
}
if (segNum == 2)
{
segNum++; continue;
}
if (segNum == 13)
{
segNum++; continue;
}
if (segNum == 14)
{
segNum++; continue;
}
//ahk.MsgBox(segNum + " | " + seg);
segNum++;
string PostTitle = "";
string PostURL = "";
string IMDbID = "";
List <string> links = imdb.Regex_IMDbLinks(seg);
if (links.Count > 0)
{
IMDbID = imdb.IMDb_ID_FromURL(links[0]);
}
// parse index page segment
List <string> lines = lst.Text_To_List(seg, true, true, false);
foreach (string line in lines)
{
if (line.Contains("postTitle"))
{
PostURL = line.Replace("<h2 class=\"postTitle\"><span></span><a href=\"", "");
PostTitle = ahk.StringSplit(PostURL, ">", 1);
PostTitle = ahk.StringSplit(PostTitle, "<", 0);
PostURL = ahk.StringSplit(PostURL, "\"", 0);
//ahk.MsgBox("IMDB: " + IMDbID + "\n\n" + PostTitle + "\n\n" + PostURL);
}
}
obj.IMDbID = IMDbID;
obj.PostTitle = PostTitle;
obj.PostURL = PostURL;
bool Updated = ReleaseBBMovies_UpdateInsert_IndexEntry(obj);
}
pageNum++;
}while (pageNum <= LastPageNum);
}
