static void Main(string[] args)
{
string url = @"http://www.1pondo.tv/static-seo/sitemap-videos.xml";
UserAgent agent = new UserAgent("hoge", 1, "*****@*****.**", DefaultAgents.Firefox14);
var reader = new HtmlReader(url, true, agent);
var xml = reader.Html;
XNamespace ns = reader.Namespace;
XNamespace video = reader.Html.Root.Attribute(XNamespace.Xmlns + "video").Value;
var list = xml.Descendants(video + "content_loc").Where(a => a.Value.Contains("092615_161"));
var item = list.First().Ancestors(ns + "url");
var vitem = list.First().Ancestors(video + "video");
}
public override TagInfo GetItem(string searchWord)
{
try {
TagInfo info = TagInfo.New();
HtmlReader reader = new HtmlReader(GetPageUrl(searchWord));
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
// タイトル
var title = xml.Descendant(ns + "h1", new XAttribute("class", "item fn"));
info.Title = (title.HasElements) ? (title.LastNode as XElement).Value.Trim() : title.Value.Trim();
// キャスト
var casts = xml.Descendant(ns + "span", new XAttribute("id", "performer"));
info.Casts = (xml.ToString().Contains("出演者：")) && casts != null
? casts.Descendants(ns + "a").Select(a => a.Value).ToArray()
: new string[0];
// シリーズ
info.Series = (xml.ToString().Contains("シリーズ："))
? xml.Descendant(ns + "td", "シリーズ：").NextElement().Value
: string.Empty;
// メーカー
if (xml.ToString().Contains("メーカー：")) {
info.Makers = xml.Descendant(ns + "td", "メーカー：").NextElement().Descendants(ns + "a").Select(a => a.Value).ToArray();
} else if (xml.ToString().Contains("サークル名：")) {
info.Makers = xml.Descendant(ns + "td", "サークル名：").NextElement().Descendants(ns + "a").Select(a => a.Value).ToArray();
} else {
info.Makers = new string[0];
}
// レーベル
info.Labels = (xml.ToString().Contains("レーベル："))
? xml.Descendant(ns + "td", "レーベル：").NextElement().Descendants(ns + "a").Select(a => a.Value).ToArray()
: new string[0];
// ジャンル
info.Genres = (xml.ToString().Contains("ジャンル："))
? xml.Descendant(ns + "td", "ジャンル：").NextElement().Descendants(ns + "a").Select(a => a.Value).ToArray()
: new string[0];
// コメント
var comment = xml.Descendant(ns + "p", new XAttribute("class", "mg-b20"));
info.Comment = (comment != null) ? comment.Value.Replace("\n", "").Replace("\r", "") : string.Empty;
if (xml.ToString().Contains("品番：")) {
// CenAvId
info.CenAvId = xml.Descendant(ns + "td", "品番：").NextElement().Value;
// Number
info.Number = Convert.ToUInt32(Regex.Match(info.CenAvId, @".*[a-zA-Z]+([0-9]+)[a-zA-Z]*$").Groups[1].Value);
}
// イメージ
var image = xml.Descendant(ns + "a", new XAttribute("name", "package-image"));
string imageUrl = (image != null)
? image.Attribute("href").Value
: xml.Descendant(ns + "img", new XAttribute("class", "tdmm")).Attribute("src").Value;
using (WebClient client = new WebClient()) {
using (MemoryStream stream = new MemoryStream(client.DownloadData(imageUrl))) {
info.Pictures = new IPicture[]{
new Picture() {
Type = PictureType.FrontCover,
MimeType = "image/jpeg",
Description = "Cover",
Data = ByteVector.FromStream(stream)
}
};
}
}
info.Images = ((Func<BitmapSource[]>)(() => {
ImageConverter ic = new ImageConverter();
return info.Pictures.Select(
p => (ic.ConvertFrom(p.Data.Data) as Bitmap).ToWPFBitmap()).ToArray();
}))();
info.IsEmpty = false;
return info;
} catch (Exception ex) {
throw new ApplicationException("アイテム要素が見つかりませんでした。", ex);
}
}
private string GetPageUrl(string searchWord)
{
string url = @"http://www.dmm.co.jp/search/=/searchstr=" + searchWord;
try {
HtmlReader reader = new HtmlReader(url);
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
var ditem = xml.Descendant(ns + "div", new XAttribute("class", "d-item"))
.Descendants(ns + "li")
.Where(a => !a.ToString().Contains("【数量限定】"))
.Where(a => !a.ToString().Contains("【アウトレット】"))
.Where(a => !a.ToString().Contains("【DMM限定】"))
.Where(a => !a.ToString().Contains("Blu-ray"))
.Where(a => !a.ToString().Contains("近日"))
.Where(a => !a.ToString().Contains("10円動画"))
.First();
return ditem.Descendants(ns + "a", "href").First().Attribute("href").Value;
} catch (Exception ex) {
throw new ApplicationException("ページが見つかりませんでした。", ex);
}
}
public override TagInfo GetItem(string searchWord)
{
try {
TagInfo info = TagInfo.New();
HtmlReader reader = new HtmlReader("http://www.heyzo.com/moviepages/" + searchWord + "/index.html");
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
var movie = xml.Descendant(ns + "div", new XAttribute("id", "movie"));
// タイトル
info.Title = string.Join(" - ", movie.Descendants("h1").First().Value.Split('-').Select(a => a.Trim()));
// キャスト
XElement cast = movie.Descendant(ns + "span", new XAttribute("class", "actor")).NextElement();
info.Casts = (cast != null) ? cast.Descendants(ns + "a").Select(a => a.Value).ToArray() : new string[0];
// シリーズ
XElement series = movie.Descendant(ns + "span", new XAttribute("class", "label")).NextElement();
info.Series = (series != null) ? series.Value.Trim() : string.Empty;
// メーカー
info.Makers = new string[] { "HEYZO" };
// レーベル
info.Labels = new string[0];
// ジャンル
XElement genres = movie.Descendant(ns + "div", new XAttribute("class", "tagkeyword"));
info.Genres = (genres != null) ? genres.Descendants(ns + "li").Select(a => a.Value).ToArray() : new string[0];
// コメント
XElement comment = movie.Descendant(ns + "p", new XAttribute("class", "memo"));
info.Comment = (comment != null) ? comment.FirstNode.ToString() : string.Empty;
// CenAvId
info.CenAvId = "HEYZO" + searchWord;
// Number
info.Number = Convert.ToUInt32(searchWord);
// イメージ
string imageUrl = "http://www.heyzo.com/contents/3000/" + searchWord + "/images/player_thumbnail.jpg";
using (WebClient client = new WebClient()) {
using (MemoryStream stream = new MemoryStream(client.DownloadData(imageUrl))) {
info.Pictures = new IPicture[]{
new Picture() {
Type = PictureType.FrontCover,
MimeType = "image/jpeg",
Description = "Cover",
Data = ByteVector.FromStream(stream)
}
};
}
}
info.Images = ((Func<BitmapSource[]>)(() => {
ImageConverter ic = new ImageConverter();
return info.Pictures.Select(
p => (ic.ConvertFrom(p.Data.Data) as Bitmap).ToWPFBitmap()).ToArray();
}))();
info.IsEmpty = false;
return info;
} catch (Exception ex) {
throw new ApplicationException("アイテム要素が見つかりませんでした。", ex);
}
}
private string GetPageUrl(string searchWord)
{
string url = @"http://www.aventertainments.com/search_Products.aspx?languageID=2&dept_id=29&keyword=" + searchWord + "&searchby=keyword";
try {
HtmlReader reader = new HtmlReader(url, agent: ua, encoding: Encoding.UTF8);
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
var table = xml.Descendant(ns + "table", new XAttribute("id", "ctl00_ContentPlaceHolder1_Rows2Items1_MyList"));
return table.Descendants(ns + "a", "href").First().Attribute("href").Value;
} catch (Exception ex) {
throw new ApplicationException("ページが見つかりませんでした。", ex);
}
}
public override TagInfo GetItem(string searchWord)
{
try {
TagInfo info = TagInfo.New();
HtmlReader reader = new HtmlReader(GetPageUrl(searchWord), agent: ua, encoding: Encoding.UTF8);
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
var sub = xml.Descendant(ns + "div", new XAttribute("class", "main-subcontent-page"));
// タイトル
info.Title = sub.Descendant(ns + "li", new XAttribute("class", "title")).Value;
// キャスト
var casts = sub.Descendant(ns + "span", "主演女優: ").Parent;
info.Casts = casts.Descendants(ns + "a").Select(a => a.Value).ToArray();
// シリーズ
info.Series = (sub.ToString().Contains("シリーズ:"))
? sub.Descendant(ns + "span", "シリーズ:").Parent.Element(ns + "a").Value
: string.Empty;
// メーカー
var makers = sub.Descendant(ns + "span", "スタジオ:").Parent;
info.Makers = makers.Descendants(ns + "a").Select(a => a.Value).ToArray();
// レーベル
info.Labels = new string[0];
// ジャンル
var genres = sub.Descendant(ns + "span", "カテゴリ一覧:").Parent;
info.Genres = genres.Descendants(ns + "a").Select(a => a.Value).ToArray();
// コメント
var comment = sub.Descendant(ns + "div", new XAttribute("class", "title2")).NextElement();
info.Comment = comment.Value;
// CenAvId
info.CenAvId = xml.Descendant(ns + "div", new XAttribute("class", "top-title")).Value.Replace("商品番号:", "").Trim();
// Number
info.Number = Convert.ToUInt32(Regex.Match(info.CenAvId, @"[0-9]+").Value);
// イメージ
var imageUrl = sub.Descendants(ns + "img").First().Attribute("src").Value.Replace("jacket_images", "bigcover");
using (WebClient client = new WebClient()) {
using (MemoryStream stream = new MemoryStream(client.DownloadData(imageUrl))) {
info.Pictures = new IPicture[]{
new Picture() {
Type = PictureType.FrontCover,
MimeType = "image/jpeg",
Description = "Cover",
Data = ByteVector.FromStream(stream)
}
};
}
}
info.Images = ((Func<BitmapSource[]>)(() => {
ImageConverter ic = new ImageConverter();
return info.Pictures.Select(
p => (ic.ConvertFrom(p.Data.Data) as Bitmap).ToWPFBitmap()).ToArray();
}))();
info.IsEmpty = false;
return info;
} catch (Exception ex) {
throw new ApplicationException("アイテム要素が見つかりませんでした。", ex);
}
}
public override TagInfo GetItem(string searchWord)
{
try {
TagInfo info = TagInfo.New();
HtmlReader reader = new HtmlReader(GetPageUrl(searchWord), encoding: enc);
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
var content = xml.Descendant(ns + "div", new XAttribute("class", "main-content-movieinfo"));
var id = new Regex("[0-9]{6}_[0-9]{3}").Match(reader.Uri).Value;
// タイトル
info.Title = content.Descendant(ns + "div", new XAttribute("class", "video-detail")).Value.Trim();
// キャスト
XElement cast = content.Descendants(ns + "dt").Where(a => a.Value == "出演:").FirstOrDefault();
info.Casts = (cast != null) ? cast.Descendants(ns + "a").Select(a => a.Value).ToArray() : new string[0];
// シリーズ
XElement series = content.Descendants(ns + "dt").Where(a => a.Value == "シリーズ:").FirstOrDefault();
info.Series = (series != null) ? series.NextElement().Value.Trim() : string.Empty;
// メーカー
info.Makers = new string[] { "カリビアンコム プレミアム" };
// レーベル
info.Labels = new string[0];
// ジャンル
XElement genres = content.Descendant(ns + "dl", new XAttribute("class", "movie-info-cat"));
info.Genres = (genres != null)
? genres.Descendants(ns + "a").Select(a => a.Value).ToArray()
: new string[0];
// コメント
XElement comment = content.Descendant(ns + "div", new XAttribute("class", "movie-comment"));
info.Comment = (comment != null) ? comment.FirstNode.ToString() : string.Empty;
// CenAvId
info.CenAvId = "CaribbeanPr-" + id;
// Number
info.Number = Convert.ToUInt32(id.Replace("_", ""));
// イメージ
string imageUrl = @"http://www.caribbeancompr.com/moviepages/" + id + "/images/l_l.jpg";
using (WebClient client = new WebClient()) {
using (MemoryStream stream = new MemoryStream(client.DownloadData(imageUrl))) {
info.Pictures = new IPicture[]{
new Picture() {
Type = PictureType.FrontCover,
MimeType = "image/jpeg",
Description = "Cover",
Data = ByteVector.FromStream(stream)
}
};
}
}
info.Images = ((Func<BitmapSource[]>)(() => {
ImageConverter ic = new ImageConverter();
return info.Pictures.Select(
p => (ic.ConvertFrom(p.Data.Data) as Bitmap).ToWPFBitmap()).ToArray();
}))();
info.IsEmpty = false;
return info;
} catch (Exception ex) {
throw new ApplicationException("アイテム要素が見つかりませんでした。", ex);
}
}
private string GetPageUrl(string searchWord)
{
if (Regex.IsMatch(searchWord, "[0-9]{6}_[0-9]{3}")) {
return @"http://www.caribbeancompr.com/moviepages/" + searchWord + "/index.html";
} else {
string url = @"http://www.caribbeancompr.com/search/?q=" + HttpUtility.UrlEncode(searchWord, enc);
try {
HtmlReader reader = new HtmlReader(url, encoding: enc);
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
var contents = xml.Descendant(ns + "div", new XAttribute("id", "main-content"));
return @"http://www.caribbeancompr.com" + contents.Descendants(ns + "a").First().Attribute("href").Value;
} catch (Exception ex) {
throw new ApplicationException("ページが見つかりませんでした。", ex);
}
}
}
public override TagInfo GetItem(string searchWord)
{
try {
TagInfo info = TagInfo.New();
if (cache == null)
cache = new HtmlReader("http://www.1pondo.tv/static-seo/sitemap-videos.xml");
HtmlReader reader = cache;
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
XNamespace video = reader.Html.Root.Attribute(XNamespace.Xmlns + "video").Value;
XElement target = (Regex.IsMatch(searchWord, "[0-9]{6}_[0-9]{3}"))
? xml.Descendants(video + "content_loc").First(a => a.Value.Contains(searchWord)).Ancestors(video + "video").First()
: xml.Descendants(video + "title").First(a => a.Value.Contains(searchWord)).Ancestors(video + "video").First();
var id = new Regex("[0-9]{6}_[0-9]{3}").Match(target.Element(video +"thumbnail_loc").Value).Value;
string cast = target.Element(video + "tag").Value;
// タイトル
info.Title = target.Element(video + "title").Value.Split('|')[0].Substring(cast.Length).Trim();
// キャスト
info.Casts = (cast != null) ? cast.Split(',') : new string[0];
// シリーズ
info.Series = string.Empty;
// メーカー
info.Makers = new string[] { "一本道" };
// レーベル
info.Labels = new string[0];
// ジャンル
info.Genres = target.Descendants(video + "tag").Skip(1).Select(a => a.Value).ToArray();
// コメント
info.Comment = target.Element(video + "description").Value;
// CenAvId
info.CenAvId = "1Pondo-" + id;
// Number
info.Number = Convert.ToUInt32(id.Replace("_", ""));
// イメージ
string imageUrl = target.Element(video + "thumbnail_loc").Value;
using (WebClient client = new WebClient()) {
using (MemoryStream stream = new MemoryStream(client.DownloadData(imageUrl))) {
info.Pictures = new IPicture[]{
new Picture() {
Type = PictureType.FrontCover,
MimeType = "image/jpeg",
Description = "Cover",
Data = ByteVector.FromStream(stream)
}
};
}
}
info.Images = ((Func<BitmapSource[]>)(() => {
ImageConverter ic = new ImageConverter();
return info.Pictures.Select(
p => (ic.ConvertFrom(p.Data.Data) as Bitmap).ToWPFBitmap()).ToArray();
}))();
info.IsEmpty = false;
return info;
} catch (Exception ex) {
throw new ApplicationException("アイテム要素が見つかりませんでした。", ex);
}
}
public override TagInfo GetItem(string searchWord)
{
try {
TagInfo info = TagInfo.New();
string url = @"http://www.xxx-av.com/mov/view/" + searchWord + ".html";
HtmlReader reader = new HtmlReader(url, true, ua);
XDocument xml = reader.Html;
XNamespace ns = reader.Namespace;
var movie = xml.Descendant(ns + "div", new XAttribute("id", "movie"));
// タイトル
info.Title = movie.Descendant(ns + "div", new XAttribute("class", "movie_tt")).Value.Trim();
// キャスト
XElement cast = movie.Descendant(ns + "p", new XAttribute("class", "act"));
info.Casts = (cast != null) ? cast.Descendants(ns + "a").Select(a => a.Value).ToArray() : new string[0];
// シリーズ
XElement series = movie.Descendant(ns + "dl", new XAttribute("class", "series"));
info.Series = (series != null) ? series.Descendants(ns + "a").Select(a => a.Value).FirstOrDefault() : string.Empty;
// メーカー
info.Makers = new string[] { "トリプルエックス" };
// レーベル
info.Labels = new string[0];
// ジャンル
XElement genres = movie.Descendant(ns + "dl", new XAttribute("class", "word"));
info.Genres = (genres != null) ? genres.Descendants(ns + "a").Select(a => a.Value).ToArray() : new string[0];
var data = xml.Descendant(ns + "div", new XAttribute("class", "movie_data"));
// コメント
XElement comment = data.Descendant(ns + "dt", "コメント").NextElement();
info.Comment = (comment != null) ? comment.Value : string.Empty;
// CenAvId
info.CenAvId = "XXX-" + searchWord;
// Number
info.Number = Convert.ToUInt32(searchWord);
// イメージ
string imageUrl = "http://image.xxx-av.com/image/" + searchWord + "/movie_main.jpg";
using (WebClient client = new WebClient()) {
using (MemoryStream stream = new MemoryStream(client.DownloadData(imageUrl))) {
info.Pictures = new IPicture[]{
new Picture() {
Type = PictureType.FrontCover,
MimeType = "image/jpeg",
Description = "Cover",
Data = ByteVector.FromStream(stream)
}
};
}
}
info.Images = ((Func<BitmapSource[]>)(() => {
ImageConverter ic = new ImageConverter();
return info.Pictures.Select(
p => (ic.ConvertFrom(p.Data.Data) as Bitmap).ToWPFBitmap()).ToArray();
}))();
info.IsEmpty = false;
return info;
} catch (Exception ex) {
throw new ApplicationException("アイテム要素が見つかりませんでした。", ex);
}
}
