public override (List <NetTask>, ExtractedInfo) Extract(string url, IExtractorOption option = null)
{
var match = ValidUrl.Match(url).Groups;
if (option == null)
{
option = RecommendOption(url);
}
if (option.Type == ExtractorType.Images)
{
var id = match["id"].Value;
// Handle Redirect
var string3 = NetTools.DownloadString(url);
RETRY_DOWNLOAD1:
if (string.IsNullOrEmpty(string3))
{
return(null, null);
}
if (string3.ToHtmlNode().SelectSingleNode("//title").InnerText == "Redirect")
{
id = string3.ToHtmlNode().SelectSingleNode("//a").GetAttributeValue("href", "").Split('/', '-').Last().Split('.')[0];
string3 = NetTools.DownloadString(string3.ToHtmlNode().SelectSingleNode("//a").GetAttributeValue("href", ""));
goto RETRY_DOWNLOAD1;
}
var sinfo = new ExtractedInfo.WorksComic();
var imgs_url = $"https://ltn.hitomi.la/galleries/{id}.js";
option.PageReadCallback?.Invoke($"https://ltn.hitomi.la/galleryblock/{id}.html");
option.PageReadCallback?.Invoke(url);
option.PageReadCallback?.Invoke(imgs_url);
var urls = new List <string> {
$"https://ltn.hitomi.la/galleryblock/{id}.html",
imgs_url
};
var strings = NetTools.DownloadStrings(urls);
if (string.IsNullOrEmpty(strings[0]) || string.IsNullOrEmpty(strings[1]))
{
return(null, null);
}
var data1 = ParseGalleryBlock(strings[0]);
var imgs = strings[1];
var string2 = NetTools.DownloadString($"https://hitomi.la{data1.Magic}");
if (string.IsNullOrEmpty(string2))
{
return(null, null);
}
var data2 = ParseGallery(string2);
option.SimpleInfoCallback?.Invoke($"[{id}] {data1.Title}");
// download.js
var number_of_frontends = 3;
var subdomain = Convert.ToChar(97 + (Convert.ToInt32(id.Last()) % number_of_frontends));
if (id.Last() == '0')
{
subdomain = 'a';
}
var arr = JToken.Parse(imgs.Substring(imgs.IndexOf('=') + 1))["files"];
var img_urls = new List <string>();
foreach (var obj in (JArray)arr)
{
var hash = obj.Value <string>("hash");
if (obj.Value <int>("haswebp") == 0 || hash == null)
{
//img_urls.Add($"https://{subdomain}a.hitomi.la/galleries/{id}/{obj.Value<string>("name")}");
var postfix = hash.Substring(hash.Length - 3);
img_urls.Add($"https://{subdomain}a.hitomi.la/images/{postfix[2]}/{postfix[0]}{postfix[1]}/{hash}.{obj.Value<string>("name").Split('.').Last()}");
}
else if (hash == "")
{
img_urls.Add($"https://{subdomain}a.hitomi.la/webp/{obj.Value<string>("name")}.webp");
}
else if (hash.Length < 3)
{
img_urls.Add($"https://{subdomain}a.hitomi.la/webp/{hash}.webp");
}
else
{
var postfix = hash.Substring(hash.Length - 3);
img_urls.Add($"https://{subdomain}a.hitomi.la/webp/{postfix[2]}/{postfix[0]}{postfix[1]}/{hash}.webp");
}
}
var result = new List <NetTask>();
var ordering = 1;
foreach (var img in img_urls)
{
var filename = Path.GetFileNameWithoutExtension(img.Split('/').Last());
if (!(option as HitomiExtractorOption).RealFilename)
{
filename = ordering++.ToString("000");
}
var task = NetTask.MakeDefault(img);
task.SaveFile = true;
task.Filename = img.Split('/').Last();
task.Format = new ExtractorFileNameFormat
{
Title = data1.Title,
Id = id,
Language = data1.Language,
UploadDate = data1.Posted,
FilenameWithoutExtension = filename,
Extension = Path.GetExtension(img.Split('/').Last()).Replace(".", "")
};
if (data1.artist != null)
{
task.Format.Artist = data1.artist[0];
}
else
{
task.Format.Artist = "NA";
}
if (data1.parody != null)
{
task.Format.Series = data1.parody[0];
}
else
{
task.Format.Series = "NA";
}
if (data2.group != null)
{
task.Format.Group = data2.group[0];
}
else
{
task.Format.Group = "NA";
}
if (data2.character != null)
{
task.Format.Character = data2.character[0];
}
else
{
task.Format.Character = "NA";
}
if (task.Format.Artist == "NA" && task.Format.Group != "NA")
{
task.Format.Artist = task.Format.Group;
}
result.Add(task);
}
option.ThumbnailCallback?.Invoke(result[0]);
sinfo.Thumbnail = result[0];
sinfo.URL = url;
sinfo.Title = data1.Title;
sinfo.Author = data1.artist?.ToArray();
sinfo.AuthorGroup = data2.group?.ToArray();
sinfo.ShortInfo = $"[{id}] {data1.Title}";
sinfo.Tags = data1.Tags?.ToArray();
sinfo.Characters = data2.character?.ToArray();
sinfo.Language = data1.Language;
sinfo.Parodies = data1.parody?.ToArray();
result.ForEach(task => task.Format.Extractor = GetType().Name.Replace("Extractor", ""));
return(result, new ExtractedInfo {
Info = sinfo, Type = ExtractedInfo.ExtractedType.WorksComic
});
}
return(null, null);
}
public override (List <NetTask>, ExtractedInfo) Extract(string url, IExtractorOption option = null)
{
var match = ValidUrl.Match(url).Groups;
if (option == null)
{
option = RecommendOption(url);
}
if (option.Type == ExtractorType.Images)
{
var sinfo = new ExtractedInfo.WorksComic();
var imgs_url = $"https://ltn.hitomi.la/galleries/{match["id"].Value}.js";
option.PageReadCallback?.Invoke($"https://ltn.hitomi.la/galleryblock/{match["id"]}.html");
option.PageReadCallback?.Invoke(url);
option.PageReadCallback?.Invoke(imgs_url);
var urls = new List <string> {
$"https://ltn.hitomi.la/galleryblock/{match["id"]}.html",
url,
imgs_url
};
var strings = NetTools.DownloadStrings(urls);
if (string.IsNullOrEmpty(strings[0]) || string.IsNullOrEmpty(strings[1]) || string.IsNullOrEmpty(strings[2]))
{
return(null, null);
}
var data1 = ParseGalleryBlock(strings[0]);
var data2 = ParseGallery(strings[1]);
var imgs = strings[2];
option.SimpleInfoCallback?.Invoke($"[{data1.Magic}] {data1.Title}");
// download.js
var number_of_frontends = 3;
var subdomain = Convert.ToChar(97 + (Convert.ToInt32(match["id"].Value.Last()) % number_of_frontends));
if (match["id"].Value.Last() == '0')
{
subdomain = 'a';
}
var arr = JArray.Parse(imgs.Substring(imgs.IndexOf('[')));
var img_urls = new List <string>();
foreach (var obj in arr)
{
if (obj.Value <int>("haswebp") == 0)
{
img_urls.Add($"https://{subdomain}a.hitomi.la/galleries/{match["id"].Value}/{obj.Value<string>("name")}");
}
else
{
img_urls.Add($"https://{subdomain}a.hitomi.la/webp/{match["id"].Value}/{obj.Value<string>("name")}.webp");
}
}
var result = new List <NetTask>();
foreach (var img in img_urls)
{
var task = NetTask.MakeDefault(img);
task.SaveFile = true;
task.Filename = img.Split('/').Last();
task.Format = new ExtractorFileNameFormat
{
Title = data1.Title,
Id = data1.Magic,
Language = data1.Language,
UploadDate = data1.Posted,
FilenameWithoutExtension = Path.GetFileNameWithoutExtension(img.Split('/').Last()),
Extension = Path.GetExtension(img.Split('/').Last()).Replace(".", "")
};
if (data1.artist != null)
{
task.Format.Artist = data1.artist[0];
}
else
{
task.Format.Artist = "N/A";
}
if (data1.parody != null)
{
task.Format.Series = data1.parody[0];
}
else
{
task.Format.Series = "N/A";
}
if (data2.group != null)
{
task.Format.Group = data2.group[0];
}
else
{
task.Format.Group = "N/A";
}
if (data2.character != null)
{
task.Format.Character = data2.character[0];
}
else
{
task.Format.Character = "N/A";
}
if (task.Format.Artist == "N/A" && task.Format.Group != "N/A")
{
task.Format.Artist = task.Format.Group;
}
result.Add(task);
}
option.ThumbnailCallback?.Invoke(result[0]);
sinfo.Thumbnail = result[0];
sinfo.URL = url;
sinfo.Title = data1.Title;
sinfo.Author = data1.artist?.ToArray();
sinfo.AuthorGroup = data2.group?.ToArray();
sinfo.ShortInfo = $"[{data1.Magic}] {data1.Title}";
sinfo.Tags = data1.Tags?.ToArray();
sinfo.Characters = data2.character?.ToArray();
sinfo.Language = data1.Language;
sinfo.Parodies = data1.parody?.ToArray();
result.ForEach(task => task.Format.Extractor = GetType().Name.Replace("Extractor", ""));
return(result, new ExtractedInfo {
Info = sinfo, Type = ExtractedInfo.ExtractedType.WorksComic
});
}
return(null, null);
}