static void ProcessEqualClustering(string[] args)
{
var iss = new ImageEqualSimilarity();
Console.Instance.WriteLine("이미지들을 해싱하는 중...");
using (var progressBar = new Console.ConsoleProgressBar())
{
var files = Directory.GetFiles(args[0]).Where(x => x.EndsWith(".gif") || x.EndsWith(".png") || x.EndsWith(".jpg") || x.EndsWith(".jpeg") || x.EndsWith(".webp") || x.EndsWith(".bmp"));
int counts = files.Count();
int complete = 0;
Parallel.ForEach(files,
x =>
{
iss.AppendImage(x);
progressBar.SetProgress(Interlocked.Increment(ref complete) / (float)counts * 100);
});
}
var clustered = iss.Clustering();
clustered.RemoveAll(x => x.Count == 1);
clustered.Sort((x, y) => y.Count.CompareTo(x.Count));
Console.Instance.WriteLine(clustered);
}
static void ProcessCollectArticles(string[] args)
{
var rstarts = Convert.ToInt32(args[1]);
var starts = Convert.ToInt32(args[1]);
var ends = Convert.ToInt32(args[2]);
bool is_minorg = !DCGalleryList.Instance.GalleryIds.Contains(args[0]);
var result = new DCInsideGalleryModel();
var articles = new List <DCInsidePageArticle>();
using (var progressBar = new Console.ConsoleProgressBar())
{
for (; starts <= ends; starts++)
{
var url = "";
if (is_minorg)
{
url = $"https://gall.dcinside.com/mgallery/board/lists/?id={args[0]}&page={starts}";
}
else
{
url = $"https://gall.dcinside.com/board/lists/?id={args[0]}&page={starts}";
}
Console.Instance.WriteLine($"Download URL: {url}");
var html = NetTools.DownloadString(url);
DCInsideGallery gall = null;
if (is_minorg)
{
gall = DCInsideUtils.ParseMinorGallery(html);
}
else
{
gall = DCInsideUtils.ParseGallery(html);
}
if (is_minorg && (gall.articles == null || gall.articles.Length == 0))
{
gall = DCInsideUtils.ParseGallery(html);
}
articles.AddRange(gall.articles);
progressBar.SetProgress((((ends - rstarts + 1) - (ends - starts)) / (float)(ends - rstarts + 1)) * 100);
}
var overlap = new HashSet <string>();
var articles_trim = new List <DCInsidePageArticle>();
foreach (var article in articles)
{
if (!overlap.Contains(article.no))
{
articles_trim.Add(article);
overlap.Add(article.no);
}
}
articles_trim.Sort((x, y) => y.no.ToInt().CompareTo(x.no.ToInt()));
result.is_minor_gallery = is_minorg;
result.gallery_id = args[0];
result.articles = articles_trim;
File.WriteAllText($"list-{args[0]}-{DateTime.Now.Ticks}.txt", JsonConvert.SerializeObject(result));
var bbb = MessagePackSerializer.Serialize(result);
using (FileStream fsStream = new FileStream($"list-{args[0]}-{DateTime.Now.Ticks}-index.txt", FileMode.Create))
using (BinaryWriter sw = new BinaryWriter(fsStream))
{
sw.Write(bbb);
}
}
}
