private static bool IsBlacklisted(DanbooruPost post, Regex regex)
{
if (regex != null)
{
return(post.TagsEntity.Any(x => regex.IsMatch(x.Name)));
}
return(false);
}
private void FinalizeDanbooru(DanbooruPost post)
{
try
{
post.MediaType = GetBooruMediaType(post.FullImageUrl);
}
catch (Exception e)
{
}
}
private static bool IsBlacklisted(DanbooruPost post, List <DanbooruTag> TagBlackList)
{
if (TagBlackList != null)
{
foreach (var tag in TagBlackList)
{
return(post.TagsEntity.Any(x => x.Name.Equals(tag.Name, StringComparison.InvariantCultureIgnoreCase)));
}
}
return(false);
}
private static string FilterTags(DanbooruPost post,
List <DanbooruTag> groupedTags,
DanbooruTagType tagType,
int tagLimit,
string tagReplacement,
string missingTagReplacement,
bool isReplaceMode,
bool isReplaceUnderScore)
{
var selectedTags = post.TagsEntity.Where <DanbooruTag>(x => x.Type == tagType).Select(x => x.Name).ToList();
var tagStr = "";
if (selectedTags != null)
{
if (isReplaceUnderScore)
{
for (int i = 0; i < selectedTags.Count; i++)
{
selectedTags[i] = selectedTags[i].Replace("_", " ").Trim();
}
}
if (selectedTags.Count() >= tagLimit)
{
if (isReplaceMode && !string.IsNullOrWhiteSpace(tagReplacement))
{
tagStr = tagReplacement;
if (isReplaceUnderScore)
{
tagStr = tagReplacement.Replace("_", " ").Trim();
}
groupedTags.RemoveAll(x => x.Type == tagType);
groupedTags.Add(new DanbooruTag()
{
Name = tagReplacement
});
}
else
{
var tempTags = selectedTags.Take(tagLimit);
tagStr = String.Join(" ", tempTags);
}
}
else
{
tagStr = string.Join(" ", selectedTags);
}
}
if (string.IsNullOrWhiteSpace(tagStr))
{
tagStr = missingTagReplacement;
}
return(tagStr);
}
private static void ReadRssMethod2(DanbooruPostDaoOption option, BindingList <DanbooruPost> posts, string xmldoc)
{
XmlReaderSettings readerSettings = new XmlReaderSettings
{
DtdProcessing = DtdProcessing.Parse,
XmlResolver = new LocalXhtmlXmlResolver()
};
using (StringReader strReader = new StringReader(xmldoc))
{
using (XmlReader reader = XmlReader.Create(strReader, readerSettings))
{
XDocument doc = XDocument.Load(reader);
var feeds = doc.Descendants("item");
XNamespace dc = "http://purl.org/dc/elements/1.1/";
foreach (var item in feeds)
{
DanbooruPost post = new DanbooruPost();
var titleData = item.Element("title").Value.Split(new char[] { '-' }, 2);
post.Id = titleData[0].Trim();
post.Tags = titleData[1].Trim();
post.Referer = AppendHttp(item.Element("link").Value, option.Provider);
post.CreatedAt = item.Element("pubDate").Value;
post.CreatorId = item.Element(dc + "creator").Value;
post.FileUrl = item.Element("enclosure").Attribute("url").Value;
try
{
HtmlAgilityPack.HtmlDocument description = new HtmlAgilityPack.HtmlDocument();
description.LoadHtml(HttpUtility.HtmlDecode(item.Element("description").Value));
//post.Width = Convert.ToInt32(matches.Groups[1].Value);
//post.Height = Convert.ToInt32(matches.Groups[2].Value);
var img = description.DocumentNode.SelectSingleNode("//img");
post.PreviewWidth = Convert.ToInt32(img.Attributes["width"].Value);
post.PreviewHeight = Convert.ToInt32(img.Attributes["height"].Value);
post.PreviewUrl = img.Attributes["src"].Value;
}
catch (Exception) { }
PostProcess(option, post);
posts.Add(post);
}
}
}
}
private static void ReparseTags(DanbooruPost post, HtmlDocument doc)
{
post.TagsEntity.Clear();
var tagsElement = doc.DocumentNode.SelectNodes("//ul[@id='tag-sidebar']/li");
foreach (var tag in tagsElement)
{
HtmlDocument tagDoc = new HtmlDocument();
tagDoc.LoadHtml(tag.OuterHtml);
var tagEntity = new DanbooruTag();
var el = tagDoc.DocumentNode.SelectSingleNode("//li");
var cls = el.Attributes["class"].Value;
switch (cls)
{
case "tag-type-general":
tagEntity.Type = DanbooruTagType.General;
break;
case "tag-type-artist":
tagEntity.Type = DanbooruTagType.Artist;
break;
case "tag-type-copyright":
tagEntity.Type = DanbooruTagType.Copyright;
break;
case "tag-type-character":
tagEntity.Type = DanbooruTagType.Character;
break;
//case "tag-type-medium":
// tagEntity.Type = DanbooruTagType.Circle;
// break;
case "tag-type-medium":
tagEntity.Type = DanbooruTagType.Faults;
break;
default:
tagEntity.Type = DanbooruTagType.Unknown;
break;
}
tagEntity.Name = tagDoc.DocumentNode.SelectSingleNode("//li/a").InnerText;
var countStr = tagDoc.DocumentNode.SelectSingleNode("//li//span[@class='post-count']").InnerText.Trim();
tagEntity.Count = Int32.Parse(countStr);
post.TagsEntity.Add(tagEntity);
}
}
private static void ReadRssMethod1(DanbooruPostDaoOption option, BindingList <DanbooruPost> posts, string xmldoc)
{
XmlReaderSettings readerSettings = new XmlReaderSettings
{
DtdProcessing = DtdProcessing.Parse,
XmlResolver = new LocalXhtmlXmlResolver()
};
using (StringReader strReader = new StringReader(xmldoc))
{
using (XmlReader reader = XmlReader.Create(strReader, readerSettings))
{
XDocument doc = XDocument.Load(reader);
string media = doc.Root.Attribute("{http://www.w3.org/2000/xmlns/}media").Value;
foreach (var item in doc.Descendants("item"))
{
DanbooruPost post = new DanbooruPost();
var titleData = item.Element("title").Value.Split(new char[] { '-' }, 2);
post.Id = titleData[0].Trim();
post.Tags = titleData[1].Trim();
post.Referer = AppendHttp(item.Element("link").Value, option.Provider);
post.CreatedAt = item.Element("pubDate").Value;
var data = item.Element("{" + media + "}thumbnail");
post.PreviewUrl = AppendHttp(data.Attribute("url").Value, option.Provider);
data = item.Element("{" + media + "}content");
post.FileUrl = AppendHttp(data.Attribute("url").Value, option.Provider);
try
{
string description = HttpUtility.HtmlDecode(item.Element("description").Value);
Match matches = imageResolutionRegex.Match(description);
post.Width = Convert.ToInt32(matches.Groups[1].Value);
post.Height = Convert.ToInt32(matches.Groups[2].Value);
post.PreviewWidth = Convert.ToInt32(matches.Groups[3].Value);
post.PreviewHeight = Convert.ToInt32(matches.Groups[4].Value);
post.CreatorId = matches.Groups[5].Value;
}
catch (Exception) { }
PostProcess(option, post);
posts.Add(post);
}
}
}
}
internal static void WriteTagFile(DanbooruPost post, string filename)
{
if (File.Exists(filename))
{
File.Delete(filename);
}
using (TextWriter tw = File.CreateText(filename))
{
foreach (var item in post.TagsEntity)
{
tw.WriteLine(item.Name);
}
}
}
public void TestDumpRawData()
{
string dump = "test124";
DanbooruPost post = new DanbooruPost();
post.Id = "123";
post.Query = "";
post.Provider = new DanbooruProvider()
{
Name = "TestProvider"
};
string filename = "Dump for Post " + post.Id + post.Provider.Name + " Query " + post.Query + ".txt";
bool result = Helper.DumpRawData(dump, filename);
Assert.IsTrue(result);
}
private static void PostProcess(DanbooruPostDaoOption option, DanbooruPost post)
{
post.TagsEntity = Helper.ParseTags(post.Tags, option.Provider);
if (option.BlacklistedTagsUseRegex)
{
post.Hidden = IsBlacklisted(post, option.BlacklistedTagsRegex);
}
else
{
post.Hidden = IsBlacklisted(post, option.BlacklistedTags);
}
post.Query = option.Query;
post.SearchTags = option.SearchTags;
post.Provider = option.Provider;
}
public static bool CheckBlacklistedTag(DanbooruPost post, DanbooruPostDaoOption option)
{
if (option.BlacklistedTagsUseRegex)
{
return(post.TagsEntity.Any(x => option.BlacklistedTagsRegex.IsMatch(x.Name)));
}
else
{
foreach (var tag in option.BlacklistedTags)
{
if (post.TagsEntity.Any(x => x.Name.Equals(tag.Name, StringComparison.InvariantCultureIgnoreCase)))
{
return(true);
}
}
}
return(false);
}
public void TestGelbooruParser()
{
DanbooruProviderDao pd = DanbooruProviderDao.GetInstance();
string target = @"../../../DanbooruDownloader3.test/TestXml/gelbooru_post.htm";
var data = File.ReadAllText(target);
var query = new DanbooruSearchParam();
query.Provider = pd.Read(sourceProvider).Where(x => x.Name == "gelbooru.com").First();
query.Tag = "";
query.OrderBy = "score";
var post = new DanbooruPost();
post.Provider = query.Provider;
GelbooruHtmlParser.ParsePost(post, data);
Assert.IsNotNull(post.FileUrl);
Assert.IsTrue(post.FileUrl == @"http://cdn2.gelbooru.com//images/1559/303b7ed1fcba0c1d9520f76ee34ec37e.jpg", "Actual: " + post.FileUrl);
}
public async Task <Post> LoadPostAsync(int postId)
{
var postHtml = await _booruLoader.LoadPageAsync(POST_URL + postId).ConfigureAwait(false);
var postJsonString = await _booruLoader.LoadPageAsync(string.Format(POST_JSON, postId)).ConfigureAwait(false);
var postJson = JsonConvert.DeserializeObject <Model.Danbooru.Json.Post>(postJsonString);
var htmlDoc = new HtmlDocument();
htmlDoc.LoadHtml(postHtml);
var post = new DanbooruPost(postId, htmlDoc.DocumentNode, postJson);
if (post.IsUgoira)
{
await LoadUgoiraMeta(post);
}
return(post);
}
private void ParsePostAttributes(XmlTextReader reader, DanbooruPost post)
{
while (reader.MoveToNextAttribute())
{
switch (reader.Name.ToLowerInvariant())
{
case "id": post.Id = reader.Value; break;
case "tags":
post.Tags = reader.Value;
post.TagsEntity = Helper.ParseTags(post.Tags, Option.Provider);
break;
case "source": post.Source = reader.Value; break;
case "creator_id": post.CreatorId = reader.Value; break;
case "file_url": post.FileUrl = AppendHttp(reader.Value); break;
case "width": post.Width = -1;
try
{
post.Width = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "height": post.Height = -1;
try
{
post.Height = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "change": post.Change = reader.Value; break;
case "score": post.Score = reader.Value; break;
case "rating": post.Rating = reader.Value; break;
case "status": post.Status = reader.Value; break;
case "has_children": post.HasChildren = Boolean.Parse(reader.Value); break;
case "created_at": post.CreatedAt = reader.Value; break;
case "md5": post.MD5 = reader.Value; break;
case "preview_url": post.PreviewUrl = AppendHttp(reader.Value); break;
case "preview_width": post.PreviewWidth = -1;
try
{
post.PreviewWidth = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "preview_height": post.PreviewHeight = -1;
try
{
post.PreviewHeight = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "parent_id": post.ParentId = reader.Value; break;
case "sample_url": post.SampleUrl = AppendHttp(reader.Value); break;
case "sample_width": post.SampleWidth = -1;
try
{
post.SampleWidth = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "sample_height": post.SampleHeight = -1;
try
{
post.SampleHeight = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "jpeg_url": post.JpegUrl = AppendHttp(reader.Value); break;
case "jpeg_width": post.JpegWidth = -1;
try
{
post.JpegWidth = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "jpeg_height": post.JpegHeight = -1;
try
{
post.JpegHeight = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
}
}
}
private void ProcessXML(string rawData)
{
RawData = rawData;
using (StringReader strReader = new StringReader(rawData))
{
using (XmlTextReader reader = new XmlTextReader(strReader))
{
if (Option.Provider.BoardType == BoardType.Shimmie2)
{
posts = Engine.ShimmieEngine.ParseRSS(rawData, Option);
}
else
{
while (reader.Read())
{
switch (reader.NodeType)
{
case XmlNodeType.Element: // The node is an element.
var nodeName = reader.Name.ToLowerInvariant();
if (nodeName.Equals("posts"))
{
while (reader.MoveToNextAttribute())
{
if (reader.Name.ToLowerInvariant().Equals("count")) // Posts Count
{
postCount = int.Parse(reader.Value);
}
else if (reader.Name.ToLowerInvariant().Equals("offset")) // Post Offset
{
offset = int.Parse(reader.Value);
}
}
}
else if (nodeName.Equals("response"))
{
Success = true;
while (reader.MoveToNextAttribute())
{
if (reader.Name.ToLowerInvariant().Equals("reason")) // Posts Count
{
ResponseMessage = reader.Value;
}
else if (reader.Name.ToLowerInvariant().Equals("success")) // Posts Count
{
Success = bool.Parse(reader.Value);
}
}
}
else if (nodeName.Equals("post"))
{
DanbooruPost post = new DanbooruPost();
ParsePostAttributes(reader, post);
post.Hidden = Helper.CheckBlacklistedTag(post, Option);
post.Provider = Option.Provider;
post.Query = Option.Query;
post.SearchTags = Option.SearchTags;
if (Option.Provider.BoardType == BoardType.Danbooru || Option.Provider.BoardType == BoardType.Shimmie2)
{
post.Referer = Option.Provider.Url + @"/post/show/" + post.Id;
}
else if (Option.Provider.BoardType == BoardType.Gelbooru)
{
post.Referer = Option.Provider.Url + @"/index.php?page=post&s=view&id=" + post.Id;
}
posts.Add(post);
actualCount++;
}
break;
case XmlNodeType.EndElement: //Display the end of the element.
//txtResult.AppendText("END");
break;
}
}
}
}
}
}
public static DanbooruPost ParsePost(DanbooruPost post, string postHtml, bool overideTagParsing)
{
try
{
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(postHtml);
string file_url = "";
string sample_url = "";
// image post
var img = doc.DocumentNode.SelectSingleNode("//img[@id='main_image']");
if (img != null)
{
sample_url = file_url = Helper.FixUrl(img.GetAttributeValue("src", ""), isHttps(post.Provider), hostname: post.Provider.Url);
}
// video post
var video = doc.DocumentNode.SelectSingleNode("//video[@id='video-id']/source");
if (video != null)
{
var dl = video.GetAttributeValue("src", "");
sample_url = file_url = Helper.FixUrl(dl, isHttps(post.Provider), hostname: post.Provider.Url);
}
post.SampleUrl = sample_url;
post.FileUrl = file_url;
post.CreatedAt = "N/A";
post.CreatedAtDateTime = DateTime.MinValue;
var imgInfo = doc.DocumentNode.SelectNodes("//table[@class='image_info form']");
foreach (var table in imgInfo)
{
if (table.HasChildNodes)
{
var time = table.SelectSingleNode("//time");
if (time != null)
{
post.CreatedAt = time.Attributes["datetime"].Value;
post.CreatedAtDateTime = DateTime.Parse(post.CreatedAt);
}
}
}
// score
var score = Regex.Matches(postHtml, @"Current Score: (\d+)");
if (score.Count > 0 && score[0].Groups.Count > 0)
{
post.Score = score[0].Groups[0].Value;
}
return(post);
}
catch (Exception ex)
{
string filename = $"Dump for Post {post.Id} - {post.Provider.Name} Query {post.Query}.txt";
bool result = Helper.DumpRawData(postHtml, filename);
if (!result)
{
Program.Logger.Error($"Failed to dump rawdata to: {filename}", ex);
}
throw;
}
}
/// <summary>
/// For constructing TAGS_FILENAME.
/// </summary>
/// <param name="baseFolder"></param>
/// <param name="format"></param>
/// <param name="post"></param>
/// <param name="limit"></param>
/// <returns></returns>
public static string MakeFilename(DanbooruFilenameFormat format, DanbooruPost post)
{
string filename = format.FilenameFormat;
string provider = post.Provider.Name;
string query = post.Query;
string searchTags = post.SearchTags;
string originalFileName = post.FileUrl.Split('/').Last();
// sanitizing the format
filename = Helper.SanitizeFilename(filename, true);
//remove extension
originalFileName = originalFileName.Substring(0, originalFileName.LastIndexOf('.'));
originalFileName = Uri.UnescapeDataString(originalFileName);
filename = filename.Replace("%provider%", Helper.SanitizeFilename(provider));
filename = filename.Replace("%id%", post.Id);
filename = filename.Replace("%rating%", post.Rating);
filename = filename.Replace("%md5%", post.MD5);
filename = filename.Replace("%searchParam%", Helper.SanitizeFilename(query));
filename = filename.Replace("%searchtag%", Helper.SanitizeFilename(searchTags));
filename = filename.Replace("%originalFilename%", Helper.SanitizeFilename(originalFileName));
// copy the tags entity to be grouped.
var groupedTags = post.TagsEntity;
groupedTags.Sort();
// remove ignored tags
groupedTags = RemoveIgnoredTags(format, groupedTags);
// artist
var artist = FilterTags(post,
groupedTags,
DanbooruTagType.Artist,
format.ArtistGroupLimit,
format.ArtistGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var artistStr = Helper.SanitizeFilename(artist).Trim();
if (String.IsNullOrEmpty(artistStr))
{
artistStr = DanbooruDownloader3.Properties.Settings.Default.tagNoArtistValue;
}
filename = filename.Replace("%artist%", artistStr);
// copyright
var copyright = FilterTags(post,
groupedTags,
DanbooruTagType.Copyright,
format.CopyrightGroupLimit,
format.CopyrightGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var copyStr = Helper.SanitizeFilename(copyright.Trim());
if (String.IsNullOrEmpty(copyStr))
{
copyStr = DanbooruDownloader3.Properties.Settings.Default.tagNoCopyrightValue;
}
filename = filename.Replace("%copyright%", copyStr);
// character
var character = FilterTags(post,
groupedTags,
DanbooruTagType.Character,
format.CharacterGroupLimit,
format.CharacterGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var charaStr = Helper.SanitizeFilename(character.Trim());
if (String.IsNullOrEmpty(charaStr))
{
charaStr = DanbooruDownloader3.Properties.Settings.Default.tagNoCharaValue;
}
filename = filename.Replace("%character%", charaStr);
// cirle
var circleSelection = post.TagsEntity.Where <DanbooruTag>(x => x.Type == DanbooruTagType.Circle).Select(x => x.Name);
var circle = FilterTags(post,
groupedTags,
DanbooruTagType.Circle,
format.CircleGroupLimit,
format.CircleGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var circleStr = Helper.SanitizeFilename(circle.Trim());
if (String.IsNullOrEmpty(circleStr))
{
circleStr = DanbooruDownloader3.Properties.Settings.Default.tagNoCircleValue;
}
filename = filename.Replace("%circle%", circleStr);
// faults
var faultsSelection = post.TagsEntity.Where <DanbooruTag>(x => x.Type == DanbooruTagType.Faults).Select(x => x.Name);
var faults = FilterTags(post,
groupedTags,
DanbooruTagType.Faults,
format.FaultsGroupLimit,
format.FaultsGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var faultStr = Helper.SanitizeFilename(faults.Trim());
if (String.IsNullOrEmpty(faultStr))
{
faultStr = DanbooruDownloader3.Properties.Settings.Default.tagNoFaultValue;
}
filename = filename.Replace("%faults%", faultStr);
// all tags
var allTempTags = groupedTags.Select(x => x.Name).ToList();
if (format.TagReplaceUnderscoreToSpace)
{
for (int i = 0; i < allTempTags.Count; i++)
{
allTempTags[i] = allTempTags[i].Replace("_", " ").Trim();
}
}
filename = filename.Replace("%tags%", Helper.SanitizeFilename(string.Join(" ", allTempTags)));
// append base folder from Save Folder text box
if (format.BaseFolder.EndsWith(@"\"))
{
filename = format.BaseFolder + filename;
}
else if (!String.IsNullOrWhiteSpace(format.BaseFolder))
{
filename = format.BaseFolder + @"\" + filename;
}
filename = filename.Substring(0, filename.Length < format.Limit ? filename.Length : format.Limit).Trim();
// check if contains subdirectory
if (filename.Contains(@"\"))
{
string dir = filename.Substring(0, filename.LastIndexOf(@"\"));
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
}
return(filename);
}
/// <summary>
/// For constructing TAGS_FILENAME.
/// </summary>
/// <param name="baseFolder"></param>
/// <param name="format"></param>
/// <param name="post"></param>
/// <param name="limit"></param>
/// <returns></returns>
public static string MakeFilename(DanbooruFilenameFormat format, DanbooruPost post)
{
string filename = format.FilenameFormat;
string provider = post.Provider.Name;
string query = post.Query;
string searchTags = post.SearchTags;
string originalFileName = post.FileUrl.Split('/').Last();
// sanitizing the format
filename = Helper.SanitizeFilename(filename, true);
//remove extension
originalFileName = originalFileName.Substring(0, originalFileName.LastIndexOf('.'));
originalFileName = Uri.UnescapeDataString(originalFileName);
filename = filename.Replace("%provider%", Helper.SanitizeFilename(provider));
filename = filename.Replace("%id%", post.Id);
filename = filename.Replace("%rating%", post.Rating);
filename = filename.Replace("%md5%", post.MD5);
filename = filename.Replace("%searchParam%", Helper.SanitizeFilename(query));
filename = filename.Replace("%searchtag%", Helper.SanitizeFilename(searchTags));
filename = filename.Replace("%originalFilename%", Helper.SanitizeFilename(originalFileName));
// copy the tags entity to be grouped.
var groupedTags = post.TagsEntity;
// remove ignored tags
foreach (DanbooruTag ignoredTag in format.IgnoredTags)
{
if (format.IgnoreTagsUseRegex)
{
if (!String.IsNullOrWhiteSpace(format.IgnoredTagsRegex))
{
//Program.Logger.Debug("Ignore Regex: " + format.IgnoredTagsRegex);
Regex re = new Regex(format.IgnoredTagsRegex, RegexOptions.IgnoreCase);
groupedTags.RemoveAll(x => re.IsMatch(x.Name));
}
}
else
{
groupedTags.RemoveAll(x => x.Name == ignoredTag.Name);
}
}
var artistSelection = post.TagsEntity.Where <DanbooruTag>(x => x.Type == DanbooruTagType.Artist).Select(x => x.Name);
var artist = "";
if (artistSelection != null)
{
if (artistSelection.Count() >= format.ArtistGroupLimit && !string.IsNullOrWhiteSpace(format.ArtistGroupReplacement))
{
artist = format.ArtistGroupReplacement;
groupedTags.RemoveAll(x => x.Type == DanbooruTagType.Artist);
groupedTags.Add(new DanbooruTag()
{
Name = format.ArtistGroupReplacement
});
}
else
{
artist = string.Join(" ", artistSelection);
}
}
if (string.IsNullOrWhiteSpace(artist))
{
artist = format.MissingTagReplacement;
}
filename = filename.Replace("%artist%", Helper.SanitizeFilename(artist).Trim());
var copyrightSelection = post.TagsEntity.Where <DanbooruTag>(x => x.Type == DanbooruTagType.Copyright).Select(x => x.Name);
var copyright = "";
if (copyrightSelection != null)
{
if (copyrightSelection.Count() >= format.CopyrightGroupLimit &&
format.CopyrightGroupLimit > 1 &&
!string.IsNullOrWhiteSpace(format.CopyrightGroupReplacement))
{
copyright = format.CopyrightGroupReplacement;
groupedTags.RemoveAll(x => x.Type == DanbooruTagType.Artist);
groupedTags.Add(new DanbooruTag()
{
Name = format.ArtistGroupReplacement
});
}
else
{
copyright = string.Join(" ", copyrightSelection);
}
}
if (string.IsNullOrWhiteSpace(copyright))
{
copyright = format.MissingTagReplacement;
}
filename = filename.Replace("%copyright%", Helper.SanitizeFilename(copyright.Trim()));
var characterSelection = post.TagsEntity.Where <DanbooruTag>(x => x.Type == DanbooruTagType.Character).Select(x => x.Name);
var character = "";
if (characterSelection != null)
{
if (characterSelection.Count() >= format.CharacterGroupLimit &&
format.CharacterGroupLimit > 1 &&
!string.IsNullOrWhiteSpace(format.CharacterGroupReplacement))
{
character = format.CharacterGroupReplacement;
groupedTags.RemoveAll(x => x.Type == DanbooruTagType.Character);
groupedTags.Add(new DanbooruTag()
{
Name = format.CharacterGroupReplacement
});
}
else
{
character = string.Join(" ", characterSelection);
}
}
if (string.IsNullOrWhiteSpace(character))
{
character = format.MissingTagReplacement;
}
filename = filename.Replace("%character%", Helper.SanitizeFilename(character.Trim()));
var circleSelection = post.TagsEntity.Where <DanbooruTag>(x => x.Type == DanbooruTagType.Circle).Select(x => x.Name);
var circle = "";
if (circleSelection != null)
{
if (circleSelection.Count() >= format.CircleGroupLimit &&
format.CircleGroupLimit > 1 &&
!string.IsNullOrWhiteSpace(format.CircleGroupReplacement))
{
circle = format.CircleGroupReplacement;
groupedTags.RemoveAll(x => x.Type == DanbooruTagType.Circle);
groupedTags.Add(new DanbooruTag()
{
Name = format.CircleGroupReplacement
});
}
else
{
circle = string.Join(" ", circleSelection);
}
}
if (string.IsNullOrWhiteSpace(circle))
{
circle = format.MissingTagReplacement;
}
filename = filename.Replace("%circle%", Helper.SanitizeFilename(circle.Trim()));
var faultsSelection = post.TagsEntity.Where <DanbooruTag>(x => x.Type == DanbooruTagType.Faults).Select(x => x.Name);
var faults = "";
if (faultsSelection != null)
{
if (faultsSelection.Count() >= format.FaultsGroupLimit &&
format.FaultsGroupLimit > 1 &&
!string.IsNullOrWhiteSpace(format.FaultsGroupReplacement))
{
faults = format.FaultsGroupReplacement;
groupedTags.RemoveAll(x => x.Type == DanbooruTagType.Faults);
groupedTags.Add(new DanbooruTag()
{
Name = format.FaultsGroupReplacement
});
}
else
{
faults = string.Join(" ", faultsSelection);
}
}
if (string.IsNullOrWhiteSpace(faults))
{
faults = format.MissingTagReplacement;
}
filename = filename.Replace("%faults%", Helper.SanitizeFilename(faults.Trim()));
groupedTags.Sort();
filename = filename.Replace("%tags%", Helper.SanitizeFilename(string.Join(" ", groupedTags.Select(x => x.Name))));
// append base folder from Save Folder text box
if (format.BaseFolder.EndsWith(@"\"))
{
filename = format.BaseFolder + filename;
}
else if (!String.IsNullOrWhiteSpace(format.BaseFolder))
{
filename = format.BaseFolder + @"\" + filename;
}
filename = filename.Substring(0, filename.Length < format.Limit ? filename.Length : format.Limit).Trim();
// check if contains subdirectory
if (filename.Contains(@"\"))
{
string dir = filename.Substring(0, filename.LastIndexOf(@"\"));
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
}
return(filename);
}
public static DanbooruPost ParsePost(DanbooruPost post, string postHtml)
{
HtmlDocument doc = new HtmlDocument();
if (String.IsNullOrEmpty(postHtml))
{
throw new Exception("No post html!");
}
doc.LoadHtml(postHtml);
string file_url = "";
string sample_url = "";
// Flash Game or bmp
// TODO: need to change the preview url
if (post.PreviewUrl == "http://chan.sankakucomplex.com/download-preview.png")
{
var links = doc.DocumentNode.SelectNodes("//a");
foreach (var link in links)
{
// flash
if (link.InnerText == "Save this flash (right click and save)")
{
file_url = link.GetAttributeValue("href", "");
break;
}
// bmp
if (link.InnerText == "Download")
{
file_url = link.GetAttributeValue("href", "");
break;
}
}
}
else
{
var image = doc.DocumentNode.SelectSingleNode("//img[@id='image']");
if (image != null)
{
sample_url = image.GetAttributeValue("src", "");
}
var links = doc.DocumentNode.SelectNodes("//a");
foreach (var link in links)
{
if (link.InnerText == "Original image")
{
file_url = link.GetAttributeValue("href", "");
break;
}
}
}
post.FileUrl = file_url;
if (!string.IsNullOrWhiteSpace(file_url) && string.IsNullOrWhiteSpace(sample_url))
{
sample_url = file_url;
}
post.SampleUrl = sample_url;
return(post);
}
/// <summary>
/// Parse search page result and return the images with initial tags.
/// </summary>
/// <param name="data"></param>
/// <param name="searchParam"></param>
/// <returns></returns>
public BindingList <DanbooruPost> Parse(string data, DanbooruSearchParam searchParam)
{
try
{
this.SearchParam = searchParam;
this.RawData = data;
BindingList <DanbooruPost> posts = new BindingList <DanbooruPost>();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
// remove popular preview
var popular = doc.DocumentNode.SelectSingleNode("//div[@id='popular-preview']");
if (popular != null)
{
popular.Remove();
}
// get all thumbs
var thumbs = doc.DocumentNode.SelectNodes("//span");
if (thumbs != null && thumbs.Count > 0)
{
foreach (var thumb in thumbs)
{
if (thumb.GetAttributeValue("class", "").Contains("thumb"))
{
DanbooruPost post = new DanbooruPost();
post.Id = thumb.GetAttributeValue("id", "-1").Substring(1);
post.Provider = searchParam.Provider;
post.SearchTags = searchParam.Tag;
post.Query = GenerateQueryString(searchParam);
int i = 0;
// get the image link
for (; i < thumb.ChildNodes.Count; ++i)
{
if (thumb.ChildNodes[i].Name == "a")
{
break;
}
}
var a = thumb.ChildNodes[i];
post.Referer = Helper.FixUrl(searchParam.Provider.Url + a.GetAttributeValue("href", ""), isHttps(post.Provider));
var img = a.ChildNodes[i];
if (img.GetAttributeValue("src", "").Contains("images/no-visibility.svg"))
{
Program.Logger.Warn(String.Format("No access for post {0}.", post.Id));
continue;
}
var title = img.GetAttributeValue("title", "");
post.Tags = title.Substring(0, title.LastIndexOf("Rating:")).Trim();
post.Tags = Helper.DecodeEncodedNonAsciiCharacters(post.Tags);
post.TagsEntity = Helper.ParseTags(post.Tags, SearchParam.Provider);
post.Hidden = Helper.CheckBlacklistedTag(post, searchParam.Option);
var status = img.GetAttributeValue("class", "").Replace("preview", "").Trim();
if (status.Contains("deleted"))
{
post.Status = "deleted";
}
else if (status.Contains("pending"))
{
post.Status = "pending";
}
else
{
post.Status = status;
}
post.PreviewUrl = Helper.FixUrl(img.GetAttributeValue("src", ""), isHttps(post.Provider));
post.PreviewHeight = img.GetAttributeValue("height", 0);
post.PreviewWidth = img.GetAttributeValue("width", 0);
// Rating:Explicit Score:4.5 Size:1080x1800 User:System
post.Source = "";
post.Score = title.Substring(title.LastIndexOf("Score:") + 6);
post.Score = post.Score.Substring(0, post.Score.IndexOf(" ")).Trim();
string resolution = title.Substring(title.LastIndexOf("Size:") + 5);
resolution = resolution.Substring(0, resolution.IndexOf(" ")).Trim();
string[] resArr = resolution.Split('x');
post.Width = Int32.Parse(resArr[0]);
post.Height = Int32.Parse(resArr[1]);
string rating = title.Substring(title.LastIndexOf("Rating:") + 7, 1);
//rating = rating.Substring(0, rating.IndexOf(" ")).Trim();
post.Rating = rating.ToLower();
post.CreatorId = title.Substring(title.LastIndexOf("User:"******"/") + 1);
post.MD5 = post.MD5.Substring(0, post.MD5.LastIndexOf("."));
posts.Add(post);
}
}
}
// idol complex
var siteTitle = doc.DocumentNode.SelectSingleNode("//*[@id='site-title']");
if (siteTitle != null)
{
var strTitle = siteTitle.InnerText.Split('\n').First();
var strCount = "-1";
if (!Regex.IsMatch(strTitle, @".* = \d+") && strTitle.LastIndexOf("(") > 0)
{
// single tag
// Sankaku Channel/ginhaha (1,198)
strCount = strTitle.Substring(strTitle.LastIndexOf("("));
}
else
{
// compound tag
// Sankaku Channel/= = (13,957) + rating:e = 585
strCount = strTitle.Split('=').Last().Trim();
}
strCount = strCount.Replace("(", "");
strCount = strCount.Replace(",", "");
strCount = strCount.Replace(".", "");
strCount = strCount.Replace(")", "");
Int32.TryParse(strCount, out int count);
TotalPost = count;
}
else
{
TotalPost = posts.Count;
}
if (!SearchParam.Page.HasValue && SearchParam.Page > 0)
{
SearchParam.Page = 1;
}
Offset = TotalPost * SearchParam.Page;
// get next id for 26th page and current page return full list (20 posts)
if (searchParam.Page >= 25 && posts.Count == 20)
{
searchParam.NextKey = posts[posts.Count - 1].Id;
}
return(posts);
}
catch (Exception ex)
{
var filename = Helper.SanitizeFilename("Dump for Sankaku Image List - " + searchParam.Tag + " - page " + searchParam.Page + ".txt");
var result = Helper.DumpRawData(data, filename);
if (!result)
{
Program.Logger.Error("Failed to dump rawdata to: " + filename, ex);
}
throw;
}
}
public BindingList <DanbooruPost> Parse(string data, DanbooruSearchParam searchParam)
{
this.SearchParam = searchParam;
this.RawData = data;
BindingList <DanbooruPost> posts = new BindingList <DanbooruPost>();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
// remove popular preview
var popular = doc.DocumentNode.SelectSingleNode("//div[@id='popular-preview']");
if (popular != null)
{
popular.Remove();
}
// get all thumbs
var thumbs = doc.DocumentNode.SelectNodes("//span");
if (thumbs != null && thumbs.Count > 0)
{
foreach (var thumb in thumbs)
{
if (thumb.GetAttributeValue("class", "").Contains("thumb"))
{
DanbooruPost post = new DanbooruPost();
post.Id = thumb.GetAttributeValue("id", "_N/A").Substring(1);
post.Provider = searchParam.Provider;
post.SearchTags = searchParam.Tag;
post.Query = GenerateQueryString(searchParam);
int i = 0;
// get the image link
for (; i < thumb.ChildNodes.Count; ++i)
{
if (thumb.ChildNodes[i].Name == "a")
{
break;
}
}
var a = thumb.ChildNodes[i];
post.Referer = searchParam.Provider.Url + "/" + System.Web.HttpUtility.HtmlDecode(a.GetAttributeValue("href", ""));
if (post.Id == "N/A")
{
post.Id = a.GetAttributeValue("id", "N/A").Substring(1);
}
var img = a.ChildNodes[i];
var title = img.GetAttributeValue("title", "");
var title2 = title.ToString();
post.Tags = title.Substring(0, title.LastIndexOf("rating:")).Trim();
post.Tags = Helper.DecodeEncodedNonAsciiCharacters(post.Tags);
post.TagsEntity = Helper.ParseTags(post.Tags, SearchParam.Provider);
post.Hidden = Helper.CheckBlacklistedTag(post, searchParam.Option);
post.PreviewUrl = img.GetAttributeValue("src", "");
post.PreviewHeight = img.GetAttributeValue("height", 0);
post.PreviewWidth = img.GetAttributeValue("width", 0);
post.Source = "";
post.Score = title.Substring(title.LastIndexOf("score:") + 6);
post.Score = post.Score.Substring(0, post.Score.LastIndexOf(" ")).Trim();
title2 = title2.Substring(title2.LastIndexOf("rating:"));
post.Rating = title2.Substring(7, 1).ToLower();
post.Status = "";
post.MD5 = post.PreviewUrl.Substring(post.PreviewUrl.LastIndexOf("/") + 1);
post.MD5 = post.MD5.Substring(0, post.MD5.LastIndexOf("."));
post.MD5 = post.MD5.Replace("thumbnail_", "");
posts.Add(post);
}
}
}
TotalPost = posts.Count;
if (!SearchParam.Page.HasValue)
{
SearchParam.Page = 0;
}
Offset = posts.Count * SearchParam.Page;
return(posts);
}
private void ProcessJson(String json)
{
if (json.Length < 4)
{
return;
}
if (json.StartsWith("{"))
{
json = json.Substring(1, json.Length - 2);
string[] splitter = { "\"," };
string[] node = json.Split(splitter, StringSplitOptions.None);
foreach (string str2 in node)
{
string[] val = str2.Split(':');
switch (val[0].ToLowerInvariant())
{
case "\"reason\"":
ResponseMessage = val[1].Replace("\"", "");
break;
case "\"success\"":
Success = bool.Parse(val[1].Replace("\"", ""));
break;
}
}
}
else
{
json = json.Substring(2, json.Length - 4);
string[] splitter = { "},{" };
string[] split = json.Split(splitter, StringSplitOptions.None);
foreach (string str in split)
{
DanbooruPost post = new DanbooruPost();
string[] node = str.Split(',');
foreach (string str2 in node)
{
char[] splitter2 = { ':' };
string[] val = str2.Split(splitter2, 2);
switch (val[0].ToLowerInvariant())
{
case "\"id\"":
post.Id = val[1].Replace("\"", "");
break;
case "\"tags\"":
post.Tags = val[1].Replace("\"", "");
post.Tags = Helper.DecodeEncodedNonAsciiCharacters(post.Tags);
post.TagsEntity = Helper.ParseTags(post.Tags, Option.Provider);
break;
case "\"source\"":
post.Source = val[1].Replace("\"", "");
break;
case "\"creator_id\"":
post.CreatorId = val[1].Replace("\"", "");
break;
case "\"file_url\"":
post.FileUrl = AppendHttp(val[1].Replace("\"", ""));
break;
case "\"width\"":
post.Width = -1;
try
{
post.Width = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"height\"":
post.Height = -1;
try
{
post.Height = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"score\"":
post.Score = val[1];
break;
case "\"rating\"":
post.Rating = val[1].Replace("\"", "");
break;
case "\"md5\"":
post.MD5 = val[1].Replace("\"", "");
break;
case "\"preview_url\"":
post.PreviewUrl = AppendHttp(val[1].Replace("\"", ""));
break;
case "\"preview_width\"":
post.PreviewWidth = -1;
try
{
post.PreviewWidth = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"preview_height\"":
post.PreviewHeight = -1;
try
{
post.PreviewHeight = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"file_size\"":
post.Filesize = -1;
try
{
post.Filesize = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"parent_id\"":
post.ParentId = val[1];
break;
case "\"status\"":
post.Status = val[1];
break;
case "\"created_at\"":
post.CreatedAt = val[1];
break;
case "\"has_children\"":
post.HasChildren = false;
try
{
post.HasChildren = Convert.ToBoolean(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"sample_width\"":
post.SampleWidth = -1;
try
{
post.SampleWidth = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"sample_height\"":
post.SampleHeight = -1;
try
{
post.SampleHeight = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"sample_url\"":
post.SampleUrl = AppendHttp(val[1].Replace("\"", ""));
break;
case "\"jpeg_width\"":
post.JpegWidth = -1;
try
{
post.JpegWidth = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"jpeg_height\"":
post.JpegHeight = -1;
try
{
post.JpegHeight = Convert.ToInt32(val[1]);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "\"jpeg_url\"":
post.JpegUrl = AppendHttp(val[1].Replace("\"", ""));
break;
default: break;
}
}
post.Hidden = Helper.CheckBlacklistedTag(post, Option);
post.Provider = Option.Provider;
post.Query = Option.Query;
post.SearchTags = Option.SearchTags;
post.Referer = Option.Provider.Url + @"/post/show/" + post.Id;
this.posts.Add(post);
actualCount++;
}
}
}
/// <summary>
/// Return string from tag filtered by tag type.
/// Replace the returned string with tagReplacement if tag count over tagLimit.
/// Replace the returned string with missingTagReplacement if tag count is 0.
/// </summary>
/// <param name="post"></param>
/// <param name="groupedTags"></param>
/// <param name="tagType"></param>
/// <param name="tagLimit"></param>
/// <param name="tagReplacement"></param>
/// <param name="missingTagReplacement"></param>
/// <param name="isReplaceMode"></param>
/// <param name="isReplaceUnderScore"></param>
/// <returns>Return string from tag filtered by tag type.</returns>
private static string FilterTags(DanbooruPost post,
List <DanbooruTag> groupedTags,
DanbooruTagType tagType,
int tagLimit,
string tagReplacement,
string missingTagReplacement,
bool isReplaceMode,
bool isReplaceUnderScore)
{
var selectedTags = post.TagsEntity.Where <DanbooruTag>(x => x.Type == tagType).Select(x => x.Name).ToList();
var tagStr = "";
if (selectedTags != null)
{
if (isReplaceUnderScore)
{
for (int i = 0; i < selectedTags.Count; i++)
{
selectedTags[i] = selectedTags[i].Replace("_", " ").Trim();
}
}
if (tagLimit > 0 && selectedTags.Count() >= tagLimit)
{
if (isReplaceMode && !string.IsNullOrWhiteSpace(tagReplacement))
{
tagStr = tagReplacement;
if (isReplaceUnderScore)
{
tagStr = tagReplacement.Replace("_", " ").Trim();
}
//groupedTags.RemoveAll(x => x.Type == tagType);
//groupedTags.Add(new DanbooruTag() { Name = tagReplacement });
}
else
{
var tempTags = selectedTags.Take(tagLimit);
tagStr = String.Join(" ", tempTags);
}
}
else
{
tagStr = string.Join(" ", selectedTags);
}
}
if (string.IsNullOrWhiteSpace(tagStr))
{
switch (tagType)
{
case DanbooruTagType.Artist:
tagStr = Properties.Settings.Default.tagNoArtistValue;
break;
case DanbooruTagType.Character:
tagStr = Properties.Settings.Default.tagNoCharacterValue;
break;
case DanbooruTagType.Circle:
tagStr = Properties.Settings.Default.tagNoCircleValue;
break;
case DanbooruTagType.Copyright:
tagStr = Properties.Settings.Default.tagNoCopyrightValue;
break;
case DanbooruTagType.Faults:
tagStr = Properties.Settings.Default.tagNoFaultValue;
break;
default:
tagStr = missingTagReplacement;
break;
}
}
return(tagStr);
}
/// <summary>
/// For constructing TAGS_FILENAME.
/// </summary>
/// <param name="baseFolder"></param>
/// <param name="format"></param>
/// <param name="post"></param>
/// <param name="limit"></param>
/// <returns></returns>
public static string MakeFilename(DanbooruFilenameFormat format, DanbooruPost post)
{
string filename = format.FilenameFormat;
string provider = post.Provider.Name;
string query = post.Query;
string searchTags = post.SearchTags;
string originalFileName = post.FileUrl.Split('/').Last();
// sanitizing the format
filename = Helper.SanitizeFilename(filename, true);
//remove extension
originalFileName = originalFileName.Substring(0, originalFileName.LastIndexOf('.'));
originalFileName = Uri.UnescapeDataString(originalFileName);
filename = filename.Replace("%provider%", Helper.SanitizeFilename(provider));
filename = filename.Replace("%id%", post.Id);
filename = filename.Replace("%rating%", post.Rating);
filename = filename.Replace("%md5%", post.MD5);
filename = filename.Replace("%searchParam%", Helper.SanitizeFilename(query));
filename = filename.Replace("%searchtag%", Helper.SanitizeFilename(searchTags));
filename = filename.Replace("%originalFilename%", Helper.SanitizeFilename(originalFileName));
string dtFormat = Properties.Settings.Default.DateTimeFormat;
if (String.IsNullOrWhiteSpace(dtFormat))
{
dtFormat = "yyyy-mm-dd HH:mm:ss";
}
filename = filename.Replace("%uploadDateTime%", Helper.SanitizeFilename(post.CreatedAtDateTime.ToString(dtFormat)));
// copy the tags entity to be grouped.
var groupedTags = post.TagsEntity;
// custom sort to prioritize some tags based on file definition
// Issue #46 and #81
// regex support
if (File.Exists(PRIORITY_TAGS))
{
var priorityTags = ReadTagsFromTextFile(PRIORITY_TAGS);
groupedTags.Sort((a, b) =>
{
var containA = priorityTags.Exists(x => Regex.IsMatch(a.Name, x.Name));
var containB = priorityTags.Exists(x => Regex.IsMatch(b.Name, x.Name));
if (containA && !containB)
{
return(-1);
}
if (!containA && containB)
{
return(1);
}
else
{
return(a.CompareTo(b));
}
});
}
else
{
groupedTags.Sort();
}
// remove ignored tags
groupedTags = RemoveIgnoredTags(format, groupedTags);
// artist
var artist = FilterTags(post,
groupedTags,
DanbooruTagType.Artist,
format.ArtistGroupLimit,
format.ArtistGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var artistStr = Helper.SanitizeFilename(artist).Trim();
filename = filename.Replace("%artist%", artistStr);
// copyright
var copyright = FilterTags(post,
groupedTags,
DanbooruTagType.Copyright,
format.CopyrightGroupLimit,
format.CopyrightGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var copyStr = Helper.SanitizeFilename(copyright.Trim());
filename = filename.Replace("%copyright%", copyStr);
// character
var character = FilterTags(post,
groupedTags,
DanbooruTagType.Character,
format.CharacterGroupLimit,
format.CharacterGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var charaStr = Helper.SanitizeFilename(character.Trim());
filename = filename.Replace("%character%", charaStr);
// cirle
var circle = FilterTags(post,
groupedTags,
DanbooruTagType.Circle,
format.CircleGroupLimit,
format.CircleGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var circleStr = Helper.SanitizeFilename(circle.Trim());
filename = filename.Replace("%circle%", circleStr);
// faults
var faults = FilterTags(post,
groupedTags,
DanbooruTagType.Faults,
format.FaultsGroupLimit,
format.FaultsGroupReplacement,
format.MissingTagReplacement,
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var faultStr = Helper.SanitizeFilename(faults.Trim());
filename = filename.Replace("%faults%", faultStr);
// general
var general = FilterTags(post,
groupedTags,
DanbooruTagType.General,
0,
"",
"",
format.IsReplaceMode,
format.TagReplaceUnderscoreToSpace);
var generalStr = Helper.SanitizeFilename(general.Trim());
filename = filename.Replace("%general%", generalStr);
// all tags
var allTempTags = groupedTags.Select(x => x.Name).ToList();
if (format.TagReplaceUnderscoreToSpace)
{
for (int i = 0; i < allTempTags.Count; i++)
{
allTempTags[i] = allTempTags[i].Replace("_", " ").Trim();
}
}
filename = filename.Replace("%tags%", Helper.SanitizeFilename(string.Join(" ", allTempTags)));
// append base folder from Save Folder text box
if (format.BaseFolder.EndsWith(@"\"))
{
filename = format.BaseFolder + filename;
}
else if (!String.IsNullOrWhiteSpace(format.BaseFolder))
{
filename = format.BaseFolder + @"\" + filename;
}
filename = filename.Substring(0, filename.Length < format.Limit ? filename.Length : format.Limit).Trim();
return(filename);
}
public static DanbooruPost ParsePost(DanbooruPost post, string postHtml)
{
HtmlDocument doc = new HtmlDocument();
if (String.IsNullOrEmpty(postHtml))
{
throw new Exception("No post html!");
}
doc.LoadHtml(postHtml);
string file_url = "";
string sample_url = "";
// Flash Game or bmp
// TODO: need to change the preview url
if (post.PreviewUrl == "http://chan.sankakucomplex.com/download-preview.png")
{
var links = doc.DocumentNode.SelectNodes("//a");
foreach (var link in links)
{
// flash
if (link.InnerText == "Save this flash (right click and save)")
{
file_url = link.GetAttributeValue("href", "");
break;
}
// bmp
if (link.InnerText == "Download")
{
file_url = link.GetAttributeValue("href", "");
break;
}
}
}
else
{
var image = doc.DocumentNode.SelectSingleNode("//img[@id='image']");
if (image != null)
{
sample_url = image.GetAttributeValue("src", "");
}
var links = doc.DocumentNode.SelectNodes("//a");
foreach (var link in links)
{
if (link.InnerText == "Original image")
{
file_url = link.GetAttributeValue("href", "");
break;
}
}
}
post.FileUrl = file_url;
if (!string.IsNullOrWhiteSpace(file_url) && string.IsNullOrWhiteSpace(sample_url))
{
sample_url = file_url;
}
post.SampleUrl = sample_url;
// parse datetime
// TODO: untested
var statsLIs = doc.DocumentNode.SelectNodes("//div[@id='stats']//li");
foreach (var item in statsLIs)
{
if (item.InnerHtml.StartsWith("Posted: "))
{
post.CreatedAt = System.Text.RegularExpressions.Regex.Match(item.InnerHtml, "Posted: (.*)<").Value;
post.CreatedAtDateTime = DanbooruPostDao.ParseDateTime(post.CreatedAt, post.Provider);
break;
}
}
return(post);
}
private void ProcessXML(string rawData)
{
RawData = rawData;
using (StringReader strReader = new StringReader(rawData))
{
using (XmlTextReader reader = new XmlTextReader(strReader))
{
if (Option.Provider.BoardType == BoardType.Shimmie2)
{
posts = Engine.ShimmieEngine.ParseRSS(rawData, Option);
}
else
{
while (reader.Read())
{
switch (reader.NodeType)
{
case XmlNodeType.Element: // The node is an element.
var nodeName = reader.Name.ToLowerInvariant();
if (nodeName.Equals("posts"))
{
while (reader.MoveToNextAttribute())
{
if (reader.Name.ToLowerInvariant().Equals("count")) // Posts Count
{
postCount = int.Parse(reader.Value);
}
else if (reader.Name.ToLowerInvariant().Equals("offset")) // Post Offset
{
offset = int.Parse(reader.Value);
}
}
}
else if (nodeName.Equals("response"))
{
Success = true;
while (reader.MoveToNextAttribute())
{
if (reader.Name.ToLowerInvariant().Equals("reason")) // Posts Count
{
ResponseMessage = reader.Value;
}
else if (reader.Name.ToLowerInvariant().Equals("success")) // Posts Count
{
Success = bool.Parse(reader.Value);
}
}
}
else if (nodeName.Equals("post"))
{
DanbooruPost post = new DanbooruPost();
while (reader.MoveToNextAttribute())
{
switch (reader.Name.ToLowerInvariant())
{
case "id": post.Id = reader.Value; break;
case "tags":
post.Tags = reader.Value;
post.TagsEntity = Helper.ParseTags(post.Tags, Option.Provider);
break;
case "source": post.Source = reader.Value; break;
case "creator_id": post.CreatorId = reader.Value; break;
case "file_url": post.FileUrl = AppendHttp(reader.Value); break;
case "width": post.Width = -1;
try
{
post.Width = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "height": post.Height = -1;
try
{
post.Height = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "change": post.Change = reader.Value; break;
case "score": post.Score = reader.Value; break;
case "rating": post.Rating = reader.Value; break;
case "status": post.Status = reader.Value; break;
case "has_children": post.HasChildren = Boolean.Parse(reader.Value); break;
case "created_at": post.CreatedAt = reader.Value; break;
case "md5": post.MD5 = reader.Value; break;
case "preview_url": post.PreviewUrl = AppendHttp(reader.Value); break;
case "preview_width": post.PreviewWidth = -1;
try
{
post.PreviewWidth = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "preview_height": post.PreviewHeight = -1;
try
{
post.PreviewHeight = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "parent_id": post.ParentId = reader.Value; break;
case "sample_url": post.SampleUrl = AppendHttp(reader.Value); break;
case "sample_width": post.SampleWidth = -1;
try
{
post.SampleWidth = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "sample_height": post.SampleHeight = -1;
try
{
post.SampleHeight = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "jpeg_url": post.JpegUrl = AppendHttp(reader.Value); break;
case "jpeg_width": post.JpegWidth = -1;
try
{
post.JpegWidth = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
case "jpeg_height": post.JpegHeight = -1;
try
{
post.JpegHeight = Int32.Parse(reader.Value);
}
catch (Exception) { if (FormMain.Debug)
{
throw;
}
}
break;
}
}
post.Hidden = Helper.CheckBlacklistedTag(post, Option);
post.Provider = Option.Provider;
post.Query = Option.Query;
post.SearchTags = Option.SearchTags;
if (Option.Provider.BoardType == BoardType.Danbooru || Option.Provider.BoardType == BoardType.Shimmie2)
{
post.Referer = Option.Provider.Url + @"/post/show/" + post.Id;
}
else if (Option.Provider.BoardType == BoardType.Gelbooru)
{
post.Referer = Option.Provider.Url + @"/index.php?page=post&s=view&id=" + post.Id;
}
posts.Add(post);
actualCount++;
}
break;
case XmlNodeType.EndElement: //Display the end of the element.
//txtResult.AppendText("END");
break;
}
}
}
}
}
}
/// <summary>
/// Reparse tags from post details.
/// </summary>
/// <param name="post"></param>
/// <param name="doc"></param>
private static void ReparseTags(DanbooruPost post, HtmlDocument doc)
{
post.TagsEntity.Clear();
var tags = doc.DocumentNode.SelectNodes("//ul[@id='tag-sidebar']/li");
foreach (var tag in tags)
{
var tagEntity = new DanbooruTag();
var cls = tag.Attributes["class"].Value; // Fix Issue #146
switch (cls)
{
case "tag-type-idol":
// idol complex
tagEntity.Type = DanbooruTagType.Artist;
break;
case "tag-type-artist":
// sankaku
tagEntity.Type = DanbooruTagType.Artist;
break;
case "tag-type-photo_set":
// idol complex: usually album name
tagEntity.Type = DanbooruTagType.Circle;
break;
case "tag-type-studio":
// sankaku: circlename
tagEntity.Type = DanbooruTagType.Circle;
break;
case "tag-type-meta":
// both
tagEntity.Type = DanbooruTagType.Faults;
break;
case "tag-type-medium":
// both
tagEntity.Type = DanbooruTagType.Faults;
break;
case "tag-type-general":
// both
tagEntity.Type = DanbooruTagType.General;
break;
case "tag-type-copyright":
// both
tagEntity.Type = DanbooruTagType.Copyright;
break;
case "tag-type-character":
// both
tagEntity.Type = DanbooruTagType.Character;
break;
default:
tagEntity.Type = DanbooruTagType.Unknown;
break;
}
tagEntity.Name = Helper.DecodeEncodedNonAsciiCharacters(tag.SelectSingleNode("//ul[@id='tag-sidebar']/li/a").InnerText);
var countStr = tag.SelectSingleNode("//ul[@id='tag-sidebar']/li//span[@class='post-count']").InnerText.Trim();
tagEntity.Count = Int32.Parse(countStr);
post.TagsEntity.Add(tagEntity);
tag.Remove();
}
post.TagsEntity = post.TagsEntity.OrderByDescending(x => x.Type).ThenBy(x => x.Name).ToList();
}
public static DanbooruPost ParsePost(DanbooruPost post, string postHtml)
{
try
{
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(postHtml);
string file_url = "";
string sample_url = "";
// Flash Game or bmp
if (post.PreviewUrl.EndsWith("download-preview.png"))
{
var links = doc.DocumentNode.SelectNodes("//a");
foreach (var link in links)
{
// flash
if (link.InnerText == "Save this file (right click and save as)")
{
file_url = Helper.FixUrl(link.GetAttributeValue("href", ""), isHttps(post.Provider));
// http://cs.sankakucomplex.com/data/f6/23/f623ea7559ef39d96ebb0ca7530586b8.swf
post.MD5 = file_url.Substring(file_url.LastIndexOf("/") + 1);
post.MD5 = post.MD5.Substring(0, 32);
break;
}
// bmp
if (link.InnerText == "Download")
{
file_url = Helper.FixUrl(link.GetAttributeValue("href", ""), isHttps(post.Provider));
break;
}
}
}
else
{
var lowresElement = doc.DocumentNode.SelectSingleNode("//a[@id='lowres']");
if (lowresElement != null)
{
sample_url = Helper.FixUrl(lowresElement.GetAttributeValue("href", ""), isHttps(post.Provider));
}
var highresElement = doc.DocumentNode.SelectSingleNode("//a[@id='highres']");
if (highresElement != null)
{
file_url = Helper.FixUrl(highresElement.GetAttributeValue("href", ""), isHttps(post.Provider));
}
}
post.FileUrl = file_url;
if (!string.IsNullOrWhiteSpace(file_url) && string.IsNullOrWhiteSpace(sample_url))
{
sample_url = file_url;
}
post.SampleUrl = sample_url;
return(post);
}
catch (Exception ex)
{
string filename = "Dump for Post " + post.Id + post.Provider.Name + " Query " + post.Query + ".txt";
bool result = Helper.DumpRawData(postHtml, filename);
if (!result)
{
Program.Logger.Error("Failed to dump rawdata to: " + filename, ex);
}
throw;
}
}
/// <summary>
/// Parse the post details after added to the download list or from batch job.
/// </summary>
/// <param name="post"></param>
/// <param name="postHtml"></param>
/// <param name="overideTagParsing"></param>
/// <returns></returns>
public static DanbooruPost ParsePost(DanbooruPost post, string postHtml, bool overideTagParsing)
{
try
{
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(postHtml);
string file_url = "";
string sample_url = "";
// reparse tags with type
if (overideTagParsing)
{
ReparseTags(post, doc);
}
// Flash Game or bmp
if (post.PreviewUrl.EndsWith("download-preview.png"))
{
var links = doc.DocumentNode.SelectNodes("//a");
foreach (var link in links)
{
// flash
if (link.InnerText == "Save this file (right click and save as)")
{
file_url = Helper.FixUrl(link.GetAttributeValue("href", ""), isHttps(post.Provider));
// http://cs.sankakucomplex.com/data/f6/23/f623ea7559ef39d96ebb0ca7530586b8.swf
post.MD5 = file_url.Substring(file_url.LastIndexOf("/") + 1);
post.MD5 = post.MD5.Substring(0, 32);
break;
}
// bmp
if (link.InnerText == "Download")
{
file_url = Helper.FixUrl(link.GetAttributeValue("href", ""), isHttps(post.Provider));
break;
}
}
}
else
{
var lowresElement = doc.DocumentNode.SelectSingleNode("//a[@id='lowres']");
if (lowresElement != null)
{
sample_url = Helper.FixUrl(lowresElement.GetAttributeValue("href", ""), isHttps(post.Provider));
}
var highresElement = doc.DocumentNode.SelectSingleNode("//a[@id='highres']");
if (highresElement != null)
{
file_url = Helper.FixUrl(highresElement.GetAttributeValue("href", ""), isHttps(post.Provider));
}
}
post.FileUrl = file_url;
if (!string.IsNullOrWhiteSpace(file_url) && string.IsNullOrWhiteSpace(sample_url))
{
sample_url = file_url;
}
post.SampleUrl = sample_url;
// Created datetime
post.CreatedAt = "N/A";
post.CreatedAtDateTime = DateTime.MinValue;
try
{
var lis = doc.DocumentNode.SelectNodes("//div[@id='post-view']//div[@id='stats']//li");
foreach (var item in lis)
{
if (item.InnerText.Contains("Posted:"))
{
var links = item.SelectNodes("//div[@id='post-view']//div[@id='stats']//li//a");
foreach (var link in links)
{
if (link.Attributes.Contains("href") && link.Attributes["href"].Value.Contains("?tags=date"))
{
post.CreatedAt = link.Attributes["title"].Value;
post.CreatedAtDateTime = DanbooruPostDao.ParseDateTime(post.CreatedAt, post.Provider);
break;
}
}
break;
}
}
}
catch (Exception ex)
{
Program.Logger.Error("Unable to parse date", ex);
}
return(post);
}
catch (Exception ex)
{
string filename = "Dump for Post " + post.Id + post.Provider.Name + " Query " + post.Query + ".txt";
bool result = Helper.DumpRawData(postHtml, filename);
if (!result)
{
Program.Logger.Error("Failed to dump rawdata to: " + filename, ex);
}
throw;
}
}
public BindingList <DanbooruPost> Parse(string data, DanbooruSearchParam searchParam)
{
try
{
this.SearchParam = searchParam;
this.RawData = data;
BindingList <DanbooruPost> posts = new BindingList <DanbooruPost>();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
// remove popular preview
var popular = doc.DocumentNode.SelectSingleNode("//div[@id='popular-preview']");
if (popular != null)
{
popular.Remove();
}
// get all thumbs
var thumbs = doc.DocumentNode.SelectNodes("//span");
if (thumbs != null && thumbs.Count > 0)
{
foreach (var thumb in thumbs)
{
if (thumb.GetAttributeValue("class", "").Contains("thumb"))
{
DanbooruPost post = new DanbooruPost();
post.Id = thumb.GetAttributeValue("id", "-1").Substring(1);
post.Provider = searchParam.Provider;
post.SearchTags = searchParam.Tag;
post.Query = GenerateQueryString(searchParam);
int i = 0;
// get the image link
for (; i < thumb.ChildNodes.Count; ++i)
{
if (thumb.ChildNodes[i].Name == "a")
{
break;
}
}
var a = thumb.ChildNodes[i];
post.Referer = Helper.FixUrl(searchParam.Provider.Url + a.GetAttributeValue("href", ""), isHttps(post.Provider));
var img = a.ChildNodes[i];
var title = img.GetAttributeValue("title", "");
post.Tags = title.Substring(0, title.LastIndexOf("Rating:")).Trim();
post.Tags = Helper.DecodeEncodedNonAsciiCharacters(post.Tags);
post.TagsEntity = Helper.ParseTags(post.Tags, SearchParam.Provider);
post.Hidden = Helper.CheckBlacklistedTag(post, searchParam.Option);
var status = img.GetAttributeValue("class", "").Replace("preview", "").Trim();
if (status.Contains("deleted"))
{
post.Status = "deleted";
}
else if (status.Contains("pending"))
{
post.Status = "pending";
}
else
{
post.Status = status;
}
post.PreviewUrl = Helper.FixUrl(img.GetAttributeValue("src", ""), isHttps(post.Provider));
post.PreviewHeight = img.GetAttributeValue("height", 0);
post.PreviewWidth = img.GetAttributeValue("width", 0);
// Rating:Explicit Score:4.5 Size:1080x1800 User:System
post.Source = "";
post.Score = title.Substring(title.LastIndexOf("Score:") + 6);
post.Score = post.Score.Substring(0, post.Score.IndexOf(" ")).Trim();
string resolution = title.Substring(title.LastIndexOf("Size:") + 5);
resolution = resolution.Substring(0, resolution.IndexOf(" ")).Trim();
string[] resArr = resolution.Split('x');
post.Width = Int32.Parse(resArr[0]);
post.Height = Int32.Parse(resArr[1]);
string rating = title.Substring(title.LastIndexOf("Rating:") + 7, 1);
//rating = rating.Substring(0, rating.IndexOf(" ")).Trim();
post.Rating = rating.ToLower();
post.CreatorId = title.Substring(title.LastIndexOf("User:"******"/") + 1);
post.MD5 = post.MD5.Substring(0, post.MD5.LastIndexOf("."));
posts.Add(post);
}
}
}
TotalPost = posts.Count;
if (!SearchParam.Page.HasValue)
{
SearchParam.Page = 1;
}
Offset = TotalPost * SearchParam.Page;
return(posts);
}
catch (Exception ex)
{
var filename = Helper.SanitizeFilename("Dump for Sankaku Image List - " + searchParam.Tag + " - page " + searchParam.Page + ".txt");
var result = Helper.DumpRawData(data, filename);
if (!result)
{
Program.Logger.Error("Failed to dump rawdata to: " + filename, ex);
}
throw;
}
}