private static List <DanbooruTag> ReadTagsFromTextFile(string filename)
{
var list = new List <DanbooruTag>();
// Read the file and display it line by line.
using (var file = new System.IO.StreamReader(filename))
{
var line = "";
while ((line = file.ReadLine()) != null)
{
var tag = new DanbooruTag()
{
Name = line,
Id = "-1",
Count = -1,
Type = DanbooruTagType.Unknown
};
list.Add(tag);
}
;
}
return(list);
}
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);
}
}
public DanbooruTag GetTag(string tag, DanbooruTagCollection tagCollection)
{
// TODO: Hot spot for perfomance
var result = tagCollection.Tag.FirstOrDefault <DanbooruTag>(x => x.Name == tag);
if (result != null)
{
return(result);
}
else
{
var unknownTag = new DanbooruTag()
{
Name = tag,
Type = DanbooruTagType.Unknown,
Count = -1,
Id = "-1"
};
return(unknownTag);
}
}
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);
}
}
public DanbooruTagCollection parseTagsPage(string data, int page)
{
DanbooruTagCollection tagCol = new DanbooruTagCollection();
List<DanbooruTag> tags = new List<DanbooruTag>();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
int index = 1 + ((page - 1) * 50);
// select all tags
var tables = doc.DocumentNode.SelectNodes("//table[contains(@class,'highlightable')]");
foreach (var table in tables)
{
if (!(table.Attributes["class"].Value == "highlightable"))
{
table.Remove();
continue;
}
var rows = table.SelectNodes("//table[contains(@class,'highlightable')]//tr");
int countIndex = 1, nameIndex = 3, typeIndex = 9;
foreach (var row in rows)
{
//if (row.ChildNodes.Count != 11 && row.ChildNodes.Count != 7) continue;
var cols = row.ChildNodes;
if (cols[1].Name == "th")
{
for (int i = 0; i < cols.Count; ++i)
{
if (cols[i].Name == "th")
{
if (cols[i].InnerText.Replace("\n", "") == "Posts")
{
countIndex = i;
continue;
}
if (cols[i].InnerText.Replace("\n", "") == "Name")
{
nameIndex = i;
continue;
}
if (cols[i].InnerText.Replace("\n", "") == "Type")
{
typeIndex = i;
continue;
}
}
}
continue;
}
if (cols[1].Name != "td") continue;
DanbooruTag tag = new DanbooruTag();
tag.Id = index.ToString();
tag.Count = Int32.Parse(cols[countIndex].InnerText);
tag.Name = Helper.RemoveControlCharacters(System.Net.WebUtility.HtmlDecode(cols[nameIndex].ChildNodes[3].InnerText.Replace("\n", "")));
string tagType = cols[typeIndex].InnerText.Replace("\n", "");
if (tagType.EndsWith("(edit)")) tagType = tagType.Substring(0, tagType.Length - 6);
tagType = tagType.ToLowerInvariant();
if (tagType == "general")
tag.Type = DanbooruTagType.General;
else if (tagType == "character")
tag.Type = DanbooruTagType.Character;
else if (tagType == "artist")
tag.Type = DanbooruTagType.Artist;
else if (tagType == "copyright")
tag.Type = DanbooruTagType.Copyright;
else if (tagType == "idol")
tag.Type = DanbooruTagType.Artist;
else if (tagType == "photo_set")
tag.Type = DanbooruTagType.Circle;
else
tag.Type = DanbooruTagType.Faults;
tags.Add(tag);
++index;
}
}
tagCol.Tag = tags.ToArray();
return tagCol;
}
public DanbooruTagCollection parseTagsPage(string data, int page)
{
DanbooruTagCollection tagCol = new DanbooruTagCollection();
List <DanbooruTag> tags = new List <DanbooruTag>();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
int index = 1 + ((page - 1) * 50);
// select all tags
var tables = doc.DocumentNode.SelectNodes("//table[contains(@class,'highlightable')]");
foreach (var table in tables)
{
if (!(table.Attributes["class"].Value == "highlightable"))
{
table.Remove();
continue;
}
var rows = table.SelectNodes("//table[contains(@class,'highlightable')]//tr");
int countIndex = 1, nameIndex = 3, typeIndex = 9;
foreach (var row in rows)
{
//if (row.ChildNodes.Count != 11 && row.ChildNodes.Count != 7) continue;
var cols = row.ChildNodes;
if (cols[1].Name == "th")
{
for (int i = 0; i < cols.Count; ++i)
{
if (cols[i].Name == "th")
{
if (cols[i].InnerText.Replace("\n", "") == "Posts")
{
countIndex = i;
continue;
}
if (cols[i].InnerText.Replace("\n", "") == "Name")
{
nameIndex = i;
continue;
}
if (cols[i].InnerText.Replace("\n", "") == "Type")
{
typeIndex = i;
continue;
}
}
}
continue;
}
if (cols[1].Name != "td")
{
continue;
}
DanbooruTag tag = new DanbooruTag();
tag.Id = index.ToString();
tag.Count = Int32.Parse(cols[countIndex].InnerText);
tag.Name = Helper.RemoveControlCharacters(System.Net.WebUtility.HtmlDecode(cols[nameIndex].ChildNodes[3].InnerText.Replace("\n", "")));
string tagType = cols[typeIndex].InnerText.Replace("\n", "");
if (tagType.EndsWith("(edit)"))
{
tagType = tagType.Substring(0, tagType.Length - 6);
}
tagType = tagType.ToLowerInvariant();
if (tagType == "general")
{
tag.Type = DanbooruTagType.General;
}
else if (tagType == "character")
{
tag.Type = DanbooruTagType.Character;
}
else if (tagType == "artist")
{
tag.Type = DanbooruTagType.Artist;
}
else if (tagType == "copyright")
{
tag.Type = DanbooruTagType.Copyright;
}
else if (tagType == "idol")
{
tag.Type = DanbooruTagType.Artist;
}
else if (tagType == "photo_set")
{
tag.Type = DanbooruTagType.Circle;
}
else
{
tag.Type = DanbooruTagType.Faults;
}
tags.Add(tag);
++index;
}
}
tagCol.Tag = tags.ToArray();
return(tagCol);
}
/// <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();
}
/// <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 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-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 = 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);
}
post.TagsEntity = post.TagsEntity.OrderByDescending(x => x.Type).ThenBy(x => x.Name).ToList();
}
private static List<DanbooruTag> ReadTagsFromTextFile(string filename)
{
var list = new List<DanbooruTag>();
// Read the file and display it line by line.
using (var file = new System.IO.StreamReader(filename))
{
var line = "";
while ((line = file.ReadLine()) != null)
{
var tag = new DanbooruTag()
{
Name = line,
Id = "-1",
Count = -1,
Type = DanbooruTagType.Unknown
};
list.Add(tag);
};
}
return list;
}
public DanbooruTag GetTag(string tag, DanbooruTagCollection tagCollection)
{
// TODO: Hot spot for perfomance
var result = tagCollection.Tag.FirstOrDefault<DanbooruTag>(x => x.Name == tag);
if (result != null) return result;
else
{
var unknownTag = new DanbooruTag()
{
Name = tag,
Type = DanbooruTagType.Unknown,
Count = -1,
Id = "-1"
};
return unknownTag;
}
}
