List <string> ParseChapter(List <string> contents)
{
if (contents == null || contents.Count < 3)
{
return(null);
}
var title = UtilityService.RemoveWhitespaces(contents[1].Trim()).Replace("\r", "").Replace("\n", "").Replace("\t", "");
var start = title.PositionOf("<h4");
start = title.PositionOf(">", start + 1);
var end = title.PositionOf("</h4>", start + 1);
if (start > 0 && end > 0)
{
title = title.Substring(start + 1, end - start - 1).Trim().Replace("♦", " ");
"Đồng tác giả|Dịch giả|Người dịch|Dịch viện|Chuyển ngữ|Dịch ra|Anh dịch|Dịch thuật|Bản dịch|Hiệu đính|Biên Tập|Biên soạn|đánh máy bổ sung|Nguyên tác|Nguyên bản|Dịch theo|Dịch từ|Theo bản|Biên dịch|Tổng Hợp|Tủ Sách|Tuyển tập|Sách Xuất Bản Tại|Chủ biên|Chủ nhiệm".Split('|')
.ForEach(excluded =>
{
start = title.IndexOf(excluded, StringComparison.OrdinalIgnoreCase);
if (start > -1)
{
end = title.IndexOf("<br>", start, StringComparison.OrdinalIgnoreCase);
if (end < 0)
{
end = title.Length - 4;
}
title = title.Remove(start, end - start + 4).Trim();
}
});
while (title.IsStartsWith("<br>"))
{
title = title.Substring(4).Trim();
}
while (title.IsEndsWith("<br>"))
{
title = title.Substring(0, title.Length - 4).Trim();
}
title = title.Replace("<br>", ": ").Replace("<BR>", " - ").Trim();
}
start = title.PositionOf("<div class=\"hr");
if (start > 0)
{
var tit = "";
do
{
start = title.PositionOf("<span", start + 1);
start = start < 0 ? -1 : title.PositionOf(">", start + 1) + 1;
end = start < 0 ? -1 : title.PositionOf("</span>", start);
var t = start > 0 && end > 0 ? title.Substring(start, end - start).Trim() : "";
if (!t.Equals("") && !t.IsEquals(this.Title) && !t.IsEquals(this.Author))
{
tit += (tit != "" ? "<br>" : "") + t;
}
} while (start > 0 && end > 0);
title = tit.Replace("<br>", ": ").Replace("<BR>", " - ").Trim();
}
title = UtilityService.ClearTag(title, "img").Trim();
title = UtilityService.RemoveTag(title, "br").Trim();
title = UtilityService.RemoveTag(title, "p").Trim();
title = UtilityService.RemoveTag(title, "i").Trim();
title = UtilityService.RemoveTag(title, "b").Trim();
title = UtilityService.RemoveTag(title, "em").Trim();
title = UtilityService.RemoveTag(title, "strong").Trim();
while (title.IndexOf(" ") > 0)
{
title = title.Replace(" ", " ");
}
while (title.IndexOf("- -") > 0)
{
title = title.Replace("- -", "-");
}
while (title.IndexOf(": -") > 0)
{
title = title.Replace(": -", ":");
}
title = title.Trim().Replace("( ", "(").Replace(" )", ")").Replace("- (", "(").Replace(": :", ":").GetNormalized();
while (title.StartsWith(")") || title.StartsWith("]"))
{
title = title.Right(title.Length - 1).Trim();
}
while (title.EndsWith("(") || title.EndsWith("["))
{
title = title.Left(title.Length - 1).Trim();
}
while (title.StartsWith(":"))
{
title = title.Right(title.Length - 1).Trim();
}
while (title.EndsWith(":"))
{
title = title.Left(title.Length - 1).Trim();
}
if (title.Equals(title.ToUpper()))
{
title = title.ToLower().GetNormalized();
}
var body = UtilityService.RemoveWhitespaces(contents[2].Trim()).Replace(StringComparison.OrdinalIgnoreCase, "\r", "").Replace(StringComparison.OrdinalIgnoreCase, "\n", "").Replace(StringComparison.OrdinalIgnoreCase, "\t", "");
body = UtilityService.RemoveTagAttributes(body, "p");
body = UtilityService.RemoveTagAttributes(body, "div");
body = UtilityService.ClearTag(body, "script");
body = UtilityService.ClearComments(body);
body = UtilityService.RemoveMsOfficeTags(body);
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<div></div>", "</p><p>").Trim();
if (body.IsStartsWith("<div") && !body.IsEndsWith("</div>"))
{
body = body.Remove(0, body.IndexOf(">") + 1);
body = "<p>" + body + "</p>";
}
while (body.IsStartsWith("<div>"))
{
body = body.Substring(5).Trim();
}
while (body.IsEndsWith("</div>"))
{
body = body.Substring(0, body.Length - 6).Trim();
}
start = body.PositionOf("<?xml");
while (start > -1)
{
end = body.PositionOf(">", start);
body = body.Remove(start, end - start + 1);
start = body.PositionOf("<?xml");
}
"strong|em|p|img".Split('|')
.ForEach(tag => body = body.Replace(StringComparison.OrdinalIgnoreCase, "<" + tag, "<" + tag).Replace(StringComparison.OrdinalIgnoreCase, "</" + tag + ">", "</" + tag + ">"));
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<DIV class=\"truyen_text\"></DIV></STRONG>", "</STRONG>\n<p>");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<DIV class=\"truyen_text\"></DIV></EM>", "</EM>\n<p>");
var headingTags = "h1|h2|h3|h4|h5|h6".Split('|');
headingTags.ForEach(tag =>
{
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "><div class=\"truyen_text\"></div>", "<" + tag + "> ").Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "><div class=\"truyen_text\"> </div>", "<" + tag + ">");
body = UtilityService.RemoveTagAttributes(body, tag);
});
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<div class=\"truyen_text\"></div>", "</p><p>").Replace(StringComparison.OrdinalIgnoreCase, "<div class=\"truyen_text\"> </div>", "</p><p>");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<div class=\"truyen_text\">", "<p>").Replace(StringComparison.OrdinalIgnoreCase, "<div", "<p").Replace(StringComparison.OrdinalIgnoreCase, "</div>", "</p>");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "</li></p>", "</li>").Replace(StringComparison.OrdinalIgnoreCase, "<p><li>", "<li>");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<p></ul></p>", "</ul>").Replace(StringComparison.OrdinalIgnoreCase, "<p></ol></p>", "</ol>");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<i class=\"calibre7\"", "<i").Replace(StringComparison.OrdinalIgnoreCase, "<img class=\"calibre1\"", "<img").Replace(StringComparison.OrdinalIgnoreCase, "<b class=\"calibre4\"", "<b");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<p> <b>", "<p><b>").Replace(StringComparison.OrdinalIgnoreCase, ". </b>", ".</b> ").Replace(StringComparison.OrdinalIgnoreCase, ". </i>", ".</i> ");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<p align=\"center\"> <", "<p align=\"center\"><").Replace(StringComparison.OrdinalIgnoreCase, "<p> <", "<p><").Replace(StringComparison.OrdinalIgnoreCase, "<p> ", "<p>");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<p><p>", "<p>").Replace(StringComparison.OrdinalIgnoreCase, "</p></p>", "</p>").Replace(StringComparison.OrdinalIgnoreCase, ". </p> ", ".</p>");
headingTags.ForEach(tag =>
{
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "> <", "<" + tag + "><").Replace(StringComparison.OrdinalIgnoreCase, "> </" + tag + ">", "></" + tag + ">");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "></" + tag + ">", "").Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "> </" + tag + ">", "");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "></p>", "<" + tag + ">").Replace(StringComparison.OrdinalIgnoreCase, "<p></" + tag + ">", "</" + tag + ">");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "><strong>", "<" + tag + ">").Replace(StringComparison.OrdinalIgnoreCase, "</strong></" + tag + ">", "</" + tag + ">");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "><em>", "<" + tag + ">").Replace(StringComparison.OrdinalIgnoreCase, "</em></" + tag + ">", "</" + tag + ">");
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<p><" + tag + ">", "<" + tag + ">").Replace(StringComparison.OrdinalIgnoreCase, "</" + tag + "></p>", "</" + tag + ">").Replace(StringComparison.OrdinalIgnoreCase, "<" + tag + "></p>", "");
});
headingTags.ForEach(tag =>
{
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<p><" + tag + ">", "<" + tag + ">").Replace(StringComparison.OrdinalIgnoreCase, "</" + tag + "></p>", "</" + tag + ">");
start = body.PositionOf("<" + tag + ">");
while (start > -1)
{
end = body.PositionOf("</" + tag + ">", start + 1);
var heading = body.Substring(start + 4, end - start - 4);
body = body.Remove(start, end - start + 5);
var pos = heading.PositionOf("<");
while (pos > -1)
{
end = heading.PositionOf(">", pos);
if (end > 0)
{
heading = heading.Remove(pos, end - pos + 1);
}
pos = heading.PositionOf("<");
}
body = body.Insert(start, "<" + tag + ">" + heading + "</" + tag + ">");
start = body.PositionOf("<" + tag + ">", start + 1);
}
});
start = body.PositionOf("<p id=\"chuhoain\"");
while (start > -1)
{
end = body.PositionOf("</span><p>", start);
var img = body.PositionOf("<img", start);
if (start > -1 && end > start && img > start)
{
var imgStart = body.PositionOf("src=\"", img) + 5;
var imgEnd = -1;
if (imgStart < 0)
{
imgStart = body.PositionOf("src='", img) + 5;
imgEnd = body.PositionOf("'", imgStart);
}
else
{
imgEnd = body.PositionOf("\"", imgStart);
}
var imgChar = body.Substring(imgStart, imgEnd - imgStart);
body = body.Remove(start, end - start + 10);
body = body.Insert(start, "<p>" + this.GetImageCharacter(imgChar));
}
start = body.PositionOf("<p id=\"chuhoain\"", start + 1);
}
start = body.PositionOf("<img");
while (start > -1)
{
end = body.PositionOf(">", start + 1);
var img = body.PositionOf("src=\"https://vnthuquan.net/userfiles/images/chu%20cai/cotich", start);
if (img < 0)
{
img = body.PositionOf("src='https://vnthuquan.net/userfiles/images/chu%20cai/cotich", start);
}
if (img > -1 && end > img)
{
end = body.PositionOf("\"", img + 5);
if (end < 0)
{
end = body.PositionOf("'", img + 5);
}
var imgChar = body.Substring(img + 5, end - img + 5);
end = body.PositionOf("<p>", start);
if (end < 0)
{
end = body.PositionOf(">", start) + 1;
}
else
{
end += 3;
}
string str = body.Substring(start, end - start);
body = body.Remove(start, end - start);
body = body.Insert(start, this.GetImageCharacter(imgChar));
}
start = body.PositionOf("<img", start + 1);
}
if (body.Equals("</p><p>"))
{
body = "";
}
else
{
body = this.NormalizeChapterBody(body);
body = body.Replace(StringComparison.OrdinalIgnoreCase, "<h1>", "<h2>").Replace(StringComparison.OrdinalIgnoreCase, "</h1>", "</h2>");
}
return(new List <string> {
title, body
});
}
