private static void createDataColumns(CQ htmlTableColumns, DataTable dataTable)
{
for (var columnIterator = 0; columnIterator < htmlTableColumns.Count(); columnIterator++)
{
var domObject = htmlTableColumns.Get(columnIterator);
string columnName;
string caption;
if (domObject.InnerText.IsNullOrEmpty())
{
columnName = domObject.HasChildren
? domObject.ChildNodes.Item(0).InnerText
: $"limiter{columnIterator}";
if (domObject.HasChildren && domObject.ChildNodes.Item(0).NodeName == "ABBR")
{
caption = domObject.ChildNodes.Item(0).OuterHTML;
caption = caption.Replace("<abbr title=", "").Replace($">{columnName}</abbr>", "");
caption = caption.Substring(1, caption.Length - 2);
}
else
{
caption = columnName;
}
}
else
{
columnName = domObject.InnerText;
caption = columnName;
}
dataTable.Columns.Add(new DataColumn {
ColumnName = columnName, Caption = caption
});
}
}
protected override void Regist(System.IO.MemoryStream contentStream, string[] args)
{
this._count++;
string content = System.Text.Encoding.UTF8.GetString(contentStream.ToArray());
CQ cq = new CQ(content, HtmlParsingMode.Auto, HtmlParsingOptions.Default, DocType.HTML5);
var poem = cq.Select(".poem");
var contents = poem.Select(".content");
List <M_Poem2> poems = new List <M_Poem2>();
List <M_PoemImage> imgs = new List <M_PoemImage>();
Regex regex = new Regex(@"(?<title>[^（]+)（(?<dynasty>[^·]+)·(?<auth>[^）]+)）\s(?<gelu>.*)(?=押)(押(?<yun>[^韵])韵){0,1}");
for (int i = 0; i < contents.Length; i++)
{
CQ titleSmallNode = poem.Select(".title:eq(" + i + ") > .small");
string smallNode = "";
if (titleSmallNode.Length > 0)
{
smallNode = titleSmallNode.Text();
titleSmallNode.Remove();
}
string titleText = poem.Select(".title:eq(" + i + ")").Text();
Match m = regex.Match(titleText);
string title = titleText;
string author = "";
string dynasty = "";
string yun = "";
if (m.Success)
{
title = m.Groups["title"].Value;
author = m.Groups["auth"].Value;
dynasty = m.Groups["dynasty"].Value;
if (m.Groups["yun"] != null)
{
yun = m.Groups["yun"].Value;
}
}
string titleNote = poem.Select(".title:eq(" + i + ") ~ .titleNote").Text();
if (!string.IsNullOrEmpty(smallNode))
{
titleNote += "\n" + smallNode;
}
string mainBody = poem.Select(".content:eq(" + i + ")").Text();
string label = poem.Select(".content:eq(" + i + ") + .footer > .label").Text();
string footer = poem.Select(".content:eq(" + i + ") + .footer > .comment").Text();
string commentBar = poem.Select("div.comment").Text();
string html = System.Web.HttpUtility.HtmlDecode(poem.Html());
//入力チェック
//Title
if (!string.IsNullOrEmpty(title) && title.Length > 250)
{
if (string.IsNullOrWhiteSpace(titleNote))
{
titleNote = title;
}
title = title.Substring(0, 250);
}
//Dynasty
if (!string.IsNullOrEmpty(title) && dynasty.Length > 50)
{
dynasty = dynasty.Substring(0, 50);
}
poems.Add(new M_Poem2()
{
ID = int.Parse(args[0]),
SubID = i,
Title = title,
Yun = yun,
TitleNote = titleNote,
MainBody = mainBody,
Author = author,
Dynasty = dynasty,
Footer = label + footer,
Comment = commentBar,
Html = html
});
if (this.Report != null)
{
Dictionary <string, string> result = new Dictionary <string, string>();
result["count"] = this._count.ToString();
result["message"] = string.Format("ID:{0}【{1}】{2}", args[0], title, author);;
result["content"] = mainBody;
this.Report(this.Parsentage, result);
}
}
//参照イメージ
CQ refImgs = poem.Select(".picSSIcon , .picSSOn");
for (int i = 0; i < refImgs.Length; i++)
{
string title = refImgs.Get(i).Attributes["title"];
string url = "http://sou-yun.com" + refImgs.Get(i).Attributes["src"];
imgs.Add(new M_PoemImage()
{
ID = int.Parse(args[0]),
SubID = 0,
ImageId = i + 1,
Title = title,
ImageUrl = url
});
}
LinqSqlHelp.AddPoem2(poems, imgs);
}
