public override DataContainer ParseHandler(DataResponse dr)
{
var html = dr.Text;
//书的名称
var name = parseTitle.Match(html).Groups[1].Value;
var baseInfoHtml = parseBaseInfoBody.Match(html).Value;
//出版日期, 没有出版日期说明不是书籍页面
#region 日期处理
var dateMatch = parsePublishDate.Match(baseInfoHtml).Groups;
var dateString = string.IsNullOrWhiteSpace(dateMatch[1].Value) ? dateMatch[2].Value : dateMatch[1].Value;
if (string.IsNullOrWhiteSpace(dateString))
{
throw new ParseErrorException("This is not a Book Page");
}
DateTime publishDate;
try
{
publishDate = DateTime.Parse(dateString);
}
catch (Exception)
{
throw new ParseErrorException("This is not a Book Page");
}
#endregion
//链接
var hrefs = parseHref.Matches(baseInfoHtml).ForEach((Match m) => new Hyperlink(m.Groups[1].Value, m.Groups[2].Value));
//介绍
var introduction = parseIntroduction.Match(baseInfoHtml).Groups[1].Value;
var metaHtml = parseMetaBody.Match(html).Value;
//分类
var categorys = parseCategory.Matches(metaHtml).ForEach((Match m) => m.Groups[1].Value);
//标签
var tags = parseTag.Matches(metaHtml).ForEach((Match m) => m.Groups[1].Value);
//生成Book对象
var book = new Book(name, publishDate, hrefs.ElementAt(0).Href) { Introduction = introduction };
book.Hrefs.AddRange(hrefs.Skip(1));
book.Categorys.AddRange(categorys);
book.Tags.AddRange(tags);
Logger.Info($"Parse book: {book.Name}");
return new DataContainer(dr.GUID, book);
}
public Book GetBook(long bookId)
{
using (var r = DB.ExecuteReaderOne("select * from book where book.id=?", bookId))
{
if (r == null)
{
return null;
}
var book = new Book(r.GetString(1), r.GetDateTime(2), r.GetString(3)) { Id = r.GetInt64(0) };
book.Hrefs.AddRange(GetHrefs(book.Id));
book.Categorys.AddRange(GetCategorys(book.Id));
book.Tags.AddRange(GetTags(book.Id));
return book;
}
}
public long Insert(Book book)
{
//添加book的基本信息
var bookId = DB.ExecuteScalar<long>("insert into book values(NULL,?,date(?),?,?); select last_insert_rowid()",
book.Name,
book.PublishDate,
book.OfficialHref,
book.Introduction);
//添加book的标签、分类和超链接
book.Hrefs.ForEach(href => AddHrefToBook(bookId, href));
book.Categorys.ForEach(category => AddCategoryToBook(bookId, category));
book.Tags.ForEach(tag => AddTagToBook(bookId, tag));
Logger.Debug($"Insert book {book.Name} successful");
return bookId;
}
public bool Equals(Book other)
{
return this.Id == other.Id;
}