public BookCardInfo Build(string content)
{
//reset
mState = BookCardInfoBuilderState.TitleAuthor;
mBookCardInfo = new BookCardInfo();
mContentAndIsbnTexts.Clear();
content = HREF_REGEX.Replace(content, "");
content = HtmlEntityCorrect.Encode(content);
using (mReader = XmlReader.Create(new MemoryStream(Encoding.UTF8.GetBytes(content))))
{
while (mReader.Read())
{
switch (mReader.NodeType)
{
case XmlNodeType.Text:
this.onText(HtmlEntityCorrect.Decode(mReader.Value));
break;
case XmlNodeType.Element:
this.onElement(HtmlEntityCorrect.Decode(mReader.Name));
break;
}
}
}
return(mBookCardInfo);
}
public void Encode()
{
string test = "Post & Put & & & &";
test = HtmlEntityCorrect.Encode(test);
Assert.AreEqual("Post & Put & & & &", test);
test = HtmlEntityCorrect.Decode(test);
Assert.AreEqual("Post & Put & & & &", test);
}
private void onElementA()
{
switch (mState)
{
case BookCardInfoBuilderState.Publisher:
mBookCardInfo.Publisher = HtmlEntityCorrect.Decode(mReader.ReadString());
mState = BookCardInfoBuilderState.ContentAndIsbn_Br;
break;
//遇到第二个a元素则内容和ISBN结束
case BookCardInfoBuilderState.ContentAndIsbn_0:
mState = BookCardInfoBuilderState.ContentAndIsbn_1;
break;
case BookCardInfoBuilderState.ContentAndIsbn_1:
getContentAndIsbn();
mState = BookCardInfoBuilderState.Other;
break;
}
}
public List <DeptInfo> Build(string content)
{
//reset
mDeptInfos = new List <DeptInfo>();
content = HtmlEntityCorrect.Encode(content);
using (mReader = XmlReader.Create(new MemoryStream(Encoding.UTF8.GetBytes(content))))
{
DeptInfo deptInfo = null;
while (mReader.Read())
{
if (mReader.NodeType == XmlNodeType.Element)
{
if (mReader.Name == "tr")
{
deptInfo = new DeptInfo();
mDeptInfos.Add(deptInfo);
mState = DeptInfoBuilderState.DeptPlace;
}
else if (mReader.Name == "a")
{
deptInfo.DeptPlace = HtmlEntityCorrect.Decode(mReader.ReadString().Trim());
mState = DeptInfoBuilderState.Index;
}
else if (mReader.Name == "td")
{
switch (mState)
{
case DeptInfoBuilderState.Index:
deptInfo.Index = HtmlEntityCorrect.Decode(mReader.ReadString()).Trim();
mState = DeptInfoBuilderState.Register;
break;
case DeptInfoBuilderState.Register:
deptInfo.Register = HtmlEntityCorrect.Decode(mReader.ReadString()).Trim();
mState = DeptInfoBuilderState.Volume;
break;
case DeptInfoBuilderState.Volume:
deptInfo.Volume = HtmlEntityCorrect.Decode(mReader.ReadString()).Trim();
mState = DeptInfoBuilderState.Year;
break;
case DeptInfoBuilderState.Year:
deptInfo.Year = HtmlEntityCorrect.Decode(mReader.ReadString()).Trim();
mState = DeptInfoBuilderState.Status;
break;
case DeptInfoBuilderState.Status:
deptInfo.Status = HtmlEntityCorrect.Decode(mReader.ReadString()).Trim();
mState = DeptInfoBuilderState.Type;
break;
case DeptInfoBuilderState.Type:
deptInfo.Type = HtmlEntityCorrect.Decode(mReader.ReadString()).Trim();
mState = DeptInfoBuilderState.Other;
break;
default:
break;
}
}
}
}
}
return(mDeptInfos);
}
public static List <Book> Parse(string tbody)
{
List <Book> books = new List <Book>();
Book book = null;
tbody = HtmlEntityCorrect.Encode(tbody);
using (var reader = XmlReader.Create(new MemoryStream(Encoding.UTF8.GetBytes(tbody))))
{
string currentElementName = null;
int tdCount = 0;
while (reader.Read())
{
switch (reader.NodeType)
{
case XmlNodeType.Element:
currentElementName = reader.Name;
if (reader.Name == "tr")
{
book = new Book();
tdCount = 0;
}
else if (reader.Name == "td")
{
tdCount++;
}
else if (reader.Name == "a")
{
book.Url = reader.GetAttribute("href");
}
break;
case XmlNodeType.Text:
if (currentElementName == "a")
{
book.Title = HtmlEntityCorrect.Decode(reader.Value);
}
else if (currentElementName == "td")
{
if (tdCount == 3)
{
book.Author = HtmlEntityCorrect.Decode(reader.Value);
}
else if (tdCount == 4)
{
try
{
book.Publisher = HtmlEntityCorrect.Decode(reader.Value);
}
catch (Exception e)
{
Console.WriteLine(e);
}
}
else if (tdCount == 5)
{
book.PublishYear = HtmlEntityCorrect.Decode(reader.Value);
}
else if (tdCount == 6)
{
book.Index = HtmlEntityCorrect.Decode(reader.Value);
}
else if (tdCount == 7)
{
book.Total = Int32.Parse(HtmlEntityCorrect.Decode(reader.Value));
}
else if (tdCount == 8)
{
book.Available = Int32.Parse(HtmlEntityCorrect.Decode(reader.Value));
}
}
break;
case XmlNodeType.EndElement:
if (reader.Name == "tr")
{
books.Add(book);
}
break;
}
}
}
return(books);
}