/// <summary> Create a new tag node.
/// Note that the attributes vector contains at least one element,
/// which is the tag name (standalone attribute) at position zero.
/// This can be used to decide which type of node to create, or
/// gate other processing that may be appropriate.
/// </summary>
/// <param name="page">The page the node is on.
/// </param>
/// <param name="start">The beginning position of the tag.
/// </param>
/// <param name="end">The ending positiong of the tag.
/// </param>
/// <param name="attributes">The attributes contained in this tag.
/// </param>
/// <returns> A tag node comprising the indicated characters from the page.
/// </returns>
public virtual ITag CreateTagNode(Page page, int start, int end, System.Collections.ArrayList attributes)
{
TagAttribute attribute;
System.String id;
ITag prototype;
ITag ret;
ret = null;
if (0 != attributes.Count)
{
attribute = (TagAttribute) attributes[0];
id = attribute.GetName();
if (null != id)
{
try
{
id = id.ToUpper(new System.Globalization.CultureInfo("en"));
if (!id.StartsWith("/"))
{
if (id.EndsWith("/"))
id = id.Substring(0, (id.Length - 1) - (0));
prototype = (ITag) mBlastocyst[id];
if (null != prototype)
{
ret = (ITag) prototype.Clone();
ret.Page = page;
ret.StartPosition = start;
ret.EndPosition = end;
ret.AttributesEx = attributes;
}
}
}
catch
{
// default to creating a generic one
}
}
}
if (null == ret)
{
// generate a generic node
try
{
ret = (ITag) TagPrototype.Clone();
ret.Page = page;
ret.StartPosition = start;
ret.EndPosition = end;
ret.AttributesEx = attributes;
}
catch
{
ret = new TagNode(page, start, end, attributes);
}
}
return (ret);
}
/// <summary>
/// 给定根节点，填充相似转发微博的信息
/// </summary>
/// <param name="currentPage">微博所在页面序号</param>
/// <param name="i">微博在所在页面中的流水号</param>
/// <param name="similarFeedDiv">包含相似微博的div标记</param>
/// <param name="originalAuthor">转发微博的原作者</param>
/// <param name="feedContent">转发微博的内容</param>
/// <returns>返回填充好的一个Feed实例</returns>
private Feed GetSimilarFeed(int currentPage, int i, TagNode similarFeedDiv, string originalAuthor, string feedContent)
{
Feed feed = new Feed();
serialNumber++;
feed.Page = currentPage;
feed.Number = serialNumber;
feed.ReFeedOrNot = true;
#region 获取转发相似微博的作者
NodeList feedAuthorNodeList = similarFeedDiv.Children.ExtractAllNodesThatMatch(feedAuthorFilter, true);
if (feedAuthorNodeList.Size() == 1)
{
ATag feedAuthorTag = (ATag)feedAuthorNodeList[0].Children[1];
string author = feedAuthorTag.GetAttribute("TITLE");
feed.Author = author;
//如果存在，则获取该作者的备注名
INode remarkNameNode = feedAuthorTag.NextSibling;
if (remarkNameNode.GetType().Equals(typeof(Span)))
{
string remarkName = ((Span)remarkNameNode).StringText;
//去掉前后括号
remarkName = remarkName.Substring(1, remarkName.Length - 2);
feed.RemarkName = remarkName;
}
}
else
{
Console.WriteLine("第" + i + "条微博中，判断转发类似微博的作者的标准出错！");
}
#endregion
#region 获取转发相似微博的转发理由
NodeList reFeedReasonNodeList = similarFeedDiv.Children.ExtractAllNodesThatMatch(feedContentFilter, true);
if (reFeedReasonNodeList.Size() == 1)
{
feed.ReFeedReason = GetContentFromChildren(feed, reFeedReasonNodeList[0], false);
}
else
{
Console.WriteLine("第" + i + "条微博中，判断转发类似微博的转发理由的标准出错！");
}
#endregion
//因为是转发相似微博，所以原微博的作者和内容就通过参数传入
feed.OriginalAuthor = originalAuthor;
feed.Content = feedContent;
//获取转发类似微博的发送地点信息
feed.Location = GetLocationInfo(i, similarFeedDiv);
//获取转发类似微博赞数
feed.LikeCount = GetFeedLikeInfo(i, similarFeedDiv);
//获取转发类似微博的转发数
feed.ReFeedCount = GetFeedForwardCount(i, similarFeedDiv);
//获取转发类似微博的评论数
feed.CommentCount = GetFeedCommentCount(i, similarFeedDiv);
//获取转发类似微博的发送时间
feed.Time = GetFeedTimeInfo(i, similarFeedDiv);
//获取转发类似微博的发送方式
feed.Device = GetFeedSendTypeInfo(i, similarFeedDiv);
return feed;
}
/// <summary> Create a tag like the one provided.</summary>
/// <param name="tag">The tag to emulate.
/// </param>
/// <param name="scanner">The scanner for this tag.
/// </param>
public TagNode(TagNode tag, TagScanner scanner):this(tag.Page, tag.TagBegin, tag.TagEnd, tag.AttributesEx)
{
ThisScanner = scanner;
}
/// <summary>
/// 给定根结点，填充转发微博相关的各种信息
/// </summary>
/// <param name="i">该微博在所在页面中的流水号</param>
/// <param name="feed">保存该微博的Feed实例</param>
/// <param name="reFeedDiv">包含转发微博的div标记</param>
/// <param name="feedDiv">包含原微博的div标记</param>
private void GetReFeedInfo(int i, Feed feed, TagNode reFeedDiv, TagNode feedDiv)
{
//标记ReFeedOrNot为true，表明是转发微博
feed.ReFeedOrNot = true;
//标识是否出现“转发微博已被删除”的情况
bool reFeedIsDeleted = false;
#region 获取原微博作者
NodeList reFeedOriginalAuthorNodeList = reFeedDiv.Children.ExtractAllNodesThatMatch(reFeedAuthorFilter, true);
if (reFeedOriginalAuthorNodeList.Size() == 1)
{
INode reFeedOriginalAuthorNode = reFeedOriginalAuthorNodeList[0];
//由于包含原微博作者的链接标记与得到的子div相对位置不定（某些情况下可能会有空文本标记，很奇怪= =），所以采用遍历判断标记类型的办法
for (int j = 0; j < reFeedOriginalAuthorNode.Children.Size(); j++)
{
INode reFeedOriginalAuthorCandidate = reFeedOriginalAuthorNode.Children[j];
if (reFeedOriginalAuthorCandidate.GetType().Equals(typeof(ATag)))
{
feed.OriginalAuthor = ((ATag)reFeedOriginalAuthorCandidate).GetAttribute("TITLE");
break;
}
}
}
else
{
NodeList deletedFeedList = reFeedDiv.Children.ExtractAllNodesThatMatch(refeedDeletedFilter1, true);
if (deletedFeedList.Size() > 0)
{
reFeedIsDeleted = true;
feed.OriginalAuthor = "Unknown";
}
else
{
Console.WriteLine("第" + i + "条微博中，判断转发微博作者的标准出错！");
}
}
#endregion
#region 获取原微博内容
NodeList reFeedContentNodeList = reFeedDiv.Children.ExtractAllNodesThatMatch(reFeedContentFilter, true);
if (reFeedContentNodeList.Size() == 1)
{
//不清楚em是什么类型的节点，所以直接传递reFeedContentNodeList.Children给函数，让函数对其中每个元素进行遍历处理
//在个人主页中，好像又没有em节点了= =瞎了……
feed.Content = GetContentFromChildren(feed, reFeedContentNodeList[0], false);
}
else
{
if (reFeedIsDeleted)
{
feed.Content = "微博已删除";
}
else
{
Console.WriteLine("第" + i + "条微博中，判断转发微博内容的标准出错！");
}
}
#endregion
#region 获取本条微博内容作为转发理由
NodeList reFeedReasonNodeList = feedDiv.Children.ExtractAllNodesThatMatch(feedContentFilter, true);
//注意：如果含有“还有X条对原微博的转发”的内容，那么此处reFeedReasonNodeList的数量应该等于(1 + similarFeedCount)，但是考虑到之前已经有了多次判断相等的过程，所以此处直接调用第一个（即下标为0的元素）即可
feed.ReFeedReason = GetContentFromChildren(feed, reFeedReasonNodeList[0], false);
#endregion
}