/// <summary>
/// 图片相对路径改为全路径
/// </summary>
/// <param name="documentNode"></param>
/// <param name="InternalRealUrl"></param>
protected static void ReplaceIncorrectImageSrc(HtmlNode documentNode, string InternalRealUrl)
{
HtmlNodeCollection imgNode = documentNode.SelectNodes("//img");
if (imgNode == null)
{
return;
}
foreach (var item in imgNode)
{
var url = item.GetAttributeValue("src", "");
var realUrl = HTMLUtil.GetFullURL(InternalRealUrl, url);
item.SetAttributeValue("src", realUrl);
//documentNode.InnerHtml=documentNode.InnerHtml.Replace(url, realUrl);
}
}
/// <summary>
/// 解析开始页
/// </summary>
/// <param name="documentNode"></param>
/// <returns></returns>
private Hashtable parseStartupPage(HtmlNode documentNode)
{
Hashtable ht = new Hashtable();
Regex tempReg = null;
Match tempMatch = null;
HtmlNode linkNodes = documentNode.SelectSingleNode("//meta[@property='og:title']");
if (linkNodes != null)
{
var title = linkNodes.GetAttributeValue("content", "");
ht.Add(CollectionFieldName.Novel_Name, title);
}
linkNodes = documentNode.SelectSingleNode("//div[@class='pic text-center']/img");
if (linkNodes != null)
{
var imgUrl = linkNodes?.GetAttributeValue("src", "");
imgUrl = HTMLUtil.GetFullURL(InternalRealUrl, imgUrl);
ht.Add(CollectionFieldName.Novel_CoverImgs, imgUrl);
}
linkNodes = documentNode.SelectSingleNode("//meta[@property='og:novel:category']");
if (linkNodes != null)
{
var tag = linkNodes?.GetAttributeValue("content", "");
ht.Add(CollectionFieldName.Novel_Tag, tag);
}
var url = InternalRealUrl;
ht.Add(CollectionFieldName.Url, url);
tempReg = new Regex(@"/(\d+).html");
tempMatch = tempReg.Match(InternalRealUrl);
if (tempMatch.Success)
{
ht.Add(CollectionFieldName.Novel_UniqueFlag, tempReg.Replace(tempMatch.Value, "$1"));
}
Hashtable returndata = new Hashtable();
if (ht != null)
{
returndata.Add(CollectionFieldName.BookInfo, ht);
}
return(returndata);
}
/// <summary>
/// 修改占位图片路径
/// </summary>
/// <param name="documentNode"></param>
/// <param name="InternalRealUrl"></param>
protected static void ReplacePlaceholderImageSrc(HtmlNode documentNode, string InternalRealUrl, string PlaceholderAttribute)
{
HtmlNodeCollection imgNode = documentNode.SelectNodes("//img");
if (imgNode == null)
{
return;
}
foreach (var item in imgNode)
{
var url = item.GetAttributeValue(PlaceholderAttribute, "");
if (!string.IsNullOrEmpty(url))
{
var realUrl = HTMLUtil.GetFullURL(InternalRealUrl, url);
item.SetAttributeValue("src", realUrl);
item.Attributes.Remove(PlaceholderAttribute);
}
}
}
/// <summary>
/// 解析开始页
/// </summary>
/// <param name="documentNode"></param>
/// <returns></returns>
private Hashtable parseStartupPage(HtmlNode documentNode)
{
Hashtable ht = new Hashtable();
Regex tempReg = null;
Match tempMatch = null;
HtmlNode linkNodes = documentNode.SelectSingleNode("//span[@class='book_name']");
if (linkNodes != null)
{
var title = linkNodes.InnerText;
ht.Add(CollectionFieldName.Novel_Name, title);
}
linkNodes = documentNode.SelectSingleNode("//div[@class='books_bar clear']");
if (linkNodes != null)
{
var imgUrl = linkNodes.SelectSingleNode("div[@class='lr_list']/img")?.GetAttributeValue("src", "");
ht.Add(CollectionFieldName.Novel_CoverImgs, imgUrl);
var tag = linkNodes.SelectSingleNode("//ul[@class='book_list']/li[4]")?.InnerText;
ht.Add(CollectionFieldName.Novel_Tag, tag.Replace("类别：", ""));
var statusName = linkNodes.SelectSingleNode("//ul[@class='book_list']/li[3]")?.InnerText;
if (!string.IsNullOrEmpty(statusName) && statusName.Contains("完结"))
{
ht.Add(CollectionFieldName.Novel_Status, BookStatus.BookStatus_Finish);
}
else
{
ht.Add(CollectionFieldName.Novel_Status, BookStatus.BookStatus_Update);
}
var ContentLenStr = linkNodes.SelectSingleNode("//ul[@class='book_list']/li[2]")?.InnerText;
int ContentLen = 0;
if (!string.IsNullOrEmpty(ContentLenStr))
{
tempReg = new Regex(@"(\d+.\d+)|(\d+)");
tempMatch = tempReg.Match(ContentLenStr);
if (tempMatch.Success)
{
if (ContentLenStr.Contains("万"))
{
ContentLen = (int)(Convert.ToDouble(tempMatch.Value) * 10000);
}
else if (ContentLenStr.Contains("千"))
{
ContentLen = (int)(Convert.ToDouble(tempMatch.Value) * 1000);
}
else
{
ContentLen = Convert.ToInt32(tempMatch.Value);
}
}
ht.Add(CollectionFieldName.Novel_ContentLen, ContentLen);
}
}
linkNodes = documentNode.SelectSingleNode("//a[@class='more_link']");
if (linkNodes != null)
{
var url = linkNodes.GetAttributeValue("href", "");
url = HTMLUtil.GetFullURL(InternalRealUrl, url);
ht.Add(CollectionFieldName.Url, url);
}
linkNodes = documentNode.SelectSingleNode("//div[@id='divDescription']");
if (linkNodes != null)
{
var intr = linkNodes.InnerHtml;
ht.Add(CollectionFieldName.Novel_Intr, intr);
}
tempReg = new Regex(@"/book/(\d+).html");
tempMatch = tempReg.Match(InternalRealUrl);
if (tempMatch.Success)
{
ht.Add(CollectionFieldName.Novel_UniqueFlag, tempReg.Replace(tempMatch.Value, "$1"));
}
Hashtable returndata = new Hashtable();
if (ht != null)
{
returndata.Add(CollectionFieldName.BookInfo, ht);
}
return(returndata);
}
/// <summary>
/// 解析开始页
/// </summary>
/// <param name="documentNode"></param>
/// <returns></returns>
private Hashtable parseStartupPage(HtmlNode documentNode)
{
Hashtable ht = new Hashtable();
Regex tempReg = null;
Match tempMatch = null;
HtmlNode linkNodes = documentNode.SelectSingleNode("//meta[@property='og:title']");
if (linkNodes != null)
{
var title = linkNodes.GetAttributeValue("content", "");
ht.Add(CollectionFieldName.Novel_Name, title);
}
linkNodes = documentNode.SelectSingleNode("//div[@class='pic text-center']/img");
if (linkNodes != null)
{
var imgUrl = linkNodes?.GetAttributeValue("src", "");
imgUrl = HTMLUtil.GetFullURL(InternalRealUrl, imgUrl);
ht.Add(CollectionFieldName.Novel_CoverImgs, imgUrl);
}
linkNodes = documentNode.SelectSingleNode("//meta[@property='og:novel:category']");
if (linkNodes != null)
{
var tag = linkNodes?.GetAttributeValue("content", "");
ht.Add(CollectionFieldName.Novel_Tag, tag);
}
//linkNodes = documentNode.SelectSingleNode("//meta[@property='og:novel:category']");
//if (linkNodes != null)
//{
// var statusName = linkNodes?.GetAttributeValue("content","");
// if (!string.IsNullOrEmpty(statusName) && statusName.Contains("完结"))
// ht.Add(CollectionFieldName.Novel_Status, BookStatus.BookStatus_Finish);
// else
// ht.Add(CollectionFieldName.Novel_Status, BookStatus.BookStatus_Update);
//}
//linkNodes = documentNode.SelectSingleNode("//meta[@property='og:novel:category']");
//if (linkNodes != null)
//{
// var ContentLenStr = linkNodes.SelectSingleNode("div[@class='fn-clear']/ul/li[4]")?.InnerText;
// int ContentLen = 0;
// if (!string.IsNullOrEmpty(ContentLenStr))
// {
// tempReg = new Regex(@"(\d+.\d+)|(\d+)");
// tempMatch = tempReg.Match(ContentLenStr);
// if (tempMatch.Success)
// {
// if (ContentLenStr.Contains("万"))
// ContentLen = (int)(Convert.ToDouble(tempMatch.Value) * 10000);
// if (ContentLenStr.Contains("千"))
// ContentLen = (int)(Convert.ToDouble(tempMatch.Value) * 1000);
// }
// ht.Add(CollectionFieldName.Novel_ContentLen, ContentLen);
// }
//}
//linkNodes = documentNode.SelectSingleNode("//div[@class='panel']/a[@class='btn block white']");
//if (linkNodes != null)
//{
// var url = linkNodes.GetAttributeValue("href", "");
// ht.Add(CollectionFieldName.Url, url);
//}
var url = InternalRealUrl;
ht.Add(CollectionFieldName.Url, url);
//linkNodes = documentNode.SelectSingleNode("//div[@class='book_intro']/p[@id='summary']");
//if (linkNodes != null)
//{
// var intr = linkNodes.InnerHtml;
// ht.Add(CollectionFieldName.Novel_Intr, intr);
//}
tempReg = new Regex(@"/(\d+).html");
tempMatch = tempReg.Match(InternalRealUrl);
if (tempMatch.Success)
{
ht.Add(CollectionFieldName.Novel_UniqueFlag, tempReg.Replace(tempMatch.Value, "$1"));
}
Hashtable returndata = new Hashtable();
if (ht != null)
{
returndata.Add(CollectionFieldName.BookInfo, ht);
}
return(returndata);
}
/// <summary>
/// 获取完整url地址
/// </summary>
/// <param name="relativeURL"></param>
/// <returns></returns>
protected string GetFullURL(string relativeURL)
{
return(HTMLUtil.GetFullURL(baseUrl, relativeURL));
}