string parse(string source)
{
StringWriter stringWriter = new StringWriter();
string result;
gen = new RSSFeedGenerator(stringWriter);
//-------------------------------------------------
// here is set values for header information for the RSS feed
//-------------------------------------------------
gen.Title = cate;
gen.Description = "";
gen.LastBuildDate = DateTime.Now;
gen.Link = "http://sbhhbs.com";
//gen.Category = "The optional category";
gen.PubDate = DateTime.Now;
// write the header of the RSS feed document
gen.WriteStartDocument();
// write the start channel for the RSS feed Items
gen.WriteStartChannel();
Regex qariRegex = new Regex("<a href='newsshow\\.aspx\\?sn=([0-9]*)'>.([^<]*)<[^(]*\\(([^)]*)\\)");
MatchCollection mc = qariRegex.Matches(source);
foreach (Match match in mc)
{
string newsid = match.Groups[1].Value;
string title = match.Groups[2].Value;
string date = match.Groups[3].Value;
string content = "";
var objList = parseEngine.GetObjectsWithQuery("RSS", new { url = newsid });
bool cateresult = false;
if (objList != null && objList.Length > 0)
{
foreach (var obj in objList)
{
if (cate.Equals(obj["category"]))
{
cateresult = true;
break;
}
}
if(cateresult)
System.Console.WriteLine("Old item:" + title);
}
if(!cateresult)
{
WebClient client = new WebClient();
client.Encoding = Encoding.UTF8;
String subURL = "http://www.tongji-uni.com/newsshow.aspx?sn=" + newsid;
string downloadString = client.DownloadString(subURL);
NSRange start = rangeOfStr(downloadString, "<table cellpadding=\"0\" cellspacing=\"2\" border=\"0\" width=\"100%\" class=\"wh\">");
NSRange end = rangeOfStr(downloadString, "<P><div align=right><a href=\"default.aspx\">【返回】</a></div></P>");
content = downloadString.Substring(start.location, end.location - start.location) + "</td></tr></table>";
var testObject = new Parse.ParseObject("RSS");
testObject["category"] = cate;
testObject["url"] = newsid;
testObject["title"] = title;
testObject["newsTime"] = String.Format("{0:yyyy-MM-dd HH:mm:ss}", DateTime.Parse(date));
testObject["content"] = saveTextToParse(content);
//Create a new object
testObject = parseEngine.CreateObject(testObject);
System.Console.WriteLine("New item:" + title);
}
gen.WriteItem(
title,
newsid,
"",
"sbhhbs",
"",
"",
"",
DateTime.Parse(date),
"",
"",
"",
"");
}
// end the channel
gen.WriteEndChannel();
try
{
// end the document
gen.WriteEndDocument();
}
catch (System.Exception ex)
{
}
finally
{
// close the rss generator and dispay to screen
gen.Close();
result = stringWriter.ToString();
}
//result = StringWriter.to
return result;
}
string parse(string source)
{
source = convertToUTF8fromGB2312(source);
StringWriter stringWriter = new StringWriter();
string result;
gen = new RSSFeedGenerator(stringWriter);
//-------------------------------------------------
// here is set values for header information for the RSS feed
//-------------------------------------------------
gen.Title = cate;
gen.Description = "";
gen.LastBuildDate = DateTime.Now;
gen.Link = "http://sbhhbs.com";
//gen.Category = "The optional category";
gen.PubDate = DateTime.Now;
// write the header of the RSS feed document
gen.WriteStartDocument();
// write the start channel for the RSS feed Items
gen.WriteStartChannel();
Regex qariRegex = new Regex("<a href=\"news_detail\\.asp\\?id=([0-9]*)\">([^<]*)");
MatchCollection mc = qariRegex.Matches(source);
foreach (Match match in mc)
{
string newsid = match.Groups[1].Value;
string title = match.Groups[2].Value;
WebClient client = new WebClient();
client.Encoding = Encoding.Default;
String subURL = "http://old.tongji-caup.org/student/news_detail.asp?id=" + newsid;
string downloadString = client.DownloadString(subURL);
downloadString = convertToUTF8fromGB2312(downloadString);
string time = "2012-10-3";
NSRange start = rangeOfStr(downloadString, "发布时间：");
NSRange end = rangeOfStr(downloadString, " 信息来源：");
time = downloadString.Substring(start.location + start.length, end.location - start.location - 5);
start = rangeOfStr(downloadString, "<td class=\"TLE\">");
end = rangeOfStr(downloadString, "<!--内容-->");
string content = downloadString.Substring(start.location, end.location - start.location) + "</td>";
var objList = parseEngine.GetObjectsWithQuery("RSS", new { url = newsid });
bool cateresult = false;
if (objList != null && objList.Length > 0)
{
foreach (var obj in objList)
{
if (cate.Equals(obj["category"]))
{
cateresult = true;
break;
}
}
if(cateresult)
System.Console.WriteLine("Old item:" + title);
}
if(!cateresult)
{
var testObject = new Parse.ParseObject("RSS");
testObject["category"] = cate;
testObject["url"] = newsid;
testObject["title"] = title;
testObject["newsTime"] = String.Format("{0:yyyy-MM-dd HH:mm:ss}", DateTime.Parse(time));
testObject["content"] = saveTextToParse(content);
//Create a new object
testObject = parseEngine.CreateObject(testObject);
System.Console.WriteLine("New item:" + title);
}
gen.WriteItem(
title,
newsid,
"",
"sbhhbs",
"",
"",
"",
DateTime.Parse(time),
"",
"",
"",
"");
}
// end the channel
gen.WriteEndChannel();
try
{
// end the document
gen.WriteEndDocument();
}
catch (System.Exception ex)
{
}
finally
{
// close the rss generator and dispay to screen
gen.Close();
result = stringWriter.ToString();
}
//result = StringWriter.to
return result;
}
string parse(string source)
{
source = convertToUTF8fromGB2312(source);
StringWriter stringWriter = new StringWriter();
string result;
gen = new RSSFeedGenerator(stringWriter);
//-------------------------------------------------
// here is set values for header information for the RSS feed
//-------------------------------------------------
gen.Title = "建筑与城市规划";
gen.Description = "";
gen.LastBuildDate = DateTime.Now;
gen.Link = "http://sbhhbs.com";
//gen.Category = "The optional category";
gen.PubDate = DateTime.Now;
// write the header of the RSS feed document
gen.WriteStartDocument();
// write the start channel for the RSS feed Items
gen.WriteStartChannel();
Regex qariRegex = new Regex("<a href=\"news_detail\\.asp\\?id=([0-9]*)\">([^<]*)");
MatchCollection mc = qariRegex.Matches(source);
foreach (Match match in mc)
{
string id = match.Groups[1].Value;
string title = match.Groups[2].Value;
WebClient client = new WebClient();
client.Encoding = Encoding.Default;
String subURL = "http://old.tongji-caup.org/student/news_detail.asp?id=" + id;
string downloadString = client.DownloadString(subURL);
downloadString = convertToUTF8fromGB2312(downloadString);
string time = "2012-10-3";
NSRange start = rangeOfStr(downloadString, "发布时间：");
NSRange end = rangeOfStr(downloadString, " 信息来源：");
time = downloadString.Substring(start.location+start.length, end.location - start.location - 5);
start = rangeOfStr(downloadString, "<td class=\"TLE\">");
end = rangeOfStr(downloadString, "<!--内容-->");
string content = downloadString.Substring(start.location, end.location - start.location)+"</td>";
gen.WriteItem(
title,
id,
"",
"sbhhbs",
"",
"",
"",
DateTime.Parse(time),
content,
"",
"",
"");
}
// end the channel
gen.WriteEndChannel();
try
{
// end the document
gen.WriteEndDocument();
}
catch (System.Exception ex)
{
}
finally
{
// close the rss generator and dispay to screen
gen.Close();
result = stringWriter.ToString();
}
//result = StringWriter.to
return result;
}