/// <summary>
/// copy file and update latest file
/// </summary>
/// <param name="fileFullName">full name of the file</param>
public static void CopyFileAndUpdateLatestFile(string fileFullName)
{
if (File.Exists(fileFullName))
{
string nameNoFolder = Path.GetFileNameWithoutExtension(fileFullName);
string shortName = nameNoFolder.Substring(0, nameNoFolder.LastIndexOf('.'));
string folder = InitInfo.FileCopyFolder;
try
{
if (!Directory.Exists(folder))
{
Directory.CreateDirectory(folder);
}
string savePath = Path.Combine(folder, Path.GetFileName(fileFullName));
string shortNameFilePath = Path.Combine(folder, shortName + Path.GetExtension(fileFullName));
InitInfo.LogMessage("Copy file to " + savePath);
File.Copy(fileFullName, savePath);
InitInfo.LogMessage("Update file " + shortNameFilePath);
File.Copy(fileFullName, shortNameFilePath, true);
}
catch (Exception ex)
{
InitInfo.LogMessage("Failed to CopyFile. Exceptoin: " + ex.GetBaseException().ToString());
}
}
}
/// <summary>
/// Crawl news and comments,save file to local file
/// </summary>
public void Crawl()
{
string content = string.Empty;
if (!GetCrawlFlag())
{
return;
}
bool isCrawled = false;
int tryCount = 0;
while (!isCrawled && tryCount++ < InitInfo.RetryCount)
{
InitInfo.LogMessage(string.Format("Starting to crawl {0} site...", GetSiteName()));
try
{
IList <NewsItem> news = GetNewsList();
if (news != null && news.Count > 0)
{
switch (Path.GetExtension(GetFileName()).ToLower())
{
case ".json":
content = CrawlAsJson(news);
break;
case ".xml":
content = CrawlAsXml(news);
break;
default:
content = CrawlAsXml(news);
break;
}
if (!string.IsNullOrEmpty(content))
{
isCrawled = true;
}
}
}
catch (Exception ex)
{
InitInfo.LogMessage("Failed to crawl. Exceptoin: " + ex.GetBaseException().ToString());
}
if (!isCrawled)
{
InitInfo.LogMessage("Failed to crawl. Retry after 10 seconds...");
System.Threading.Thread.Sleep(10 * 1000);
continue;
}
else
{
string fileName = GetFileName();
InitInfo.LogMessage("Save file to " + fileName);
ArticleParserHelper.SaveToLocalFile(fileName, content);
ArticleParserHelper.CopyFileAndUpdateLatestFile(fileName);
}
}
if (!isCrawled)
{
InitInfo.LogMessage("Error occurs when crawl news and comments!");
}
}