private static void StoreDataToDBGitDataPart(List <DBGitDataModel> inputList, int startIndex, int endIndex)
{
if (startIndex >= endIndex)
{
return;
}
string dbInsertSql = "INSERT INTO [" + Configuration.DBName + "].[dbo].[" + Configuration.TableName + "]([repositoryPath],[downloadURL],[impressionCount],[clickCount],[fileName],[dirName])VALUES";
for (int i = 0; i < inputList.Count; i++)
{
DBGitDataModel model = inputList[i];
dbInsertSql += "('" + model.repositoryPath + "','" + model.downloadURL + "','" + model.impressionCount + "','" + model.clickCount + "','" + model.fileName + "','" + model.dirName + "'),";
}
bool success = ExecuteByText(dbInsertSql.Remove(dbInsertSql.Length - 1, 1));
if (!success)
{
if (startIndex + 1 < endIndex)
{
int midIndex = (startIndex + endIndex) / 2;
StoreDataToDBGitDataPart(inputList, startIndex, midIndex);
StoreDataToDBGitDataPart(inputList, midIndex, endIndex);
}
}
}
public static void StoreDataToDBAll(List <DBGitDataModel> inputList)
{
Connection = new SqlConnection("Server=" + Configuration.DBserver + ";DataBase=" + Configuration.DBName + ";uid=" + Configuration.DBUID + ";pwd=" + Configuration.DBPWD);
Connection.Open();
string dbInsertSql = "INSERT INTO [" + Configuration.DBName + "].[dbo].[" + Configuration.TableName + "]([repositoryPath],[downloadURL],[impressionCount],[clickCount],[fileName],[dirName])VALUES";
int lineCount = 0;
for (int i = 0; i < inputList.Count; i++)
{
DBGitDataModel model = inputList[i];
dbInsertSql += "('" + model.repositoryPath + "','" + model.downloadURL + "','" + model.impressionCount + "','" + model.clickCount + "','" + model.fileName + "','" + model.dirName + "'),";
lineCount++;
if (lineCount == Configuration.DBExecuteCountEveryTime)
{
ExecuteByText(dbInsertSql.Remove(dbInsertSql.Length - 1, 1));
dbInsertSql = "INSERT INTO [" + Configuration.DBName + "].[dbo].[" + Configuration.TableName + "]([repositoryPath],[downloadURL],[impressionCount],[clickCount],[fileName],[dirName])VALUES";
lineCount = 0;
}
}
if (lineCount > 0)
{
ExecuteByText(dbInsertSql.Remove(dbInsertSql.Length - 1, 1));
}
Connection.Close();
}
public static void CheckZipFile(string inputStrFirst, string InputStrSecond)
{
HashSet <string> fileSetLocal = FileUtils.TraverseFile(Configuration.DownloadZipDir);
HashSet <string> dirSetLocal = FileUtils.TraverseFolder(Configuration.ZipExtractDir);
Dictionary <string, DBGitDataModel> existDownLoadURLData;
DBUtils.GetExistZipData(out existDownLoadURLData);
List <string> fileListDBExtra = new List <string>();
List <string> dirListDBExtra = new List <string>();
foreach (string downLoadURLData in existDownLoadURLData.Keys)
{
DBGitDataModel model = existDownLoadURLData[downLoadURLData];
if (fileSetLocal.Contains(model.fileName))
{
fileSetLocal.Remove(model.fileName);
}
else
{
fileListDBExtra.Add(model.fileName);
}
if (dirSetLocal.Contains(model.dirName))
{
dirSetLocal.Remove(model.dirName);
}
else
{
dirListDBExtra.Add(model.dirName);
}
}
Console.WriteLine("Extra local file: ");
foreach (string fileName in fileSetLocal)
{
Console.WriteLine(fileName);
}
Console.WriteLine("Extra local directory: ");
foreach (string dirName in dirSetLocal)
{
Console.WriteLine(dirName);
}
Console.WriteLine("Extra DB file: ");
foreach (string fileName in fileListDBExtra)
{
Console.WriteLine(fileName);
}
Console.WriteLine("Extra DB directory: ");
foreach (string dirName in dirListDBExtra)
{
Console.WriteLine(dirName);
}
}
public static HashSet <string> GetExistZipData(out Dictionary <string, DBGitDataModel> existDownLoadURLData)
{
existDownLoadURLData = new Dictionary <string, DBGitDataModel>();
HashSet <string> pkNameSet = new HashSet <string>();
Connection = new SqlConnection("Server=" + Configuration.DBserver + ";DataBase=" + Configuration.DBName + ";uid=" + Configuration.DBUID + ";pwd=" + Configuration.DBPWD);
Connection.Open();
string dbQuerySql = "SELECT * FROM [" + Configuration.DBName + "].[dbo].[" + Configuration.TableName + "] order by id";
try
{
SqlCommand cmd = new SqlCommand(dbQuerySql, Connection);
cmd.CommandType = CommandType.Text;
SqlDataReader reader = cmd.ExecuteReader();
while (reader.Read())
{
string repositoryPath = reader["repositoryPath"].ToString();
string downloadURL = reader["downloadURL"].ToString();
int impressionCount = Convert.ToInt32(reader["impressionCount"].ToString());
int clickCount = Convert.ToInt32(reader["clickCount"].ToString());
string fileName = reader["fileName"].ToString();
string dirName = reader["dirName"].ToString();
pkNameSet.Add(repositoryPath);
DBGitDataModel model = new DBGitDataModel(repositoryPath, downloadURL, impressionCount, clickCount, fileName, dirName);
if (!existDownLoadURLData.ContainsKey(downloadURL))
{
existDownLoadURLData.Add(downloadURL, model);
}
}
Connection.Close();
return(pkNameSet);
}
catch (Exception ex)
{
Connection.Close();
Console.WriteLine(ex.Message);
return(pkNameSet);
}
}
public DBCrawlerGitDetailDataModel(
string repositoryPath,
string downloadURL,
int impressionCount,
int clickCount,
string repositoryContent,
List <string> topicsList,
string readmeFileName)
{
this.repositoryPath = DBGitDataModel.CheckDBStringData(repositoryPath);
this.downloadURL = DBGitDataModel.CheckDBStringData(downloadURL);
this.impressionCount = impressionCount;
this.clickCount = clickCount;
this.repositoryContent = DBGitDataModel.CheckDBStringData(repositoryContent);
this.topicsList = topicsList;
if (this.topicsList != null)
{
for (int i = 0; i < this.topicsList.Count; i++)
{
this.topicsList[i] = DBGitDataModel.CheckDBStringData(this.topicsList[i]);
}
}
this.readmeFileName = readmeFileName;
}
public static void CrawlAndStoreGitData(string inputStrFirst, string InputStrSecond)
{
List <string> inputList = FileUtils.ReadFileLine(Configuration.URLFile);
List <DBGitDataModel> insertData = new List <DBGitDataModel>();
Dictionary <string, DBGitDataModel> existDownLoadURLData;
HashSet <string> existPKData = DBUtils.GetExistZipData(out existDownLoadURLData);
for (int i = 0; i < inputList.Count; i++)
{
string[] eles = inputList[i].Split(new char[] { '\t' });
string repositoryPath = eles[0];
Console.WriteLine(repositoryPath);
if (existPKData.Contains(repositoryPath))
{
continue;
}
int impressionCount = Convert.ToInt32(eles[1]);
int clickCount = Convert.ToInt32(eles[2]);
HttpStatusCode statusCode;
Dictionary <string, string> header;
string htmlContent;
try
{
htmlContent = CrawlerClass.Crawl(repositoryPath, out statusCode, out header);
}
catch (Exception e)
{
Logger.WriteLog("error: crawl \"" + repositoryPath + "\" " + e.Message);
continue;
}
string downloadRelativeURL = HtmlResolve.GetGitDownloadURL(htmlContent);
if (downloadRelativeURL == null || downloadRelativeURL == "")
{
continue;
}
string downloadURL = Configuration.RootURL + downloadRelativeURL;
string fileName;
string dirName;
if (existDownLoadURLData.ContainsKey(downloadURL))
{
DBGitDataModel modelTemp = existDownLoadURLData[downloadURL];
fileName = modelTemp.fileName;
dirName = modelTemp.dirName;
}
else
{
fileName = CrawlerClass.HttpDownloadFile(downloadURL, Configuration.DownloadZipDir);
dirName = FileUtils.ZipExtractToDirectory(Path.Combine(Configuration.DownloadZipDir, fileName), Configuration.ZipExtractDir);
if (dirName == "")
{
continue;
}
}
DBGitDataModel model = new DBGitDataModel(repositoryPath, downloadURL, impressionCount, clickCount, fileName, dirName);
insertData.Add(model);
if (insertData.Count == Configuration.DBInsertCountEveryTime)
{
DBUtils.StoreDataToDBGitDataPart(insertData);
insertData = new List <DBGitDataModel>();
Console.WriteLine("Store Data To DBGitData Part End!!!");
}
}
if (insertData.Count > 0)
{
DBUtils.StoreDataToDBGitDataPart(insertData);
}
}