Ejemplo n.º 1
0
        /// <summary> Process the IIS web logs to SQL insert commands </summary>
        public void Process_IIS_Logs()
        {
            // **** READ THE LOOKUP TABLES FROM THE DATABASE **** //
            DataSet lookupTables = SobekCM_Database.Get_Statistics_Lookup_Tables();

            // ***** CODE BELOW READS ALL THE LOG FILES AND THEN WRITES THEM AS XML DATASETS *****//
            SobekCM_Log_Reader ufdc_log_reader = new SobekCM_Log_Reader(lookupTables.Tables[0], sobekcm_web_location);

            //string[] files = Directory.GetFiles(sobekcm_log_location, "*.log");
            //foreach (string thisFile in files)
            //{
            //    On_New_Status("Processing " + (new FileInfo(thisFile)).Name);
            //    ufdc_log_reader.Read_Log(thisFile).Write_XML(dataset_location);
            //}

            // ***** CODE BELOW READS ALL THE DAILY XML DATASETS AND COMBINES THEM INTO MONTHLY *****//
            // ***** DATASETS WHICH ARE SUBSEQUENTLY WRITTEN AS XML DATASETS AS WELL            *****//
            On_New_Status("Combining daily datasets into a monthly dataset");
            foreach (string year_month in year_months)
            {
                string[] year_month_files = Directory.GetFiles(dataset_location, year_month + "*.xml");
                if (year_month_files.Length > 0)
                {
                    SobekCM_Stats_DataSet combined = new SobekCM_Stats_DataSet();
                    foreach (string file in year_month_files)
                    {
                        if ((new FileInfo(file)).Name.IndexOf(year_month + ".xml") < 0)
                        {
                            SobekCM_Stats_DataSet daily = new SobekCM_Stats_DataSet();
                            daily.Read_XML(file);

                            combined.Merge(daily);
                        }
                    }

                    // Write the complete data set
                    combined.Write_XML(dataset_location, year_month + ".xml");

                    // Just write the highest users in a seperate, more readable, file
                    combined.Write_Highest_Users(dataset_location, "users_" + year_month + ".xml");
                }
            }

            //// ***** CODE BELOW READS THE MONTHLY DATASETS AND THEN WRITES THE SQL INSERTION SCRIPTS ***** //
            // Read all the data lists first for id lookups
            Dictionary <string, int> aggregationHash = Table_To_Hash(lookupTables.Tables[2]);
            Dictionary <string, int> bibHash         = Table_To_Hash(lookupTables.Tables[1]);
            Dictionary <string, int> portalHash      = new Dictionary <string, int>();

            foreach (DataRow thisRow in lookupTables.Tables[3].Rows)
            {
                if (!portalHash.ContainsKey(thisRow[2].ToString().ToUpper()))
                {
                    portalHash[thisRow[2].ToString().ToUpper()] = Convert.ToInt32(thisRow[0]);
                }
            }

            On_New_Status("Writing SQL insert commands");
            foreach (string yearmonth in year_months)
            {
                SobekCM_Stats_DataSet monthly;
                string thisFile = dataset_location + "\\" + yearmonth + ".xml";
                if (File.Exists(thisFile))
                {
                    monthly = new SobekCM_Stats_DataSet();
                    monthly.Read_XML(thisFile);
                    int year  = Convert.ToInt32(yearmonth.Substring(0, 4));
                    int month = Convert.ToInt32(yearmonth.Substring(4));
                    monthly.Write_SQL_Inserts(sql_output_location + "\\" + yearmonth + ".sql", year, month, aggregationHash, bibHash, portalHash);
                }
            }

            On_New_Status("COMPLETE!");
        }