/// <summary> Read a IIS web log, analyze completely, and return the corresponding <see cref="SobekCM_Stats_DataSet"/> object </summary>
/// <param name="Log_File"> Location for the log file to read </param>
/// <returns> Object with all the analyzed hits and sessions from the web log </returns>
public SobekCM_Stats_DataSet Read_Log(string Log_File)
{
// Create the list of hits
hits = new SortedList <SobekCM_Hit, SobekCM_Hit>();
// Create the list of sessions
sessions = new Dictionary <string, SobekCM_Session>();
// Create the return set
SobekCM_Stats_DataSet returnValue = new SobekCM_Stats_DataSet();
// Get the date of the log file
FileInfo fileInfo = new FileInfo(Log_File);
string name = fileInfo.Name.Replace(fileInfo.Extension, "");
DateTime logDate = new DateTime(Convert.ToInt32("20" + name.Substring(4, 2)),
Convert.ToInt32(name.Substring(6, 2)), Convert.ToInt32(name.Substring(8, 2)));
returnValue.Date = logDate;
// Open a connection to the log file and save each hit
StreamReader reader = new StreamReader(Log_File);
string line = reader.ReadLine();
while (line != null)
{
parse_line(line);
line = reader.ReadLine();
}
// Now, step through each hit in the list
foreach (SobekCM_Hit hit in hits.Values)
{
if (hit.SobekCM_URL.ToUpper().IndexOf(".ASPX") < 0)
{
// Always increment the hits
returnValue.Increment_Hits();
// Add this IP hit
returnValue.Add_IP_Hit(hit.IP, hit.UserAgent);
// Shouldn't start with '/'
if (hit.SobekCM_URL[0] == '/')
{
hit.SobekCM_URL = hit.SobekCM_URL.Substring(1);
}
hit.SobekCM_URL = hit.SobekCM_URL.ToLower();
if (hit.SobekCM_URL.IndexOf("design/webcontent/") == 0)
{
hit.SobekCM_URL = hit.SobekCM_URL.Substring(18);
}
// Add this as a webcontent hit
returnValue.Add_WebContent_Hit(hit.SobekCM_URL);
}
else
{
// parse the url
string[] splitter = hit.Query_String.ToLower().Split("&".ToCharArray());
NameValueCollection queryStringCollection = new NameValueCollection();
foreach (string thisSplit in splitter)
{
int equals_index = thisSplit.IndexOf("=");
if ((equals_index > 0) && (equals_index < thisSplit.Length - 1))
{
string query_name = thisSplit.Substring(0, equals_index);
string query_value = thisSplit.Substring(equals_index + 1);
queryStringCollection[query_name] = query_value;
if (query_name.ToLower() == "portal")
{
hit.SobekCM_URL = query_value;
}
}
}
// Now, get the navigation object using the standard SobekCM method
try
{
Navigation_Object currentMode = new Navigation_Object();
QueryString_Analyzer.Parse_Query(queryStringCollection, currentMode, hit.SobekCM_URL,
new string[] { "en" }, Engine_ApplicationCache_Gateway.Codes, Engine_ApplicationCache_Gateway.Collection_Aliases,
Engine_ApplicationCache_Gateway.URL_Portals, Engine_ApplicationCache_Gateway.WebContent_Hierarchy,
Engine_ApplicationCache_Gateway.Settings.System.Custom_BibID_RegEx, null);
if (currentMode != null)
{
currentMode.Set_Robot_Flag(hit.UserAgent, hit.IP);
}
if ((currentMode != null) && (!currentMode.Is_Robot))
{
// Always increment the hits
returnValue.Increment_Hits();
// Add this IP hit
returnValue.Add_IP_Hit(hit.IP, hit.UserAgent);
// Increment the portal hits
returnValue.Add_Portal_Hit(currentMode.Instance_Name.ToUpper());
// Check for pre-existing session
SobekCM_Session thisSession;
if (sessions.ContainsKey(hit.IP))
{
SobekCM_Session possibleSession = sessions[hit.IP];
TimeSpan difference = hit.Time.Subtract(possibleSession.Last_Hit);
if (difference.TotalMinutes >= 60)
{
thisSession = new SobekCM_Session(hit.IP, hit.Time);
sessions[hit.IP] = thisSession;
returnValue.Increment_Sessions();
}
else
{
possibleSession.Last_Hit = hit.Time;
thisSession = possibleSession;
}
}
else
{
thisSession = new SobekCM_Session(hit.IP, hit.Time);
sessions.Add(hit.IP, thisSession);
returnValue.Increment_Sessions();
}
if ((currentMode.Mode == Display_Mode_Enum.Item_Display) ||
(currentMode.Mode == Display_Mode_Enum.Item_Print))
{
if (((currentMode.ItemID_DEPRECATED.HasValue) && (currentMode.ItemID_DEPRECATED > 0)) ||
((!String.IsNullOrEmpty(currentMode.VID)) && (!String.IsNullOrEmpty(currentMode.BibID))))
{
if ((!currentMode.ItemID_DEPRECATED.HasValue) || (currentMode.ItemID_DEPRECATED < 0))
{
if (bib_vid_itemid_dictionary.ContainsKey(currentMode.BibID + ":" + currentMode.VID))
{
currentMode.ItemID_DEPRECATED = bib_vid_itemid_dictionary[currentMode.BibID + ":" + currentMode.VID];
}
}
int itemid = -1;
if (currentMode.ItemID_DEPRECATED.HasValue)
{
itemid = currentMode.ItemID_DEPRECATED.Value;
}
returnValue.Add_Item_Hit(itemid, currentMode.BibID,
currentMode.VID, currentMode.ViewerCode,
currentMode.Text_Search, thisSession.SessionID);
}
else if (!String.IsNullOrEmpty(currentMode.BibID))
{
returnValue.Add_Bib_Hit(currentMode.BibID.ToUpper(), thisSession.SessionID);
}
}
else
{
string code = currentMode.Aggregation;
string institution = String.Empty;
if ((!String.IsNullOrEmpty(code)) && (code.ToUpper()[0] == 'I'))
{
institution = code;
code = String.Empty;
}
if ((!String.IsNullOrEmpty(institution)) && (institution.ToUpper()[0] != 'I'))
{
institution = "i" + institution;
}
// For some collections we are counting the institution hit and collection
// hit just so the full use of the site is recorded
if (!String.IsNullOrEmpty(code))
{
returnValue.Add_Collection_Hit(code.ToLower(), currentMode.Mode, currentMode.Aggregation_Type, thisSession.SessionID);
}
// Was this an institutional level hit?
if (!String.IsNullOrEmpty(institution))
{
returnValue.Add_Institution_Hit(institution.ToLower(), currentMode.Mode, currentMode.Aggregation_Type, thisSession.SessionID);
}
// Is this a static "webcontent" top-level page?
if (currentMode.Mode == Display_Mode_Enum.Simple_HTML_CMS)
{
if ((currentMode.Info_Browse_Mode != "unknown") &&
(currentMode.Info_Browse_Mode != "default"))
{
returnValue.Add_WebContent_Hit(currentMode.Info_Browse_Mode.ToLower());
}
}
// Add the write type, if not normal HTML stuff
switch (currentMode.Writer_Type)
{
case Writer_Type_Enum.DataSet:
case Writer_Type_Enum.XML:
returnValue.Add_XML_Hit();
break;
case Writer_Type_Enum.OAI:
returnValue.Add_OAI_Hit();
break;
case Writer_Type_Enum.JSON:
returnValue.Add_JSON_Hit();
break;
}
}
}
else
{
if ((currentMode != null) && (currentMode.Is_Robot))
{
returnValue.Add_Robot_Hit();
}
}
}
catch (Exception ee)
{
if (ee.Message.Length > 0)
{
return(null);
}
// Do nothing.. not important?
}
}
}
return(returnValue);
}
/// <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!");
}
/// <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 = new DataSet();
//lookupTables.ReadXml(@"C:\Users\mark.v.sullivan\Documents\Visual Studio 2012\Projects\SobekDbInfoSaver\SobekDbInfoSaver\bin\Release\uf\lookup.xml");
try
{
lookupTables = Engine_Database.Get_Statistics_Lookup_Tables();
On_New_Status("Retrieved statistics lookup tables from the database", false);
}
catch (Exception ee)
{
On_New_Status("Error getting statistics lookup tables from the database. " + ee.Message, true);
return;
}
// Ensure the lookup tables were not null
if (lookupTables == null)
{
On_New_Status("Error getting statistics lookup tables from the database. Table is NULL", true);
return;
}
// Determine, from the year_month, which logs to read
List <string> logs_start = new List <string>();
foreach (string thisYearMonth in year_months)
{
logs_start.Add("u_ex" + thisYearMonth.Substring(2, 2) + thisYearMonth.Substring(4, 2));
}
// ***** CODE BELOW READS ALL THE LOG FILES AND THEN WRITES THEM AS XML DATASETS *****//
On_New_Status("Read all needed log files and write them as XML datasets", false);
SobekCM_Log_Reader sobekcm_log_reader = new SobekCM_Log_Reader(lookupTables.Tables[0], sobekcm_web_location);
string[] files = Directory.GetFiles(sobekcm_log_location, "u_ex*.log");
foreach (string thisFile in files)
{
try
{
string filename_lower = Path.GetFileName(thisFile).ToLower().Substring(0, 8);
if (logs_start.Contains(filename_lower))
{
On_New_Status("Processing " + (new FileInfo(thisFile)).Name, false);
FileInfo fileInfo = new FileInfo(thisFile);
string name = fileInfo.Name.Replace(fileInfo.Extension, "");
DateTime logDate = new DateTime(Convert.ToInt32("20" + name.Substring(4, 2)),
Convert.ToInt32(name.Substring(6, 2)), Convert.ToInt32(name.Substring(8, 2)));
string resultant_file = dataset_location + "\\" + logDate.Year.ToString() + logDate.Month.ToString().PadLeft(2, '0') + logDate.Day.ToString().PadLeft(2, '0') + ".xml";
if (!File.Exists(resultant_file))
{
sobekcm_log_reader.Read_Log(thisFile).Write_XML(dataset_location);
}
}
}
catch (Exception ee)
{
On_New_Status("Error reading a log file and writing as XML dataset. " + ee.Message, true);
}
}
// ***** 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 monthly datasets", false);
try
{
foreach (string year_month in year_months)
{
On_New_Status("Combining " + year_month + " daily datasets into one month", false);
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");
}
}
}
catch (Exception ee)
{
On_New_Status("Error combining daily datasets into monthly datasets. " + ee.Message, true);
return;
}
//// ***** 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("Insert new statistics into database", false);
try
{
foreach (string yearmonth in year_months)
{
On_New_Status("Writing statistics for " + yearmonth, false);
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.Perform_SQL_Inserts(year, month, aggregationHash, bibHash, portalHash);
}
}
}
catch (Exception ee)
{
On_New_Status("Error saving new usage statistics into the database. " + ee.Message, true);
On_New_Status("Trace from previous error: " + ee.StackTrace, true);
return;
}
On_New_Status("COMPLETE!", false);
}
