private AccessDataDetail DeserializeJsonDataInLocal(string deviceId, string accessDate)
{
string jsonName = ComputeJsonName(deviceId, accessDate);
if (!Directory.Exists(localPath))
{
return(null);
}
string fullPath = Path.Combine(localPath, jsonName);
if (File.Exists(fullPath))
{
var content = File.ReadAllText(fullPath);
AccessDataDetail deserializedObj = JsonConvert.DeserializeObject <AccessDataDetail>(content);
return(deserializedObj);
}
return(null);
}
public AccessDataDetail DeserializeJsonDataInCloud(string deviceId, string accessDate)
{
string jsonName = ComputeJsonName(deviceId, accessDate);
// connect to our storage account and create a blob client
var storageAccount = CloudStorageAccount.Parse(connectionString);
var blobClient = storageAccount.CreateCloudBlobClient();
// get a reference to the container
var blobcontainer = blobClient.GetContainerReference(containerName);
blobcontainer.CreateIfNotExists();
foreach (var blob in blobcontainer.ListBlobs())
{
string flName = blob.Uri.Segments.Last().ToString();
Console.WriteLine(flName);
if (jsonName == flName)
{
CloudBlob cblob = blobcontainer.GetBlobReference(flName);
using (StreamReader reader = new StreamReader(cblob.OpenRead()))
{
string content = "";
StringBuilder strjson = new StringBuilder();
while ((content = reader.ReadLine()) != null)
{
strjson.Append(content);
}
AccessDataDetail overlapAddObj = JsonConvert.DeserializeObject <AccessDataDetail>(strjson.ToString());
return(overlapAddObj);
}
}
}
return(null);
}
public IList <string> Parse(string blobName, Stream accessLogBlob)
{
string deviceId = GetDeviceId(blobName);
IList <string> result = new List <string>();
string lineEntry = string.Empty;
//Initialize the prev record to some random date
DateTime prevRecord = DateTime.Parse("01/01/1900");
var file = new StreamReader(accessLogBlob);
while ((lineEntry = file.ReadLine()) != null)
{
string[] tokens = lineEntry.Split(' ');
bool isMainModule = false;
//TODO: We skip this for now - need to check if this is an error and log appropriately
if (tokens.Length <= 6)
{
continue;
}
//If the first token (token[0]) is not of type IP address - skip processing - evaluate via Regex
string clientIP = tokens[0];
Regex ipRegex = new Regex(@"\b\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}\b");
MatchCollection regexResult = ipRegex.Matches(clientIP);
if (regexResult.Count == 0)
{
continue;
}
//Ignore Token 2(token[1]) & Token 3(token[2]) - Irrelevant
//Token 4(token[3]) - Date information
string dateValue = tokens[3].Substring(1);
CultureInfo provider = CultureInfo.InvariantCulture;
DateTime dtObj = DateTime.ParseExact(dateValue, "dd/MMM/yyyy:HH:mm:ss", provider);
//TODO return if date is not valid
string formattedDateValue = dtObj.ToString("yyyyMMdd");
//Ignore Token 5(token[4]) & Token 6(token[5]) - Irrelevant
//Token 7(token[6]) Contains Module Information.
string module = tokens[6];
string[] moduleTokens = module.Split('/', '?');
//TODO: We skip this for now - need to check if this is an error and log appropriately
if (moduleTokens.Length <= 3)
{
continue;
}
//Specification says, the module url should start with "modules" - if that's not the case skip
if (!moduleTokens.Contains("modules"))
{
continue;
}
//Ignore all the Image based URLs
if (moduleTokens[moduleTokens.Length - 1].Contains(".png") || moduleTokens[moduleTokens.Length - 1].Contains(".jpg") ||
moduleTokens[moduleTokens.Length - 1].Contains(".bmp") || moduleTokens[moduleTokens.Length - 1].Contains(".gif") ||
moduleTokens[moduleTokens.Length - 1].Contains(".js") || moduleTokens[moduleTokens.Length - 1].Contains(".css"))
{
continue;
}
//Ignore all the request that came for the same time - An html page has css, js, and lot of references - they dont count towards an article that was read - round off to 1 event for 1 second.
if (prevRecord == dtObj)
{
continue;
}
else
{
prevRecord = dtObj;
}
//If we reached here - we are certainly processing the record - Save any metadata information about the record.
if (!dateRangeList.ContainsKey(formattedDateValue))
{
//If there is no record for a date - create a record and register start and end time as that of current log entry
dateRangeList.Add(formattedDateValue, new DateRange()
{
StartTime = DateTime.Parse(dtObj.ToString("HH:mm:ss")), EndTime = DateTime.Parse(dtObj.ToString("HH:mm:ss"))
});
}
else
{
//If there exist a record, update only the end time as we process each record
DateRange rangeObj = dateRangeList[formattedDateValue];
rangeObj.EndTime = DateTime.Parse(dtObj.ToString("HH:mm:ss"));
}
//If there are no records for the current date - check from cloud or local and load it
if (!existingRecords.ContainsKey(formattedDateValue))
{
//Check if there is a JSON for this date - if exist load the JSON in memory
AccessDataDetail existingObj = DeserializeJsonData(deviceId, formattedDateValue);
if (existingObj != null)
{
existingRecords.Add(formattedDateValue, existingObj);
}
else
{
existingRecords.Add(formattedDateValue, null);
}
}
//If there are records for current date - check if there is an overlap in time and ignore
if (existingRecords[formattedDateValue] != null)
{
DateTime startTime = existingRecords[formattedDateValue].StartTime;
DateTime endTime = existingRecords[formattedDateValue].EndTime;
DateTime currrentTime = DateTime.Parse(dtObj.ToString("HH:mm:ss"));
//If the current log entry is a overlapping with existing record - ignore
if (currrentTime >= startTime && currrentTime <= endTime)
{
continue;
}
}
//Check if the request is for Main Module
if (moduleTokens[3].Contains(".htm") || moduleTokens[3].Contains("html"))
{
isMainModule = true;
}
if (!accessDataList.ContainsKey(formattedDateValue))
{
//Add new value object (Dictionary) for each date
IDictionary <string, AccessData> accessDataKVPair = new Dictionary <string, AccessData>();
accessDataList.Add(formattedDateValue, accessDataKVPair);
}
IDictionary <string, AccessData> dictionaryObj = accessDataList[formattedDateValue];
string moduleName = moduleTokens[2];
if (!dictionaryObj.ContainsKey(moduleName))
{
//Add new value object (AccessData) for each record in access log
AccessData obj = new AccessData()
{
ModuleName = moduleName, MainModuleCount = 0, SubModuleCount = 0, UpLoadTime = dtObj
};
dictionaryObj.Add(moduleName, obj);
}
AccessData accessDataObj = dictionaryObj[moduleName];
if (isMainModule)
{
accessDataObj.MainModuleCount = accessDataObj.MainModuleCount + 1;
}
else
{
accessDataObj.SubModuleCount = accessDataObj.SubModuleCount + 1;
}
}
//In the loop above - we just skip the overlapping record. Here is where the merge happens
//Iterate AccessDataDetails for each date
foreach (string dateValue in accessDataList.Keys)
{
AccessDataDetail addObj = new AccessDataDetail();
addObj.AccessDate = dateValue;
addObj.DeviceId = deviceId;
//Add the Start and End Date we derived from processing the current log
if (dateRangeList.ContainsKey(dateValue))
{
addObj.StartTime = dateRangeList[dateValue].StartTime;
addObj.EndTime = dateRangeList[dateValue].EndTime;
}
//Add the Unmerged record from processing the current log
IDictionary <string, AccessData> dictionaryObj = accessDataList[dateValue];
foreach (string moduleName in dictionaryObj.Keys)
{
addObj.AccessDetails.Add(dictionaryObj[moduleName]);
}
//Merge existing record with current record
if (existingRecords.ContainsKey(dateValue))
{
if (existingRecords[dateValue] != null)
{
if (existingRecords[dateValue].StartTime != null)
{
//Update Start Time with minimum of both
if (existingRecords[dateValue].StartTime <= addObj.StartTime)
{
addObj.StartTime = existingRecords[dateValue].StartTime;
}
}
if (existingRecords[dateValue].EndTime != null)
{
//Update End Time with maximum of both
if (existingRecords[dateValue].EndTime >= addObj.EndTime)
{
addObj.EndTime = existingRecords[dateValue].EndTime;
}
}
foreach (AccessData existingaddObj in existingRecords[dateValue].AccessDetails)
{
string moduleName = existingaddObj.ModuleName;
//Check if record exist for this module
bool matchFound = false;
foreach (AccessData newaddObj in addObj.AccessDetails)
{
if (moduleName == newaddObj.ModuleName)
{
matchFound = true;
newaddObj.MainModuleCount += existingaddObj.MainModuleCount;
newaddObj.SubModuleCount += existingaddObj.SubModuleCount;
break;
}
}
if (matchFound == false)
{
addObj.AccessDetails.Add(existingaddObj);
}
}
}
}
//Serialize AccessDataDetail into JSON File
string jsonValue = JsonConvert.SerializeObject(addObj, Formatting.Indented);
Console.WriteLine(jsonValue);
Console.WriteLine();
try
{
string tempPath = Path.GetTempPath();
string fileName = ComputeJsonName(addObj.DeviceId, addObj.AccessDate);
string filePath = Path.Combine(tempPath, fileName);
StreamWriter sw = new StreamWriter(filePath);
sw.WriteLine(jsonValue);
sw.Close();
result.Add(filePath);
}
catch (Exception)
{
return(null);
}
}
return(result);
}
