add hash check for downloaded files

SQL/Loan_ChargeOff.ps1

@@ -48,6 +48,7 @@ $dataSize = ""
 
 $scriptPath = Get-Location
 $filePath = $scriptPath.Path+ "\"
+$dataFilePath = $dataPath + "\"
 
 ##########################################################################
 # Script level variables
@@ -147,11 +148,11 @@ if ($uninterrupted -eq 'y' -or $uninterrupted -eq 'Y')
 		# upload csv files into SQL tables
         foreach ($dataFile in $dataList)
         {
-            $destination = $dataPath + $dataFile + $table_suffix + ".csv"
-			$error_file = $dataPath + $dataFile + $table_suffix + ".error"			
-            Write-Host -ForeGroundColor 'magenta'("    Populate SQL table: {0}..." -f $dataFile)
+            $destination = $dataFilePath + $dataFile + $table_suffix + ".csv"
+			$error_file = $dataFilePath + $dataFile + $table_suffix + ".error"			
+            Write-Host -ForeGroundColor 'magenta'("    Populate SQL table: {0}... from {1}" -f $dataFile, $destination)
             $tableName = $DBName + ".dbo." + $dataFile + $table_suffix
-            $tableSchema = $dataPath + $dataFile + $table_suffix + ".xml"
+            $tableSchema = $dataFilePath + $dataFile + $table_suffix + ".xml"
             bcp $tableName format nul -c -x -f $tableSchema  -U $username -S $ServerName -P $password  -t ','
             Write-Host -ForeGroundColor 'magenta'("    Loading {0} to SQL table..." -f $dataFile)
             bcp $tableName in $destination -t ',' -S $ServerName -f $tableSchema -F 2 -C "RAW" -b 100000 -U $username -P $password -e $error_file
@@ -236,11 +237,11 @@ if ($ans -eq 'y' -or $ans -eq 'Y')
 		# upload csv files into SQL tables
         foreach ($dataFile in $dataList)
         {
-            $destination = $dataPath + $dataFile + $table_suffix + ".csv"
-			$error_file = $dataPath + $dataFile + $table_suffix + ".error"
-            Write-Host -ForeGroundColor 'magenta'("    Populate SQL table: {0}..." -f $dataFile)
+            $destination = $dataFilePath + $dataFile + $table_suffix + ".csv"
+			$error_file = $dataFilePath + $dataFile + $table_suffix + ".error"
+            Write-Host -ForeGroundColor 'magenta'("    Populate SQL table: {0} from {1}..." -f $dataFile, $destination)
             $tableName = $DBName + ".dbo." + $dataFile + $table_suffix
-            $tableSchema = $dataPath + $dataFile + $table_suffix + ".xml"
+            $tableSchema = $dataFilePath + $dataFile + $table_suffix + ".xml"
             bcp $tableName format nul -c -x -f $tableSchema  -U $username -S $ServerName -P $password  -t ','
             Write-Host -ForeGroundColor 'magenta'("    Loading {0} to SQL table..." -f $dataFile)
             bcp $tableName in $destination -t ',' -S $ServerName -f $tableSchema -F 2 -C "RAW" -b 100000 -U $username -P $password -e $error_file

SQL/Setup.ps1

@@ -10,6 +10,7 @@
 ################################################################################################
 param([string]$serverName,[string]$baseurl,[string]$username,[string]$password)
 
+$originalLocation = Get-Location
 # This is the directory for the data/code download
 $solutionTemplateSetupDir = "LoanChargeOffSolution"
 $solutionTemplateSetupPath = "D:\" + $solutionTemplateSetupDir
@@ -27,27 +28,45 @@ cd $dataDirPath
 $helpShortCutFile = "LoanChargeOffHelp.url"
 
 # List of files to be downloaded
-$dataList = "loan_info_10k.csv", "member_info_10k.csv", "payments_info_10k.csv", "loan_info_100k.csv", "member_info_100k.csv", "payments_info_100k.csv", "loan_info_1m.csv", "member_info_1m.csv", "payments_info_1m.csv"
+$dataList = "loan_info_10k", "member_info_10k", "payments_info_10k", "loan_info_100k", "member_info_100k", "payments_info_100k", "loan_info_1m", "member_info_1m", "payments_info_1m"
+$dataExtn = ".csv"
+$hashExtn = ".hash"
 foreach ($dataFile in $dataList)
 {
-    $down = $baseurl + '/' + $dataFile
+    $down = $baseurl + '/' + $dataFile + $dataExtn
     Write-Host $down
     Start-BitsTransfer -Source $down  
 }
-# making sure that the data files conform to windows style of line ending. 
-foreach ($dataFile in $dataList)
-{
-    unix2dos $dataFile 
-}
 
 #checkout setup scripts/code from github
 cd $solutionTemplateSetupPath
+Remove-Item $checkoutDir -Force -Recurse
 git clone -n https://github.com/Microsoft/r-server-loan-chargeoff $checkoutDir
 cd $checkoutDir
 git config core.sparsecheckout true
 echo "/*`r`n!HDI" | out-file -encoding ascii .git/info/sparse-checkout
 git checkout master
 
+$sqlsolutionCodePath = $solutionTemplateSetupPath + "\" + $checkoutDir + "\SQL"
+cd $sqlsolutionCodePath
+
+# make sure the hashes match for data files
+foreach ($dataFile in $dataList)
+{
+	$dataFileHash = $dataDirPath + "\" + $dataFile + $dataExtn | Get-Hash -Algorithm SHA512
+	$storedHash = $dataFile + $hashExtn | Get-Content
+	if ($dataFileHash.Hash -ne $storedHash)
+	{
+		Write-Host -ForeGroundColor 'Red' "Data file has been corrupted. Please try again."
+		throw
+	}
+}
+# making sure that the data files conform to windows style of line ending. 
+foreach ($dataFile in $dataList)
+{
+    unix2dos $dataFile + $dataExtn
+}
+
 # Start the script for DB creation. Due to privilege issues with SYSTEM user (the user that runs the 
 # extension script), we use ps-remoting to login as admin use and run the DB creation scripts
 
@@ -57,9 +76,10 @@ $command1 = "runDB.ps1"
 $command2 ="setupHelp.ps1"
 
 Enable-PSRemoting -Force
-Invoke-Command  -Credential $credential -ComputerName $serverName -FilePath $command1 -ArgumentList $dataDirPath
+Invoke-Command  -Credential $credential -ComputerName $serverName -FilePath $command1 -ArgumentList $dataDirPath, $sqlsolutionCodePath
 Invoke-Command  -Credential $credential -ComputerName $serverName -FilePath $command2 -ArgumentList $helpShortCutFile
 Disable-PSRemoting -Force
 
+cd $originalLocation.Path
 Stop-Transcript
 

SQL/runDB.ps1

@@ -12,7 +12,8 @@
 #            datadir - directory where raw csv data has been downloaded
 #            datasize - size of the data to train on (10k, 100k, 1m)
 ##############################################################################################
-Param([string]$dbuser, [string]$dbpass, [bool]$createuser = $true, [string]$datadir, [ValidateSet("10k", "100k", "1m")][string]$datasize=10k)
+Param([string]$datadir, [string]$scriptdir, [string]$dbuser, [string]$dbpass, [bool]$createuser = $true, [ValidateSet("10k", "100k", "1m")][string]$datasize="10k")
+cd $scriptdir
 # Function to generate a temporary password for SQL Server
 Function Get-TempPassword()
 {