BCP update database table base on output from powershell - sql

Question

I have 4 files with the same csv header as following
Column1,Column2,Column3,Column4
But I only required data from Column2,Column3,Column4 for import the data into SQL database using BCP . I am using the PowerShell to select the columns that I want and import the required data using BCP but my powershell executed with no error and there are not data updated in my database table. May I know how to set the BCP to import the output from Powershell to database table. Here are my powershell script
$filePath = Get-ChildItem -Path 'D:\test\*' -Include $filename
$desiredColumn = 'Column2','Column3','Column4'
foreach($file in $filePath)
{
write-host $file
$test = import-csv $file | select $desiredColumn
write-host $test
$action = bcp <myDatabaseTableName> in $test -T -c -t";" -r"\n" -F2 -S <MyDatabase>
}
These are the output from the powershell script
D:\test\sample1.csv
#{column2=111;column3=222;column4=333} #{column2=444;column3=555;column4=666}
D:\test\sample2.csv
#{column2=777;column3=888;column4=999} #{column2=aaa;column3=bbb;column4=ccc}

Answer 1

First off, you can't update a table with bcp. It is used to bulk load data. That is, it will either insert new rows or export existing data into a flat file. Changing existing rows, usually called as updating, is out of scope for bcp. If that's what you need, you need to use another a tool. Sqlcmd works fine, and Powershell's got Invoke-Sqlcmd for running arbitary TSQL statements.
Anyway, the BCP utility has notoriously tricky syntax. As far as I know, one cannot bulk load data by passing the data as parameter to bcp, a source file must be used. Thus you need to save the filtered file and pass its name to bcp.
Exporting a filtered CSV is easy enough, just remember to use -NoTypeInformation switch, lest you'll get #TYPE Selected.System.Management.Automation.PSCustomObject as your first row of data. Assuming the bcp arguments are well and good (why -F2 though? And Unix newlines?).
Stripping double quotes requires another an edit to the file. Scrpting Guy has a solution.
foreach($file in $filePath){
write-host $file
$test = import-csv $file | select $desiredColumn
# Overwrite filtereddata.csv, should one exist, with filtered data
$test | export-csv -path .\filtereddata.csv -NoTypeInformation
# Remove doulbe quotes
(gc filtereddata.csv) | % {$_ -replace '"', ''} | out-file filtereddata.csv -Fo -En ascii
$action = bcp <myDatabaseTableName> in filtereddata.csv -T -c -t";" -r"\n" -F2 -S <MyDatabase>
}
Depending on your locale, column separator might be semicolon, colon or something else. Use -Delimiter '<character>' switch to pass whatever you need or change bcp's argument.
Erland's got a helpful page about bulk operations. Also, see Redgate's advice.

Answer 2

Without need to modify the file first, there is an answer here about how bcp can handle quoted data.
BCP in with quoted fields in source file
Essentially, you need to use the -f option and create/use a format file to tell SQL your custom field delimiter (in short, it is no longer a lone comma (,) but it is now (",")... comma with two double quotes. Need to escape the dblquotes and a small trick to handle the first doulbe quote on a line. But it works like a charm.
Also, need the format file to ignore column(s)... just set the destination column number to zero. All with no need to modify the file before load. Good luck!