Why does the acute gets separated from its base letter when making an insert to an oracle database via Powershell?

Question

i am using Powershell to execute an insert statement and insert a string into a database table. The text I want to insert I get from an HTTP-Request to Confluence's REST API and containts Czech characters. Following code snipet does the job of inserting the data to the database

$DAOControllerClass | Add-member -MemberType ScriptMethod -Name Get-DataBaseConnection -Value {
[OutputType([System.Data.OracleClient.OracleConnection])]

$username = $this.username
$password = $this.password
$data_source = $this.data_source 

log("Executing Get-DataBaseConnection")

$connection_string = "User Id=$username;Password=$password;Data Source=$data_source"
$con = New-Object System.Data.OracleClient.OracleConnection($connection_string)
try {
    $con.Open()
} catch {
    throw "Could not open database connection"
}
log("Connectiong opened")
return $con
}

$DAOControllerClass | Add-Member -MemberType ScriptMethod -Name Update-CNFLPageIntoOldWorld -Value {
param(
    [Parameter(Mandatory=$true)][String[]]$values
)
log("Executing Update-CNFLPageIntoBaseLayer")


try{
    $con = $this.'Get-DataBaseConnection'()

    $command             = $con.CreateCommand()
    $command.Connection  = $con
    $command.CommandText = [IO.File]::ReadAllText(".\Database queries\Data dictionary - Core layer queries\Update_cnfl_page_old_world.sql")

    $null = $command.Parameters.Add("cnfl_page_id", $values[0])
    $null = $command.Parameters.Add("label", $values[1]) 
    $null = $command.Parameters.Add("business_pojem_html", $values[2]) 
    $null = $command.Parameters.Add("popis_html",$values[3]) # The issue is with $values[3]

    $null = $command.ExecuteNonQuery()

    log("The cnfl page with the id: " + $values[0] + " got updated in the table confluence_page_old_world")
} catch {
    throw (“Database Exception: " + $con.ConnectionString + ": " + $_.Exception.ToString())
} finally{
    if ($con.State -eq ‘Open’) { 
        $con.close() 
        $command.Dispose()
    }
}

}

Now the text that I get passed as parameter when downloaded from a Confluence page is as follows: "Reportingové statusy a příchody/odchody klientů."

When I print this text in Powershell, everything looks fine. All letters are represented as they should be. When I debug this code and see what text is assigned to the $command.CommandText then it also looks fine.

But when I see the result in the Database it looks like following:

So all the letters are fine except that the acutes are split away from its base letter. I tried different encodings in powershell, I tried to change the NLS settings in the Database. I also tried to write to a .txt file, encode it with utf-8, with unicode and with ISO/IEC 8859-2 only to read it from the file but this also did not work.

The only thing that works is when I hardcode the text into Powershell like this:

$null = $command.Parameters.Add("popis_html","Reportingové statusy a příchody/odchody klientů.")

Then I get the expected result. So it seems to me that when passing the string as an argument some conversion or encoding is happing but I have no clue what it could be because the letters actually get represented, it is just about the acutes. I have following Encoding settings in Powershell

IsSingleByte      : True
BodyName          : iso-8859-2
EncodingName      : Central European (Windows)
HeaderName        : windows-1250
WebName           : windows-1250
WindowsCodePage   : 1250
IsBrowserDisplay  : True
IsBrowserSave     : True
IsMailNewsDisplay : True
IsMailNewsSave    : True
EncoderFallback   : System.Text.InternalEncoderBestFitFallback
DecoderFallback   : System.Text.InternalDecoderBestFitFallback
IsReadOnly        : True
CodePage          : 1250

And following nls_session_parameters

NLS_LANGUAGE    CZECH
NLS_TERRITORY   CZECH REPUBLIC
NLS_CURRENCY    Kč
NLS_ISO_CURRENCY    CZECH REPUBLIC
NLS_NUMERIC_CHARACTERS  ,.
NLS_CALENDAR    GREGORIAN
NLS_DATE_FORMAT DD.MM.RR
NLS_DATE_LANGUAGE   CZECH
NLS_SORT    CZECH
NLS_TIME_FORMAT HH24:MI:SSXFF
NLS_TIMESTAMP_FORMAT    DD.MM.RR HH24:MI:SSXFF
NLS_TIME_TZ_FORMAT  HH24:MI:SSXFF TZR
NLS_TIMESTAMP_TZ_FORMAT DD.MM.RR HH24:MI:SSXFF TZR
NLS_DUAL_CURRENCY   Kč
NLS_COMP    BINARY
NLS_LENGTH_SEMANTICS    BYTE
NLS_NCHAR_CONV_EXCP FALSE

I tried setting the Powershell Encoding to Unicode with this line

$OutputEncoding = [System.Text.Encoding]::Unicode

The result in the database was the same. What else could I try? Thank you!

Answer 1

Perhaps the problem is the following (I cannot personally verify this):

It sounds like what you're receiving from the Confluence REST API are strings in decomposed Unicode normalization form (NFD), in which accented characters are represented by two code points: the base letter (e.g., e), followed by a combining diacritic (e.g., ́, the combining acute accent, U+0301)

And it looks like Oracle perhaps has trouble with this decomposed normal form and supports only a composed form (NFC), where accented letters have a direct representation as a single code point (e.g., é, the Latin small letter with acute, U+00E9 ),

Therefore, you can try to convert the strings to the composed form (NFC) using the String.Normalize() method:

$values[3].Normalize()  # Converts string $values[3] to NFC