Escaping non-ASCII characters (or how to remove the BOM?)

Question

I need to create an ANSI text file from an Access recordset that outputs to JSON and YAML. I can write the file, but the output is coming out with the original characters, and I need to escape them. For example, an umlaut-O (ö) should be "\u00f6".

I thought encoding the file as UTF-8 would work, but it doesn't. However, having looked at the file coding again, if you write "UTF-8 without BOM" then everything works.

Does anyone know how to either

a) Write text out as UTF-8 without BOM, or b) Write in ANSI but escaping the non-ASCII characters?

Public Sub testoutput()

Set db = CurrentDb()

str_filename = "anothertest.json"
MyFile = CurrentProject.Path & "\" & str_filename
str_temp = "Hello world here is an ö"

fnum = FreeFile

Open MyFile For Output As fnum
Print #fnum, str_temp
Close #fnum

End Sub

Kutzenco

Uh, to most database developers, "BOM" means "bill of materials," I think. It would have made your question more comprehensible if you'd defined your abbreviation upon first usage. – Function

Answer 1

... ok .... i found some example code on how to remove the BOM. I would have thought it would be possible to do this more elegantly when actually writing the text in the first place. Never mind. The following code removes the BOM.

(This was originally posted by Simon Pedersen at http://www.imagemagick.org/discourse-server/viewtopic.php?f=8&t=12705)

' Removes the Byte Order Mark - BOM from a text file with UTF-8 encoding
' The BOM defines that the file was stored with an UTF-8 encoding.

Public Function RemoveBOM(filePath)

    ' Create a reader and a writer
            Dim writer, reader, fileSize
            Set writer = CreateObject("Adodb.Stream")
            Set reader = CreateObject("Adodb.Stream")

    ' Load from the text file we just wrote
            reader.Open
            reader.LoadFromFile filePath

    ' Copy all data from reader to writer, except the BOM
            writer.Mode = 3
            writer.Type = 1
            writer.Open
            reader.Position = 5
            reader.CopyTo writer, -1

    ' Overwrite file
            writer.SaveToFile filePath, 2

    ' Return file name
            RemoveBOM = filePath

    ' Kill objects
            Set writer = Nothing
            Set reader = Nothing
    End Function

It might be useful for someone else.

Kutzenco

Answer 2

Late to the game here, but I can't be the only coder who got got fed up with my SQL imports being broken by text files with a Byte Order Marker. There are very few 'Stack questions that touch on the problem - this is one of closest - so I'm posting an overlapping answer here.

I say 'overlapping' because the code below is solving a slightly different problem to yours - the primary purpose is writing a Schema file for a folder with a heterogeneous collection of files - but the BOM-handling segment is clearly marked.

The key functionality is that we iterate through all the '.csv' files in a folder, and we test each file with a quick nibble of the first four bytes: and we only only strip out the Byte Order Marker if we see one.

After that, we're working in low-level file-handling code from the primordial C. We have to, all the way down to using byte arrays, because everything else that you do in VBA will deposit the Byte Order Markers embedded in the structure of a string variable.

So, without further adodb, here's the code:

BOM-Disposal code for text files in a schema.ini file:

Public Sub SetSchema(strFolder As String)
On Error Resume Next 

' Write a Schema.ini file to the data folder. 

' This is necessary if we do not have the registry privileges to set the
' correct 'ImportMixedTypes=Text' registry value, which overrides IMEX=1 

' The code also checks for ANSI or UTF-8 and UTF-16 files, and applies a
' usable setting for CharacterSet ( UNICODE|ANSI ) with a horrible hack. 

' OEM codepage-defined text is not supported: further coding is required 

' ...And we strip out Byte Order Markers, if we see them - the OLEDB SQL
' provider for textfiles can't deal with a BOM in a UTF-16 or UTF-8 file 

' Not implemented: handling tab-delimited files or other delimiters. The
' code assumes a header row with columns, specifies 'scan all rows', and
' imposes 'read the column as text' if the data types are mixed. 

Dim strSchema   As String
Dim strFile     As String
Dim hndFile     As Long
Dim arrFile()  As Byte
Dim arrBytes(0 To 4) As Byte 

If Right(strFolder, 1) <> "\" Then strFolder = strFolder & "\" 

' Dir() is an iterator function when you call it with a wildcard: 

strFile = VBA.FileSystem.Dir(strFolder & "*.csv") 

Do While Len(strFile) > 0 

    hndFile = FreeFile
    Open strFolder & strFile For Binary As #hndFile
    Get #hndFile, , arrBytes
    Close #hndFile 

    strSchema = strSchema & "[" & strFile & "]" & vbCrLf
    strSchema = strSchema & "Format=CSVDelimited" & vbCrLf
    strSchema = strSchema & "ImportMixedTypes=Text" & vbCrLf
    strSchema = strSchema & "MaxScanRows=0" & vbCrLf 

    If arrBytes(2) = 0 Or arrBytes(3) = 0 Then   ' this is a hack
        strSchema = strSchema & "CharacterSet=UNICODE" & vbCrLf
    Else
        strSchema = strSchema & "CharacterSet=ANSI" & vbCrLf
    End If 

    strSchema = strSchema & "ColNameHeader = True" & vbCrLf
    strSchema = strSchema & vbCrLf 


    ' BOM disposal - Byte order marks confuse OLEDB text drivers: 

    If arrBytes(0) = &HFE And arrBytes(1) = &HFF _
    Or arrBytes(0) = &HFF And arrBytes(1) = &HFE Then 

        hndFile = FreeFile
        Open strFolder & strFile For Binary As #hndFile
        ReDim arrFile(0 To LOF(hndFile) - 1)
        Get #hndFile, , arrFile
        Close #hndFile 

        BigReplace arrFile, arrBytes(0) & arrBytes(1), "" 

        hndFile = FreeFile
        Open strFolder & strFile For Binary As #hndFile
        Put #hndFile, , arrFile
        Close #hndFile
        Erase arrFile 

    ElseIf arrBytes(0) = &HEF And arrBytes(1) = &HBB And arrBytes(2) = &HBF Then 

        hndFile = FreeFile
        Open strFolder & strFile For Binary As #hndFile
        ReDim arrFile(0 To LOF(hndFile) - 1)
        Get #hndFile, , arrFile
        Close #hndFile
        BigReplace arrFile, arrBytes(0) & arrBytes(1) & arrBytes(2), "" 

        hndFile = FreeFile
        Open strFolder & strFile For Binary As #hndFile
        Put #hndFile, , arrFile
        Close #hndFile
        Erase arrFile 

    End If 


    strFile = ""
    strFile = Dir 

Loop 

If Len(strSchema) > 0 Then 

    strFile = strFolder & "Schema.ini" 

    hndFile = FreeFile
    Open strFile For Binary As #hndFile
    Put #hndFile, , strSchema
    Close #hndFile 

End If 


End Sub 
 

Public Sub BigReplace(ByRef arrBytes() As Byte, ByRef SearchFor As String, ByRef ReplaceWith As String)
On Error Resume Next 

Dim varSplit As Variant 

varSplit = Split(arrBytes, SearchFor)
arrBytes = Join$(varSplit, ReplaceWith) 

Erase varSplit 

End Sub

The code's easier to understand if you know that a Byte Array can be assigned to a VBA.String, and vice versa. The BigReplace() function is a hack that sidesteps some of VBA's inefficient string-handling, especially allocation: you'll find that large files cause serious memory and performance problems if you do it any other way.

Heshum