>>>Are the first row elements are reliable indicators of the rest of the data?
>>>
>>>You could test the first row to discern column types, then use a replace all or an update to convert.
>>
>>In a word- No, sir.
>>
>>All you know of the data that you are getting, is:
>>
>>It´s a bunch of strings in a CSV file, i.e "2000","12345","91.21" asoasf.
>>There is quite bit more to it, but nothing concerning the issue at hand.
>
>oReg=CreateObj('VBScript.RegExp')
>oReg.Global = .T.
>oReg.Pattern = '("\d*)(\d{2}")'
>cFileContent = FILETOSTR(cFileNameToImport)
>cFileFixed = oReg.Replace(cFileContent,'$1.$2')
>=STRTOFILE(cFileFixed,cFileNameToImport+'.fixed')
>
>Now everything is already converted to exactly the correct decimals. The output for your sample line is:
>
>"20.00","123.45","91.21"
>
>Just import your CSV. I suspect that you really need to rip off the quotes surrounding numbers, so I'd do:
>
>oReg=CreateObj('VBScript.RegExp')
>oReg.Global = .T.
>oReg.Pattern = '("\d*)(\d{2}")'
>cFileContent = FILETOSTR(cFileNameToImport)
>cFixedUpIntegers = oReg.Replace(cFileContent,'$1.$2')
>oReg.Pattern = '"(\d*\.\d*)"'
>cAndRippedAwayUnwantedQuotes = oReg.Replace(cFixedUpIntegers,'$1')
>=STRTOFILE(cAndRippedAwayUnwantedQuotes,cFileNameToImport+'.fixed')
>
>That produces:
>
>20.00,123.45,91.21
>
>If I try the string "2000","99 and 44/100ths","12345","ABCDGoldFish","91.21"
>
>It produces:
>
>20.00,"99 and 44/100ths",123.45,"ABCDGoldFish",91.21
>
>It requires the WSH; see my sig block for details on the WSH. Do what you like; the WSH is non-negotiable; if you want to use this approach, you need it.


Ed,

I´ve tried it, and it´s fast enough, thats for sure.

As I wrote earlier,there is quite bit more to it...
For reasons better not talked about, I have to step through the file SUBSTR() by SUBSTR() in order to read the values between the commas, all in quotes.
I then have to look up the SAS INFORMAT (16.3,16.5,13.3 a.s.o. if you are familiar with SAS) in a table, and mimick the behaviour of SAS.
16.3 seems to mean divide by 100, if it´s not already a decimal value, 16.5 indicates a division by 10000, which would apply to nanoseconds for instance.

I guess what I´m trying to get at here is, I sort of need something like


DO CASE
>>CASE cFormat="16.3"
>>>>oReg.Pattern ='("\d*)(\d{2}")'
>>>>lcReplacePattern='$1.$2'

>>CASE cFormat="16.5"
>>>>oReg.Pattern =????
>>>>lcReplacePattern=?????
ENDCASE
oReg.Replace(cVar,lcReplacePattern)

Where do I find info on the Pattern property? How does it work for a division by 1000,10000?

And finally why does my text show up in bold, ´cause I don´t mean to shout.

Thank you for taking the trouble.