Powerbuilder treats strings as UTF-16LE so you can convert it into a byte array and loop through that in steps of two, which is much quicker than converting each individual character to its ascii value. This works reasonably well for strings with a a few million characters:

ulong lul_len
ulong lul_index
string ls_output
blob lblb_string
byte lby_bytearray[]
lblb_string = blob(as_input)
lby_bytearray = GetByteArray(lblb_string)
lul_len = upperbound(lby_bytearray)
for lul_index = 1 to lul_len step 2
	if lby_bytearray[lul_index] > 126 or lby_bytearray[lul_index + 1] > 0 then
		lby_bytearray[lul_index] = 32
		lby_bytearray[lul_index + 1] = 0
	end if
next 
lblb_string = blob(lby_bytearray)
ls_output = String( lblb_string )

return ls_output

Crashed at 26 Million characters, though. For extremely large strings you could split the input into multiple strings and process the parts sequentially. Or look into multithreading.

Hi John,

As a follow up on my previous comment "Instead of checking on every character of the pasted text, loop through all characters above asc(126). I'll post a new answer ..." here you go (see attached):

Probably too late, since you needed it last Friday, but I've made a DLL that works (and it's pretty fast too).
I tried to build an MSI installer for it also, but I think it doesn't work (yet), so for now you'd have to register it using regasm.exe.

I've zipped everything up, including a small app (pb 12.6) with a main window and a button where I do 2 calls to the replace function).

To register the DLL you have to run this command from an ADMINISTRATOR commandshell:

"%WINDIR%\Microsoft.NET\Framework\v4.0.30319\regasm.exe" "C:\fullpathToTheDLL\CSReplace.dll" /codebase

Microsoft .NET Framework Assembly Registration Utility version 4.8.4084.0
for Microsoft .NET Framework version 4.8.4084.0
Copyright (C) Microsoft Corporation. All rights reserved.

Types registered successfully

(might be different paths, depending on where your regasm.exe is and where you copy the CSReplace.dll).

If you're interested, tomorrow I'll have a look if I can get the MSI installer working so you don't have to run regasm on your end users' computers.

In my test app I've used a file called "dirs.txt" which is not included.

regards

John -

 DO NOT attempt this task in PowerBuilder. Use Java. Create a Java Class:
       https://stackoverflow.com/questions/39800324/java-replace-ascii-char

Use the RUN command to execute the Java class.

Done.


Olan

Hi guys,

i am at my wits end :(

anyone can help me with changing the string functions into blob functions please ?

i need to come up with a solution before tomorrow !!!!

https://onlineasciitools.com/validate-ascii   this site does almost what i want , they validate a copied string of 2MB in less than 2 seconds for non - ascii characters.

this would be step 1 , check if there are any

second step would be , replacing them with any ascii character eg. space

tia

John

Do you have anybody on your team that is comfortable with C/C++?  That would be a pretty easy function to code in C and the string processing would verily blaze away compared to PowerScript. You then call the DLL function using an external function declaration.

Hi John;

1. The command "Asc(character) > 0" does not make sense if your App is only looking for > 126 values.
2. The POS () command would be faster than checking every character in a Loop.
3. Try to convert the String datum into a Blob data type and then use the BlobXxxxx() functions to process the replacements.

HTH

Regards ... Chris

use blobs instead of string if you want 100% PB.

if using 2019rx, you can use a .net assembly to do the work

or do it in a c dll.