Swapping the characters could be a matter of find/replace, however this is highly dependent on the actual contents of your text variable.
If what you're getting from the webpage is the relevant number(s) only, this should work:
{TXTREPLACEVAR:"{TXTREPLACEVAR:"{TXTREPLACEVAR:textSource:",":"§"}":".":","}":"§":"."}
First, commas are replaced with "§", then dots are replaced with commas, lastly "§" is replaced with dots(I'm using "§" as I believe it to be unlikely to appear in regular text).
If you tried turning commas into dots, and dots into commas or vice-versa, you'd always end up with the second replacement replacing the characters replaced by the first, so you'd end up with either all dots or all commas.
This assumes you don't have any dots or commas that shouldn't be replaced in the text you're getting from that webpage. Otherwise, you'll have to use the "{TXTSUBSTR:}" token to separate those parts.