This is a disgusting solution based on an undocumented and unsupported formula which means YOU COULD BREAK YOUR DOC (so make a copy). Also I hate regex passionately. But it works!
Formula: RegexExtract((_Merge(thisRow.Name)+""),'(?<="url"\ *:\ *")(?:\\"|[^"])*')
If you want to go spelunking down the data structure of other stuff in coda, that _Merge(thisRow.Name)+"" bit will show you what’s hiding under the hood. Copy/paste the result into a JSON formatter to visualize the hierarchy.