I have a very big string (several Mb) I have to search in using regexp in a loop, advancing from string elements to string elements. For example, searching A, then B fellowing A, then 5 following B, etc :

[...]A[...]B[...]5[...]

The logical way would be to search for A, then search B from A, then 5 from B, etc. But that method, in JS, require to substring at every loop (as find/search/match/exec don't permit to search from a specific position).

My question, in that regard, is quite simple: is it better (in term of speed) to work with a "temporary string" which will be subtringed at every loop:

let tmpString = stringToParse
loop 0..n:
  match = tmpString.find()
  tmpString = tmpString.substring(match.index)

resulting on a big string copy (even if a little smaller at each iteration) in memory, or rather regex from the start every time:

loop 0..n:
  match = stringToParse.find()

but resulting here, of course, of a parsing of string parts which doesn't need to be parsed anymore.