jthulhu / beans Goto Github PK

There are currently two functions that could blow-up the stack. One in the regex module, the other in the parser.

Both of these could be replaced with a loop, which could prevent the danger.

The regex engine needs to be able to detect at least prefix static strings from regex to do an initial step in which it removes from the available possible matches the regex that do not even start with the good prefix.

Generated files should begin with something like // This file has been generated using 'beans' from the files xx and yy; DO NOT EDIT THIS FILE DIRECTLY, so people unfamiliar with the library can still know what to look at to do changes to those files.

The regex engine's matching function should take as parameter the location of the last matched string in the stream, and the Match object returned should have the location of the matched string in the stream.
This would add no asymptotic overhead to creating locations, as opposed to the current way which adds a O(ln(n)) overhead for each token.