Fix end-of-line bounds handling.

The existing implementation misses ending lines and as a result, fails
to match certain protocols (SMTP for example, which needs to see at
least 3 commain/response matches).  It appears from the commit history
that an attempt was made (end-1) to prevent reads past payload length.

This can be ensured by simply not reading any payload that is < 3 bytes.

The updated logic for this loop is:
- Payload length is >= 3 bytes, or return.
- Loop over payload, compare for EOL (CR + NL) sequence.
- If found, process string.
- If index 'a' plus two is less than payload length, increment 'a' by
  one and continue.
- Loop return always increments index 'a' by one.

0 files changed, 0 insertions, 0 deletions