A regular expression that parses RFC2812 (IRC protocol)

ircparse.coffee

# original work by Michael F. Lamb. License: GPLv3.
# added optional tilde before !user by kupp

RFC2812Matcher = ///
  ^ # We'll match the whole line. Start.
    # Optional prefix and the space that separates it
    # from the next thing. Prefix can be a servername,
    # or nick[[!user]@host]
  (?::(               # This whole set is optional but if it's
                      # here it begins with : and ends with space
    ([^@!\ ]*)        # nick
    (?:               # then, optionally user/host
      (?:             # but user is optional if host is given
        !~?([^@]*)    # !(~)user
      )?              # (user was optional)
      @([^\ ]*)       # @host
    )?                # (host was optional)
  )\ )?               # ":nick!user@host " ends
  ([^\ ]+)            # IRC command (required)
  # Optional args, max 15, space separated. Last arg is
  # the only one that may contain inner spaces. More than
  # 15 words means remainder of words are part of 15th arg.
  # Last arg may be indicated by a colon prefix instead.
  # Pull the leading and last args out separately; we have
  # to split the former on spaces.
  (
    (?:
      \ [^:\ ][^\ ]* # space, no colon, non-space characters
    ){0,14}          # repeated up to 14 times
  )                  # captured in one reference
  (?:\ :?(.*))?      # the rest, does not capture colon.
  $ # EOL
///

parse = (line) ->
  res = RFC2812Matcher.exec line
  if (! res) return invalid: line  
  raw:      line   # Whole line
  prefix:   res[1] # complete prefix
  nick:     res[2] # or servername
  username: res[3]
  hostname: res[4]
  command:  res[5]
  params:   res[6]
    .split ' '
    .slice 1       # First char of args is always ' '
    .concat if res[7] then res[7] else []

ircparser.py

import re

RFC2812 = re.compile(r"""
  ^   # We'll match the whole line. Start.
      # Optional prefix and the space that separates it
      # from the next thing. Prefix can be a servername,
      # or nick[[!user]@host]
  (?::(            # This whole set is optional but if it's
                   # here it begins with : and ends with space
    ([^@!\ ]*)     # nick
    (?:            # then, optionally user/host
      (?:          # but user is optional if host is given
        !~?([^@]*) # !user
      )?           # (user was optional)
      @([^\ ]*)    # @host
    )?             # (host was optional)
  )\ )?            # ":nick!user@host " ends
  ([^\ ]+)         # IRC command (required)
  # Optional args, max 15, space separated. Last arg is
  # the only one that may contain inner spaces. More than
  # 15 words means remainder of words are part of 15th arg.
  # Last arg may be indicated by a colon prefix instead.
  # Pull the leading and last args out separately; we have
  # to split the former on spaces.
  (
    (?:
      \ [^:\ ][^\ ]* # space, no colon, non-space characters
    ){0,14}          # repeated up to 14 times
  )                  # captured in one reference
  (?:\ :?(.*))?      # the rest, does not capture colon.
  $ # EOL""", re.X)

def parse(data):
    recognized_data = RFC2812.search(data).groups()
    info = {}
    info['raw'] = data
    info['ident'] = recognized_data[0]
    info['nick'] = recognized_data[1]
    info['username'] = recognized_data[2]
    info['host'] = recognized_data[3]
    info['command'] = recognized_data[4]
    info['params'] = recognized_data[5][1:]
    return info