A script for tcpdump; generates a oui-generated.c from http://standards-oui.ieee.org/oui.txt and http://www.iana.org/assignments/enterprise-numbers

make-oui.py

#!/usr/bin/env python
#
# Generates a 'oui-generated.c' for tcpdump.
#
from __future__ import print_function

import sys, os, time, re, getopt, codecs

OUI_URL = "http://standards-oui.ieee.org/oui.txt"
ENT_URL = "http://www.iana.org/assignments/enterprise-numbers"

OUI_TXT = 'oui-generated.txt'
OUI_C   = 'oui-generated.c'
ENT_NUM = 'enterprise-numbers'

OUI_C_HEAD = r"""
/*
 * %s: "Organizationally Unique Identifier" list
 * obtained from %s.
 * This file was generated by %s at
 * %s.
 * DO NOT EDIT!
 */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#include <netdissect-stdinc.h>
#include "netdissect.h"
#include "oui.h"

const struct tok oui_values[] = {
"""

SMI_C_HEAD = r"""
const struct tok smi_values[] = {
     { SMI_IETF, "IETF (reserved)"},
"""

OUI_C_BOTTOM = r"""

/* Comparision routine needed by bsearch() routines.
 * Don't use unsigned arithmetic.
 */
static int compare (const struct tok *a, const struct tok *b)
{
  return ((long)a->v - (long)b->v);
}

typedef int (*CompareFunc) (const void *, const void *);

const char *oui_val_to_name (unsigned oui)
{
  size_t num = sizeof(oui_values) / sizeof(oui_values[0]) - 1;
  const struct tok *t = bsearch (&oui, oui_values, num, sizeof(oui_values[0]),
                                 (CompareFunc)compare);
  return (t ? t->s : "Unknown");
}

const char *smi_val_to_name (unsigned smi)
{
  size_t num = sizeof(smi_values) / sizeof(smi_values[0]) - 1;
  const struct tok *t = bsearch (&smi, smi_values, num, sizeof(smi_values[0]),
                                 (CompareFunc)compare);
  return (t ? t->s : "Unknown");
}
"""

prefixes    = dict()
enterprises = dict()

def write_oui_c():
  f = codecs.open (OUI_C, 'w+b', 'UTF-8')
  f.write (OUI_C_HEAD % (OUI_C, OUI_URL, __file__, time.ctime()))

  for p in sorted(prefixes):
    f.write ("    { 0x%.2s%.2s%.2s, \"%s\" },\n" % (p[0:], p[3:], p[6:], prefixes[p]))
  f.write ("    { 0xFFFFFF, NULL } /* Since XEROX CORPORATION has value 0, use this */\n  };\n")
  f.close()
  print ("Wrote %d OUI records to %s" % (len(prefixes), OUI_C))

def append_oui_c():
  f = codecs.open (OUI_C, 'a+b', 'UTF-8')
  f.write (SMI_C_HEAD)
  for e in sorted(enterprises):
    width = len("SMI_IETF") - len(str(e))
    f.write ("     { %d,%*s \"%s\" },\n" % (e, width, "", enterprises[e]))
  f.write ("     { 0,     NULL }\n};\n")
  f.write (OUI_C_BOTTOM)
  f.close()
  print ("Appended %d SMI records to %s" % (len(enterprises), OUI_C))

#
# The progress callback for 'urllib.urlretrieve()'.
#
def url_progress (blocks, block_size, total_size):
  if blocks:
    percent = 100 * (blocks*block_size)/total_size
    print ("Got %d kBytes (%u%%)\r" % ((block_size*blocks)/1024, percent), end="")

#
# Parse the lines from the download 'oui-generated.txt' file.
# Extract only the first line from each record.
# Grow the 'prefixes[]' dictionary as we walk the list of 'lines'.
#
def parse_oui_txt (lines):
  #
  # The format of the oui-generated.txt looks like:
  #
  #  OUI/MA-L                                                    Organization
  #  company_id                                                  Organization
  #                                                              Address
  #
  #  E0-43-DB   (hex) Shenzhen ViewAt Technology Co.,Ltd.
  #  E043DB     (base 16) Shenzhen ViewAt Technology Co.,Ltd.
  #             9A,Microprofit,6th Gaoxin South Road, High-Tech Industrial Park, Nanshan, Shenzhen, CHINA.
  #             shenzhen  guangdong  518057
  #             CN
  #
  # We want to parse only "(hex)" lines.
  #
  for line in lines:
    line = line.rstrip()
    if not re.match("^[0-9A-F]{2}-[0-9A-F]{2}-[0-9A-F]{2}[\t ]*\(hex\)[\t ]*", line):
      continue
    prefix = line [0:8]
    vendor = line [line.rindex('(hex)')+5:].lstrip()
    prefixes [prefix] = vendor

#
# Parse the lines from the download 'enterprise-numbers' file.
# Grow the 'enterprises[]' dictionary as we walk the list of 'lines'.
#
def parse_ent_txt (lines):
  r = re.compile (r"^(\d+)$")
  got_number = 0
  for line in lines:
    match = r.search(line)
    if got_number == 0 and match:
      #
      # Org 0 ("Reserved") is already in 'smi_values[0]'.
      # Expect "Organization" on the next line.
      #
      got_number = int(match.group(1))
      continue
    if got_number:
      #
      # Convert (", \, and some \x codes to C-compatible strings.
      #
      org = line.strip()
      org = org.replace (r'\\', r"\\")
      org = org.replace (r'"', r'\"')
      org = org.replace (r'\T', r'\\T')
      org = org.replace (r'\C', r'\\C')
      org = org.replace (r'\D', r'\\D')
      enterprises [got_number] = org
      got_number = 0

#
# Check if a local 'fname' exist. Otherwise download it from 'url'.
# Return the contents as a list.
#
def get_local_file_or_download (fname, url):
  if os.path.exists(fname):
    print ("A local %s already exist." % fname)
  else:
    try:
      from urllib import urlretrieve as url_get
    except ImportError:
      from urllib.request import urlretrieve as url_get

    print ("Downloading %s from %s" % (fname, url))
    url_get (url, filename=fname, reporthook=url_progress)
    print ("")

  #
  # Now read 'fname' and return it as a list.
  #
  data = codecs.open (fname, 'rb', 'UTF-8')
  lines = data.read().splitlines()
  data.close()
  return lines

def usage (err=""):
  print ("%s%s [-h | --help]:" % (err, os.path.realpath(sys.argv[0])))
  print (r"""
  This script does 2 things:
    Creates a %s file from a local %s file
    or from one downloaded from here:
      %s""" % (OUI_C, OUI_TXT, OUI_URL))

  print (r"""
  and:
    appends an 'smi_values[]' array at the end of %s.
    This is taken from a local %s or from one downloaded from here:
      %s""" % (OUI_C, ENT_NUM, ENT_URL))

  sys.exit (0)

##  __main__

try:
  opts, args = getopt.getopt (sys.argv[1:], "?h", ["help"])
except getopt.GetoptError as e:
  usage (e.msg + "\n")

for o, a in opts:
  if o in ["-h", "-?", "--help"]:
    usage()

l = get_local_file_or_download (OUI_TXT, OUI_URL)
parse_oui_txt (l)
write_oui_c()

l = get_local_file_or_download (ENT_NUM, ENT_URL)
parse_ent_txt (l)
append_oui_c()