[tools] add script to automatically generate bacnet vendors

Requires beautifulsoup4 for parsing, translitcodec for converting accents/diaeresis etc to regular text

Change-Id: I4d400f15916ee3ab07fa14b09fca12beac2010f3
Reviewed-on: https://code.wireshark.org/review/4518
Reviewed-by: Evan Huus <eapache@gmail.com>
Reviewed-by: Anders Broman <a.broman58@gmail.com>

1 files changed, 36 insertions, 0 deletions

diff --git a/tools/generate-bacnet-vendors.py b/tools/generate-bacnet-vendors.py
new file mode 100755
index 0000000000..66263cfcd1
--- /dev/null
+++ b/tools/generate-bacnet-vendors.py
@@ -0,0 +1,36 @@
+#!/usr/bin/env python2
+
+from bs4 import BeautifulSoup
+import urllib
+import sys
+import string
+# Required to convert accents/diaeresis etc.
+import translitcodec
+
+f = urllib.urlopen("http://www.bacnet.org/VendorID/BACnet%20Vendor%20IDs.htm")
+html = f.read()
+soup = BeautifulSoup(''.join(html))
+
+entry = "static const value_string\nBACnetVendorIdentifiers [] = {"
+table = soup.find('table')
+
+rows = table.findAll('tr')
+
+for tr in rows:
+  cols = tr.findAll('td')
+  for index,td in enumerate(cols[0:2]):
+    text = ''.join(td.find(text=True))
+    if index == 0:
+      entry = "    { %3s" % text
+    else:
+      entry += ", \"%s\" }," % text.rstrip()
+  # Required specially for "Dorsette's Inc." due to malformed html
+  entry = entry.replace(u'\u0092', u'\'')
+  # Required to convert accents/diaeresis etc.
+  entry = entry.encode('translit/long')
+  # Encode to ascii so we can out to file
+  entry = entry.encode("ascii",'ignore')
+  print entry
+
+entry = "    { 0, NULL }\n};"
+print entry.encode("ascii")