Add support for (canonical) server name matching.

apachedex/__init__.py

@@ -630,6 +630,11 @@ class ERP5SiteStats(GenericSiteStats):
 DURATION_US_FORMAT = '%D'
 DURATION_S_FORMAT = '%T'
 
+server_name_group_dict = {
+  '%v': lambda x: x.group('servername') + '/',
+  '%V': lambda x: x.group('canonical_servername') + '/',
+}
+
 logformat_dict = {
   '%h': r'(?P<host>[^ ]*)',
   '%l': r'(?P<ident>[^ ]*)',
@@ -643,6 +648,8 @@ logformat_dict = {
   DURATION_US_FORMAT: r'(?P<duration>[0-9]*)',
   DURATION_S_FORMAT: r'(?P<duration_s>[0-9]*)',
   '%%': r'%',
+  '%v': r'(?P<servername>[^ ]*)',
+  '%V': r'(?P<canonical_servername>[^ ]*)',
   # TODO: add more formats
 }
 
@@ -975,6 +982,8 @@ def main():
   group.add_argument('--skip-base', dest='path', nargs='+',
     action=AggregateSiteUrl,
     help='Absolute base url(s) to ignore.')
+  group.add_argument('--match-servername', choices=server_name_group_dict,
+    help='Prefix URL with (canonical) server name.')
 
   group = parser.add_argument_group('filtering')
   group.add_argument('--skip-user-agent', nargs='+', default=[],
@@ -991,6 +1000,13 @@ def main():
     print >> sys.stderr, 'Neither %D nor %T are present in logformat, apdex ' \
       'cannot be computed.'
     sys.exit(1)
+  if args.match_servername is not None and \
+      args.match_servername not in args.logformat:
+    print >> sys.stderr, '--match-servername %s requested, but missing ' \
+      'from logformat.' % args.match_servername
+    sys.exit(1)
+  get_url_prefix = server_name_group_dict.get(args.match_servername,
+    lambda _: '')
   line_regex = ''
   expensive_line_regex = ''
   try:
@@ -1125,6 +1141,7 @@ def main():
       url = url_match.group('url')
       if url.startswith('http'):
         url = splithost(splittype(url)[1])[1]
+      url = get_url_prefix(match) + url
       for site, prefix_match, action in site_list:
         if prefix_match(url) is not None:
           break