Make user agent section optional.

apachedex/__init__.py

@@ -361,7 +361,8 @@ _APDEXDateDictAsJSONState = lambda date_dict: dict(((y, z.asJSONState())
   for y, z in date_dict.iteritems()))
 
 class GenericSiteStats(object):
-  def __init__(self, threshold, getDuration, suffix, error_detail=False):
+  def __init__(self, threshold, getDuration, suffix, error_detail=False,
+      user_agent_detail=False):
     self.threshold = threshold
     self.suffix = suffix
     self.error_detail = error_detail
@@ -371,6 +372,7 @@ class GenericSiteStats(object):
       self.error_url_count = defaultdict(partial(defaultdict, Counter))
     self.url_apdex = defaultdict(partial(APDEXStats, threshold, getDuration))
     self.apdex = defaultdict(partial(APDEXStats, threshold, getDuration))
+    self.user_agent_detail = user_agent_detail
     self.user_agent_counter = Counter()
 
   def rescale(self, convert, getDuration):
@@ -397,7 +399,8 @@ class GenericSiteStats(object):
     if self.error_detail and statusIsError(status):
       # XXX: can eat memory if there are many errors on many different urls
       self.error_url_count[status][url][match.group('referer')] += 1
-    self.user_agent_counter[match.group('agent')] += 1
+    if self.user_agent_detail:
+      self.user_agent_counter[match.group('agent')] += 1
 
   def getApdexData(self):
     return getDataPoints(self.apdex, self.status)
@@ -426,12 +429,13 @@ class GenericSiteStats(object):
       append(data.asHTML(self.threshold))
       append('<td class="text">%s</td></tr>' % unquoteToHtml(url))
     append('</table>')
-    append('<h2>User agents</h2><table class="stats"><tr><th>hits</th>'
-      '<th>user agent</th></tr>')
-    for user_agent, hit in self.user_agent_counter.most_common(N_USER_AGENT):
-      # XXX: s/escape/unquoteToHtml/ ?
-      append('<tr><td>%s</td><td class="text">%s</td></tr>' % (hit, escape(user_agent)))
-    append('</table>')
+    if self.user_agent_detail:
+      append('<h2>User agents</h2><table class="stats"><tr><th>hits</th>'
+        '<th>user agent</th></tr>')
+      for user_agent, hit in self.user_agent_counter.most_common(N_USER_AGENT):
+        # XXX: s/escape/unquoteToHtml/ ?
+        append('<tr><td>%s</td><td class="text">%s</td></tr>' % (hit, escape(user_agent)))
+      append('</table>')
     column_set = set()
     filtered_status = defaultdict(partial(defaultdict, int))
     for status, date_dict in self.status.iteritems():
@@ -498,7 +502,8 @@ class GenericSiteStats(object):
   @classmethod
   def fromJSONState(cls, state, getDuration, suffix):
     error_detail = state['error_detail']
-    result = cls(state['threshold'], getDuration, suffix, error_detail)
+    result = cls(state['threshold'], getDuration, suffix, error_detail,
+      state.get('user_agent_detail', True))
     if error_detail:
       error_url_count = result.error_url_count
       for state_status, state_url_dict in state['error_url_count'].iteritems():
@@ -524,10 +529,12 @@ class GenericSiteStats(object):
       'apdex': _APDEXDateDictAsJSONState(self.apdex),
       'status': self.status,
       'user_agent_counter': self.user_agent_counter,
+      'user_agent_detail': self.user_agent_detail,
     }
 
   def accumulateFrom(self, other):
-    # XXX: ignoring: threshold, getDuration, suffix, error_detail.
+    # XXX: ignoring: threshold, getDuration, suffix, error_detail,
+    # user_agent_detail.
     # Assuming they are consistently set.
     if self.error_detail:
       for status, other_url_dict in other.error_url_count.iteritems():
@@ -554,9 +561,10 @@ class ERP5SiteStats(GenericSiteStats):
   - If a line belongs to a module and has at least 2 slashes after module,
     count line as belonging to a document of that module
   """
-  def __init__(self, threshold, getDuration, suffix, error_detail=False):
+  def __init__(self, threshold, getDuration, suffix, error_detail=False,
+      user_agent_detail=False):
     super(ERP5SiteStats, self).__init__(threshold, getDuration, suffix,
-      error_detail=error_detail)
+      error_detail=error_detail, user_agent_detail=user_agent_detail)
     # Key levels:
     # - module id (string)
     # - is document (bool)
@@ -1188,6 +1196,9 @@ def main():
       'Default: %(default).2fs')
   group.add_argument('-e', '--error-detail', action='store_true',
     help='Include detailed report (url & referers) for error statuses.')
+  group.add_argument('-u', '--user-agent-detail', action='store_true',
+    help='Include report of most frequent user agents.')
+
   group.add_argument('-f', '--format', choices=format_generator,
     default='html', help='Format in which output should be generated.')
   group.add_argument('-p', '--period', choices=period_parser,
@@ -1307,6 +1318,7 @@ def main():
   quiet = args.quiet
   threshold = args.apdex
   error_detail = args.error_detail
+  user_agent_detail = args.user_agent_detail
   file_count = len(infile_list)
   per_site = {}
   if '-' in args.state_file and '-' in infile_list:
@@ -1437,7 +1449,7 @@ def main():
         site_data = per_site[site]
       except KeyError:
         site_data = per_site[site] = action(threshold, getDuration,
-          error_detail=error_detail)
+          error_detail=error_detail, user_agent_detail=user_agent_detail)
       try:
         site_data.accumulate(match, url_match, hit_date)
       except Exception: