[tor-commits] [nyx/master] Move deduplication to log util

[atagar at torproject.org]

commit 4135a5498a289a6b75320535ab5ecb20020ac54b
Author: Damian Johnson <atagar at torproject.org>
Date:   Sun Apr 12 14:39:49 2015 -0700

    Move deduplication to log util
    
    This fits very nicely as a method of the LogEntry. This will also make it much
    easier to add tests.
---
 nyx/log_panel.py |   42 +-----------------------------------------
 nyx/util/log.py  |   53 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+), 41 deletions(-)

diff --git a/nyx/log_panel.py b/nyx/log_panel.py
index 75fa59d..cdb2346 100644
--- a/nyx/log_panel.py
+++ b/nyx/log_panel.py
@@ -24,12 +24,6 @@ from nyx import __version__
 from nyx.util import panel, tor_controller, ui_tools
 from nyx.util.log import LogEntry, read_tor_log
 
-try:
-  # added in python 3.2
-  from functools import lru_cache
-except ImportError:
-  from stem.util.lru_cache import lru_cache
-
 DAYBREAK_EVENT = 'DAYBREAK'  # special event for marking when the date changes
 TIMEZONE_OFFSET = time.altzone if time.localtime()[8] else time.timezone
 
@@ -104,23 +98,6 @@ def days_since(timestamp = None):
   return int((timestamp - TIMEZONE_OFFSET) / 86400)
 
 
- at lru_cache()
-def common_log_messages():
-  """
-  Fetches a mapping of common log messages to their runlevels from the config.
-  """
-
-  nyx_config = conf.get_config('nyx')
-  messages = {}
-
-  for conf_key in nyx_config.keys():
-    if conf_key.startswith('dedup.'):
-      event_type = conf_key[4:].upper()
-      messages[event_type] = nyx_config.get(conf_key, [])
-
-  return messages
-
-
 def log_file_path():
   for log_entry in tor_controller().get_conf('Log', [], True):
     entry_comp = log_entry.split()  # looking for an entry like: notice file /var/log/tor/notices.log
@@ -240,24 +217,7 @@ def is_duplicate(event, event_set, get_duplicates = False):
       break
 
     if event.type == forward_entry.type:
-      is_duplicate = False
-
-      if event.msg == forward_entry.msg:
-        is_duplicate = True
-      else:
-        for common_msg in common_log_messages().get(event.type, []):
-          # if it starts with an asterisk then check the whole message rather
-          # than just the start
-
-          if common_msg[0] == '*':
-            is_duplicate = common_msg[1:] in event.msg and common_msg[1:] in forward_entry.msg
-          else:
-            is_duplicate = event.msg.startswith(common_msg) and forward_entry.msg.startswith(common_msg)
-
-          if is_duplicate:
-            break
-
-      if is_duplicate:
+      if event.is_duplicate(forward_entry):
         if get_duplicates:
           duplicate_indices.append(i)
         else:
diff --git a/nyx/util/log.py b/nyx/util/log.py
index 2d8f523..a753390 100644
--- a/nyx/util/log.py
+++ b/nyx/util/log.py
@@ -5,14 +5,41 @@ runlevels.
 
 import time
 
+import stem.util.conf
 import stem.util.log
 import stem.util.system
 
 import nyx.util
 
+try:
+  # added in python 3.2
+  from functools import lru_cache
+except ImportError:
+  from stem.util.lru_cache import lru_cache
+
 TOR_RUNLEVELS = ['DEBUG', 'INFO', 'NOTICE', 'WARN', 'ERR']
 
 
+ at lru_cache()
+def _common_log_messages():
+  """
+  Provides a mapping of message types to its common log messages. These are
+  message prefixes unless it starts with an asterisk, in which case it can
+  appear anywhere in the message.
+
+  :returns: **dict** of the form {event_type => [msg1, msg2...]}
+  """
+
+  nyx_config, messages = stem.util.conf.get_config('nyx'), {}
+
+  for conf_key in nyx_config.keys():
+    if conf_key.startswith('dedup.'):
+      event_type = conf_key[4:]
+      messages[event_type] = nyx_config.get(conf_key, [])
+
+  return messages
+
+
 class LogEntry(object):
   """
   Individual tor or nyx log entry.
@@ -34,6 +61,32 @@ class LogEntry(object):
     entry_time = time.localtime(self.timestamp)
     self.display_message = '%02i:%02i:%02i [%s] %s' % (entry_time[3], entry_time[4], entry_time[5], self.type, self.message)
 
+  @lru_cache()
+  def is_duplicate(self, entry):
+    """
+    Checks if we are a duplicate of the given message or not.
+
+    :returns: **True** if the given log message is a duplicate of us and **False** otherwise
+    """
+
+    if self.message == entry.message:
+      return True
+    elif self.type != entry.type:
+      return False
+
+    for common_msg in _common_log_messages().get(self.type, []):
+      # if it starts with an asterisk then check the whole message rather
+      # than just the start
+
+      if common_msg[0] == '*':
+        if common_msg[1:] in self.message and common_msg[1:] in entry.message:
+          return True
+      else:
+        if self.message.startswith(common_msg) and entry.message.startswith(common_msg):
+          return True
+
+    return False
+
   def __eq__(self, other):
     if isinstance(other, LogEntry):
       return hash(self) == hash(other)