[tor-commits] [stem/master] Don't clone immutables when parsing descriptors

[atagar at torproject.org]

commit 1a50ce30e95dd11dbc80167f4c4df2df0b05f55b
Author: Damian Johnson <atagar at torproject.org>
Date:   Tue Jan 24 10:44:01 2017 -0800

    Don't clone immutables when parsing descriptors
    
    Avoiding an unnecessary copy call on types which are immutable. Seems even for
    empty lists and dictionaries calling the constructor rather than copy() is a
    tad faster.
    
    On my wee little netbook this speeds up reading microdescriptors by ~5% when
    there's validation and speeds our integ tests a bit...
    
      Before
      descriptor.microdescriptor...                        success (15.38s)
      descriptor.networkstatus...                          success (23.25s)
    
      After
      descriptor.microdescriptor...                        success (13.96s)
      descriptor.networkstatus...                          success (22.77s)
    
    This should help all descriptor reading though when validation is disabled this
    is boon is only when accessing attributes.
---
 docs/change_log.rst         |  1 +
 stem/descriptor/__init__.py | 12 +++++++++++-
 2 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/docs/change_log.rst b/docs/change_log.rst
index e16a948..477db2e 100644
--- a/docs/change_log.rst
+++ b/docs/change_log.rst
@@ -45,6 +45,7 @@ The following are only available within Stem's `git repository
 
  * **Descriptors**
 
+  * Sped descriptor reading by ~5% by not cloning immutable fields
   * Support for protocol descriptor fields (:spec:`eb4fb3c`)
   * Shared randomness properties weren't being read in votes (:trac:`21102`)
 
diff --git a/stem/descriptor/__init__.py b/stem/descriptor/__init__.py
index d7b1aea..67ea374 100644
--- a/stem/descriptor/__init__.py
+++ b/stem/descriptor/__init__.py
@@ -78,6 +78,7 @@ KEYWORD_LINE = re.compile('^([%s]+)(?:[%s]+(.*))?$' % (KEYWORD_CHAR, WHITESPACE)
 SPECIFIC_KEYWORD_LINE = '^(%%s)(?:[%s]+(.*))?$' % WHITESPACE
 PGP_BLOCK_START = re.compile('^-----BEGIN ([%s%s]+)-----$' % (KEYWORD_CHAR, WHITESPACE))
 PGP_BLOCK_END = '-----END %s-----'
+EMPTY_COLLECTION = ([], {}, set())
 
 DocumentHandler = stem.util.enum.UppercaseEnum(
   'ENTRIES',
@@ -519,7 +520,16 @@ class Descriptor(object):
 
     for attr in self.ATTRIBUTES:
       if not hasattr(self, attr):
-        setattr(self, attr, copy.copy(self.ATTRIBUTES[attr][0]))
+        value = self.ATTRIBUTES[attr][0]
+
+        if value is None or isinstance(value, (bool, stem.exit_policy.ExitPolicy)):
+          pass  # immutable
+        elif value in EMPTY_COLLECTION:
+          value = type(value)()  # collection construction tad faster than copy
+        else:
+          value = copy.copy(value)
+
+        setattr(self, attr, value)
 
     for keyword, values in list(entries.items()):
       try: