pytracemalloc用于跟踪Python内存的分配情况，在我们的项目中用于在后期优化性能使用。
1.安装
pytracemalloc的安装需要修改Python的源码，我们创建一个名为pytracemalloc的目录，进入目录，运行如下命令：
> wget http://www.python.org/ftp/python/2.7.8/Python-2.7.8.tgz
> wget https://pypi.python.org/packages/source/p/pytracemalloc/pytracemalloc-1.2.tar.gz
> tar -xf Python-2.7.8.tgz
> tar -xf pytracemalloc-1.2.tar.gz
> cd Python-2.7.8
> patch -p1 < ../pytracemalloc-1.2/patches/2.7/pep445.patch
> ./configure --prefix=../build/
> make && make install
> cd ../pytracemalloc-1.2
> ../bin/python setup.py install
2.简单的实例
显示最占用内存的10行代码
1 # -*- coding:utf-8 -*-
2
3 import os
4 import sys
5 import tracemalloc
6
7 tracemalloc.start()
8
9 li = []
10 for i in xrange(0, 100000):
11 li.append(i)
12
13 class A(object):
14 def __init__(self):
15 self.m = {}
16 for i in xrange(0, 10):
17 self.m[i] = i * i
18
19 obj_li = []
20 for i in xrange(0, 100000):
21 obj_li.append(A())
22
23 snapshot = tracemalloc.take_snapshot()
24 top_stats = snapshot.statistics("lineno")
25
26 for line in top_stats[:10]:
27 print line
运行结果如下：
test.py:17: size=73.0 MiB, count=100000, average=768 B
test.py:15: size=53.0 MiB, count=199999, average=280 B
test.py:21: size=7055 KiB, count=100004, average=72 B
test.py:10: size=2356 KiB, count=2432, average=992 B
test.py:11: size=805 KiB, count=1, average=805 KiB
test.py:13: size=1952 B, count=7, average=278 B
test.py:23: size=456 B, count=1, average=456 B
test.py:14: size=120 B, count=1, average=120 B
test.py:16: size=72 B, count=1, average=72 B
比较两个snapshot的不同
1 # -*- coding:utf-8 -*-
2
3 import os
4 import sys
5 import tracemalloc
6
7 tracemalloc.start()
8
9 li = []
10 for i in xrange(0, 100000):
11 li.append(i)
12
13 class A(object):
14 def __init__(self):
15 self.m = {}
16 for i in xrange(0, 10):
17 self.m[i] = i * i
18
19 obj_li = []
20 for i in xrange(0, 100000):
21 obj_li.append(A())
22
23 snapshot1 = tracemalloc.take_snapshot()
24
25 for i in xrange(0, 1000):
26 obj_li.append(A())
27
28 snapshot2 = tracemalloc.take_snapshot()
29
30 top_stats = snapshot2.compare_to(snapshot1, "lineno")
31
32 for line in top_stats[:10]:
33 print line
下面是输出结果：
test.py:17: size=73.0 MiB (+750 KiB), count=101000 (+1000), average=768 B
test.py:15: size=53.0 MiB (+546 KiB), count=201999 (+2000), average=280 B
test.py:26: size=62.0 KiB (+62.0 KiB), count=1000 (+1000), average=64 B
/Users/whosemario/MyApplications/pytracemalloc/build/lib/python2.7/site-packages/tracemalloc.py:380: size=1008 B (+1008 B), count=6 (+6), average=168 B
/Users/whosemario/MyApplications/pytracemalloc/build/lib/python2.7/site-packages/tracemalloc.py:518: size=672 B (+672 B), count=4 (+4), average=168 B
/Users/whosemario/MyApplications/pytracemalloc/build/lib/python2.7/site-packages/tracemalloc.py:308: size=280 B (+280 B), count=1 (+1), average=280 B
/Users/whosemario/MyApplications/pytracemalloc/build/lib/python2.7/site-packages/tracemalloc.py:306: size=64 B (+64 B), count=1 (+1), average=64 B
test.py:21: size=7055 KiB (+0 B), count=100004 (+0), average=72 B
test.py:10: size=2356 KiB (+0 B), count=2432 (+0), average=992 B
test.py:11: size=805 KiB (+0 B), count=1 (+0), average=805 KiB
果然python的dict是很耗内存的，这让我想到了原来的项目的优化。
3.Reference
- pytracemalloc官网 - http://pytracemalloc.readthedocs.org/index.html#