python 内存泄漏查找方法

方法

import tracemalloc

def test():
	tracemalloc.start()
	snapshot1 = tracemalloc.take_snapshot()
	## 你的各种数据操作
	………………
	snapshot2 = tracemalloc.take_snapshot()
    top_stats = snapshot2.compare_to(snapshot1, 'lineno')
    print(top_stats[0:10])

打印结果：

[<StatisticDiff traceback=<Traceback (<Frame filename='/Users/opt/anaconda3/envs/python36/lib/python3.6/json/decoder.py' lineno=355>,)> size=10123317 (+9533677) count=304448 (+297771)>, 
<StatisticDiff traceback=<Traceback (<Frame filename='/Users/opt/anaconda3/envs/python36/lib/python3.6/re.py' lineno=191>,)> size=44658 (+44658) count=387 (+387)>,
<StatisticDiff traceback=<Traceback (<Frame filename='/Users/PycharmProjects/algo-question/src/main.py' lineno=87>,)> size=25024 (+22784) count=391 (+356)>,
 <StatisticDiff traceback=<Traceback (<Frame filename='/Users/PycharmProjects/algo-question/src/main.py' lineno=90>,)> size=3240 (+3240) count=1 (+1)>, 
 <StatisticDiff traceback=<Traceback (<Frame filename='/Users/opt/anaconda3/envs/python36/lib/python3.6/site-packages/sklearn/svm/_base.py' lineno=616>,)> size=3176 (+3176) count=2 (+2)>,
  <StatisticDiff traceback=<Traceback (<Frame filename='/Users/opt/anaconda3/envs/python36/lib/python3.6/site-packages/pandas/core/indexes/base.py' lineno=409>,)> size=3840 (+1920) count=6 (+3)>,
   <StatisticDiff traceback=<Traceback (<Frame filename='/Users/opt/anaconda3/envs/python36/lib/python3.6/site-packages/requests/utils.py' lineno=313>,)> size=2424 (+1344) count=26 (+21)>, 
   <StatisticDiff traceback=<Traceback (<Frame filename='/Users/opt/anaconda3/envs/python36/lib/python3.6/http/client.py' lineno=1209>,)> size=1344 (+1344) count=21 (+21)>, 
   <StatisticDiff traceback=<Traceback (<Frame filename='/Users/opt/anaconda3/envs/python36/lib/python3.6/site-packages/requests/structures.py' lineno=46>,)> size=1928 (+1200) count=13 (+10)>, 
 <StatisticDiff traceback=<Traceback (<Frame filename='/Users/opt/anaconda3/envs/python36/lib/python3.6/threading.py' lineno=237>,)> size=2320 (+1056) count=6 (+2)>]

结果显示的是你一次调用后，内存新增情况，比如：size=3176 (+3176)
就代表内存为3176，比上次新增3176，说明初始内存为0；
size=2424 (+1344) 表示初始内存 为1080， 内存有残留，当然最终要多运行几次来查看内存情况。

其他方法：

显示变量引用次数

print('sentences:{}'.format(sys.getrefcount(sentences)))

保存各个类型内存增长情况：

with open('log_del.txt', 'a+') as f:
        objgraph.show_most_common_types(limit=50, file=f)
        f.close()

打印内存增长情况：

objgraph.show_growth()