Python struct.pack 导致内存泄漏，即使在删除对象之后

Question

我正在使用 PyAudio 检测音频源的 activity。我正在从经常死的流上的每个音频事件创建 WAV 文件。使用 memory_profiler 我注意到 record_to_file() 方法中的 pack 方法使用的内存量通常是我通过删除数据对象重新获得的内存量的 4 倍。此代码取自 Detect & Record Audio in Python

from sys import byteorder
import sys
from array import array
import struct

import gc

import pyaudio
import wave
import subprocess
import objgraph
from memory_profiler import profile
from guppy import hpy

THRESHOLD = 5000
CHUNK_SIZE = 1024
FORMAT = pyaudio.paInt16
RATE = 44100

def is_silent(snd_data):
    "Returns 'True' if below the 'silent' threshold"
    return max(snd_data) < THRESHOLD


def normalize(snd_data):
    "Average the volume out"
    MAXIMUM = 16384
    times = float(MAXIMUM)/max(abs(i) for i in snd_data)

    r = array('h')
    for i in snd_data:
        r.append(int(i*times))
    return r


def trim(snd_data):
    "Trim the blank spots at the start and end"
    def _trim(snd_data):
        snd_started = False
        r = array('h')

        for i in snd_data:
            if not snd_started and abs(i)>THRESHOLD:
                snd_started = True
                r.append(i)

            elif snd_started:
                r.append(i)
        return r

    # Trim to the left
    snd_data = _trim(snd_data)

    # Trim to the right
    snd_data.reverse()
    snd_data = _trim(snd_data)
    snd_data.reverse()
    return snd_data


def add_silence(snd_data, seconds):
    "Add silence to the start and end of 'snd_data' of length 'seconds' (float)"
    r = array('h', [0 for i in xrange(int(seconds*RATE))])
    r.extend(snd_data)
    r.extend([0 for i in xrange(int(seconds*RATE))])
    return r


def record():
    """g
    Record a word or words from the microphone and 
    return the data as an array of signed shorts.

    Normalizes the audio, trims silence from the 
    start and end, and pads with 0.5 seconds of 
    blank sound to make sure VLC et al can play 
    it without getting chopped off.
    """
    p = pyaudio.PyAudio()
    stream = p.open(format=FORMAT, channels=1, rate=RATE,
        input=True, output=True,
        frames_per_buffer=CHUNK_SIZE)

    num_silent = 0
    snd_started = False

    r = array('h')

    while 1:
        # little endian, signed short
        snd_data = array('h', stream.read(CHUNK_SIZE))
        if byteorder == 'big':
            snd_data.byteswap()
        r.extend(snd_data)

        silent = is_silent(snd_data)

        if silent and snd_started:
            num_silent += 1
        elif not silent and not snd_started:
            snd_started = True

        if snd_started and num_silent > 400:
            break

    sample_width = p.get_sample_size(FORMAT)
    stream.stop_stream()
    stream.close()
    p.terminate()

    #r = normalize(r)
    r = trim(r)
    r = add_silence(r, 0.5)

    return sample_width, r

@profile
def record_to_file(path):
    "Records from the microphone and outputs the resulting data to 'path'"
    sample_width, data = record()

    data = struct.pack('<' + ('h'*len(data)), *data)
    print(sys.getsizeof(data))
    wf = wave.open(path, 'wb')
    wf.setnchannels(1)
    wf.setsampwidth(sample_width)
    wf.setframerate(RATE)
    wf.writeframes(data)
    wf.close()
    del data
    gc.collect()


if __name__ == '__main__':
    count = 1
    h = hpy()
    f = open('heap.txt','w')
    objgraph.show_growth(limit=3) 
    while(1):
        filename = 'demo' + str(count) + '.wav'
        print("please speak a word into the microphone")
        record_to_file(filename)
        print("done - result written to {0}".format(filename))
        cmd = 'cd "C:\Users\user\Desktop\Raudio" & ffmpeg\bin\ffmpeg -i {0} -acodec libmp3lame {1}.mp3'.format(filename, filename)
        "subprocess.call(cmd, shell=True)"
        count += 1
        objgraph.show_growth()
        print h.heap()

这是内存分析器模块输出的一次迭代：

Line #    Mem usage    Increment   Line Contents
================================================
   117     19.6 MiB      0.0 MiB   @profile
   118                             def record_to_file(path):
   119                                 "Records from the microphone and outputs the resulting data to 'path'"
   120     22.4 MiB      2.8 MiB       sample_width, data = record()
   125     31.3 MiB      8.9 MiB       data = struct.pack('<' + ('h'*len(data)), *data)
   126     31.3 MiB      0.0 MiB       print(sys.getsizeof(data))
   127     31.3 MiB      0.0 MiB       wf = wave.open(path, 'wb')
   128     31.3 MiB      0.0 MiB       wf.setnchannels(1)
   129     31.3 MiB      0.0 MiB       wf.setsampwidth(sample_width)
   130     31.4 MiB      0.0 MiB       wf.setframerate(RATE)
   131     31.4 MiB      0.0 MiB       wf.writeframes(data)
   132     31.4 MiB      0.0 MiB       wf.close()
   133     30.4 MiB     -0.9 MiB       del data
   134     27.9 MiB     -2.5 MiB       gc.collect()

使用 VS 调试进程时，我在系统内存中看到大量 'h' 个字符，这可能是发生泄漏的原因。任何帮助将不胜感激

Answer 1

结构 class 保留项目缓存以便更快地访问。清除结构缓存的唯一方法是调用 struct._clearcache() 方法。可在 here 中找到使用此方法的示例。

警告！这是一个 _ 方法，可以随时更改。请参阅 here 了解这些类型的方法及其说明。

python 论坛 here and here 上有关于此 'memory leak' 的讨论。

Python struct.pack 导致内存泄漏，即使在删除对象之后

Python struct.pack causing a memory leak, even after deleting object

python

memory-leaks

pack

pyaudio