Python Dict Switcher导致内存泄漏

时间:2018-06-22 09:09:56

标签: python dictionary garbage-collection immutability

几个月以来,我已经广泛阅读了有关Python中不可变和可变对象的知识,而且我似乎已经开始理解这个概念。我仍然无法发现下面我的代码为什么会产生内存泄漏的问题。字典用作对特定类型的不可变记录的引用。在许多情况下,我会更新现有记录,在这种情况下,仅当两个记录(oldrecordnewrecord)不相等时才会更新现有记录。但是,我觉得newrecordoldrecord匹配时newrecord不会被删除,尽管在这种情况下所有引用似乎都不存在。

我的问题: 下面的代码是基于记录类型选择对字典的引用的良好实践吗?还是应该以其他方式(例如通过dictSwitcher)来代替?

class myRecordDicts():
    def __init__(self, type1Dict=dict(), type2Dict=dict(),
                type3Dict=dict(),type4Dict=dict(),type5Dict=dict(),type6Dict=dict(), type7Dict=dict()):
        self.type1Dict = type1Dict
        self.type2Dict = type2Dict
        self.type3Dict = type3Dict
        self.type4Dict = type4Dict
        self.type5Dict = type5Dict
        self.type6Dict = type6Dict
        self.type7Dict = type7Dict

    def dictSelector(self, record):
        dictSwitcher = {
            myCustomRecordType1().name: self.type1Dict,
            myCustomRecordType2().name: self.type2Dict,
            myCustomRecordType3().name: self.type3Dict,
            myCustomRecordType4().name: self.type4Dict,
            myCustomRecordType5().name: self.type5Dict,
            myCustomRecordType6().name: self.type6Dict,
            myCustomRecordType7().name: self.type7Dict,
        }
        return dictSwitcher.get(record.name)

    def AddRecordToDict(self, newrecord):
        dict = self.dictSelector(newrecord)
        recordID = newrecord.id
        if recordID in dict:
            oldrecord = dict[recordID]
            self.MergeExistingRecords(oldrecord,newrecord)
        else:                                                                                        
            dict[recordID] = newrecord

    def MergeExistingRecords(self, oldrecord, newrecord):
        # Basic Compare function
        oldRecordString = oldrecord.SerializeToString()
        newRecordString = newrecord.SerializeToString()
        # no need to do anything if same length
        if not len(oldRecordString) == len(newRecordString):
            oldrecord.CustomMergeFrom(newrecord)

1 个答案:

答案 0 :(得分:0)

好吧,似乎总是这样:我在这个问题上工作了几个小时,无法取得进展。在StackExchange上正确提出问题后5分钟,我发现了我的问题:

我需要删除 init 中的引用,因为在实例化myRecordsDicts()时从未传递过命令,因此以下代码不会泄漏内存:

class myRecordDicts():
    def __init__(self):
        self.type1Dict = dict()
        self.type2Dict = dict()
        self.type3Dict = dict()
        self.type4Dict = dict()
        self.type5Dict = dict()
        self.type6Dict = dict()
        self.type7Dict = dict()