Python实现最大堆（大顶堆）

最大堆是指最大的元素在堆顶的堆。

Python自带的heapq模块实现的是最小堆，没有提供最大堆的实现。虽然有些文章通过把元素取反再放入堆，出堆时再取反，把问题转换为最小堆问题也能间接实现最大堆，但是这样的实现只适合数值型的元素，不适合自定义类型。

下面给出实现代码：

# -*- coding: UTF-8 -*- 
                
import random


class MaxHeap(object):

    def __init__(self):
        self._data = []
        self._count = len(self._data)

    def size(self):
        return self._count

    def isEmpty(self):
        return self._count == 0

    def add(self, item):
        # 插入元素入堆
        self._data.append(item)
        self._count += 1
        self._shiftup(self._count-1)

    def pop(self):
        # 出堆
        if self._count > 0:
            ret = self._data[0]
            self._data[0] = self._data[self._count-1]
            self._count -= 1
            self._shiftDown(0)
            return ret
        
    def _shiftup(self, index):
        # 上移self._data[index]，以使它不大于父节点
        parent = (index-1)>>1
        while index > 0 and self._data[parent] < self._data[index]:
            # swap
            self._data[parent], self._data[index] = self._data[index], self._data[parent]
            index = parent
            parent = (index-1)>>1

    def _shiftDown(self, index):
        # 上移self._data[index]，以使它不小于子节点
        j = (index << 1) + 1
        while j < self._count :
            # 有子节点
            if j+1 < self._count and self._data[j+1] > self._data[j]:
                # 有右子节点，并且右子节点较大
                j += 1
            if self._data[index] >= self._data[j]:
                # 堆的索引位置已经大于两个子节点，不需要交换了
                break
            self._data[index], self._data[j] = self._data[j], self._data[index]
            index = j
            j = (index << 1) + 1

# 元素是数值类型                
def testIntValue():
    for iTimes in range(10):
        iLen = random.randint(1,300)
        allData= random.sample(range(iLen*100), iLen)
#         allData = [1, 4, 3, 2, 5, 7, 6]
#         iLen = len(allData)
        print('\nlen =',iLen)
        
        oMaxHeap = MaxHeap()
        print('_data:\t   ', allData)
        arrDataSorted = sorted(allData, reverse=True)
        print('dataSorted:', arrDataSorted)
        for i in allData:
            oMaxHeap.add(i)
        heapData = []    
        for i in range(iLen):
            iExpected = arrDataSorted[i]
            iActual = oMaxHeap.pop()
            heapData.append(iActual)
            print('{0}, expected: {1}, actual: {2}'.format(iExpected==iActual, iExpected, iActual))
            assert iExpected==iActual, ""
        print('dataSorted:', arrDataSorted)
        print('heapData:  ',heapData)
        
# 元素是元祖类型
def testTupleValue():
    for iTimes in range(10):
        iLen = random.randint(1,300)
        listData= random.sample(range(iLen*100), iLen)
#         listData = [1, 4, 3, 2, 5, 7, 6]
#         iLen = len(listData)
        # 注意：key作为比较大小的关键
        allData = dict(zip(listData, [str(e) for e in listData]))
        print('\nlen =',iLen)
        print('allData: ', allData)
        
        oMaxHeap = MaxHeap()
        arrDataSorted = sorted(allData.items(), key=lambda d:d[0], reverse=True)
#         arrDataSorted = sorted(allData, reverse=True)
        print('dataSorted:', arrDataSorted)
        for (k,v) in allData.items():
            oMaxHeap.add((k,v)) # 元祖的第一个元素作为比较点
        heapData = []    
        for i in range(iLen):
            iExpected = arrDataSorted[i]
            iActual = oMaxHeap.pop()
            heapData.append(iActual)
            print('{0}, expected: {1}, actual: {2}'.format(iExpected==iActual, iExpected, iActual))
            assert iExpected==iActual, ""
        print('dataSorted:', arrDataSorted)
        print('heapData:  ',heapData)
        
# 元素是自定义类    
def testClassValue():
    
    class Model4Test(object):
        '''
        用于放入到堆的自定义类。注意要重写__lt__、__ge__、__le__和__cmp__函数。
        '''
        def __init__(self, sUid, value):
            self._sUid = sUid
            self._value = value
        
        def getUid(self):
            return self._sUid
        
        def getValue(self):
            return self._value
        
        # 类类型，使用的是小于号_lt_
        def __lt__(self, other):#operator < 
#             print('in __lt__(self, other)')
            return self.getValue() < other.getValue()
       
        def __ge__(self,other):#oprator >=
            return self.getValue() >= other.getValue()
     
        #下面两个方法重写一个就可以了
        def __le__(self,other):#oprator <=
            return self.getValue() <= other.getValue()
         
        def __cmp__(self,other):
            #call global(builtin) function cmp for int
            return super.cmp(self.getValue(),other.getValue())
        
        def __str__(self):
            return '({0}, {1})'.format(self._value, self._sUid)
            
    for iTimes in range(10):
        iLen = random.randint(1,300)
        listData = random.sample(range(iLen*100), iLen)
#         listData = [1, 4, 3, 2, 5, 7, 6]
        allData = [Model4Test(str(value), value) for value in listData]
        print('allData:   ', [str(e) for e in allData])
        iLen = len(allData)
        print('\nlen =',iLen)

        oMaxHeap = MaxHeap()
        arrDataSorted = sorted(allData, reverse=True)
        print('dataSorted:', [str(e) for e in arrDataSorted])
        for i in allData:
            oMaxHeap.add(i)
        heapData = []    
        for i in range(iLen):
            iExpected = arrDataSorted[i]
            iActual = oMaxHeap.pop()
            heapData.append(iActual)
            print('{0}, expected: {1}, actual: {2}'.format(iExpected==iActual, iExpected, iActual))
            assert iExpected==iActual, ""
        print('dataSorted:', [str(e) for e in arrDataSorted])
        print('heapData:  ', [str(e) for e in heapData])
                        
if __name__ == '__main__':
    testIntValue()
    testTupleValue()
    testClassValue()