Merge pull request #325 from Song2017/master

skip list

Author: wangzheng0822

python/17_skiplist/skip_list_comments.py

@@ -0,0 +1,118 @@
+import random
+
+
+class SkipListNode(object):
+    def __init__(self, val, high=1):
+        # 节点存储的值
+        self.data = val
+        # 节点对应索引层的深度
+        self.deeps = [None] * high
+
+
+class SkipList(object):
+    """
+        An implementation of skip list.
+        The list stores positive integers without duplicates.
+        跳表的一种实现方法。
+        跳表中储存的是正整数，并且储存的是不重复的。
+        Author: Ben
+    """
+
+    def __init__(self):
+        # 索引层的最大深度
+        self.__MAX_LEVEL = 16
+        # 跳表的高度
+        self._high = 1
+        # 每一索引层的首节点, 默认值为None
+        self._head = SkipListNode(None, self.__MAX_LEVEL)
+
+    def find(self, val):
+        cur = self._head
+        # 从索引的顶层, 逐层定位要查找的值
+        # 索引层上下是对应的, 下层的起点是上一个索引层中小于插入值的最大值对应的节点
+        for i in range(self._high - 1, -1, -1):
+            # 同一索引层内, 查找小于插入值的最大值对应的节点
+            while cur.deeps[i] and cur.deeps[i].data < val:
+                cur = cur.deeps[i]
+
+        if cur.deeps[0] and cur.deeps[0].data == val:
+            return cur.deeps[0]
+        return None
+
+    def insert(self, val):
+        '''
+        新增时, 通过随机函数获取要更新的索引层数,
+        要对低于给定高度的索引层添加新结点的指针
+        '''
+        high = self.randomLevel()
+        if self._high < high:
+            self._high = high
+        # 申请新结点
+        newNode = SkipListNode(val, high)
+        # cache用来缓存对应索引层中小于插入值的最大节点
+        cache = [self._head] * high
+        cur = self._head
+
+        # 在低于随机高度的每一个索引层寻找小于插入值的节点
+        for i in range(high - 1, -1, -1):
+            # 每个索引层内寻找小于带插入值的节点
+            # ! 索引层上下是对应的, 下层的起点是上一个索引层中小于插入值的最大值对应的节点
+            while cur.deeps[i] and cur.deeps[i].data < val:
+                cur = cur.deeps[i]
+            cache[i] = cur
+
+        # 在小于高度的每个索引层中插入新结点
+        for i in range(high):
+            # new.next = prev.next \ prev.next = new.next
+            newNode.deeps[i] = cache[i].deeps[i]
+            cache[i].deeps[i] = newNode
+
+    def delete(self, val):
+        '''
+        删除时, 要将每个索引层中对应的节点都删掉
+        '''
+        # cache用来缓存对应索引层中小于插入值的最大节点
+        cache = [None] * self._high
+        cur = self._head
+        # 缓存每一个索引层定位小于插入值的节点
+        for i in range(self._high - 1, -1, -1):
+            while cur.deeps[i] and cur.deeps[i].data < val:
+                cur = cur.deeps[i]
+            cache[i] = cur
+        # 如果给定的值存在, 更新索引层中对应的节点
+        if cur.deeps[0] and cur.deeps[0].data == val:
+            for i in range(self._high):
+                if cache[i].deeps[i] and cache[i].deeps[i].data == val:
+                    cache[i].deeps[i] = cache[i].deeps[i].deeps[i]
+
+    def randomLevel(self, p=0.25):
+        '''
+        #define ZSKIPLIST_P 0.25      /* Skiplist P = 1/4 */
+        https://github.com/antirez/redis/blob/unstable/src/t_zset.c
+        '''
+        high = 1
+        for _ in range(self.__MAX_LEVEL - 1):
+            if random.random() < p:
+                high += 1
+        return high
+
+    def __repr__(self):
+        vals = []
+        p = self._head
+        while p.deeps[0]:
+            vals.append(str(p.deeps[0].data))
+            p = p.deeps[0]
+        return '->'.join(vals)
+
+
+if __name__ == '__main__':
+    sl = SkipList()
+    for i in range(100):
+        sl.insert(i)
+    print(sl)
+    p = sl.find(7)
+    print(p.data)
+    sl.delete(37)
+    print(sl)
+    sl.delete(37.5)
+    print(sl)