修改点 :

因为做排序需要互换结构体里面的内容, 导致 hashnode 里面保存的 filePtr 指向的内容发生改变， 现在把 hashnode 和 file node 结合在一起 (本来应该合在一起，但之前觉得分开管理清晰一点)

 

#include "common.h"

#define nullptr 0

#define u64 unsigned long long int
#define u32 unsigned int
#define maxstringsize 10

typedef struct LinkNode
{
	u32 depth; // 这个 node 所在的深度， 根目录是0
	u32 mNum;
	u32 create_tm;
	bool is_del;
	char str[maxstringsize];
	LinkNode *father;
	LinkNode *sonlist;
	LinkNode *next; // 用于 sonlist 中链接各个子 node
	//for hash
	u64 hash_value;
	LinkNode *hash_next;
}LinkNode;

#define maxmemsize 1007
static LinkNode g_n_mem[maxmemsize];
static int g_n_mem_cnt = 0;

#define maxhashsize 107
static LinkNode *m_hashmap[maxhashsize];

static char filename[20][maxstringsize]
{
	{"root"},
	{"aaaa"},
	{"bbbb"},
	{"april"},
	{"friday"},
	{"monday"},
	{"love"},
	{"kiss"},
	{"autum"},
	{"weather"},
	{"water"},
	{"leak"},
	{"mouth"},
	{"leader"},
	{"gohome"},
	{"shafa"},
	{"season"},
	{"global"},
	{"see"},
	{"sea"},
};

static LinkNode *root = nullptr;
static u32 g_create_tm = 0;
#define m_max_depth 5 // there is 5 level only

u64 my_hash2(const char str[])
{
	u64 h = 0;
	while (*str != '\0')
	{
		h = h * 26 + (*str++) - 'a' + 1; // 字符串全是小写的情况, 可根据情况改成 31 / 131
	}
	return h & 0xFFFFFFFFFFFFFFFF;
}

u64 my_hash(const char str[])
{
	u64 h = 0;
	u64 p = 1;
	while (*str != '\0')
	{
		h += p * (*str++);
		p *= 2531011;
	}
	return h & 0xFFFFFFFFFFFFFFFF;
}

void add_to_hash_list(LinkNode **phead, LinkNode *pNode)
{
	pNode->hash_next = *phead;
	*phead = pNode;
}

LinkNode *is_exist_in_hash_list(u64 hash_value)
{
	LinkNode *p = m_hashmap[hash_value % maxhashsize];
	while (p != nullptr)
	{
		if (p->hash_value == hash_value)
		{
			return p;
		}
		p = p->hash_next;
	}
	return nullptr;
}

void del_from_hash_list(u64 hash_value)
{
	LOGE("%lld will be del from hash list idx = %d", hash_value, (hash_value % maxhashsize));
	LinkNode *p = m_hashmap[hash_value % maxhashsize];
	if (p->hash_value == hash_value)
	{
		m_hashmap[hash_value % maxhashsize] = p->hash_next;
	}
	else
	{
		LinkNode *pre = p;
		p = p->hash_next;
		while (p != nullptr)
		{
			if (p->hash_value == hash_value)
			{
				pre->hash_next = p->hash_next;
				break; //must has break
			}
			pre = p;
			p = p->hash_next;
		}
	}

	p->hash_value = 0;
	p->hash_next = nullptr;
}

void add_to_father_son_list(LinkNode *father, LinkNode *pNode)
{
	pNode->next = father->sonlist;
	father->sonlist = pNode;
	pNode->father = father;
}

void del_from_father_list(LinkNode *pNode)
{
	LinkNode *p = pNode->father->sonlist;
	if (p == pNode)
	{
		pNode->father->sonlist = p->next;
	}
	else
	{
		LinkNode *pre = p;
		p = p->next;
		while (p != nullptr)
		{
			if (p == pNode)
			{
				pre->next = p->next;
				break; //must has break
			}
			pre = p;
			p = p->next;
		}
	}
}

void init(int mNum)
{
	LOGE("mNum = %d", mNum);
	g_create_tm = 0;

	for (int i = 0; i < maxhashsize; i++)
	{
		m_hashmap[i] = nullptr;
	}
	g_n_mem_cnt = 0;

	//add node
	u64 hash_value = my_hash(filename[0]);
	root = &g_n_mem[g_n_mem_cnt++];
	strcpy_s(root->str, filename[0]);
	root->hash_value = hash_value;
	root->sonlist = nullptr;
	root->next = nullptr;
	root->father = nullptr;
	root->depth = 0;
	root->mNum = mNum;
	root->create_tm = g_create_tm++;
	root->is_del = false;

	//add to hash list
	root->hash_next = nullptr;
	add_to_hash_list(&m_hashmap[root->hash_value % maxhashsize], root);
}

void add_file(char upper[], char newFile[], int mNum)
{
	u64 hash_value = my_hash(upper);
	LinkNode * father = is_exist_in_hash_list(hash_value);
	if (father == nullptr)
	{
		LOGE("there is no upper= %s mNum=%d return %s", upper, mNum, newFile);
		return;
	}
	else if (mNum >= father->mNum)
	{
		LOGE("can not add to upper= %s mNum=%d > %d return %s", upper, mNum, father->mNum, newFile);
		return;
	}

	father->mNum -= mNum;

	LOGE("father = %s %s , son = %s", father->str, upper, newFile);

	//add node
	u64 son_hash_value = my_hash(newFile);
	LinkNode *son = &g_n_mem[g_n_mem_cnt++];
	strcpy_s(son->str, newFile);
	son->hash_value = son_hash_value;
	son->sonlist = nullptr;
	son->next = nullptr;
	son->mNum = mNum;
	son->depth = father->depth + 1;
	son->create_tm = g_create_tm++;
	son->is_del = false;

	//add to hash list
	son->hash_next = nullptr;
	add_to_hash_list(&m_hashmap[son->hash_value % maxhashsize], son);

	//add to father son list
	add_to_father_son_list(father, son);
}

void update_depth(LinkNode *father, u32 *max_depth)
{
	if (father->sonlist == nullptr)
	{
		if (father->depth > *max_depth)
		{
			*max_depth = father->depth;
		}
		return;
	}

	LinkNode *p = father->sonlist;
	while (p != nullptr) {
		p->depth = father->depth + 1;
		update_depth(p, max_depth);
		p = p->next;
	}
}

//如果file 已经存在于 upper 下, 直接返回
//假设深度最大是5, 如果并入 upper 后，总的深度超过5，则直接返回
//合并规则: file 的 mNum 并入 upper 节点, file 的子链表变成 upper 的子链表
void merge_file(char upper[], char file[])
{
	u64 hash_value = my_hash(upper);
	LinkNode * father = is_exist_in_hash_list(hash_value);

	u64 son_hash_value = my_hash(file);
	LinkNode * son = is_exist_in_hash_list(son_hash_value);

	if (father == nullptr || son == nullptr)
		return;

	LOGE("father = %s %s , son = %s %s", father->str, upper, son->str, file);

	u32 max_depth = 0;
	update_depth(son, &max_depth);
	if (father->depth + max_depth - son->depth > m_max_depth)
	{
		LOGE("depth is beyond limit, return depth: %s(%d) %s(%d) max_depth(%d)",
			father->str, father->depth, son->str, son->depth, max_depth);
		return;
	}
	LOGE("max_depth of %s = %d", son->str, max_depth);

	//remove from hash list
	del_from_hash_list(son->hash_value);

	//remove from son list (这里要注意, son 的father 不是当前的 father)
	del_from_father_list(son);

	//add to father son list
	LinkNode *p = son->sonlist;
	while (p != nullptr) {
		// 这里有 BUG, 因为添加到父节点的子链表会导致 next指向改变, 再指行 p=p->next 会错误
		LinkNode *q = p;
		p = p->next;
		add_to_father_son_list(father, q);
	}

	max_depth = 0;
	update_depth(father, &max_depth);

	//remove node
	son->sonlist = nullptr;
	son->father = nullptr;
	son->next = nullptr;
	son->hash_value = 0;
	son->mNum = 0;
	son->depth = 0;
	son->is_del = true;
}

//给排名前 depthNum 的文件增加个数 (除了 root 之外)
//排名规则: 人数最少,优先级最高, 如果人数一样,创建时间越晚，优先级越高
//这里有一个问题, 当直接对 g_n_mem 数组进行冒泡排序, 将会改下对于数组里面的内容
//但由于这个数组对应的 hash 节点里的 *filePtr 还是指向之前的内存, 因此需要同步改 hash

void recuit(u32 depthNum, u32 mNum)
{
	LOGE("depthNum = %d mNum = %d", depthNum, mNum);
	int count = 0;
	for (int i = 1; i < g_n_mem_cnt - 1; i++)
	{
		if (g_n_mem[i].is_del == false)
		{
			count++;
			for (int j = i + 1; j < g_n_mem_cnt; j++)
			{
				if (g_n_mem[j].is_del == false
					&& (g_n_mem[i].mNum > g_n_mem[j].mNum
						|| (g_n_mem[i].mNum > g_n_mem[j].mNum && g_n_mem[i].create_tm < g_n_mem[j].create_tm)))
				{
					LinkNode tmp = g_n_mem[i];
					g_n_mem[i] = g_n_mem[j];
					g_n_mem[j] = tmp;
				}
			}
		}
		if (count >= depthNum)
		{
			break;
		}
	}

	count = 0;
	for (int i = 1; i < g_n_mem_cnt - 1; i++)
	{
		if (g_n_mem[i].is_del == false)
		{
			g_n_mem[i].mNum += mNum;
			count++;
		}
		if (count >= depthNum)
		{
			break;
		}
	}
}

//root->1->3(april)->5(monday)->6(love)->7(kiss)
//root->1->4(friday)
//root->2(bbb)->8(autum)->9(weather)->10(water)

extern void test_hash_optimize()
{
	init(100);
	add_file(filename[0], filename[1], 50);
	add_file(filename[0], filename[2], 40);
	add_file(filename[1], filename[3], 10);
	add_file(filename[1], filename[4], 15);

	add_file(filename[3], filename[5], 5);
	add_file(filename[5], filename[6], 3);
	add_file(filename[6], filename[7], 2);

	add_file(filename[2], filename[8], 7);

	//add fail
	add_file(filename[8], filename[9], 9); //父只有7个,添加失败
	add_file(filename[9], filename[10], 4); //没有父节点，添加失败

	add_file(filename[8], filename[9], 6); //父有7个
	add_file(filename[9], filename[10], 4); // okay
	add_file(filename[9], filename[11], 1); // okay

	//recuit(3, 10);
	merge_file(filename[6], filename[8]);
	merge_file(filename[6], filename[9]);

	//检查 hash 链表 filePtr 是否已经被修改
	for (int i = 0; i < g_n_mem_cnt; i++)
	{
		u64 hash_value = my_hash(filename[i]);
		LinkNode * pNode = is_exist_in_hash_list(hash_value);
		if (pNode != nullptr)
			LOGE("filename[%d] = %s, node str = %s", i, filename[i], pNode->str);
	}
}

 

Log：

e:\c++\c_test\c_test\hash_optimize.cpp init 162 : mNum = 100
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = root root , son = aaaa
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = root root , son = bbbb
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = aaaa aaaa , son = april
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = aaaa aaaa , son = friday
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = april april , son = monday
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = monday monday , son = love
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = love love , son = kiss
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = bbbb bbbb , son = autum
e:\c++\c_test\c_test\hash_optimize.cpp add_file 200 : can not add to upper= autum mNum=9 > 7 return weather
e:\c++\c_test\c_test\hash_optimize.cpp add_file 195 : there is no upper= weather mNum=4 return water
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = autum autum , son = weather
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = weather weather , son = water
e:\c++\c_test\c_test\hash_optimize.cpp add_file 206 : father = weather weather , son = leak
e:\c++\c_test\c_test\hash_optimize.cpp merge_file 261 : father = love love , son = autum autum
e:\c++\c_test\c_test\hash_optimize.cpp merge_file 268 : depth is beyond limit, return depth: love(4) autum(2) max_depth(4)
e:\c++\c_test\c_test\hash_optimize.cpp merge_file 261 : father = love love , son = weather weather
e:\c++\c_test\c_test\hash_optimize.cpp merge_file 271 : max_depth of weather = 4
e:\c++\c_test\c_test\hash_optimize.cpp del_from_hash_list 103 : 2575973297045806516 will be del from hash list idx = 69
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[0] = root, node str = root
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[1] = aaaa, node str = aaaa
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[2] = bbbb, node str = bbbb
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[3] = april, node str = april
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[4] = friday, node str = friday
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[5] = monday, node str = monday
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[6] = love, node str = love
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[7] = kiss, node str = kiss
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[8] = autum, node str = autum
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[10] = water, node str = water
e:\c++\c_test\c_test\hash_optimize.cpp test_hash_optimize 384 : filename[11] = leak, node str = leak
e:\c++\c_test\c_test\c_test.cpp main 37 : process_time = 7.000000