查重

1/源.cpp

+#define _CRT_SECURE_NO_WARNINGS
+#include <stdio.h>
+#include <stdlib.h>
+#include <io.h>
+#include <string>
+#include <iostream>
+#include <conio.h>
+
+using namespace std;
+
+struct filenm;
+struct file1;
+struct sentence;
+struct node;
+struct con;
+
+struct filenm
+{
+    string name;
+    filenm* next = NULL;
+    node* head = NULL;
+
+};
+
+
+struct file1
+{
+    filenm* path = NULL;
+    sentence* head = NULL;
+    int num = 0;
+};
+
+struct sentence
+{
+
+    string s1;
+    string s2;
+    int len1 = 0;
+    int len2 = 0;
+    sentence* next = NULL;
+};
+
+
+struct node
+{
+
+    int flags = 0;
+    file1* a_file = NULL;
+    file1* b_file = NULL;
+    string path1;
+    string path2;
+    con* head = NULL;
+    int num = 0;
+    node* next = NULL;
+
+    double rate = 0;
+
+    int code = 0;
+};
+
+struct con
+{
+
+    sentence* a_sen = NULL;
+    sentence* b_sen = NULL;
+    double similar = 0;
+    int score = 0;
+
+    con* next = NULL;
+};
+
+filenm* create_filenm();
+filenm* create_filenm(const string s);
+filenm* findlast_filenm(filenm* head);
+void create_filenm(filenm* head, const string s);
+filenm* dir_Allfile(string path);
+filenm* get_files(string path);
+sentence* create_sentence();
+sentence* create_sentence(string s);
+sentence* findlast_sentence(sentence* head);
+void create_sentence(sentence* head, const string s);
+file1* create_file1();
+file1* get_file(string path);
+con* create_con();
+node* create_node();
+
+bool is_exist(const char* s);
+bool is_dir(const char* fileName);
+bool istxt(string s);
+string& replace_all(string& src, const string& old_value, const string& new_value);
+void inil_sentence(sentence* s, int);
+void inil_sentence(sentence* s);
+int ex_dis(string q1, string q2);
+int calc_dis(string s1, string s2);
+double rate_dis(string s1, string s2, int score);
+con* find_best(sentence* one, sentence* many, int);
+con* find_best(sentence* head1, sentence* head2);
+con* find_best(file1* a, file1* b);
+void inil_node(node* n);
+void inil_node(node* n, filenm* f1, filenm* f2);
+void one_to_many(filenm* f1, filenm* head);
+void many_to_many(filenm* head1, filenm* head2);
+int get_length(con* m);
+int get_length(sentence* s);
+int get_length(filenm* f);
+void repeat_rate(node* n);
+
+void print_percent(double d);
+void print_filenm(filenm* f);
+void print(const string& s);
+void print_sentence1(sentence* f);
+void print_sentence2(sentence* f);
+void print(sentence* s, int flags);
+void print(const char* s, double num);
+void print(con* m, int);
+void print(con* m);
+void print(file1* a);
+void print(file1* a, int);
+void print(node* n);
+void print(filenm* f);
+void print_info(node* n);
+void print_info(filenm* f);
+bool print_info(filenm* f, int code);
+
+void impo02();
+void impo03();
+
+
+
+
+filenm* create_filenm()
+{
+    filenm* f = new filenm();
+    if (f == NULL)
+    {
+        printf("创建失败！\n");
+        return NULL;
+    }
+    else
+    {
+        f->name = "";
+        f->next = NULL;
+        f->head = NULL;
+        return f;
+    }
+}
+
+
+filenm* create_filenm(const string s)
+{
+    filenm* f = create_filenm();
+    f->name = s;
+    return f;
+}
+
+filenm* findlast_filenm(filenm* head)
+{
+    if (head == NULL)
+    {
+        printf("有误\n");
+        return NULL;
+    }
+    else
+    {
+        filenm* f1 = head;
+        while (f1->next != NULL)
+        {
+            f1 = f1->next;
+        }
+        return f1;
+    }
+}
+
+void create_filenm(filenm* head, const string s)
+{
+    filenm* f = findlast_filenm(head);
+    filenm* f2 = create_filenm(s);
+    f->next = f2;
+}
+
+filenm* dir_Allfile(string path)
+{
+
+    filenm* f1 = create_filenm();
+    intptr_t hFile = 0;
+
+    struct _finddata_t fileinfo;
+    string p;
+    if ((hFile = _findfirst(p.assign(path).append("\\*").c_str(), &fileinfo)) != -1)
+    {
+        do
+        {
+            string q1 = p.assign(path).append("\\").append(fileinfo.name);
+            if (istxt(q1))
+            {
+                create_filenm(f1, q1);
+            }
+        } while (_findnext(hFile, &fileinfo) == 0);
+        _findclose(hFile);
+    }
+    return f1->next;
+}
+
+filenm* get_files(string path)
+{
+    if (istxt(path))
+        return create_filenm(path);
+    return dir_Allfile(path);
+}
+
+sentence* create_sentence()
+{
+    sentence* f = new sentence();
+    if (f == NULL)
+    {
+        printf("创建失败！\n");
+        return NULL;
+    }
+    else
+    {;
+        f->s1 = "";
+        f->s2 = "";
+        f->len1 = 0;
+        f->len2 = 0;
+        f->next = NULL;
+        return f;
+    }
+}
+
+sentence* create_sentence(string s)
+{
+    sentence* f = create_sentence();
+    f->s1 = s;
+    return f;
+}
+
+sentence* findlast_sentence(sentence* head)
+{
+    if (head == NULL)
+    {
+        printf("传入的sentence指针有误\n");
+        return NULL;
+    }
+    else
+    {
+        sentence* f1 = head;
+        while (f1->next != NULL)
+        {
+            f1 = f1->next;
+        }
+        return f1;
+    }
+}
+
+
+void create_sentence(sentence* head, const string s)
+{
+    sentence* f = findlast_sentence(head);
+    sentence* f2 = create_sentence(s);
+    f->next = f2;
+}
+
+
+file1* create_file1()
+{
+    file1* f = new file1();
+    if (f == NULL)
+    {
+        printf("创建失败！\n");
+        return NULL;
+    }
+    else
+    {
+        f->head = NULL;
+        f->path = NULL;
+        f->num = 0;
+        return f;
+    }
+}
+
+
+file1* get_file(string path)
+{
+
+    FILE* fp;
+    if ((fp = fopen(path.c_str(), "r")) == NULL) 
+    {
+        printf("读取文件失败!\n");
+        return NULL;
+    }
+
+    file1* a = create_file1();
+    sentence* sen = create_sentence();
+
+    int num = 0;
+    char ch = ' ';
+    string str = "";
+    while (fp != NULL && (ch = fgetc(fp)) != EOF)
+    {
+        str.push_back(ch);
+        if (ch == ';')
+        {
+            create_sentence(sen, str);
+            num++;
+            str = "";
+        }
+    }
+    create_sentence(sen, str);
+    num++;
+
+    a->head = sen->next;
+    a->num = num;
+    a->path = create_filenm();
+    a->path->name = path;
+
+    if (fp != NULL)
+    {
+        fclose(fp);  
+    }
+
+    return a;
+}
+
+
+con* create_con()
+{
+    con* f = new con();
+    if (f == NULL)
+    {
+        printf("创建失败！\n");
+        return NULL;
+    }
+    else
+    {
+        f->a_sen = NULL;
+        f->b_sen = NULL;
+        f->next = NULL;
+        f->similar = 0;
+        return f;
+    }
+}
+
+
+node* create_node()
+{
+    node* f = new node();
+    if (f == NULL)
+    {
+        printf("创建失败！\n");
+        return NULL;
+    }
+    else
+    {
+        f->flags = 0;
+        f->a_file = NULL;
+        f->b_file = NULL;
+        f->head = NULL;
+        f->num = 0;
+        f->next = NULL;
+        return f;
+    }
+}
+
+
+bool is_exist(const char* s)
+{
+    if (!_access(s, 0))
+    {
+        return true;
+    }
+    printf("%s不存在\n", s);
+    return false;
+}
+
+bool is_dir(const char* fileName)
+{
+    struct stat buf;
+    int result;
+    result = stat(fileName, &buf);
+    if (S_IFDIR & buf.st_mode) {
+
+        return true;
+    }
+    else 
+    return false;
+}
+
+bool istxt(string s)
+{
+    if (!is_exist(s.c_str()))
+        return false;
+    string::size_type idx;
+    idx = s.find(".txt"); 
+    if (idx == string::npos) 
+    {
+        return false;
+    }
+    else
+    {
+        return true;
+    }
+}
+
+
+string& replace_all(string& src, const string& old_value, const string& new_value)
+{
+    for (string::size_type pos(0); pos != string::npos; pos += new_value.length())
+    {
+        if ((pos = src.find(old_value, pos)) != string::npos) {
+            src.replace(pos, old_value.length(), new_value);
+        }
+        else break;
+    }
+    return src;
+}
+
+
+
+void inil_sentence(sentence* s, int)
+{
+    s->s2 = s->s1;
+
+}
+
+void inil_sentence(sentence* s)
+{
+    while (s != NULL)
+    {
+        inil_sentence(s, 1);
+        s = s->next;
+    }
+}
+
+int ex_dis(string q1, string q2)
+{
+    int len1 = (int)q1.length();
+    int len2 = (int)q2.length();
+    const char* s1 = q1.c_str();
+    const char* s2 = q2.c_str();
+    int dp[60][60];
+
+
+    for (int i = 0; i < 60; i++)
+    {
+        dp[i][0] = i;
+        dp[0][i] = i;
+    }
+
+
+    for (int i = 1; i <= len1; i++)
+        for (int j = 1; j <= len2; j++)
+            dp[i][j] = s1[i - 1] == s2[j - 1] ? dp[i - 1][j - 1] : min(dp[i - 1][j - 1], min(dp[i - 1][j], dp[i][j - 1])) + 1;
+    return dp[len1][len2];
+}
+
+
+int calc_dis(string s1, string s2)
+{
+
+    int len1 = (int)s1.length();
+    int len2 = (int)s2.length();
+    int x = 0;
+    int dis = 58;
+    int score = 0;
+
+    while (x < len1 || x < len2)
+    {
+
+        if (x >= len1)
+        {
+            score += len2 - len1;
+            break;
+        }
+        if (x >= len2)
+        {
+            score += len1 - len2;
+            break;
+        }
+
+
+        string s3(s1.substr(x, dis));
+        string s4(s2.substr(x, dis));
+        x += dis;
+
+        score += ex_dis(s3, s4);
+
+    }
+    return score;
+}
+
+double rate_dis(string s1, string s2, int score)
+{
+    int len1 = (int)s1.length();
+    int len2 = (int)s2.length();
+
+    double x1 = score * 1.0 / len1;
+    double x2 = score * 1.0 / len2;
+    return x1 < x2 ? x1 : x2;
+}
+
+con* find_best(sentence* one, sentence* many, int)
+{
+    string s = one->s2;
+
+    int score = calc_dis(s, many->s2);
+    double min_rate = rate_dis(s, many->s2, score);
+    sentence* best_sen = many;
+    many = many->next;
+
+    while (many != NULL)
+    {
+        int new_score = calc_dis(s, many->s2);
+        double new_rate = rate_dis(s, many->s2, new_score);
+        if (new_rate < min_rate)
+        {
+            score = new_score;
+            min_rate = new_rate;
+            best_sen = many;
+        }
+        many = many->next;
+    }
+    con* m = create_con();
+    m->a_sen = one;
+    m->b_sen = best_sen;
+    m->similar = min_rate;
+    m->score = score;
+
+    return m;
+}
+
+
+con* find_best(sentence* head1, sentence* head2)
+{
+    con* m_head = create_con();
+    con* m = m_head;
+
+    while (head1 != NULL)
+    {
+        m->next = find_best(head1, head2, 1);
+        m = m->next;
+        head1 = head1->next;
+
+    }
+    return m_head->next;
+}
+
+con* find_best(file1* a, file1* b)
+{
+    return find_best(a->head, b->head);
+}
+
+
+void inil_node(node* n)
+{
+    n->a_file = get_file(n->path1);
+    n->b_file = get_file(n->path2);
+
+    inil_sentence(n->a_file->head);
+    inil_sentence(n->b_file->head);
+
+    n->head = find_best(n->a_file, n->b_file);
+
+
+    repeat_rate(n);
+
+}
+
+void inil_node(node* n, filenm* f1, filenm* f2)
+{
+    n->path1 = f1->name;
+    n->path2 = f2->name;
+    inil_node(n);
+}
+
+
+void one_to_many(filenm* f1, filenm* head)
+{
+
+    node* n = create_node();
+    inil_node(n, f1, head);
+    f1->head = n;
+
+    filenm* f2 = head;
+    while (f2->next != NULL)
+    {
+        f2 = f2->next;
+        node* new_node = create_node();
+        inil_node(new_node, f1, f2);
+
+        n->next = new_node;
+        n = n->next;
+
+    }
+
+}
+
+void many_to_many(filenm* head1, filenm* head2)
+{
+
+    while (head1 != NULL)
+    {
+        one_to_many(head1, head2);
+        head1 = head1->next;
+    }
+}
+
+int get_length(con* m)
+{
+    int l = 0;
+    while (m != NULL)
+    {
+        m = m->next;
+        l++;
+    }
+
+    return l;
+}
+
+int get_length(sentence* s)
+{
+    int num = 0;
+    while (s != NULL)
+    {
+        s = s->next;
+        num++;
+    }
+    return num;
+}
+
+int get_length(filenm* f)
+{
+    int num = 0;
+    while (f != NULL)
+    {
+        f = f->next;
+        num++;
+    }
+    return num;
+}
+
+void repeat_rate(node* n)
+{
+    int s1_num = get_length(n->a_file->head);
+    int s2_num = get_length(n->b_file->head);
+
+    int m_num = get_length(n->head);
+
+    n->rate = 1.0 * m_num / s1_num;
+
+}
+
+
+void print_percent(double d)
+{
+    d *= 100;
+    printf("%.2f%%", d);
+}
+void print_filenm(filenm* f)
+{
+
+    while (f != NULL)
+    {
+        printf("%s\n", f->name.c_str());
+        f = f->next;
+    }
+    printf("\n");
+}
+void print(const string& s)
+{
+    printf("%s\n", s.c_str());
+}
+void print_sentence1(sentence* f)
+{
+    while (f != NULL)
+    {
+        printf("%s\n", f->s1.c_str());
+        f = f->next;
+    }
+    printf("\n");
+}
+void print_sentence2(sentence* f)
+{
+    while (f != NULL)
+    {
+        printf("%s\n", f->s2.c_str());
+        f = f->next;
+    }
+    printf("\n");
+}
+void print(sentence* s, int flags)
+{
+    if (flags == 1)
+    {
+        print(s->s1);
+    }
+    if (flags == 2)
+    {
+        print(s->s2);
+    }
+}
+void print(const char* s, double num)
+{
+    printf("%s%lf\n", s, num);
+}
+void print(con* m, int)
+{
+    print(m->a_sen, 1);
+    print(m->b_sen, 1);
+    int l1 = (int)m->a_sen->s2.length();
+    int l2 = (int)m->b_sen->s2.length();
+    print_percent(1 - m->similar);
+    printf("\n\n");
+}
+void print(con* m)
+{
+    printf("\n\n");
+    int num = 1;
+    while (m != NULL)
+    {
+        print(m, 1);
+        m = m->next;
+
+        num++;
+    }
+}
+void print(file1* a)
+{
+    print_sentence1(a->head);
+    get_length(a->head);
+}
+void print(file1* a, int)
+{
+    print_sentence2(a->head);
+    get_length(a->head);
+}
+void print(node* n)
+{
+    if (n == NULL)
+    {
+        printf("这个文件是空的！\n");
+    }
+    else
+    {
+        while (n)
+        {
+            printf("\n\n\n");
+            print(n->a_file->path->name);
+            printf("和");
+            print(n->b_file->path->name);
+            printf("\n");
+            print(n->head);
+
+            print(n->a_file->path->name);
+            print(n->b_file->path->name);
+            printf("\n\n\n");
+            break;
+
+            n = n->next;
+        }
+    }
+}
+void print(filenm* f)
+{
+    print(f->head);
+}
+void print_info(node* n)
+{
+    printf("%s\n%s\n", n->a_file->path->name.c_str(), n->b_file->path->name.c_str());
+    printf("查重率为%2f\n", n->rate);
+}
+void print_info(filenm* f)
+{
+    int code = 0;
+    while (f != NULL)
+    {
+        node* n = f->head;
+        while (n != NULL)
+        {
+            n->code = code;
+            print_info(n);
+
+            code++;
+            n = n->next;
+        }
+        f = f->next;
+    }
+}
+bool print_info(filenm* f, int code)
+{
+    while (f != NULL)
+    {
+        node* n = f->head;
+        while (n != NULL)
+        {
+            if (code == n->code)
+            {
+                print(n);
+                return true;
+            }
+            n = n->next;
+        }
+        f = f->next;
+    }
+    return false;
+}
+
+
+
+void start(filenm* f1, filenm* f2)
+{
+    system("cls");
+    many_to_many(f1, f2);
+
+loou:
+    system("cls");
+    print_info(f1);
+    printf("按 0 退出\n");
+
+    while (true)
+    {
+        int code = 0;
+        cin >> code;
+        if (print_info(f1, code))
+        {
+            system("pause");
+            goto loou;
+            return;
+        }
+        else if (code == 0)
+        {
+            exit(0);
+        }
+        else
+        {
+            printf("您的输入不合法\n");
+        }
+    }
+
+}
+
+void impo03()
+{
+    printf("\n\n");
+    printf("\t请输入第一组文件路径\n");
+    printf("\t请输入第二组文件路径\n");
+    printf("\t按任意键继续"); 
+    char c = _getch();
+    impo02();
+
+
+}
+
+void impo02()
+{
+    printf("\n");
+    printf("\t请输入第一组文件路径\n");
+    printf("\t请输入第二组文件路径\n");
+    printf("\t 按 0 退出\n");
+
+    string s1;
+    string s2;
+    filenm* f1;
+    filenm* f2;
+    int num1 = 0;
+    int num2 = 0;
+
+
+    while (true)
+    {
+
+        printf("\n请输入第一组文件路径及名称：");
+        cin >> s1;
+        if (s1 == "0")
+        {
+            exit(0);
+        }
+       
+        if (istxt(s1))
+        {
+            f1 = get_files(s1);
+            printf("第一组文件为%s\n", s1.c_str());
+            break;
+        }
+        else if (is_dir(s1.c_str()))
+        {
+            f1 = get_files(s1);
+            num1 = get_length(f1);
+            if (num1 == 0)
+            {
+                printf("此文件中.txt文件份数为0 请重新输入\n");
+            }
+            else
+            {
+                printf("此文件中有 %d 份.txt文件\n", num1);
+                print_filenm(f1);
+                break;
+            }
+        }
+    }
+    while (true)
+    {
+        printf("\n请输入第二组文件路径及名称：");
+        cin >> s2;
+        if (s2 == "0")
+        {
+            exit(0);
+        }
+        if (istxt(s2))
+        {
+            f2 = get_files(s2);
+            printf("第2组文件为%s\n", s2.c_str());
+            break;
+        }
+        else if (is_dir(s2.c_str()))
+        {
+            f2 = get_files(s2);
+            num2 = get_length(f2);
+            if (num2 == 0)
+            {
+                printf("此文件中.txt文件份数为0 请重新输入\n");
+            }
+            else
+            {
+                printf("此文件中有 %d 份.txt文件\n", num2);
+                print_filenm(f2);
+                break;
+            }
+        }
+    }
+    system("pause");
+    start(f1, f2);
+}
+
+
+
+
+
+int main()
+{
+
+    printf("\n\n\n\n\n");
+    printf("\t\t\t\t\t\t查重系统\n\n\n");
+    printf("\t\t\t\t\t\t 输入 1 启动程序\n");
+
+    int c = 0;
+    while (1)
+    {
+        c = _getch();
+        if (c == '1')
+        {
+            impo02();
+            break;
+        }
+
+    }
+
+}