UvaLive 4917 Abstract Extract （模拟）

  1 #include <bits/stdc++.h>
  2 using namespace std;
  3 char a[512][256];
  4 bool para(int st, int ed){
  5     string s;
  6     set<string> dic;
  7     for(int i = st; i < ed; i++){//将很多行的段落 转换为只有一行的string
  8         s += a[i], s += ‘\n‘;
  9     }
 10     int len = s.size();
 11     vector<int> sen_be; //标记句子开头
 12     vector<int> sen_ed; //句子结尾
 13     vector<set<string> > sen_word;// 每个句子的distinct单词
 14     int k = 0, l = 0;
 15     //闭区间[k,l] 为句子
 16     int once = 0, t = 3;
 17     for(;;){
 18 
 19         while(s[k] == ‘ ‘ || s[k] == ‘\n‘) {
 20             k++;
 21             if(k >= len) break;
 22         }
 23         if(k >= len) break;
 24         while(s[l] != ‘?‘ && s[l] != ‘!‘ && s[l] != ‘.‘){
 25             l++;
 26             if(l >= len) break;
 27         }
 28         if(l >= len) break;
 29         sen_be.push_back(k);
 30         sen_ed.push_back(l);
 31         l++;
 32         k = l;
 33     }
 34 
 35     /*-----------下面处理这段中所有的句子-----------*/
 36 
 37     int n_sen = sen_be.size();
 38     if(n_sen < 3){
 39         return false;
 40     }
 41     for(int i = 0; i < n_sen; i++){
 42         int k1, l1;
 43         k1 = sen_be[i], l1 = sen_ed[i];
 44         string temp = s.substr(k1,l1-k1);
 45         set<string> temp1;
 46         stringstream ss(temp);
 47         string temp_word;
 48         while(ss >> temp_word){
 49             string word = "";
 50             for(int i = 0; i < temp_word.size(); i++){
 51                 if(isalnum(temp_word[i]))
 52                     word += tolower(temp_word[i]);
 53             }
 54             temp1.insert(word);
 55         }
 56         sen_word.push_back(temp1);
 57     }
 58     int mcnt = 0, k2 = sen_be[0], l2 = sen_ed[0];
 59     for(int i = 0; i < n_sen; i++){
 60         int cnt = 0;
 61         for(set<string> :: iterator it = sen_word[i].begin(); it != sen_word[i].end(); it ++){
 62             for(int j = i + 1; j < n_sen; j++){
 63                 if(sen_word[j].count(*it)){
 64                     cnt++;
 65                     break;
 66                 }
 67             }
 68         }
 69         if(cnt > mcnt){
 70             mcnt = cnt;
 71             k2 = sen_be[i];
 72             l2 = sen_ed[i];
 73         }
 74     }
 75 
 76     for(int i = k2; i <= l2; i++){
 77         putchar(s[i]);
 78     }
 79     printf("\n");
 80     return true;
 81 }
 82 void art(int n){
 83     int last = 0;
 84     bool flag = false; //标记是否所有段落都小于3个句子， 如果是要输出空行
 85     for(int i = 0; i < n; i++){
 86         if(a[i][0] == 0){
 87             if(para(last,i))
 88                 flag = true;
 89             last = i + 1;
 90         }
 91     }
 92     if(last != n)
 93         if(para(last, n))
 94             flag = true;
 95     if(!flag) printf("\n");
 96     puts("======");
 97 }
 98 int main(){
 99     int line = 0;
100     while(gets(a[line])){//gets是不会读入回车符的， 但fgets会
101         if(strcmp(a[line],"***") == 0){ // article
102             art(line);
103             line = 0;//每读完一篇文章就让line 归0
104         }
105         else if(strcmp(a[line], "******") == 0){
106             art(line);
107             break;
108         }
109         else line++;
110     }
111 }