#include<iostream>
#include<fstream>
#include<sstream>
#include<string>
#include<iomanip>
usingnamespace std;
constint ArrayMax = 50000;
int readData(string *words)
{
ifstream infile;
string line;
string fileName;
cout << "Type the name of text to read ";
cin >> fileName;
cout << "Opening File: " << "\""<< fileName <<"\""<<endl;
infile.open (fileName); //name of file here. plz mention Complete path if file is not at root
int count=0;
string data;
if (infile.is_open()) //if file opened
{
while(infile>>data)
{
words[count] = data;
count++;
}
return count;
}
else//if file not found show the below message
{
cout << "Sorry, we could not find the file." << endl;
}
infile.close();
return -1;
}
void removeSpecialCharacter(string s)
{
for (int i = 0; i < s.size(); i++)
{
// Finding the character whose
// ASCII value fall under this
// range
if (s[i] < 'A' || s[i] > 'Z' && s[i] < 'a' || s[i] > 'z')
{
// erase function to erase
// the character
s.erase(i, 1);
i--;
}
}
}
string cleanWord(string in)
{
string data = "";
char ch;
for(unsignedint i = 0; i < in.length(); i++)
{
ch = in[i];
//convert upper case to lower
if(ch >= 'A' && ch <= 'Z' )
{
data += ch-'A'+'a';
}
elseif(ch>='0' && ch<='9' )
{
data += ch;
}
elseif(ch>='a' && ch<='z' )
{
data += ch;
}
///////////// else
//////////////// {
/////////////////// data.erase(i, 1);
//////////////// }
}
return data;
}
void cleanAndStoreToFile(string *word,int count)
{
ofstream myfile ("ast1log.txt");
cout << "Exporting data to file......" << endl;
for(int i=0; i < count; i++)
{
myfile << word[i] << " ===> " << cleanWord(word[i]) << endl;
}
}
int main()
{
string words[ArrayMax];
cout << "Array size is : " << ArrayMax << endl;
int count = readData(words);
cleanAndStoreToFile(words,count);
cout << "Finished Reading " << count <<" words. Closing File." << endl;
cout << "Commands" << endl;
cout << "Unigram: search for unigram" << endl;
cout << "Bigram : search for bigram" << endl;
cout << "Save : save 1gms.txt and 2gms.txt" << endl;
cout << "Index : print index of unigram or bigram " << endl;
cout << "Quit/q : terminate program" << endl;
}
string cleanWord(string in)
에서
파일 읽으면
일단 숫자 포함 대문자 소문자로 빠꾸고 싶은데 어렵네요....
파일 읽는게
*** START OF THIS PROJECT GUTENBERG EBOOK THE TIME MACHINE *** Release Date: October 2, 2004 [EBook #35] Author: H. G. Wells
이라면
*** ===> START ===> start OF ===> of THIS ===> this PROJECT ===> project GUTENBERG ===> gutenberg EBOOK ===> ebook THE ===> the TIME ===> time MACHINE ===> machine *** ===> Release ===> release Date: ===> date October ===> october 2, ===> 2 2004 ===> 2004 [EBook ===> ebook #35] ===> 35 Author: ===> author H. ===> h G. ===> g Wells ===> wells
자꾸 이렇게 되네요.
START ===> start OF ===> of THIS ===> this PROJECT ===> project GUTENBERG ===> gutenberg EBOOK ===> ebook THE ===> the TIME ===> time MACHINE ===> machine Release ===> release Date: ===> date October ===> october 2, ===> 2 2004 ===> 2004 [EBook ===> ebook #35] ===> 35 Author: ===> author H. ===> h G. ===> g Wells ===> wells
원래라면 이렇게 되야하는데....
도움 부탁드려요
