#include "grader.h"
#include "lang.h"
#include <vector>
#include <map>
#include <cmath>
using namespace std;
#define MAX_LANG 56
#define MAX_SYMB 65536
// We store the frequency of n-grams for each language.
// To save memory, we use maps or small arrays for higher n-grams.
// For this optimized version, we focus on 1-grams and 2-grams for speed/memory.
double cnt[MAX_LANG][MAX_SYMB];
map<pair<int, int>, int> bigrams[MAX_LANG];
void excerpt(int *E) {
double best_score = -1.0;
int best_lang = 0;
for (int l = 0; l < MAX_LANG; l++) {
double score = 0;
// 1-gram scoring
for (int i = 0; i < 100; i++) {
score += cnt[l][E[i]];
}
// 2-gram scoring (Higher weight usually yields better accuracy)
for (int i = 0; i < 99; i++) {
if (bigrams[l].count({E[i], E[i+1]})) {
score += bigrams[l][{E[i], E[i+1]}] * 5.0;
}
}
if (score > best_score) {
best_score = score;
best_lang = l;
}
}
// Guess the language and get the actual answer
int actual = language(best_lang);
// Update our statistical model with the correct answer
for (int i = 0; i < 100; i++) {
cnt[actual][E[i]]++;
}
for (int i = 0; i < 99; i++) {
bigrams[actual][{E[i], E[i+1]}]++;
}
}
