infer.cpp

#include <cstdlib>
#include <ctime>
#include <fstream>
#include <iostream>
#include <sstream>
#include <unordered_map>
#include <vector>
#include <algorithm>

using namespace std;

unordered_map<string, uint32_t> vocab;
unordered_map<uint64_t, vector<uint32_t>> hasilToOutput;

void preprocess(vector<string> &theString) {
	vector<string> tmp;
	for(auto s : theString) {
		string tmpp;
		for(auto c : s /*1.6 :P*/) {
			if(!isalnum(c)) {
				if(tmpp.length() == 0) {
					tmp.push_back(string(1, c));
				} else {
					tmp.push_back(tmpp);
					tmpp = "";
					tmp.push_back(string(1, c));
				}
			} else {
			tmpp += tolower(c);
			}
		}
		if(tmpp != "") tmp.push_back(tmpp);
	}
	theString = tmp;
}

// === Fungsi: Load Model dari File ===
void loadModel(const string &filename) {
  ifstream file(filename);
  if (!file) {
    cerr << "Gagal membuka file model.\n";
    exit(1);
  }

  string line;
  bool readingVocab = false;
  bool readingMatch = false;

  while (getline(file, line)) {
    if (line == "Vocabs:") {
      readingVocab = true;
      readingMatch = false;
      continue;
    }
    if (line == "Matchs:") {
      readingVocab = false;
      readingMatch = true;
      continue;
    }

    if (readingVocab) {
      size_t pos = line.find(": ");
      if (pos != string::npos) {
        string word = line.substr(0, pos);
        uint32_t id = stoi(line.substr(pos + 2));
        vocab[word] = id;
      }
    } else if (readingMatch) {
      if (line.back() == ':') {
        uint64_t key = stoull(line.substr(0, line.size() - 1));
        getline(file, line); // [
        vector<uint32_t> targets;
        while (getline(file, line) && line != "]") {
          if (!line.empty()) {
            targets.push_back(stoi(line));
          }
        }
        hasilToOutput[key] = targets;
      }
    }
  }

  file.close();
}

// === Fungsi: Prediksi Kata Berikutnya ===
string inferNextWord(const vector<string> &contextWords) {
  uint64_t total = 0;
  for (size_t i = 0; i < contextWords.size(); ++i) {
    const string &word = contextWords[i];
    if (vocab.count(word)) {
      total += vocab[word] * (i + 1); // Bobot posisi
    } else {
      return "<unknown word: " + word + ">";
    }
  }

  if (hasilToOutput.count(total) == 0) {
    // Cari key terdekat
    uint64_t closestKey = 0;
    uint64_t minDiff = UINT64_MAX;

    for (const auto &[key, _] : hasilToOutput) {
      uint64_t diff = (key > total) ? key - total : total - key;
      if (diff < minDiff) {
        minDiff = diff;
        closestKey = key;
      }
    }

    if (minDiff == UINT64_MAX)
      return "<no prediction>";
    total = closestKey;
  }
  const auto &candidates = hasilToOutput[total];
  // uint32_t predictedID = candidates[rand() % candidates.size()]
  unordered_map<uint32_t, int> freq;
  for (auto id : candidates) {
    freq[id]++;
  }

  uint32_t predictedID = max_element(freq.begin(), freq.end(),
                                     [](const pair<uint32_t, int> &a,
                                        const pair<uint32_t, int> &b) {
                                       return a.second < b.second;
                                     })
                             ->first;

  // Balikkan ID ke kata
  for (const auto &[word, id] : vocab) {
    if (id == predictedID)
      return word;
  }

  return "<not found>";
}

// === Main ===
int main() {
  srand(time(0));
  loadModel("model.txt");

  cout << "Masukkan kalimat sebagai konteks:\n";
  vector<string> context;
  string word;
  string words;
  getline(cin, words);
  stringstream ss(words);
  while (ss >> word) {
    context.push_back(word);
  }
  preprocess(context);
  auto newContext = context;

  string prediction;

  int i = 0;

  while (prediction != "[AKHIR]" && i < 50) {
    prediction = inferNextWord(newContext);
    newContext.push_back(prediction);
    i++;
  }
  cout << "Prediksi kata berikutnya:";
  for (auto m : newContext) {
    cout << " " << m;
    if (m.find("<unknown word:") != string::npos)
      break;
  }
  cout << endl;

  return 0;
}