Main.cpp

#include <iostream>
#include <fstream>
#include <string>
#include <vector>
#include "Aminoacid.h"
#include "Protein.h"
#include "ORF.h"

using namespace std;

string readSequence(const string& path) {
    string seq;
    ifstream file(path);
    if (file.is_open()) {
        getline(file, seq);
        file.close();
    }
    return seq;
}

string transcribe(string dna) {
    for (char& c : dna) {
        if (c == 'T') c = 'U';
    }
    return dna;
}

// zwraca wektor wskaznikow do dynamicznie alokowanych orfow.
vector<ORF*> findProteins(const string& seq) {
    vector<ORF*> proteins;
    const char* mrna = seq.c_str(); // wskaznik na poczatek bufora mrna

    for (int frame = 0; frame < 3; ++frame) {
        size_t pos = frame;

        while (true) {
            size_t start = seq.find("AUG", pos);
            if (start == string::npos) break;

            if ((start - frame) % 3 != 0) {
                pos = start + 1;
                continue;
            }

            Protein p;
            bool foundStop = false;

            for (size_t i = start; i + 3 <= seq.size(); i += 3) {
                Aminoacid a(mrna + i);
                if (a.aminoacid == '*') {
                    foundStop = true;
                    pos = i + 3;
                    break;
                }
                p.push(a);
            }

            proteins.push_back(new ORF{frame, p, foundStop}); // alokacja dynamiczna

            if (!foundStop) break;
        }
    }

    return proteins;
}


void printProteins(const vector<ORF*>& proteins) {
    cout << "Liczba bialek: " << proteins.size() << endl << endl;

    for (size_t i = 0; i < proteins.size(); ++i) {
        const ORF* orf = proteins[i];

        cout << "Bialko " << i + 1 << " (ORF " << orf->frame + 1 << ")";
        if (!orf->complete) cout << " (brak kodonu STOP)";
        cout << ":" << endl;

        string seqProt = orf->protein.getProteinSequence();
        cout << "Sekwencja: " << seqProt << endl;
        cout << "Liczba aminokwasow: " << seqProt.size() << endl;

        cout << "Czestosci aminokwasow:" << endl;
        for (const auto& [aa, count] : orf->protein.countAminoacids()) {
            cout << aa << " " << count << endl;
        }

        cout << endl;
    }
}

// showincomplete pomija bialka bez kodonu stop
void printProteins(const vector<ORF*>& proteins, bool showIncomplete) {
    size_t shown = 0;
    for (const ORF* orf : proteins) {
        if (orf->complete || showIncomplete) shown++;
    }

    cout << "Liczba bialek: " << shown << endl << endl;

    int index = 0;
    for (size_t i = 0; i < proteins.size(); ++i) {
        const ORF* orf = proteins[i];

        if (!orf->complete && !showIncomplete) continue;
        ++index;
        cout << "Bialko " << index << " (Ramka " << orf->frame + 1 << ")";
        if (!orf->complete) cout << " [brak kodonu STOP]";
        cout << ":" << endl;

        string seqProt = orf->protein.getProteinSequence();
        cout << "Sekwencja: " << seqProt << endl;
        cout << "Liczba aminokwasow: " << seqProt.size() << endl;

        cout << "Czestosci aminokwasow:" << endl;
        for (const auto& [aa, count] : orf->protein.countAminoacids()) {
            cout << aa << " " << count << endl;
        }

        cout << endl;
    }
}

void freeProteins(vector<ORF*>& proteins) {
    for (ORF* orf : proteins) {
        delete orf;
    }
    proteins.clear();
}

// postanowilem ze bardziej "biologicznym" podejsciem bedzie zrobienie po prostu trzech ramek odczytu zamiast
// tworzenia jednego ORFa w miejscu gdzie pojawia sie pierwszy kodon start
// poslugiwalem sie tym toolem do weryfikacji wynikow: https://web.expasy.org/translate/

int main(int argc, char* argv[]) {
    if (argc < 2) {
        cerr << "Usage: " << argv[0]
             << " <file.txt> [--show-incomplete]" << endl;
        return 1;
    }

    string path = argv[1];
    bool showIncomplete = false; // domyslnie nie pokazujemy niekompletnych

    for (int i = 2; i < argc; ++i) {
        string arg = argv[i];
        if (arg == "--show-incomplete") {
            showIncomplete = true;
        } else {
            cerr << "Unknown option: " << arg << endl;
            return 1;
        }
    }

    string dna = readSequence(path);
    if (dna.empty()) {
        cerr << "Unable to read file: " << path << endl;
        return 1;
    }

    string mrna = transcribe(dna); // transkrypcja

    vector<ORF*> proteins = findProteins(mrna);
    printProteins(proteins, showIncomplete); // wypisywanie wyniku
    freeProteins(proteins);   // trzeba po sobie posprzatac

    return 0;
}