tipplektionen/main.cpp

#include <algorithm>
#include <filesystem>
#include <fstream>
#include <iostream>
#include <locale>
#include <random>
#include <string>
#include <vector>

#include <boost/program_options.hpp>

#include <unicode/unistr.h>
#include <unicode/ustream.h>

namespace po = boost::program_options;

int main(int argc, char *argv[])
{
    icu::UnicodeString letters;
    icu::UnicodeString mustHaveLetters;
    int wordLengthMin{1};
    int wordLengthMax{1024};
    int lineLength{60};
    int numLines{4};
    std::string wordsFile{"/usr/share/dict/ngerman"};
    bool lowercase{false};
    bool startsCapital{false};

    po::options_description desc("Erlaubte Optionen");
    desc.add_options()("help,h", "Ausgabe der Hilfe")(
        "characters,c", po::value<icu::UnicodeString>(&letters), "Erlaubte Buchstaben")(
        "must-have,m", po::value<icu::UnicodeString>(&mustHaveLetters),
        "Buchstaben, die im Wort vorhanden sein *müssen* (nur in Verbindung mit --characters; "
        "Buchstaben müssen davon Subset sein.)")("min-length", po::value<int>(&wordLengthMin),
                                                 "Minimale Wortlänge")(
        "max-length", po::value<int>(&wordLengthMax),
        "Maximale Wortlänge")("line-length", po::value<int>(&lineLength), "Maximale Zeilenlänge")(
        "num-lines,n", po::value<int>(&numLines), "Anzahl Zeilen")(
        "dictfile,d", po::value<std::string>(&wordsFile), "Pfad zur Wörterbuchdatei")(
        "lowercase,l", po::bool_switch(&lowercase),
        "Ausgabe nur klein geschrieben")("starts-capital,s", po::bool_switch(&startsCapital),
                                         "Wort muss mit einem Großbuchstaben beginnen");

    po::variables_map vm;
    try {
        po::store(po::parse_command_line(argc, argv, desc), vm);
        po::notify(vm);
    } catch (po::error &err) {
        std::cerr << "Fehler beim Verarbeiten der Optionen:\n";
        std::cerr << err.what() << std::endl;
        return 1;
    }

    if (vm.contains("help")) {
        std::cout << desc << std::endl;
        return 1;
    }

    if (mustHaveLetters.length() > 0) {
        for (int i{0}; i < mustHaveLetters.length(); i++) {
            auto ch = mustHaveLetters.charAt(i);
            if (letters.indexOf(ch) == -1) {
                std::cout
                    << "Die Buchstaben von --must-have müssen ein Subset von --characters sein."
                    << std::endl;
                return 1;
            }
        }
    }

    std::ifstream infile(wordsFile);
    std::string line;
    icu::UnicodeString uline;
    std::vector<icu::UnicodeString> words;
    icu::UnicodeString lettersUpper = letters;
    lettersUpper.toUpper();
    icu::UnicodeString mustHaveLettersUpper = mustHaveLetters;
    mustHaveLettersUpper.toUpper();
    long counter{0};

    if (!std::filesystem::exists(wordsFile)) {
        std::cerr << "Datei existiert nicht: " << wordsFile << std::endl;
        return 1;
    }

    while (std::getline(infile, line)) {
        uline = icu::UnicodeString::fromUTF8(line);
        int length = uline.length();

        if (length < wordLengthMin || length > wordLengthMax)
            continue;

        if (mustHaveLetters.length() > 0) {
            bool valid = true;
            for (int i{0}; i < mustHaveLetters.length(); i++) {
                auto ch = mustHaveLetters.charAt(i);
                auto chU = mustHaveLettersUpper.charAt(i);
                if (uline.indexOf(ch) == -1 && uline.indexOf(chU) == -1) {
                    valid = false;
                    break;
                }
            }
            if (!valid)
                continue;
        }

        if (letters.length() > 0) {
            bool valid = true;
            for (int i{0}; i < length; i++) {
                auto ch = uline.charAt(i);
                if (startsCapital && i == 0) {
                    if (lettersUpper.indexOf(ch) == -1) {
                        valid = false;
                        break;
                    }
                } else if (letters.indexOf(ch) == -1 && lettersUpper.indexOf(ch) == -1) {
                    valid = false;
                    break;
                }
            }
            if (!valid)
                continue;
        }
        if (lowercase) {
            uline.toLower();
        }

        words.push_back(uline);
        counter++;
    }

    if (counter == 0) {
        std::cout << "Keine Wörter mit den angegebenen Kriterien gefunden." << std::endl;
        return 0;
    }

    std::cout << "Anzahl infrage kommender Wörter: " << counter << "\n" << std::endl;

    std::random_device rdev;
    std::mt19937 rng(rdev());
    std::uniform_int_distribution<std::mt19937::result_type> distrib(0, words.size() - 1);

    for (int i{0}; i < numLines; i++) {
        icu::UnicodeString lectionLine;
        while (lectionLine.length() < lineLength) {
            if (lectionLine.isEmpty())
                lectionLine.append(words.at(distrib(rng)));
            else
                lectionLine.append(" " + words.at(distrib(rng)));
        }

        if (lectionLine.length() > lineLength) {
            lectionLine.retainBetween(0, lectionLine.lastIndexOf(" ") + 1);
            lectionLine.trim();
        }

        std::cout << lectionLine << std::endl;
    }
}