labs-edaf30/lab2/dictionary.cc

#include "dictionary.h"
#include "edit_distance.h"
#include "word.h"
#include <algorithm>
#include <fstream>
#include <iostream>
#include <set>
#include <string>
#include <vector>

using std::string;
using std::vector;

Dictionary::Dictionary() {}

bool Dictionary::contains(const string &word) const {
    auto l = word.length();
    Word w = Word(word);
    if (std::find(this->words[l].begin(), this->words[l].end(), w) !=
        std::end(this->words[l])) {
        return true;
    }
    return false;
}

std::vector<string> Dictionary::get_suggestions(const string &word) const {
    vector<string> suggestions;
    add_trigram_suggestions(suggestions, word);
    trim_suggestions(suggestions, word);
    rank_suggestions(suggestions, word);
    return suggestions;
}

void Dictionary::add_trigram_suggestions(std::vector<std::string> &suggestions,
                                         const std::string &word) const {
    // Get trigrams of the input word
    Word input_word(word);
    const std::vector<std::string> &input_trigrams = input_word.get_trigrams();

    // Iterate through all words in the dictionary
    for (int i = 0; i < MAXLEN; ++i) {
        for (const Word &dict_word : words[i]) {
            // Get the trigrams of the dictionary word
            const std::vector<std::string> &dict_word_trigrams =
                dict_word.get_trigrams();

            // Count how many trigrams match
            unsigned int match_count = dict_word.get_matches(input_trigrams);

            // If there are any matches, add the word to suggestions
            if (match_count > 0) {
                suggestions.push_back(dict_word.get_word());
            }
        }
    }
}

void Dictionary::rank_suggestions(std::vector<std::string> &suggestions,
                                  const std::string &word) const {
    // Sort suggestions based on the levenshtein distance
    std::sort(suggestions.begin(), suggestions.end(),
              [&](const std::string &a, const std::string &b) {
                  auto dist_a = edit_distance(a, word);
                  auto dist_b = edit_distance(b, word);
                  return dist_a < dist_b;
              });
}

void Dictionary::trim_suggestions(std::vector<std::string> &suggestions,
                                  const std::string &word) const {
    // Remove duplicates using a set
    std::set<std::string> unique_suggestions(suggestions.begin(),
                                             suggestions.end());
    suggestions.assign(unique_suggestions.begin(), unique_suggestions.end());

    // Remove the input word from the suggestions list (if present)
    suggestions.erase(std::remove(suggestions.begin(), suggestions.end(), word),
                      suggestions.end());

    auto l = word.length();
    std::cout << "WTF" << l << std::endl;

    // Example: Remove any suggestions that are not within 1 string length
    suggestions.erase(std::remove_if(suggestions.begin(), suggestions.end(),
                                     [l](const std::string &s) {
                                         return s.length() > (l + 1) ||
                                                s.length() < (l - 1);
                                     }),
                      suggestions.end());
}

int Dictionary::spit(path p) {
    std::ofstream file(p);

    if (!file.is_open()) {
        std::cerr << "Error opening file! " << std::endl;
        return 1;
    }

    for (int a = 0; a < MAXLEN; a++) {
        for (auto &word : words[a]) {
            file << word;
            file << std::endl;
        }
    }

    file.flush();
    file.close();
    return 0;
}

int Dictionary::slurp(path p) {
    std::ifstream file(p.string());

    if (!file.is_open()) {
        std::cerr << "Error opening file! " << std::endl;
        return 1;
    }

    std::string line;
    while (std::getline(file, line)) {
        if (line.empty())
            continue;
        // Words larger  than max gets placed in the topmost bucket
        words[std::min(line.size(), static_cast<size_t>(MAXLEN) - 1)].push_back(
            Word(line));
    }

    file.close();
    return 0;
}