From 48f05ddb4fde33b24ca02dcc5eb5a9f5821d5464 Mon Sep 17 00:00:00 2001
From: Michael Mandl <mandlm@molez.net>
Date: Thu, 21 Mar 2024 08:55:44 +0100
Subject: [PATCH] refactor: add WordList type

---
 QVectorSearch/mainwindow.cpp                  | 19 ++----
 QVectorSearch/mainwindow.h                    |  3 +-
 lib_vector_search/CMakeLists.txt              |  4 +-
 lib_vector_search/include/grouped_finder.h    |  3 +-
 lib_vector_search/include/linear_finder.h     | 10 ++--
 lib_vector_search/include/parallel_finder.h   |  7 +--
 .../include/sorted_linear_finder.h            |  3 +-
 lib_vector_search/include/tree_finder.h       |  6 +-
 lib_vector_search/include/word_list.h         | 14 +++++
 .../include/word_list_generator.h             | 13 ----
 lib_vector_search/src/grouped_finder.cpp      |  2 +-
 lib_vector_search/src/linear_finder.cpp       |  3 +-
 lib_vector_search/src/parallel_finder.cpp     |  4 +-
 .../src/sorted_linear_finder.cpp              |  4 +-
 lib_vector_search/src/tree_finder.cpp         |  7 +--
 lib_vector_search/src/word_list.cpp           | 59 +++++++++++++++++++
 lib_vector_search/src/word_list_generator.cpp | 32 ----------
 vector_search_cli/main.cpp                    | 10 ++--
 18 files changed, 111 insertions(+), 92 deletions(-)
 create mode 100644 lib_vector_search/include/word_list.h
 delete mode 100644 lib_vector_search/include/word_list_generator.h
 create mode 100644 lib_vector_search/src/word_list.cpp
 delete mode 100644 lib_vector_search/src/word_list_generator.cpp

diff --git a/QVectorSearch/mainwindow.cpp b/QVectorSearch/mainwindow.cpp
index 69fb33c..4967bba 100644
--- a/QVectorSearch/mainwindow.cpp
+++ b/QVectorSearch/mainwindow.cpp
@@ -8,9 +8,7 @@
 #include "sorted_linear_finder.h"
 #include "timer.h"
 #include "tree_finder.h"
-#include "word_list_generator.h"
 
-#include <fstream>
 #include <sstream>
 
 MainWindow::MainWindow(QWidget *parent)
@@ -55,7 +53,9 @@ void MainWindow::setupWordList() {
 
 void MainWindow::generateWordList() {
   Timer timer;
-  word_list_ = WordListGenerator::generate(ui->wordListSizeSelector->value());
+  word_list_ = WordList::fourCaps()
+                   .multiply(ui->wordListSizeSelector->value())
+                   .shuffle();
   timer.stop();
 
   std::stringstream status_message;
@@ -64,17 +64,10 @@ void MainWindow::generateWordList() {
 }
 
 void MainWindow::loadWordList(std::filesystem::path path) {
-  word_list_.clear();
-
   Timer timer;
-  std::ifstream ifs(path);
-
-  std::string line;
-  while (std::getline(ifs, line)) {
-    for (auto i = 0; i < ui->wordListSizeSelector->value(); ++i) {
-      word_list_.emplace_back(line);
-    }
-  }
+  word_list_ = WordList::fromFile("words.txt")
+                   .multiply(ui->wordListSizeSelector->value())
+                   .shuffle();
   timer.stop();
 
   std::stringstream status_message;
diff --git a/QVectorSearch/mainwindow.h b/QVectorSearch/mainwindow.h
index 84976b6..33f7d24 100644
--- a/QVectorSearch/mainwindow.h
+++ b/QVectorSearch/mainwindow.h
@@ -2,6 +2,7 @@
 #define MAINWINDOW_H
 
 #include "finder.h"
+#include "word_list.h"
 #include <QMainWindow>
 #include <QStandardItemModel>
 #include <QStringListModel>
@@ -17,7 +18,7 @@ class MainWindow : public QMainWindow {
   Q_OBJECT
 
 private:
-  std::vector<std::string> word_list_;
+  WordList word_list_;
   QStringListModel result_model_;
   QStandardItemModel search_algorithms_;
   QStandardItemModel word_list_sources_;
diff --git a/lib_vector_search/CMakeLists.txt b/lib_vector_search/CMakeLists.txt
index 38f41b6..daa98b2 100644
--- a/lib_vector_search/CMakeLists.txt
+++ b/lib_vector_search/CMakeLists.txt
@@ -7,8 +7,8 @@ project(
 
 add_library(
   vector_search STATIC
-  src/word_list_generator.cpp
-  include/word_list_generator.h
+  src/word_list.cpp
+  include/word_list.h
   src/linear_finder.cpp
   include/linear_finder.h
   src/sorted_linear_finder.cpp
diff --git a/lib_vector_search/include/grouped_finder.h b/lib_vector_search/include/grouped_finder.h
index 060b437..b562330 100644
--- a/lib_vector_search/include/grouped_finder.h
+++ b/lib_vector_search/include/grouped_finder.h
@@ -1,6 +1,7 @@
 #pragma once
 
 #include "finder.h"
+#include "word_list.h"
 
 #include <map>
 #include <vector>
@@ -10,7 +11,7 @@ private:
   std::map<char, std::vector<const std::string *>> groups_;
 
 public:
-  GroupedFinder(const std::vector<std::string> &word_list);
+  GroupedFinder(const WordList &word_list);
 
   virtual std::forward_list<const std::string *>
   find_prefix(std::string_view search_term) const override;
diff --git a/lib_vector_search/include/linear_finder.h b/lib_vector_search/include/linear_finder.h
index 35201d9..7aabb35 100644
--- a/lib_vector_search/include/linear_finder.h
+++ b/lib_vector_search/include/linear_finder.h
@@ -1,17 +1,15 @@
 #pragma once
 
 #include "finder.h"
-
-#include <vector>
-
-using std::vector;
+#include "word_list.h"
 
 class LinearFinder : public Finder {
 private:
-  const vector<std::string> &word_list_;
+  const WordList &word_list_;
 
 public:
-  LinearFinder(const vector<std::string> &word_list);
+  LinearFinder(const WordList &word_list);
+
   std::forward_list<const std::string *>
   find_prefix(std::string_view search_term) const override;
 };
diff --git a/lib_vector_search/include/parallel_finder.h b/lib_vector_search/include/parallel_finder.h
index 1e8b527..55c776b 100644
--- a/lib_vector_search/include/parallel_finder.h
+++ b/lib_vector_search/include/parallel_finder.h
@@ -1,15 +1,14 @@
 #pragma once
 
 #include "finder.h"
-
-#include <vector>
+#include "word_list.h"
 
 class ParallelFinder : public Finder {
 private:
-  const std::vector<std::string> &word_list_;
+  const WordList &word_list_;
 
 public:
-  ParallelFinder(const std::vector<std::string> &word_list);
+  ParallelFinder(const WordList &word_list);
 
   std::forward_list<const std::string *>
   find_prefix(std::string_view search_term) const override;
diff --git a/lib_vector_search/include/sorted_linear_finder.h b/lib_vector_search/include/sorted_linear_finder.h
index f49c27e..866c2b2 100644
--- a/lib_vector_search/include/sorted_linear_finder.h
+++ b/lib_vector_search/include/sorted_linear_finder.h
@@ -1,6 +1,7 @@
 #pragma once
 
 #include "finder.h"
+#include "word_list.h"
 
 #include <vector>
 
@@ -9,7 +10,7 @@ private:
   std::vector<const std::string *> word_list_;
 
 public:
-  SortedLinearFinder(const std::vector<std::string> &word_list);
+  SortedLinearFinder(const WordList &word_list);
 
   std::forward_list<const std::string *>
   find_prefix(std::string_view search_term) const override;
diff --git a/lib_vector_search/include/tree_finder.h b/lib_vector_search/include/tree_finder.h
index ffac545..707a7b2 100644
--- a/lib_vector_search/include/tree_finder.h
+++ b/lib_vector_search/include/tree_finder.h
@@ -1,9 +1,9 @@
 #pragma once
 
 #include "finder.h"
+#include "word_list.h"
 
 #include <map>
-#include <vector>
 
 class SearchTreeNode {
 private:
@@ -19,7 +19,7 @@ public:
 
 class SearchTree : public SearchTreeNode {
 public:
-  SearchTree(const std::vector<std::string> &word_list);
+  SearchTree(const WordList &word_list);
 };
 
 class TreeFinder : public Finder {
@@ -27,7 +27,7 @@ private:
   SearchTree search_tree_;
 
 public:
-  TreeFinder(const std::vector<std::string> &word_list);
+  TreeFinder(const WordList &word_list);
 
   virtual std::forward_list<const std::string *>
   find_prefix(std::string_view search_term) const override;
diff --git a/lib_vector_search/include/word_list.h b/lib_vector_search/include/word_list.h
new file mode 100644
index 0000000..ca52d9a
--- /dev/null
+++ b/lib_vector_search/include/word_list.h
@@ -0,0 +1,14 @@
+#pragma once
+
+#include <filesystem>
+#include <string>
+#include <vector>
+
+class WordList : public std::vector<std::string> {
+public:
+  WordList &multiply(size_t factor);
+  WordList &shuffle();
+
+  static WordList fourCaps();
+  static WordList fromFile(const std::filesystem::path &path);
+};
diff --git a/lib_vector_search/include/word_list_generator.h b/lib_vector_search/include/word_list_generator.h
deleted file mode 100644
index 605b749..0000000
--- a/lib_vector_search/include/word_list_generator.h
+++ /dev/null
@@ -1,13 +0,0 @@
-#pragma once
-
-#include <string>
-#include <vector>
-
-class WordListGenerator {
-private:
-  static const std::string charset_;
-
-public:
-  WordListGenerator() = delete;
-  static std::vector<std::string> generate(const size_t multiplier = 1);
-};
diff --git a/lib_vector_search/src/grouped_finder.cpp b/lib_vector_search/src/grouped_finder.cpp
index 20e4d67..39b8b43 100644
--- a/lib_vector_search/src/grouped_finder.cpp
+++ b/lib_vector_search/src/grouped_finder.cpp
@@ -8,7 +8,7 @@
 using std::mutex, std::vector, std::thread, std::lock_guard, std::string,
     std::forward_list, std::string_view;
 
-GroupedFinder::GroupedFinder(const std::vector<string> &word_list) {
+GroupedFinder::GroupedFinder(const WordList &word_list) {
   for (const auto &word : word_list) {
     groups_[word.front()].push_back(&word);
   }
diff --git a/lib_vector_search/src/linear_finder.cpp b/lib_vector_search/src/linear_finder.cpp
index 45525ac..f2d910e 100644
--- a/lib_vector_search/src/linear_finder.cpp
+++ b/lib_vector_search/src/linear_finder.cpp
@@ -4,8 +4,7 @@
 
 using std::string, std::forward_list, std::string_view;
 
-LinearFinder::LinearFinder(const vector<string> &word_list)
-    : word_list_(word_list) {}
+LinearFinder::LinearFinder(const WordList &word_list) : word_list_(word_list) {}
 
 forward_list<const string *>
 LinearFinder::find_prefix(string_view search_term) const {
diff --git a/lib_vector_search/src/parallel_finder.cpp b/lib_vector_search/src/parallel_finder.cpp
index 441cadb..1b7effe 100644
--- a/lib_vector_search/src/parallel_finder.cpp
+++ b/lib_vector_search/src/parallel_finder.cpp
@@ -6,7 +6,7 @@
 using std::mutex, std::thread, std::lock_guard, std::vector, std::forward_list,
     std::string, std::string_view;
 
-ParallelFinder::ParallelFinder(const vector<string> &word_list)
+ParallelFinder::ParallelFinder(const WordList &word_list)
     : word_list_(word_list) {}
 
 forward_list<const string *>
@@ -28,7 +28,7 @@ ParallelFinder::find_prefix(string_view search_term) const {
             : (thread_index + 1) * (word_list_size / thread_count);
 
     threads.emplace_back(
-        [](const vector<string> &word_list, const string_view &search_term,
+        [](const WordList &word_list, const string_view &search_term,
            forward_list<const string *> &result, size_t start_index,
            size_t end_index, mutex &result_mutex) {
           forward_list<const string *> thread_results;
diff --git a/lib_vector_search/src/sorted_linear_finder.cpp b/lib_vector_search/src/sorted_linear_finder.cpp
index e904d6f..db11154 100644
--- a/lib_vector_search/src/sorted_linear_finder.cpp
+++ b/lib_vector_search/src/sorted_linear_finder.cpp
@@ -3,9 +3,9 @@
 #include <algorithm>
 #include <iterator>
 
-using std::vector, std::forward_list, std::string, std::string_view;
+using std::forward_list, std::string, std::string_view;
 
-SortedLinearFinder::SortedLinearFinder(const vector<string> &word_list) {
+SortedLinearFinder::SortedLinearFinder(const WordList &word_list) {
   std::transform(word_list.cbegin(), word_list.cend(),
                  std::back_inserter(word_list_),
                  [](const string &word) { return &word; });
diff --git a/lib_vector_search/src/tree_finder.cpp b/lib_vector_search/src/tree_finder.cpp
index 2367d98..611738e 100644
--- a/lib_vector_search/src/tree_finder.cpp
+++ b/lib_vector_search/src/tree_finder.cpp
@@ -1,6 +1,6 @@
 #include "tree_finder.h"
 
-using std::string, std::string_view, std::forward_list, std::vector;
+using std::string, std::string_view, std::forward_list;
 
 void SearchTreeNode::insert(string_view partial_word,
                             const string *original_word) {
@@ -35,14 +35,13 @@ forward_list<const string *> SearchTreeNode::words() const {
   return results;
 };
 
-SearchTree::SearchTree(const vector<string> &word_list) {
+SearchTree::SearchTree(const WordList &word_list) {
   for (const auto &word : word_list) {
     insert(string_view(word), &word);
   }
 }
 
-TreeFinder::TreeFinder(const vector<string> &word_list)
-    : search_tree_(word_list) {}
+TreeFinder::TreeFinder(const WordList &word_list) : search_tree_(word_list) {}
 
 forward_list<const string *>
 TreeFinder::find_prefix(string_view search_term) const {
diff --git a/lib_vector_search/src/word_list.cpp b/lib_vector_search/src/word_list.cpp
new file mode 100644
index 0000000..db01a19
--- /dev/null
+++ b/lib_vector_search/src/word_list.cpp
@@ -0,0 +1,59 @@
+#include "word_list.h"
+
+#include <algorithm>
+#include <fstream>
+#include <random>
+
+WordList &WordList::multiply(size_t factor) {
+  WordList result;
+  result.reserve(size() * factor);
+
+  for (auto time = 0; time < factor; ++time) {
+    std::copy(cbegin(), cend(), std::back_inserter(result));
+  }
+
+  std::swap(*this, result);
+
+  return *this;
+};
+
+WordList &WordList::shuffle() {
+  std::random_device random_device;
+  std::mt19937 random_number_generator(random_device());
+  std::shuffle(begin(), end(), random_number_generator);
+
+  return *this;
+}
+
+WordList WordList::fourCaps() {
+  const static std::string charset_ = "ABCDEFGHIJKLMNOPQRSTUVWXYZ";
+
+  WordList word_list;
+  word_list.reserve(std::pow(charset_.length(), 4));
+
+  for (auto char_1 : charset_) {
+    for (auto char_2 : charset_) {
+      for (auto char_3 : charset_) {
+        for (auto char_4 : charset_) {
+          word_list.emplace_back(
+              std::initializer_list<char>({char_1, char_2, char_3, char_4}));
+        }
+      }
+    }
+  }
+
+  return word_list;
+};
+
+WordList WordList::fromFile(const std::filesystem::path &path) {
+  WordList word_list;
+
+  std::ifstream ifs(path);
+
+  std::string line;
+  while (std::getline(ifs, line)) {
+    word_list.emplace_back(line);
+  }
+
+  return word_list;
+}
diff --git a/lib_vector_search/src/word_list_generator.cpp b/lib_vector_search/src/word_list_generator.cpp
deleted file mode 100644
index ba4cf06..0000000
--- a/lib_vector_search/src/word_list_generator.cpp
+++ /dev/null
@@ -1,32 +0,0 @@
-#include "word_list_generator.h"
-
-#include <algorithm>
-#include <cmath>
-#include <random>
-
-const std::string WordListGenerator::charset_ = "ABCDEFGHIJKLMNOPQRSTUVWXYZ";
-
-std::vector<std::string> WordListGenerator::generate(const size_t multiplier) {
-  std::vector<std::string> word_list;
-  word_list.reserve(multiplier * std::pow(charset_.length(), 4));
-
-  for (auto char_1 : charset_) {
-    for (auto char_2 : charset_) {
-      for (auto char_3 : charset_) {
-        for (auto char_4 : charset_) {
-          for (auto i = 0; i < multiplier; ++i) {
-            word_list.emplace_back(
-                std::initializer_list<char>({char_1, char_2, char_3, char_4}));
-          }
-        }
-      }
-    }
-  }
-
-  std::random_device random_device;
-  std::mt19937 random_number_generator(random_device());
-
-  std::shuffle(word_list.begin(), word_list.end(), random_number_generator);
-
-  return word_list;
-}
diff --git a/vector_search_cli/main.cpp b/vector_search_cli/main.cpp
index e915f2e..9922090 100644
--- a/vector_search_cli/main.cpp
+++ b/vector_search_cli/main.cpp
@@ -4,18 +4,18 @@
 #include "sorted_linear_finder.h"
 #include "timer.h"
 #include "tree_finder.h"
-#include "word_list_generator.h"
+#include "word_list.h"
 
 #include <cstdlib>
 #include <iostream>
 
-using std::string, std::string_view, std::vector, std::cout, std::endl;
+using std::string, std::string_view, std::cout, std::endl;
 
-vector<string> generate_word_list(size_t size_factor = 1) {
+WordList generate_word_list(size_t size_factor = 1) {
   cout << endl << "generating word list (" << size_factor << "x)" << endl;
 
   Timer generator_timer;
-  auto word_list = WordListGenerator::generate(size_factor);
+  auto word_list = WordList::fourCaps().multiply(size_factor).shuffle();
   generator_timer.stop();
 
   cout << "word list generator took " << generator_timer << endl;
@@ -36,7 +36,7 @@ void test_finder_search(Finder &finder, string_view name,
 }
 
 template <typename FINDER>
-void test_finder(const vector<string> &word_list, string_view finder_name) {
+void test_finder(const WordList &word_list, string_view finder_name) {
   cout << endl << "running " << finder_name << endl;
 
   Timer constructor_timer;