Initial commit

Author: andronov-alexey

.gitignore

@@ -0,0 +1,34 @@
+# Prerequisites
+*.d
+
+# Compiled Object files
+*.slo
+*.lo
+*.o
+*.obj
+
+# Precompiled Headers
+*.gch
+*.pch
+
+# Compiled Dynamic libraries
+*.so
+*.dylib
+*.dll
+
+# Fortran module files
+*.mod
+*.smod
+
+# Compiled Static libraries
+*.lai
+*.la
+*.a
+*.lib
+
+# Executables
+*.exe
+*.out
+*.app
+
+!bin

CMakeLists.txt

@@ -0,0 +1,12 @@
+project(WordCountSorterProject)
+
+cmake_minimum_required(VERSION 3.10)
+
+FILE(GLOB DOC_FILES "README.md" "docs/*")
+add_custom_target(docs SOURCES ${DOC_FILES})
+
+add_subdirectory(src)
+
+if (MSVC)
+	set_property(DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR} PROPERTY VS_STARTUP_PROJECT WordCountSorter)
+endif()

docs/Problem statement.txt

@@ -0,0 +1,25 @@
+Напишите на C++ нехитрую считалку частотного словаря.
+Важно: напишите примерно так, как сегодня пишете код для боя, по-взрослому (часть особо редких проверок в стиле “а не кончился ли диск” можно и нужно не писать; но все проверки не писать нельзя). Важно: напишите и/или упакуйте вашу C++ программу так, чтобы мы ее тоже сумели собрать и запустить. (Например, на боевых серверах у нас сегодня Debian 9.) Важно: не надо копипастить простыню кода в docx/pdf документ, человечество недавно таки придумало аттачи.
+
+Словом считается набор латинских букв, a-zA-Z.
+Любой другой символ считается пробелом. Регистр нужно привести.
+Сортировать нужно сначала по частоте, потом по алфавиту.
+Выводить тоже сначала частоту, потом слово.
+
+Вот пример.
+$ cat in.txt
+The time has come, the Walrus said,
+to talk of many things...
+$ freq in.txt out.txt
+$ cat out.txt
+  2 the
+  1 come
+  1 has
+  1 many
+  1 of
+  1 said
+  1 talk
+  1 things
+  1 time
+  1 to
+  1 walrus

src/CMakeLists.txt

@@ -0,0 +1,15 @@
+project(WordCountSorter)
+
+cmake_minimum_required(VERSION 3.10)
+
+set(SOURCES
+	main.cpp
+)
+
+add_executable(${PROJECT_NAME}
+	${SOURCES}
+)
+
+set_property(TARGET ${PROJECT_NAME} PROPERTY OUTPUT_NAME "freq")
+set_property(TARGET ${PROJECT_NAME} PROPERTY CXX_STANDARD 17)
+set_property(TARGET ${PROJECT_NAME} PROPERTY CXX_STANDARD_REQUIRED ON)

src/main.cpp

@@ -0,0 +1,120 @@
+#include <iostream>
+#include <fstream>
+#include <cctype>
+#include <cassert>
+#include <unordered_map>
+#include <vector>
+#include <algorithm>
+
+using namespace std;
+
+constexpr auto INPUT_ARGS_COUNT = 3;
+
+using WordToCountDictionary = unordered_map<string, unsigned int>;
+using CountToWordsDictionary = unordered_map<unsigned int, vector<string>>;
+using SortedCountToWordsDictionary = vector<pair<unsigned int, vector<string>*>>;
+
+WordToCountDictionary ReadWords(ifstream & inputStream)
+{
+	WordToCountDictionary wordToCountDictionary;
+
+	string wordBuf;
+	using buf_iter = std::istreambuf_iterator<char>;
+	const buf_iter eof;
+	for (buf_iter it = inputStream; it != eof; ++it)
+	{
+		const auto symbol = tolower(static_cast<unsigned char>(*it));
+		if (isalpha(symbol))
+		{
+			wordBuf += symbol;
+		}
+		else if (!wordBuf.empty())
+		{
+			++wordToCountDictionary[wordBuf];
+			wordBuf.clear();
+		}
+	}
+
+	return wordToCountDictionary;
+}
+
+CountToWordsDictionary GetCountToWordsDictionary(const WordToCountDictionary & wordToCountDictionary)
+{
+	CountToWordsDictionary countToWordsDictionary;
+	for (const auto & [word, count] : wordToCountDictionary)
+	{
+		countToWordsDictionary[count].push_back(word);
+	}
+
+	return countToWordsDictionary;
+}
+
+SortedCountToWordsDictionary SortWordsByCount(CountToWordsDictionary & countToWordsDictionary)
+{
+	SortedCountToWordsDictionary sortedCountToWordsDictionary;
+	sortedCountToWordsDictionary.reserve(countToWordsDictionary.size());
+
+	for (auto & [count, words] : countToWordsDictionary)
+	{
+		sort(begin(words), end(words));
+		sortedCountToWordsDictionary.push_back({ count, &words });
+	}
+
+	sort(begin(sortedCountToWordsDictionary), end(sortedCountToWordsDictionary),
+		[] (const auto & left, const auto & right)
+		{
+			return left.first > right.first;
+		});
+
+	return sortedCountToWordsDictionary;
+}
+
+void PrintResults(ofstream & outputStream, const SortedCountToWordsDictionary & sortedCountToWords)
+{
+	for (const auto & [count, pWords] : sortedCountToWords)
+	{
+		assert(pWords != nullptr);
+		for (const auto & word : *pWords)
+		{
+			outputStream << count << " " << word << endl;
+		}
+	}
+}
+
+int main(int _argc, char * _argv[])
+{
+	if (_argc != INPUT_ARGS_COUNT)
+	{
+		cerr << "Usage: " << _argv[0] << " <inputfile> <outputfile>" << endl;
+		return 1;
+	}
+
+	const string inputFileName(_argv[1]);
+	ifstream inputStream{ inputFileName };
+	if (!inputStream)
+	{
+		cerr << "Cannot open input file: " << inputFileName << endl;
+		return 2;
+	}
+
+	const string outputFileName(_argv[2]);
+	ofstream outputStream{ outputFileName };
+	if (!outputStream)
+	{
+		cerr << "Cannot create output file: " << outputFileName << endl;
+		return 3;
+	}
+
+	const auto wordToCountDictionary = ReadWords(inputStream);
+	auto countToWordsDictionary = GetCountToWordsDictionary(wordToCountDictionary);
+	const auto sortedCountToWordsDictionary = SortWordsByCount(countToWordsDictionary);
+
+	PrintResults(outputStream, sortedCountToWordsDictionary);
+
+	inputStream.close();
+	outputStream.close();
+
+	cout << "Results were successfully written to file: \"" << outputFileName << "\"" << endl;
+
+	return 0;
+}