From b71a49cfbeacdf44fa442663e44f367938b4b319 Mon Sep 17 00:00:00 2001
From: Pcornat <florent.denef@protonmail.com>
Date: Sun, 21 Feb 2021 19:02:32 +0100
Subject: [PATCH] Caching of files is done : not tested yet.

---
 include/cache_files.hpp | 27 +++++++++++++++++++++++++-
 src/cache_files.cpp     | 43 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 69 insertions(+), 1 deletion(-)

diff --git a/include/cache_files.hpp b/include/cache_files.hpp
index 9aa1a0d..8034ef7 100644
--- a/include/cache_files.hpp
+++ b/include/cache_files.hpp
@@ -1,13 +1,38 @@
 #ifndef GEMINISERVER_CACHE_FILES_HPP
 #define GEMINISERVER_CACHE_FILES_HPP
 
+#include <string>
+#include <unordered_set>
+#include <unordered_map>
+#include <filesystem>
 
 namespace gemini {
+	struct Information;
+
 	/**
 	 * \brief This class is used to store the files in cache.
 	 */
-	class CacheFiles {
+	class CacheFiles final {
+	private:
+		std::unordered_set<std::string> files;
+		std::unordered_map<std::string, std::string> content;
 
+	public:
+		const Information &information;
+
+		CacheFiles() = delete;
+
+		CacheFiles(const std::filesystem::path &folder, const Information &infos);
+
+		~CacheFiles() noexcept = default;
+
+		[[nodiscard, gnu::always_inline]] inline const std::unordered_set<std::string> &get_files() const {
+			return files;
+		}
+
+		[[nodiscard, gnu::always_inline]] inline const std::unordered_map<std::string, std::string> &get_content() const {
+			return content;
+		}
 	};
 }
 
diff --git a/src/cache_files.cpp b/src/cache_files.cpp
index 9366e7e..06700ae 100644
--- a/src/cache_files.cpp
+++ b/src/cache_files.cpp
@@ -1 +1,44 @@
 #include "../include/cache_files.hpp"
+#include <spdlog/spdlog.h>
+#include "../include/information.hpp"
+#include <fstream>
+
+namespace fs = std::filesystem;
+
+gemini::CacheFiles::CacheFiles(const std::filesystem::path &folder, const Information &infos) : information(infos) {
+	if (!fs::is_directory(folder)) {
+		spdlog::error("Path {} is not a folder", folder.string());
+		throw std::runtime_error("Stopped in cache files constructor, see log.");
+	}
+	std::size_t global_cache = 0;
+	for (const auto &iterator : fs::directory_iterator(folder, fs::directory_options::follow_directory_symlink)) {
+		// Each time it finds a file, it is going to load it in memory and check if infos.cache_size is reached in terms of size.
+		if (!iterator.is_regular_file())
+			continue;
+		const std::size_t file_size = [&]() -> std::size_t {
+			std::ifstream in(iterator.path());
+			const std::size_t begin = in.tellg();
+			in.seekg(0, std::ios::end);
+			return static_cast<std::size_t>(in.tellg()) - begin;
+		}();
+		global_cache += file_size;
+
+		if (global_cache > infos.cache_size)
+			break;
+
+		// I want to have the filename and not the whole path to be stored (less characters, less memory used).
+		const std::string filename = [&]() -> std::string {
+			const std::string_view path = iterator.path().string();
+			const std::size_t pos = path.rfind('/');
+			return iterator.path().string().substr(pos != std::string::npos ? pos + 1 : pos);
+		}();
+
+		if (files.insert(filename.empty() ? iterator.path().string() : filename).second)
+			spdlog::warn("File {} could not be registered", iterator.path().string());
+
+		std::ifstream file{ iterator.path() };
+		if (!content.emplace(filename.empty() ? iterator.path().string() : filename,
+							 std::string{ std::istreambuf_iterator<char>(file), std::istreambuf_iterator<char>() }).second)
+			spdlog::warn("File {} could not be loaded in cache", iterator.path().string());
+	}
+}