From 72d52860f60a97c1d3e293e39592cc7c661e0ae7 Mon Sep 17 00:00:00 2001
From: King_DuckZ <king_duckz@posteo.net>
Date: Sat, 5 Apr 2025 19:59:21 +0100
Subject: [PATCH] Refactor visitors into separate files

This also includes some small improvements that I don't remember right now
---
 src/main.cpp                    | 222 ++++----------------------------
 src/meson.build                 |   2 +
 src/parser.cpp                  |  34 +++++
 src/parser.hpp                  |   8 +-
 src/visitors/debug_visitor.cpp  |  88 +++++++++++++
 src/visitors/debug_visitor.hpp  |  25 ++++
 src/visitors/find_t_visitor.cpp |  91 +++++++++++++
 src/visitors/find_t_visitor.hpp |  55 ++++++++
 8 files changed, 323 insertions(+), 202 deletions(-)
 create mode 100644 src/visitors/debug_visitor.cpp
 create mode 100644 src/visitors/debug_visitor.hpp
 create mode 100644 src/visitors/find_t_visitor.cpp
 create mode 100644 src/visitors/find_t_visitor.hpp

diff --git a/src/main.cpp b/src/main.cpp
index 9b8cfec..74967fc 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -1,18 +1,12 @@
 #include "parser.hpp"
-#include "split.hpp"
+#include "visitors/debug_visitor.hpp"
+#include "visitors/find_t_visitor.hpp"
 
 #include <iostream>
 #include <fstream>
 #include <string>
-#include <cctype>
 #include <algorithm>
-#include <ciso646>
 #include <cstdint>
-#include <array>
-#include <stdexcept>
-#include <type_traits>
-#include <climits>
-#include <boost/variant/apply_visitor.hpp>
 #include <sha1/sha1.hpp>
 #include <sha1/BufferSource.hpp>
 
@@ -27,186 +21,6 @@ std::string load_file (const std::string& path) {
 	);
 }
 
-bool is_printable(std::string_view s) {
-    return std::find_if(s.cbegin(), s.cend(),
-		 [](unsigned char c){ return not std::isprint(c); }
-	) == s.cend();
-}
-
-struct PrintVisitor {
-	PrintVisitor(int depth=0, bool skip_first=true) :
-		depth(depth), first_line(skip_first), skip_first(skip_first) {}
-
-	void print_tabs(int extra=0) {
-		if (!first_line) {
-			for (int z = 0; z < depth+extra; ++z) std::cout << '\t';
-		}
-		else {
-			first_line = false;
-		}
-	}
-
-	void reset() {
-		first_line = skip_first;
-	}
-
-	void operator() (long long value) {
-		print_tabs();
-		std::cout << value << '\n';
-	}
-
-	void operator() (const duck::TorrentStringType& value) {
-		constexpr std::size_t max_size = 41;
-
-		print_tabs();
-
-		std::string_view str{value};
-		std::string_view ellip;
-		if (is_printable(value)) {
-			if (value.size() > max_size + 3) {
-				str = str.substr(0, max_size);
-				ellip = "...";
-			}
-		}
-		else {
-			str = "<binary_data>";
-		}
-
-		std::cout << "\"" << str << ellip << "\"[" << value.size() << "]\n";
-	}
-
-	void operator() (const std::vector<duck::TorrentValue>& value) {
-		print_tabs();
-		std::cout << "[\n";
-
-		PrintVisitor new_visitor(depth+1, false);
-		for (const auto& item : value) {
-			boost::apply_visitor(new_visitor, item);
-		}
-
-		print_tabs();
-		std::cout << "]\n";
-	}
-
-	void operator() (const std::map<duck::TorrentStringType, duck::TorrentValue>& value) {
-		print_tabs();
-		std::cout << "{\n";
-
-		PrintVisitor new_visitor(depth+1);
-		for (const auto& item : value) {
-			print_tabs(1);
-			std::cout << item.first << " => ";
-
-			new_visitor.reset();
-			boost::apply_visitor(new_visitor, item.second);
-		}
-		print_tabs();
-		std::cout << "}\n";
-	}
-
-	int depth;
-	bool first_line;
-	bool skip_first;
-};
-
-template <typename T>
-struct FindTVisitor : boost::static_visitor<T> {
-	FindTVisitor (std::string_view search_path) {
-		auto split = dincore::split(search_path, '/', true, true);
-		search.reserve(split.size());
-		for (const auto& itm : split) {
-			search.emplace_back(itm.data(), itm.size());
-		}
-	}
-
-	void reset() {
-		search_error = false;
-		search_match_count = 0;
-	}
-
-	bool is_path_found() const {
-		return search.size() == search_match_count;
-	}
-
-	T operator() (long long value) {
-		if constexpr (std::is_same<long long, T>::value)
-			return value;
-		search_error = true;
-		return {};
-	}
-
-	T operator() (const duck::TorrentStringType& value) {
-		if constexpr (std::is_same<duck::TorrentStringType, T>::value) {
-			if (is_path_found())
-				return value;
-		}
-		search_error = true;
-		return {};
-	}
-
-	T operator() (const std::vector<duck::TorrentValue>&) {
-		search_error = true;
-		return {};
-	}
-
-	T operator() (const std::map<duck::TorrentStringType, duck::TorrentValue>& value) {
-		if (is_path_found()) {
-			search_error = true;
-			return {};
-		}
-
-		auto it_found = value.find(search[search_match_count]);
-		if (value.end() == it_found)
-			return {};
-
-		++search_match_count;
-		return boost::apply_visitor(*this, it_found->second);
-	}
-
-	std::vector<std::string> search;
-	std::size_t search_match_count{0};
-	bool search_error{false};
-};
-
-std::vector<std::array<std::uint32_t, 5>> collect_hashes (std::string_view hashes) {
-	typedef std::array<std::uint32_t, 5> array_type;
-
-	if (hashes.empty())
-		return {};
-
-	constexpr std::size_t hash_size = 20;
-	constexpr std::size_t uint_count = std::tuple_size<array_type>();
-	static_assert(sizeof(array_type::value_type) * uint_count == hash_size);
-
-	if (hashes.size() % hash_size != 0)
-		throw std::runtime_error("Bad pieces array size " + std::to_string(hashes.size()));
-
-	std::vector<array_type> retval(hashes.size() / hash_size);
-	const char* src = hashes.data();
-	for (std::size_t z = 0; z < retval.size(); ++z, src+=hash_size) {
-		//std::copy_n(src, hash_size, reinterpret_cast<char*>(retval[z].data()));
-		for (std::size_t n = 0; n < uint_count; ++n) {
-			char* const out_uint = reinterpret_cast<char*>(retval[z].data() + n);
-			for (std::size_t u = 0; u < sizeof(std::uint32_t); ++u) {
-				out_uint[u] = src[n * sizeof(std::uint32_t) + sizeof(std::uint32_t) - u - 1];
-			}
-		}
-	}
-	return retval;
-}
-
-template <typename T>
-T find_item(std::string_view path, const std::vector<duck::TorrentValue>& values) {
-	FindTVisitor<T> visitor(path);
-	for (const auto& value : values) {
-		visitor.reset();
-		T found = boost::apply_visitor(visitor, value);
-		if (not visitor.search_error)
-			return found;
-	}
-	return {};
-}
-
 SHA1::MessageDigest to_hash160_digest (const std::array<std::uint32_t, 5>& arr) {
 	SHA1::MessageDigest hash_found;
 	std::copy_n(arr.begin(), 5, hash_found.hash);
@@ -215,9 +29,6 @@ SHA1::MessageDigest to_hash160_digest (const std::array<std::uint32_t, 5>& arr)
 } //unnamed namespace
 
 int main(int argc, const char* argv[]) {
-	using duck::parse_torrent;
-	typedef duck::TorrentStringType torrent_string;
-
 	if (argc != 2) {
 		std::cerr << "Wrong number of parameters. Usage:\n"
 			<< argv[0] << " <path_to_torrent>\n";
@@ -226,23 +37,27 @@ int main(int argc, const char* argv[]) {
 
 	std::string full_torrent = load_file(argv[1]);
 	std::cout << "Loaded file into string of size " << full_torrent.size() << '\n';
-	std::vector<duck::TorrentValue> values = parse_torrent(full_torrent);
+	std::vector<duck::TorrentValue> values = duck::parse_torrent(full_torrent);
 
 	for (const auto& value : values) {
-		PrintVisitor visitor;
+		duck::DebugVisitor visitor;
 		boost::apply_visitor(visitor, value);
 	}
 
-	auto source_hashes = find_item<torrent_string>("/info/pieces", values);
-	std::cout << "Got source_hashes with size " << source_hashes.size() << '\n';
-	auto hashes = collect_hashes(source_hashes);
+	auto hashes = duck::collect_hashes(values);
 	std::cout << "Got " << hashes.size() << " hashes\n";
 
-	std::string file_name = std::string{find_item<torrent_string>("/info/name", values)};
-	std::cout << "Found file name \"" << file_name << "\"\n";
+	const auto file_count = duck::find_int("/info/files/[[size]]", values);
+	if (file_count)
+		std::cout << "Input has " << file_count << " file entries\n";
+	else
+		std::cout << "Input seems to contain a single file only\n";
 
-	{
-		const auto piece_length = find_item<long long int>("/info/piece length", values);
+	if (0 == file_count) {
+		std::string file_name = std::string{duck::find_string("/info/name", values)};
+		std::cout << "Found file name \"" << file_name << "\"\n";
+
+		const auto piece_length = duck::find_int("/info/piece length", values);
 		std::ifstream istream(file_name, std::ios::in|std::ios::binary);
 
 		if (istream.is_open()) {
@@ -267,4 +82,11 @@ int main(int argc, const char* argv[]) {
 			std::cout << "Error: opening \"" << file_name << "\" failed\n";
 		}
 	}
+	else {
+		std::string search("/info/files/[[");
+		for (std::int_fast32_t z = 0; z < static_cast<std::int_fast32_t>(file_count); ++z) {
+			const auto path = duck::find_string(search + std::to_string(z) + "]]/path/[[0]]", values);
+			std::cout << "path index " << z << '\t' << path << '\n';
+		}
+	}
 }
diff --git a/src/meson.build b/src/meson.build
index 65789b9..cc19b39 100644
--- a/src/meson.build
+++ b/src/meson.build
@@ -2,6 +2,8 @@ executable(meson.project_name(),
   'main.cpp',
   'parser.cpp',
   'split.cpp',
+  'visitors/debug_visitor.cpp',
+  'visitors/find_t_visitor.cpp',
   dependencies: [
 	boost_dep,
     libstriezel_dep,
diff --git a/src/parser.cpp b/src/parser.cpp
index 65443b1..aca2c40 100644
--- a/src/parser.cpp
+++ b/src/parser.cpp
@@ -3,12 +3,15 @@
 #endif
 
 #include "parser.hpp"
+#include "visitors/find_t_visitor.hpp"
 
 #include <boost/spirit/home/x3.hpp>
 #include <boost/fusion/adapted/std_pair.hpp>
 #if !defined(NDEBUG)
 #	include <iostream>
 #endif
+#include <stdexcept>
+#include <cstddef>
 
 namespace boost::spirit::x3::traits {
 template <typename Char, typename Trait>
@@ -128,4 +131,35 @@ std::vector<TorrentValue> parse_torrent (std::string_view binary_data) {
 
 	return retval;
 }
+
+std::vector<std::array<std::uint32_t, 5>> collect_hashes (std::string_view hashes) {
+	typedef std::array<std::uint32_t, 5> array_type;
+
+	if (hashes.empty())
+		return {};
+
+	constexpr std::size_t hash_size = 20;
+	constexpr std::size_t uint_count = std::tuple_size<array_type>();
+	static_assert(sizeof(array_type::value_type) * uint_count == hash_size);
+
+	if (hashes.size() % hash_size != 0)
+		throw std::runtime_error("Bad pieces array size " + std::to_string(hashes.size()));
+
+	std::vector<array_type> retval(hashes.size() / hash_size);
+	const char* src = hashes.data();
+	for (std::size_t z = 0; z < retval.size(); ++z, src+=hash_size) {
+		//std::copy_n(src, hash_size, reinterpret_cast<char*>(retval[z].data()));
+		for (std::size_t n = 0; n < uint_count; ++n) {
+			char* const out_uint = reinterpret_cast<char*>(retval[z].data() + n);
+			for (std::size_t u = 0; u < sizeof(std::uint32_t); ++u) {
+				out_uint[u] = src[n * sizeof(std::uint32_t) + sizeof(std::uint32_t) - u - 1];
+			}
+		}
+	}
+	return retval;
+}
+
+std::vector<std::array<std::uint32_t, 5>> collect_hashes (const std::vector<TorrentValue>& values) {
+	return collect_hashes(find_string("/info/pieces", values));
+}
 } //namespace duck
diff --git a/src/parser.hpp b/src/parser.hpp
index 5bfd91c..c3eb10b 100644
--- a/src/parser.hpp
+++ b/src/parser.hpp
@@ -4,12 +4,15 @@
 #include <boost/spirit/home/x3/support/ast/variant.hpp>
 #include <map>
 #include <vector>
+#include <array>
+#include <cstdint>
 
 namespace duck {
 typedef std::string_view TorrentStringType;
+typedef long long int TorrentIntType;
 
 struct TorrentValue : public boost::spirit::x3::variant<
-	long long,
+	TorrentIntType,
 	TorrentStringType,
 	boost::spirit::x3::forward_ast< std::vector<TorrentValue> >,
 	boost::spirit::x3::forward_ast< std::map<TorrentStringType, TorrentValue> >
@@ -19,5 +22,6 @@ struct TorrentValue : public boost::spirit::x3::variant<
 };
 
 std::vector<TorrentValue> parse_torrent (std::string_view binary_data);
-
+std::vector<std::array<std::uint32_t, 5>> collect_hashes (std::string_view hashes);
+std::vector<std::array<std::uint32_t, 5>> collect_hashes (const std::vector<TorrentValue>& values);
 } //namespace duck
diff --git a/src/visitors/debug_visitor.cpp b/src/visitors/debug_visitor.cpp
new file mode 100644
index 0000000..6d03fc2
--- /dev/null
+++ b/src/visitors/debug_visitor.cpp
@@ -0,0 +1,88 @@
+#include "debug_visitor.hpp"
+
+#include <iostream>
+#include <ciso646>
+#include <boost/variant/apply_visitor.hpp>
+#include <algorithm>
+#include <cctype>
+
+namespace duck {
+namespace {
+bool is_printable(std::string_view s) {
+    return std::find_if(s.cbegin(), s.cend(),
+		 [](unsigned char c){ return not std::isprint(c); }
+	) == s.cend();
+}
+
+} //unnamed namespace
+
+DebugVisitor::DebugVisitor(int depth, bool skip_first) :
+	m_depth(depth), m_first_line(skip_first), m_skip_first(skip_first) {}
+
+void DebugVisitor::reset() {
+	m_first_line = m_skip_first;
+}
+
+void DebugVisitor::operator() (TorrentIntType value) {
+	print_tabs();
+	std::cout << value << '\n';
+}
+
+void DebugVisitor::operator() (const TorrentStringType& value) {
+	constexpr std::size_t max_size = 41;
+
+	print_tabs();
+
+	std::string_view str{value};
+	std::string_view ellip;
+	if (is_printable(value)) {
+		if (value.size() > max_size + 3) {
+			str = str.substr(0, max_size);
+			ellip = "...";
+		}
+	}
+	else {
+		str = "<binary_data>";
+	}
+
+	std::cout << "\"" << str << ellip << "\"[" << value.size() << "]\n";
+}
+
+void DebugVisitor::operator() (const std::vector<TorrentValue>& value) {
+	print_tabs();
+	std::cout << "[\n";
+
+	DebugVisitor new_visitor(m_depth+1, false);
+	for (const auto& item : value) {
+		boost::apply_visitor(new_visitor, item);
+	}
+
+	print_tabs();
+	std::cout << "]\n";
+}
+
+void DebugVisitor::operator() (const std::map<TorrentStringType, TorrentValue>& value) {
+	print_tabs();
+	std::cout << "{\n";
+
+	DebugVisitor new_visitor(m_depth+1);
+	for (const auto& item : value) {
+		print_tabs(1);
+		std::cout << item.first << " => ";
+
+		new_visitor.reset();
+		boost::apply_visitor(new_visitor, item.second);
+	}
+	print_tabs();
+	std::cout << "}\n";
+}
+
+void DebugVisitor::print_tabs(int extra) {
+	if (!m_first_line) {
+		for (int z = 0; z < m_depth+extra; ++z) std::cout << '\t';
+	}
+	else {
+		m_first_line = false;
+	}
+}
+} //namespace duck
diff --git a/src/visitors/debug_visitor.hpp b/src/visitors/debug_visitor.hpp
new file mode 100644
index 0000000..2ce049e
--- /dev/null
+++ b/src/visitors/debug_visitor.hpp
@@ -0,0 +1,25 @@
+#pragma once
+
+#include "../parser.hpp"
+
+#include <cstddef>
+
+namespace duck {
+class DebugVisitor {
+public:
+	DebugVisitor(int depth=0, bool skip_first=true);
+
+	void reset();
+	void operator() (TorrentIntType value);
+	void operator() (const TorrentStringType& value);
+	void operator() (const std::vector<duck::TorrentValue>& value);
+	void operator() (const std::map<duck::TorrentStringType, duck::TorrentValue>& value);
+
+private:
+	void print_tabs(int extra=0);
+
+	int m_depth;
+	bool m_first_line;
+	bool m_skip_first;
+};
+} //namespace duck
diff --git a/src/visitors/find_t_visitor.cpp b/src/visitors/find_t_visitor.cpp
new file mode 100644
index 0000000..e16ae97
--- /dev/null
+++ b/src/visitors/find_t_visitor.cpp
@@ -0,0 +1,91 @@
+#include "find_t_visitor.hpp"
+#include "split.hpp"
+
+#include <type_traits>
+#include <string>
+#include <regex>
+
+namespace duck::detail {
+template <typename T>
+FindTVisitor<T>::FindTVisitor (std::string_view search_path) {
+	auto split = dincore::split(search_path, '/', true, true);
+	m_search.reserve(split.size());
+	for (const auto& itm : split) {
+		m_search.emplace_back(itm.data(), itm.size());
+	}
+}
+
+template <typename T>
+void FindTVisitor<T>::reset() {
+	m_search_error = false;
+	m_search_match_count = 0;
+}
+
+template <typename T>
+bool FindTVisitor<T>::has_error() const noexcept {
+	return m_search_error;
+}
+
+template <typename T>
+T FindTVisitor<T>::operator() (TorrentIntType value) {
+	if constexpr (std::is_same<TorrentIntType, T>::value)
+		return value;
+	m_search_error = true;
+	return {};
+}
+
+template <typename T>
+T FindTVisitor<T>::operator() (const TorrentStringType& value) {
+	if constexpr (std::is_same<TorrentStringType, T>::value) {
+		if (is_path_found())
+			return value;
+	}
+	m_search_error = true;
+	return {};
+}
+
+template <typename T>
+T FindTVisitor<T>::operator() (const std::vector<TorrentValue>& values) {
+	if (is_path_found())
+		return {};
+
+	if constexpr (std::is_same<TorrentIntType, T>::value) {
+		if (m_search[m_search_match_count] == "[[size]]")
+			return static_cast<TorrentIntType>(values.size());
+	}
+
+	std::regex reg_index(R"(^\[\[(\d+)\]\]$)");
+	std::smatch match;
+	const bool matched = std::regex_match(m_search[m_search_match_count], match, reg_index);
+	if (matched) {
+		++m_search_match_count;
+		const auto index = static_cast<std::size_t>(std::stoul(match[1]));
+		if (index < values.size())
+			return boost::apply_visitor(*this, values[index]);
+	}
+	return {};
+}
+
+template <typename T>
+T FindTVisitor<T>::operator() (const std::map<TorrentStringType, TorrentValue>& value) {
+	if (is_path_found()) {
+		m_search_error = true;
+		return {};
+	}
+
+	auto it_found = value.find(m_search[m_search_match_count]);
+	if (value.end() == it_found)
+		return {};
+
+	++m_search_match_count;
+	return boost::apply_visitor(*this, it_found->second);
+}
+
+template <typename T>
+bool FindTVisitor<T>::is_path_found() const {
+	return m_search.size() == m_search_match_count;
+}
+
+template class FindTVisitor<TorrentStringType>;
+template class FindTVisitor<TorrentIntType>;
+} //namespace duck::detail
diff --git a/src/visitors/find_t_visitor.hpp b/src/visitors/find_t_visitor.hpp
new file mode 100644
index 0000000..aa529aa
--- /dev/null
+++ b/src/visitors/find_t_visitor.hpp
@@ -0,0 +1,55 @@
+#pragma once
+
+#include "../parser.hpp"
+#include <string_view>
+#include <cstddef>
+#include <boost/variant/apply_visitor.hpp>
+#include <vector>
+#include <ciso646>
+
+namespace duck {
+namespace detail {
+template <typename T>
+class FindTVisitor : boost::static_visitor<T> {
+public:
+	FindTVisitor (std::string_view search_path);
+
+	void reset();
+	bool has_error() const noexcept;
+	T operator() (TorrentIntType value);
+	T operator() (const TorrentStringType& value);
+	T operator() (const std::vector<TorrentValue>&);
+	T operator() (const std::map<TorrentStringType, TorrentValue>& value);
+
+private:
+	bool is_path_found() const;
+
+	std::vector<std::string> m_search;
+	std::size_t m_search_match_count{0};
+	bool m_search_error{false};
+};
+} //namespace detail
+
+typedef detail::FindTVisitor<TorrentStringType> FindStringVisitor;
+typedef detail::FindTVisitor<TorrentIntType> FindIntVisitor;
+
+template <typename T>
+inline T find_item(std::string_view path, const std::vector<TorrentValue>& values) {
+	detail::FindTVisitor<T> visitor(path);
+	for (const auto& value : values) {
+		visitor.reset();
+		T found = boost::apply_visitor(visitor, value);
+		if (not visitor.has_error())
+			return found;
+	}
+	return {};
+}
+
+inline TorrentIntType find_int(std::string_view path, const std::vector<TorrentValue>& values) {
+	return find_item<TorrentIntType>(path, values);
+}
+
+inline TorrentStringType find_string(std::string_view path, const std::vector<TorrentValue>& values) {
+	return find_item<TorrentStringType>(path, values);
+}
+} //namespace duck