rpcsx/rpcs3/Loader/TAR.cpp

426 lines
9.5 KiB
C++
Raw Normal View History

2020-12-05 13:08:24 +01:00
#include "stdafx.h"
#include "Emu/VFS.h"
#include "Emu/System.h"
#include "Crypto/unself.h"
#include "TAR.h"
#include "util/asm.hpp"
#include "util/serialization.hpp"
#include <charconv>
LOG_CHANNEL(tar_log, "TAR");
tar_object::tar_object(const fs::file& file)
2017-02-22 14:08:53 +01:00
: m_file(file)
{
}
TARHeader tar_object::read_header(u64 offset) const
{
TARHeader header{};
if (m_file.seek(offset) != offset)
{
return header;
}
if (!m_file.read(header))
{
std::memset(&header, 0, sizeof(header));
}
return header;
}
u64 octal_text_to_u64(std::string_view sv)
{
u64 i = -1;
const auto ptr = std::from_chars(sv.data(), sv.data() + sv.size(), i, 8).ptr;
// Range must be terminated with either NUL or space
if (ptr == sv.data() + sv.size() || (*ptr && *ptr != ' '))
{
i = -1;
}
return i;
}
std::vector<std::string> tar_object::get_filenames()
{
std::vector<std::string> vec;
get_file("");
for (auto it = m_map.cbegin(); it != m_map.cend(); ++it)
{
vec.push_back(it->first);
}
return vec;
}
fs::file tar_object::get_file(const std::string& path)
{
2017-02-22 14:08:53 +01:00
if (!m_file) return fs::file();
if (auto it = m_map.find(path); it != m_map.end())
{
u64 size = 0;
std::memcpy(&size, it->second.second.size, sizeof(size));
std::vector<u8> buf(size);
m_file.seek(it->second.first);
m_file.read(buf, size);
return fs::make_stream(std::move(buf));
}
else //continue scanning from last file entered
{
const u64 max_size = m_file.size();
while (largest_offset < max_size)
2017-02-22 14:08:53 +01:00
{
TARHeader header = read_header(largest_offset);
u64 size = -1;
std::string filename;
if (std::memcmp(header.magic, "ustar", 5) == 0)
{
const std::string_view size_sv{header.size, std::size(header.size)};
size = octal_text_to_u64(size_sv);
// Check for overflows and if surpasses file size
if ((header.name[0] || header.prefix[0]) && size + 512 > size && max_size >= size + 512 && max_size - size - 512 >= largest_offset)
{
// Cache size in native u64 format
static_assert(sizeof(size) < sizeof(header.size));
std::memcpy(header.size, &size, 8);
std::string_view prefix_name{header.prefix, std::size(header.prefix)};
std::string_view name{header.name, std::size(header.name)};
prefix_name = prefix_name.substr(0, prefix_name.find_first_of('\0'));
name = name.substr(0, name.find_first_of('\0'));
filename += prefix_name;
filename += name;
// Save header and offset
m_map.insert_or_assign(filename, std::make_pair(largest_offset + 512, header));
}
else
{
// Invalid
size = -1;
tar_log.error("tar_object::get_file() failed to convert header.size=%s, filesize=0x%x", size_sv, max_size);
}
}
else
{
tar_log.trace("tar_object::get_file() failed to parse header: offset=0x%x, filesize=0x%x", largest_offset, max_size);
}
if (size == umax)
{
largest_offset += 512;
continue;
}
// Advance offset to next block
largest_offset += utils::align(size, 512) + 512;
if (!path.empty() && path == filename)
{
// Path is equal, read file and advance offset to start of next block
std::vector<u8> buf(size);
if (m_file.read(buf, size))
{
return fs::make_stream(std::move(buf));
}
tar_log.error("tar_object::get_file() failed to read file entry %s (size=0x%x)", filename, size);
largest_offset -= utils::align(size, 512);
}
2017-02-22 14:08:53 +01:00
}
2018-06-23 08:26:11 +02:00
return fs::file();
}
}
bool tar_object::extract(std::string prefix_path, bool is_vfs)
{
2017-02-22 14:08:53 +01:00
if (!m_file) return false;
get_file(""); // Make sure we have scanned all files
for (auto& iter : m_map)
{
const TARHeader& header = iter.second.second;
const std::string& name = iter.first;
2018-06-23 08:26:11 +02:00
std::string result = name;
if (!prefix_path.empty())
{
result = prefix_path + '/' + result;
}
else
{
// Must be VFS here
is_vfs = true;
result.insert(result.begin(), '/');
}
if (is_vfs)
2018-06-23 08:26:11 +02:00
{
result = vfs::get(result);
if (result.empty())
{
tar_log.error("Path of entry is not mounted: '%s' (prefix_path='%s')", name, prefix_path);
return false;
}
2018-06-23 08:26:11 +02:00
}
u64 mtime = octal_text_to_u64({header.mtime, std::size(header.mtime)});
2021-07-30 08:31:36 +02:00
// Let's use it for optional atime
u64 atime = octal_text_to_u64({header.padding, 12});
// This is a fake timestamp, it can be invalid
if (atime == umax)
{
// Set to mtime if not provided
atime = mtime;
}
2017-02-22 14:08:53 +01:00
switch (header.filetype)
{
case '\0':
case '0':
{
// Create the directories which should have been mount points if prefix_path is not empty
if (!prefix_path.empty() && !fs::create_path(fs::get_parent_dir(result)))
{
tar_log.error("TAR Loader: failed to create directory for file %s (%s)", name, fs::g_tls_error);
return false;
}
auto data = get_file(name).release();
2018-06-23 08:26:11 +02:00
fs::file file(result, fs::rewrite);
if (file)
{
file.write(static_cast<fs::container_stream<std::vector<u8>>*>(data.get())->obj);
file.close();
if (mtime != umax && !fs::utime(result, atime, mtime))
{
tar_log.error("TAR Loader: fs::utime failed on %s (%s)", result, fs::g_tls_error);
return false;
}
tar_log.notice("TAR Loader: written file %s", name);
break;
}
const auto old_error = fs::g_tls_error;
tar_log.error("TAR Loader: failed to write file %s (%s) (fs::exists=%s)", name, old_error, fs::exists(result));
return false;
}
2018-06-23 08:26:11 +02:00
case '5':
{
if (!fs::create_path(result))
{
tar_log.error("TAR Loader: failed to create directory %s (%s)", name, fs::g_tls_error);
return false;
}
if (mtime != umax && !fs::utime(result, atime, mtime))
{
tar_log.error("TAR Loader: fs::utime failed on %s (%s)", result, fs::g_tls_error);
return false;
}
break;
}
default:
tar_log.error("TAR Loader: unknown file type: 0x%x", header.filetype);
return false;
}
}
return true;
2017-02-22 14:08:53 +01:00
}
void tar_object::save_directory(const std::string& src_dir, utils::serial& ar, const process_func& func, std::string full_path)
{
const std::string& target_path = full_path.empty() ? src_dir : full_path;
fs::stat_t stat{};
2023-07-18 23:30:36 +02:00
if (!fs::get_stat(target_path, stat))
{
return;
}
if (stat.is_directory)
{
bool has_items = false;
for (auto& entry : fs::dir(target_path))
{
if (entry.name.find_first_not_of('.') == umax) continue;
save_directory(src_dir, ar, func, target_path + '/' + entry.name);
has_items = true;
}
if (has_items)
{
return;
}
}
auto write_octal = [](char* ptr, u64 i)
{
if (!i)
{
*ptr = '0';
return;
}
2022-11-09 19:07:15 +01:00
ptr += utils::aligned_div(static_cast<u32>(std::bit_width(i)), 3) - 1;
for (; i; ptr--, i /= 8)
{
*ptr = static_cast<char>('0' + (i % 8));
}
};
std::string saved_path{target_path.data() + src_dir.size(), target_path.size() - src_dir.size()};
const u64 old_size = ar.data.size();
ar.data.resize(old_size + sizeof(TARHeader));
if (!stat.is_directory)
{
fs::file fd(target_path);
const u64 old_size2 = ar.data.size();
if (func)
{
// Use custom function for file saving if provided
2021-07-30 08:31:36 +02:00
// Allows for example to compress PNG files as JPEG in the TAR itself
if (!func(fd, saved_path, ar))
{
// Revert (this entry should not be included if func returns false)
ar.data.resize(old_size);
return;
}
}
else
{
ar.data.resize(ar.data.size() + stat.size);
ensure(fd.read(ar.data.data() + old_size2, stat.size) == stat.size);
}
// Align
ar.data.resize(old_size2 + utils::align(ar.data.size() - old_size2, 512));
fd.close();
fs::utime(target_path, stat.atime, stat.mtime);
}
TARHeader header{};
std::memcpy(header.magic, "ustar ", 6);
// Prefer saving to name field as much as we can
// If it doesn't fit, save 100 characters at name and 155 characters preceding to it at max
2021-07-30 08:31:36 +02:00
const u64 prefix_size = std::clamp<usz>(saved_path.size(), 100, 255) - 100;
std::memcpy(header.prefix, saved_path.data(), prefix_size);
const u64 name_size = std::min<usz>(saved_path.size(), 255) - prefix_size;
std::memcpy(header.name, saved_path.data() + prefix_size, name_size);
write_octal(header.size, stat.is_directory ? 0 : stat.size);
write_octal(header.mtime, stat.mtime);
write_octal(header.padding, stat.atime);
header.filetype = stat.is_directory ? '5' : '0';
std::memcpy(ar.data.data() + old_size, &header, sizeof(header));
// TAR is an old format which does not depend on previous data so memory ventilation is trivial here
ar.breathe();
}
bool extract_tar(const std::string& file_path, const std::string& dir_path, fs::file file)
{
tar_log.notice("Extracting '%s' to directory '%s'...", file_path, dir_path);
if (!file)
{
file.open(file_path);
}
if (!file)
{
tar_log.error("Error opening file '%s' (%s)", file_path, fs::g_tls_error);
return false;
}
std::vector<fs::file> vec;
if (SCEDecrypter self_dec(file); self_dec.LoadHeaders())
{
// Encrypted file, decrypt
self_dec.LoadMetadata(SCEPKG_ERK, SCEPKG_RIV);
if (!self_dec.DecryptData())
{
tar_log.error("Failed to decrypt TAR.");
return false;
}
vec = self_dec.MakeFile();
if (vec.size() < 3)
{
tar_log.error("Failed to decrypt TAR.");
return false;
}
}
else
{
// Not an encrypted file
tar_log.warning("TAR is not encrypted, it may not be valid for this tool. Encrypted TAR are known to be found in PS3 Firmware files only.");
}
if (!vfs::mount("/tar_extract", dir_path))
{
tar_log.error("Failed to mount '%s'", dir_path);
return false;
}
tar_object tar(vec.empty() ? file : vec[2]);
const bool ok = tar.extract("/tar_extract", true);
if (ok)
{
tar_log.success("Extraction complete!");
}
else
{
tar_log.error("TAR contents are invalid.");
}
// Unmount
Emu.Init();
return ok;
}