feat(bindless): Refactor DLL parser

This commit is contained in:
PancakeTAS 2026-04-25 19:54:09 +02:00
parent bd77a7917f
commit 8a47213020
No known key found for this signature in database
2 changed files with 235 additions and 0 deletions

View file

@ -0,0 +1,212 @@
/* SPDX-License-Identifier: GPL-3.0-or-later */
#include "dll.hpp"
#include <array>
#include <cstddef>
#include <cstdint>
#include <filesystem>
#include <fstream>
#include <ios>
#include <iostream>
#include <optional>
#include <span>
#include <stdexcept>
#include <unordered_map>
#include <utility>
#include <vector>
namespace {
/// DOS file header
struct DOSHeader {
uint16_t magic; // 0x5A4D
std::array<uint16_t, 29> pad;
int32_t pe_offset; // File offset
};
/// PE header
struct PEHeader {
uint32_t signature; // "PE\0\0"
std::array<uint16_t, 1> pad1;
uint16_t sect_count;
std::array<uint16_t, 6> pad2;
uint16_t opt_hdr_size;
std::array<uint16_t, 1> pad3;
};
/// (partial!) PE optional header
struct PEOptionalHeader {
uint16_t magic; // 0x20B
std::array<uint16_t, 63> pad4;
std::pair<uint32_t, uint32_t> resource_table; // File offset/size
};
/// Section header
struct SectionHeader {
std::array<uint16_t, 4> pad1;
uint32_t vsize; // Virtual
uint32_t vaddress;
uint32_t fsize; // Raw
uint32_t foffset;
std::array<uint16_t, 8> pad2;
};
/// Resource directory
struct ResourceDirectory {
std::array<uint16_t, 6> pad;
uint16_t name_count;
uint16_t id_count;
};
/// Resource directory entry
struct ResourceDirectoryEntry {
uint32_t id;
uint32_t offset; // High bit = Directory
};
/// Resource data entry
struct ResourceDataEntry {
uint32_t offset;
uint32_t size;
std::array<uint32_t, 2> pad;
};
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Wunknown-warning-option"
#pragma clang diagnostic ignored "-Wunsafe-buffer-usage-in-container"
/// Safely cast a vector to a pointer of type T
template<typename T>
const T* safe_cast(const std::vector<uint8_t>& data, size_t offset) {
const size_t end{offset + sizeof(T)};
if (end > data.size() || end < offset)
throw std::runtime_error("Buffer overflow/underflow during safe cast");
return reinterpret_cast<const T*>(&data.at(offset)); // NOLINT (unsafe cast)
}
/// Safely cast a vector to a span of T
template<typename T>
std::span<const T> span_cast(const std::vector<uint8_t>& data, size_t offset, size_t count) {
const size_t end{offset + (count * sizeof(T))};
if (end > data.size() || end < offset)
throw std::runtime_error("Buffer overflow/underflow during safe cast");
return{ reinterpret_cast<const T*>(&data.at(offset)), count }; // NOLINT (unsafe cast)
}
#pragma clang diagnostic pop
}
using namespace lsfgvk::library;
std::unordered_map<uint32_t, std::vector<uint32_t>> priv::parseDll(
const std::filesystem::path& dll
) {
std::ifstream file(dll, std::ios::binary | std::ios::ate);
if (!file.is_open())
throw std::runtime_error("Unable to open file");
const std::streamsize size{static_cast<std::streamsize>(file.tellg())};
file.seekg(0, std::ios::beg);
std::vector<uint8_t> data(static_cast<size_t>(size));
if (!file.read(reinterpret_cast<char*>(data.data()), size)) // NOLINT (unsafe cast)
throw std::runtime_error("Unable to read file");
// Parse dos header
size_t fileOffset{0};
const auto* dosHdr{safe_cast<const DOSHeader>(data, 0)};
if (dosHdr->magic != 0x5A4D)
throw std::runtime_error("Magic number in DOS header is incorrect");
// Parse pe header
fileOffset += static_cast<size_t>(dosHdr->pe_offset);
const auto* peHdr{safe_cast<const PEHeader>(data, fileOffset)};
if (peHdr->signature != 0x00004550)
throw std::runtime_error("Signature in PE header is incorrect");
// Parse optional pe header
fileOffset += sizeof(PEHeader);
const auto* peOptHdr{safe_cast<const PEOptionalHeader>(data, fileOffset)};
if (peOptHdr->magic != 0x20B)
throw std::runtime_error("PE format is not PE32+");
const auto& [rsrc_rva, rsrc_size] = peOptHdr->resource_table;
// }Locate section containing resources
std::optional<size_t> rsrc_offset;
fileOffset += peHdr->opt_hdr_size;
const auto sectHdrs{span_cast<const SectionHeader>(data, fileOffset, peHdr->sect_count)};
for (const auto& sectHdr : sectHdrs) {
if (rsrc_rva < sectHdr.vaddress || rsrc_rva > (sectHdr.vaddress + sectHdr.vsize))
continue;
rsrc_offset.emplace((rsrc_rva - sectHdr.vaddress) + sectHdr.foffset);
break;
}
if (!rsrc_offset)
throw std::runtime_error("Unable to locate resource section");
// Parse resource directory
fileOffset = rsrc_offset.value();
const auto* rsrcDir{safe_cast<const ResourceDirectory>(data, fileOffset)};
if (rsrcDir->id_count < 3)
throw std::runtime_error("Resource directory does not have enough entries");
// Find resource table with data type
std::optional<size_t> rsrc_tbl_offset;
fileOffset = rsrc_offset.value() + sizeof(ResourceDirectory);
const auto rsrcDirEntries{span_cast<const ResourceDirectoryEntry>(
data, fileOffset, rsrcDir->name_count + rsrcDir->id_count)};
for (const auto& rsrcDirEntry : rsrcDirEntries) {
if (rsrcDirEntry.id != 10) // RT_RCDATA
continue;
if ((rsrcDirEntry.offset & 0x80000000) == 0)
throw std::runtime_error("Expected resource directory, found data entry");
rsrc_tbl_offset.emplace(rsrcDirEntry.offset & 0x7FFFFFFF);
}
if (!rsrc_tbl_offset)
throw std::runtime_error("Unable to locate RT_RCDATA directory");
// Parse data type resource directory
fileOffset = rsrc_offset.value() + rsrc_tbl_offset.value();
const auto* rsrcTbl{safe_cast<const ResourceDirectory>(data, fileOffset)};
if (rsrcTbl->id_count < 1)
throw std::runtime_error("RT_RCDATA directory does not have enough entries");
// Collect all resources
fileOffset += sizeof(ResourceDirectory);
const auto rsrcTblEntries{span_cast<const ResourceDirectoryEntry>(
data, fileOffset, rsrcTbl->name_count + rsrcTbl->id_count)};
std::unordered_map<uint32_t, std::vector<uint32_t>> resources;
resources.reserve(rsrcTbl->id_count);
for (const auto& rsrcTblEntry : rsrcTblEntries) {
if ((rsrcTblEntry.offset & 0x80000000) == 0)
throw std::runtime_error("Expected resource directory, found data entry");
// Skip over language directory
fileOffset = rsrc_offset.value() + (rsrcTblEntry.offset & 0x7FFFFFFF);
const auto* langDir = safe_cast<const ResourceDirectory>(data, fileOffset);
if (langDir->id_count < 1)
throw std::runtime_error("Malformed language directory");
fileOffset += sizeof(ResourceDirectory);
const auto* langDirEntry{safe_cast<const ResourceDirectoryEntry>(data, fileOffset)};
if ((langDirEntry->offset & 0x80000000) != 0)
throw std::runtime_error("Expected resource data entry, found directory");
// Parse resource data entry
fileOffset = rsrc_offset.value() + (langDirEntry->offset & 0x7FFFFFFF);
const auto* entry{safe_cast<const ResourceDataEntry>(data, fileOffset)};
if (entry->offset < rsrc_rva || entry->offset > (rsrc_rva + rsrc_size))
throw std::runtime_error("Resource data entry points outside resource section");
// Extract resource
fileOffset = (entry->offset - rsrc_rva) + rsrc_offset.value();
const auto rdata{span_cast<const uint32_t>(
data, fileOffset, entry->size / sizeof(uint32_t))};
resources.emplace(rsrcTblEntry.id, std::vector<uint32_t>(rdata.begin(), rdata.end()));
}
return resources;
}

View file

@ -0,0 +1,23 @@
/* SPDX-License-Identifier: GPL-3.0-or-later */
#pragma once
#include <cstdint>
#include <filesystem>
#include <unordered_map>
#include <vector>
namespace lsfgvk::library::priv {
///
/// Parse all resources from a DLL file
///
/// @param dll File path
/// @returns Map of resource ID to data
/// @throws std::runtime_error if the file is invalid or cannot be read
///
std::unordered_map<uint32_t, std::vector<uint32_t>> parseDll(
const std::filesystem::path& dll
);
}