enhancement(nodeps): remove pe-parse dependency

.gitmodules

@@ -1,3 +0,0 @@
-[submodule "thirdparty/pe-parse"]
-	path = thirdparty/pe-parse
-	url = https://github.com/trailofbits/pe-parse

CMakeLists.txt

@@ -12,7 +12,6 @@ add_compile_options(-fPIC
     -Wno-deprecated-declarations
     -Wno-unused-template)
 
-add_subdirectory(thirdparty/pe-parse/pe-parser-library EXCLUDE_FROM_ALL)
 add_subdirectory(framegen)
 
 if(LSFGVK_EXCESS_DEBUG)
@@ -44,7 +43,6 @@ target_include_directories(lsfg-vk SYSTEM
 target_include_directories(lsfg-vk
     PUBLIC include)
 target_link_libraries(lsfg-vk PUBLIC
-    pe-parse
     lsfg-vk-framegen)
 
 # diagnostics

include/extract/dll.hpp

@@ -0,0 +1,22 @@
+#pragma once
+
+#include <unordered_map>
+#include <cstdint>
+#include <string>
+#include <vector>
+
+namespace DLL {
+
+    ///
+    /// Parse all resources from a DLL file.
+    ///
+    /// *Shouldn't* cause any segmentation faults.
+    ///
+    /// @param filename Path to the DLL file.
+    /// @return A map of resource IDs to their binary data.
+    ///
+    /// @throws std::runtime_error on various failure points.
+    ///
+    std::unordered_map<uint32_t, std::vector<uint8_t>> parse_dll(const std::string& filename);
+
+}

src/extract/dll.cpp

@@ -0,0 +1,203 @@
+#include "extract/dll.hpp"
+
+#include <unordered_map>
+#include <stdexcept>
+#include <algorithm>
+#include <iostream>
+#include <optional>
+#include <cstddef>
+#include <cstdint>
+#include <fstream>
+#include <utility>
+#include <string>
+#include <vector>
+#include <array>
+#include <span>
+
+/// 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 "-Wunsafe-buffer-usage-in-container"
+namespace {
+    /// 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 during safe cast");
+        return reinterpret_cast<const T*>(&data.at(offset));
+    }
+
+    /// 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 during safe cast");
+        return std::span<const T>(reinterpret_cast<const T*>(&data.at(offset)), count);
+    }
+}
+#pragma clang diagnostic pop
+
+std::unordered_map<uint32_t, std::vector<uint8_t>> DLL::parse_dll(const std::string& filename) {
+    std::ifstream file(filename, std::ios::binary | std::ios::ate);
+    if (!file.is_open())
+        throw std::runtime_error("Failed to open Lossless.dll");
+
+    const auto size = 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))
+        throw std::runtime_error("Failed to read Lossless.dll");
+
+    // parse dos header
+    size_t fileOffset = 0;
+    const auto* dosHdr = safe_cast<const DOSHeader>(data, 0);
+    if (dosHdr->magic != 0x5A4D)
+        throw std::runtime_error("Invalid DOS header magic number");
+
+    // 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("Invalid PE header signature");
+
+    // parse optional pe header
+    fileOffset += sizeof(PEHeader);
+    const auto* peOptHdr = safe_cast<const PEOptionalHeader>(data, fileOffset);
+    if (peOptHdr->magic != 0x20B)
+        throw std::runtime_error("Unsupported PE format (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("Failed 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("Incorrect resource directory");
+
+    // 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, but found data entry");
+
+        rsrc_tbl_offset.emplace(rsrcDirEntry.offset & 0x7FFFFFFF);
+    }
+    if (!rsrc_tbl_offset)
+        throw std::runtime_error("Failed 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("Incorrect RT_RCDATA directory");
+
+    // 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<uint8_t>> resources;
+    for (const auto& rsrcTblEntry : rsrcTblEntries) {
+        if ((rsrcTblEntry.offset & 0x80000000) == 0)
+            throw std::runtime_error("Expected resource directory, but 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("Incorrect 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, but 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
+        std::vector<uint8_t> resource(entry->size);
+        fileOffset = (entry->offset - rsrc_rva) + rsrc_offset.value();
+        if (fileOffset + entry->size > data.size())
+            throw std::runtime_error("Resource data entry points outside file");
+        std::copy_n(&data.at(fileOffset), entry->size, resource.data());
+        resources.emplace(rsrcTblEntry.id, std::move(resource));
+    }
+
+    return resources;
+}

src/extract/extract.cpp

@@ -1,16 +1,14 @@
 #include "extract/extract.hpp"
 #include "config/config.hpp"
+#include "extract/dll.hpp"
 
-#include <pe-parse/parse.h>
-
+#include <iostream>
 #include <unordered_map>
 #include <filesystem>
-#include <algorithm>
 #include <stdexcept>
 #include <cstdint>
 #include <cstdlib>
 #include <string>
-#include <utility>
 #include <vector>
 #include <array>
 
@@ -73,22 +71,11 @@ const std::unordered_map<std::string, uint32_t> nameIdxTable = {{
 }};
 
 namespace {
-    auto& pshaders() {
+    auto& shaders() {
         static std::unordered_map<uint32_t, std::array<std::vector<uint8_t>, 2>> shaderData;
         return shaderData;
     }
 
-    int on_resource(void* ptr, const peparse::resource& res) {
-        if (res.type != peparse::RT_RCDATA || res.buf == nullptr || res.buf->bufLen <= 0)
-            return 0;
-        std::vector<uint8_t> resource_data(res.buf->bufLen);
-        std::copy_n(res.buf->buf, res.buf->bufLen, resource_data.data());
-
-        auto* shaders = reinterpret_cast<std::unordered_map<uint32_t, std::vector<uint8_t>>*>(ptr);
-        shaders->emplace(res.name, std::move(resource_data));
-        return 0;
-    }
-
     const std::vector<std::filesystem::path> PATHS{{
         ".local/share/Steam/steamapps/common",
         ".steam/steam/steamapps/common",
@@ -121,44 +108,39 @@ namespace {
 }
 
 void Extract::extractShaders() {
-    if (!pshaders().empty())
+    if (!shaders().empty())
         return;
 
-    std::unordered_map<uint32_t, std::vector<uint8_t>> shaders{};
-
     // parse the dll
-    peparse::parsed_pe* dll = peparse::ParsePEFromFile(getDllPath().c_str());
-    if (!dll)
-        throw std::runtime_error("Unable to read Lossless.dll, is it installed?");
-    peparse::IterRsrc(dll, on_resource, reinterpret_cast<void*>(&shaders));
-    peparse::DestructParsedPE(dll);
+    const auto resources = DLL::parse_dll(getDllPath());
+    std::cerr << "lsfg-vk: Extracted " << resources.size() << " resources from dll.\n";
 
     // ensure all shaders are present
     for (const auto& [name, idx] : nameIdxTable) {
-        auto fp16 = shaders.find(idx);
-        if (fp16 == shaders.end())
+        auto fp16 = resources.find(idx);
+        if (fp16 == resources.end())
             throw std::runtime_error("Shader not found: " + name + " (FP16).\n- Is Lossless Scaling up to date?");
-        auto fp32 = shaders.find(idx + FP);
-        if (fp32 == shaders.end())
+        auto fp32 = resources.find(idx + FP);
+        if (fp32 == resources.end())
             throw std::runtime_error("Shader not found: " + name + " (FP32).\n- Is Lossless Scaling up to date?");
 
-        pshaders().emplace(idx, std::array<std::vector<uint8_t>, 2>{
-            std::move(fp32->second),
-            std::move(fp16->second)
+        shaders().emplace(idx, std::array<std::vector<uint8_t>, 2>{
+            fp32->second,
+            fp16->second
         });
     }
 }
 
 std::vector<uint8_t> Extract::getShader(const std::string& name, bool fp16) {
-    if (pshaders().empty())
+    if (shaders().empty())
         throw std::runtime_error("Shaders are not loaded.");
 
     auto hit = nameIdxTable.find(name);
     if (hit == nameIdxTable.end())
         throw std::runtime_error("Shader hash not found: " + name);
 
-    auto sit = pshaders().find(hit->second);
-    if (sit == pshaders().end())
+    auto sit = shaders().find(hit->second);
+    if (sit == shaders().end())
         throw std::runtime_error("Shader not found: " + name);
     return fp16 ? sit->second.at(1) : sit->second.at(0);
 }

thirdparty/pe-parse

@@ -1 +0,0 @@
-Subproject commit 31ac5966503689d5693cd9fb520bd525a8710e17