spice2x-r3d/util/sigscan.cpp

#include "sigscan.h"

#include <format>
#include <fstream>
#include <sstream>
#include <vector>

#include "util/logging.h"
#include "util/memutils.h"
#include "util/utils.h"

intptr_t find_pattern(std::span<const unsigned char> data, intptr_t base, const uint8_t *pattern,
        const char *mask, intptr_t offset, intptr_t usage, bool reverse)
{
    return find_pattern_from(data, base, pattern, mask, offset, usage, 0, reverse);
}

intptr_t find_pattern(const std::vector<uint8_t> &data, intptr_t base, const uint8_t *pattern,
        const char *mask, intptr_t offset, intptr_t usage, bool reverse)
{
    return find_pattern_from(std::span<const unsigned char>(data), base, pattern, mask, offset, usage, 0, reverse);
}

intptr_t find_pattern(HMODULE module, const uint8_t *pattern, const char *mask,
        intptr_t offset, intptr_t result_usage, bool reverse)
{
    return find_pattern_from(module, pattern, mask, offset, result_usage, 0, reverse);
}

intptr_t find_pattern(HMODULE module, const std::string &pattern, const char *mask,
        intptr_t offset, intptr_t result_usage, bool reverse)
{
    return find_pattern_from(module, pattern, mask, offset, result_usage, 0, reverse);
}

///

intptr_t find_pattern_from(std::span<const unsigned char> data, intptr_t base, const uint8_t *pattern,
        const char *mask, intptr_t offset, intptr_t usage, intptr_t start_from, bool reverse)
{
    // build pattern
    std::vector<std::pair<uint8_t, bool>> pattern_vector;
    size_t mask_size = strlen(mask);
    for (size_t i = 0; i < mask_size; i++) {
        pattern_vector.emplace_back(pattern[i], mask[i] == 'X');
    }
    if (reverse) std::reverse(pattern_vector.begin(), pattern_vector.end());

    // the scan loop
    auto on_search = [&](uint8_t c, std::pair<uint8_t, bool> pat)
    {
        return (!pat.second) || c == pat.first;
    };
    auto cur_usage = 0;
    if (!reverse)
    {
        auto data_begin = data.begin() + start_from;
        while (true)
        {
            // search for the pattern
            auto search_result = std::search(data_begin, data.end(), pattern_vector.begin(), pattern_vector.end(), on_search);

            // check for a match
            if (search_result != data.end())
            {

                // return the result if we hit the usage count
                if (cur_usage == usage)
                {
                    return (std::distance(data.begin(), search_result) + base) + offset;
                }

                // increment the found count
                ++cur_usage;
                data_begin = ++search_result;
            }
            else break;
        }
    }
    else
    {
        auto data_begin = std::make_reverse_iterator(start_from ? data.begin() + start_from + 1 : data.end());
        while(true)
        {
            // search for the pattern
            auto search_result = std::search(data_begin, data.rend(), pattern_vector.begin(), pattern_vector.end(), on_search);

            // check for a match
            if (search_result != data.rend())
            {

                // return the result if we hit the usage count
                if (cur_usage == usage)
                {
                    return (std::distance(data.begin(), search_result.base()) - pattern_vector.size() + base) + offset;
                }

                // increment the found count
                ++cur_usage;
                data_begin = ++search_result;
            }
            else break;
        }

    }

    return 0;
}

intptr_t find_pattern_from(const std::vector<uint8_t> &data, intptr_t base, const uint8_t *pattern,
        const char *mask, intptr_t offset, intptr_t usage, intptr_t start_from, bool reverse)
{
    return find_pattern_from(std::span<const unsigned char>(data), base, pattern, mask, offset, usage, start_from, reverse);
}

intptr_t find_pattern_from(HMODULE module, const uint8_t *pattern, const char *mask,
        intptr_t offset, intptr_t result_usage, intptr_t start_from, bool reverse)
{
    intptr_t base;
    auto data = get_module_data(module, &base);
    return find_pattern_from(
            data,
            base,
            pattern,
            mask,
            offset,
            result_usage,
            start_from,
            reverse);
}

intptr_t find_pattern_from(HMODULE module, const std::string &pattern, const char *mask,
        intptr_t offset, intptr_t result_usage, intptr_t start_from, bool reverse)
{
    auto pattern_bin = std::make_unique<uint8_t[]>(pattern.length() / 2);
    if (!hex2bin(pattern.c_str(), pattern_bin.get())) {
        log_warning("sigscan", "hex2bin failed");
        return false;
    }

    return find_pattern_from(module, pattern_bin.get(), mask, offset, result_usage, start_from, reverse);
}


intptr_t replace_pattern_from(HMODULE module, const uint8_t *pattern, const char *mask, intptr_t offset,
        intptr_t usage, intptr_t start_from, const uint8_t *replace_data, const char *replace_mask, bool reverse)
{

    // find result
    auto result = find_pattern_from(module, pattern, mask, offset, usage, start_from, reverse);

    // check result
    if (!result) {
        return 0;
    }

    // unprotect memory
    auto replace_mask_len = strlen(replace_mask);
    memutils::VProtectGuard guard((void *) result, replace_mask_len);

    // replace data
    for (size_t i = 0; i < replace_mask_len; i++) {
        if (replace_mask[i] == 'X') {
            *((unsigned char *) (result + i)) = replace_data[i];
        }
    }

    // success
    return result;
}

intptr_t replace_pattern_from(HMODULE module, const std::string &signature,
        const std::string &replacement, intptr_t offset, intptr_t usage, intptr_t start_from, bool reverse)
{
    // build pattern
    std::string pattern_str(signature);
    strreplace(pattern_str, "??", "00");
    auto pattern_bin = std::make_unique<uint8_t[]>(signature.length() / 2);
    if (!hex2bin(pattern_str.c_str(), pattern_bin.get())) {
        return false;
    }

    // build signature mask
    std::ostringstream signature_mask;
    for (size_t i = 0; i < signature.length(); i += 2) {
        if (signature[i] == '?') {
            if (signature[i + 1] == '?') {
                signature_mask << '?';
            } else {
                return false;
            }
        } else {
            signature_mask << 'X';
        }
    }

    // build replace data
    std::string replace_data_str(replacement);
    strreplace(replace_data_str, "??", "00");
    auto replace_data_bin = std::make_unique<uint8_t[]>(replacement.length() / 2);
    if (!hex2bin(replace_data_str.c_str(), replace_data_bin.get())) {
        return false;
    }

    // build replace mask
    std::ostringstream replace_mask;
    for (size_t i = 0; i < replacement.length(); i += 2) {
        if (replacement[i] == '?') {
            if (replacement[i + 1] == '?') {
                replace_mask << '?';
            } else {
                return false;
            }
        } else {
            replace_mask << 'X';
        }
    }

    // do the replacement
    return replace_pattern_from(
            module,
            pattern_bin.get(),
            signature_mask.str().c_str(),
            offset,
            usage,
            start_from,
            replace_data_bin.get(),
            replace_mask.str().c_str(),
            reverse
    );
}

intptr_t replace_pattern(HMODULE module, const std::string &signature,
        const std::string &replacement, intptr_t offset, intptr_t usage, bool reverse)
{
    return replace_pattern_from(module, signature, replacement, offset, usage, 0, reverse);
}

intptr_t replace_pattern(HMODULE module, const uint8_t *pattern, const char *mask, intptr_t offset,
        intptr_t usage, const uint8_t *replace_data, const char *replace_mask, bool reverse)
{
    return replace_pattern_from(module, pattern, mask, offset, usage, 0, replace_data, replace_mask, reverse);
}

bool get_pe_identifier(const std::filesystem::path& dll_path, uint32_t* time_date_stamp, uint32_t* address_of_entry_point) {
    std::ifstream file(dll_path, std::ios::binary);
    if (!file) {
        log_warning("sigscan", "Failed to open file: {}", dll_path.string().c_str());
        return false;
    }

    // read the DOS header
    IMAGE_DOS_HEADER dos_header;
    file.read(reinterpret_cast<char*>(&dos_header), sizeof(dos_header));
    if (dos_header.e_magic != IMAGE_DOS_SIGNATURE) {
        log_warning("sigscan", "Invalid DOS signature: {}", dll_path.string().c_str());
        return false;
    }

    // move to the NT headers
    file.seekg(dos_header.e_lfanew);

    // read the NT headers
    IMAGE_NT_HEADERS nt_headers;
    file.read(reinterpret_cast<char*>(&nt_headers), sizeof(nt_headers));
    if (nt_headers.Signature != IMAGE_NT_SIGNATURE) {
        log_warning("sigscan", "Invalid NT signature: {}", dll_path.string().c_str());
        return false;
    }

    // get the TimeDateStamp and AddressOfEntryPoint from the file header
    *time_date_stamp = nt_headers.FileHeader.TimeDateStamp;
    *address_of_entry_point = nt_headers.OptionalHeader.AddressOfEntryPoint;

    return true;
}

intptr_t find_asm_leacstr_from(HMODULE module, reg_t reg, const char *cstr)
{
    // get module information
    intptr_t base;
    auto data = get_module_data(module, &base);
    const unsigned char *modbegp = data.data();
    const unsigned char *modendp = data.data() + data.size();

    // search cstr
    const unsigned char *cstr_addr = std::search(
            modbegp,
            modendp,
            reinterpret_cast<const unsigned char *>(cstr),
            reinterpret_cast<const unsigned char *>(cstr) + strlen(cstr)
    );
    if (cstr_addr == modendp)
    {
        return 0;
    }

#if defined(__i386__)
    // find LEA REG, [abs32]
    if (reg.t != GPR_32) return 0;
    unsigned char asmop[6] = { 0x8D, reg.v, 0x00, 0x00, 0x00, 0x00};
    int32_t dist32 = reinterpret_cast<int32_t>(cstr_addr);
    memcpy(asmop + (sizeof(asmop) - 4), &dist32, 4);
    return reinterpret_cast<intptr_t>(std::search(
            modbegp,
            modendp,
            asmop,
            asmop + sizeof(asmop)
    )) % reinterpret_cast<intptr_t>(modendp);
#elif defined(__x86_64__)
    // find LEA REG, [RIP + rel32]
    unsigned char rex = reg.t == GPR_EXT ? 0x4C : 0x48;
    unsigned char asmop[7] = {rex, 0x8D, reg.v, 0x00, 0x00, 0x00, 0x00};
    for (const unsigned char *addr = modbegp; addr < modendp - sizeof(asmop); addr++)
    {
        auto dist32 = static_cast<int32_t>(reinterpret_cast<intptr_t>(cstr_addr) - reinterpret_cast<intptr_t>(addr + sizeof(asmop)));
        memcpy(asmop + (sizeof(asmop) - 4), &dist32, 4);
        if (memcmp(addr, asmop, sizeof(asmop)) == 0) return reinterpret_cast<intptr_t>(addr);
    }
#endif

    return 0;
}

std::span<const unsigned char> get_module_data(HMODULE module, intptr_t *base)
{
    // get module information
    size_t sz = 0;
    MODULEINFO module_info {};
    if (GetModuleInformation(GetCurrentProcess(), module, &module_info, sizeof(module_info))) {
        sz = static_cast<size_t>(module_info.SizeOfImage);
    }

    // wrap data
    std::span<const unsigned char> data(static_cast<const unsigned char*>(module_info.lpBaseOfDll), sz);
    *base = reinterpret_cast<intptr_t>(module_info.lpBaseOfDll);

    return data;
}