Skip to content

Instantly share code, notes, and snippets.

@wheremyfoodat
Created May 5, 2024 22:45
Show Gist options
  • Save wheremyfoodat/ae2c2268da8f015168b862e268d28dbe to your computer and use it in GitHub Desktop.
Save wheremyfoodat/ae2c2268da8f015168b862e268d28dbe to your computer and use it in GitHub Desktop.
Single-header patcher for IPS. UPS and BPS files
#pragma once
#include <algorithm>
#include <array>
#include <climits>
#include <cstdio>
#include <cstring>
#include <type_traits>
#include <utility>
#include <vector>
namespace Hips {
using u8 = std::uint8_t;
using u16 = std::uint16_t;
using u32 = std::uint32_t;
using u64 = std::uint64_t;
using usize = std::size_t;
using s8 = std::int8_t;
using s16 = std::int16_t;
using s32 = std::int32_t;
using s64 = std::int64_t;
enum class PatchType : u32 {
IPS,
UPS,
BPS,
};
enum class Result : u32 {
Success = 0,
InvalidPatch,
UnknownFormat,
SizeMismatch,
ChecksumMismatch,
};
namespace Detail {
// Read "size" bytes, returning 0 if we're going to go out of bounds
template <typename T = u64, usize size>
T readBE(const u8* data, usize& offset, usize patchSize) {
static_assert(std::is_integral<T>() && sizeof(T) >= size);
offset += size;
if (offset > patchSize) {
// We're going to go out of bounds, return 0
return T(0);
}
T ret = T(0);
int shift = 0;
// Read byte-by-byte
for (int i = 0; i < size; i++) {
ret |= T(data[offset - (i + 1)]) << shift;
shift += CHAR_BIT;
}
return ret;
}
template <typename T = u64, usize size>
T readLE(const u8* data, usize& offset, usize patchSize) {
static_assert(std::is_integral<T>() && sizeof(T) >= size);
offset += size;
if (offset > patchSize) {
// We're going to go out of bounds, return 0
return T(0);
}
T ret = T(0);
int shift = (size - 1) * 8;
// Read byte-by-byte
for (int i = 0; i < size; i++) {
ret |= T(data[offset - (i + 1)]) << shift;
shift -= CHAR_BIT;
}
return ret;
}
// Formats like UPS and BPS use run-length encoded integers
// Regrettably, handling anything > 64 bits is not easy, or particularly worth it
// Until files start being larger than 18 exabytes that is
template <typename T = u64>
T readRunLength(const u8* data, usize& offset, usize patchSize) {
u64 ret = 0;
u64 shift = 1;
while (true) {
const u64 byte = readLE<u64, 1>(data, offset, patchSize);
ret += (byte & 0x7F) * shift;
// If the msb is set then the encoding ends on this byte
if (byte & 0x80) {
break;
}
shift <<= 7;
ret += shift;
}
return T(ret);
}
static u32 crc32(const u8* data, usize length, u32 crc = 0) {
// 8-bit CRC table
static constexpr u32 crcTable[] = {
0x00000000, 0x77073096, 0xEE0E612C, 0x990951BA, 0x076DC419, 0x706AF48F, 0xE963A535, 0x9E6495A3, 0x0EDB8832, 0x79DCB8A4, 0xE0D5E91E,
0x97D2D988, 0x09B64C2B, 0x7EB17CBD, 0xE7B82D07, 0x90BF1D91, 0x1DB71064, 0x6AB020F2, 0xF3B97148, 0x84BE41DE, 0x1ADAD47D, 0x6DDDE4EB,
0xF4D4B551, 0x83D385C7, 0x136C9856, 0x646BA8C0, 0xFD62F97A, 0x8A65C9EC, 0x14015C4F, 0x63066CD9, 0xFA0F3D63, 0x8D080DF5, 0x3B6E20C8,
0x4C69105E, 0xD56041E4, 0xA2677172, 0x3C03E4D1, 0x4B04D447, 0xD20D85FD, 0xA50AB56B, 0x35B5A8FA, 0x42B2986C, 0xDBBBC9D6, 0xACBCF940,
0x32D86CE3, 0x45DF5C75, 0xDCD60DCF, 0xABD13D59, 0x26D930AC, 0x51DE003A, 0xC8D75180, 0xBFD06116, 0x21B4F4B5, 0x56B3C423, 0xCFBA9599,
0xB8BDA50F, 0x2802B89E, 0x5F058808, 0xC60CD9B2, 0xB10BE924, 0x2F6F7C87, 0x58684C11, 0xC1611DAB, 0xB6662D3D, 0x76DC4190, 0x01DB7106,
0x98D220BC, 0xEFD5102A, 0x71B18589, 0x06B6B51F, 0x9FBFE4A5, 0xE8B8D433, 0x7807C9A2, 0x0F00F934, 0x9609A88E, 0xE10E9818, 0x7F6A0DBB,
0x086D3D2D, 0x91646C97, 0xE6635C01, 0x6B6B51F4, 0x1C6C6162, 0x856530D8, 0xF262004E, 0x6C0695ED, 0x1B01A57B, 0x8208F4C1, 0xF50FC457,
0x65B0D9C6, 0x12B7E950, 0x8BBEB8EA, 0xFCB9887C, 0x62DD1DDF, 0x15DA2D49, 0x8CD37CF3, 0xFBD44C65, 0x4DB26158, 0x3AB551CE, 0xA3BC0074,
0xD4BB30E2, 0x4ADFA541, 0x3DD895D7, 0xA4D1C46D, 0xD3D6F4FB, 0x4369E96A, 0x346ED9FC, 0xAD678846, 0xDA60B8D0, 0x44042D73, 0x33031DE5,
0xAA0A4C5F, 0xDD0D7CC9, 0x5005713C, 0x270241AA, 0xBE0B1010, 0xC90C2086, 0x5768B525, 0x206F85B3, 0xB966D409, 0xCE61E49F, 0x5EDEF90E,
0x29D9C998, 0xB0D09822, 0xC7D7A8B4, 0x59B33D17, 0x2EB40D81, 0xB7BD5C3B, 0xC0BA6CAD, 0xEDB88320, 0x9ABFB3B6, 0x03B6E20C, 0x74B1D29A,
0xEAD54739, 0x9DD277AF, 0x04DB2615, 0x73DC1683, 0xE3630B12, 0x94643B84, 0x0D6D6A3E, 0x7A6A5AA8, 0xE40ECF0B, 0x9309FF9D, 0x0A00AE27,
0x7D079EB1, 0xF00F9344, 0x8708A3D2, 0x1E01F268, 0x6906C2FE, 0xF762575D, 0x806567CB, 0x196C3671, 0x6E6B06E7, 0xFED41B76, 0x89D32BE0,
0x10DA7A5A, 0x67DD4ACC, 0xF9B9DF6F, 0x8EBEEFF9, 0x17B7BE43, 0x60B08ED5, 0xD6D6A3E8, 0xA1D1937E, 0x38D8C2C4, 0x4FDFF252, 0xD1BB67F1,
0xA6BC5767, 0x3FB506DD, 0x48B2364B, 0xD80D2BDA, 0xAF0A1B4C, 0x36034AF6, 0x41047A60, 0xDF60EFC3, 0xA867DF55, 0x316E8EEF, 0x4669BE79,
0xCB61B38C, 0xBC66831A, 0x256FD2A0, 0x5268E236, 0xCC0C7795, 0xBB0B4703, 0x220216B9, 0x5505262F, 0xC5BA3BBE, 0xB2BD0B28, 0x2BB45A92,
0x5CB36A04, 0xC2D7FFA7, 0xB5D0CF31, 0x2CD99E8B, 0x5BDEAE1D, 0x9B64C2B0, 0xEC63F226, 0x756AA39C, 0x026D930A, 0x9C0906A9, 0xEB0E363F,
0x72076785, 0x05005713, 0x95BF4A82, 0xE2B87A14, 0x7BB12BAE, 0x0CB61B38, 0x92D28E9B, 0xE5D5BE0D, 0x7CDCEFB7, 0x0BDBDF21, 0x86D3D2D4,
0xF1D4E242, 0x68DDB3F8, 0x1FDA836E, 0x81BE16CD, 0xF6B9265B, 0x6FB077E1, 0x18B74777, 0x88085AE6, 0xFF0F6A70, 0x66063BCA, 0x11010B5C,
0x8F659EFF, 0xF862AE69, 0x616BFFD3, 0x166CCF45, 0xA00AE278, 0xD70DD2EE, 0x4E048354, 0x3903B3C2, 0xA7672661, 0xD06016F7, 0x4969474D,
0x3E6E77DB, 0xAED16A4A, 0xD9D65ADC, 0x40DF0B66, 0x37D83BF0, 0xA9BCAE53, 0xDEBB9EC5, 0x47B2CF7F, 0x30B5FFE9, 0xBDBDF21C, 0xCABAC28A,
0x53B39330, 0x24B4A3A6, 0xBAD03605, 0xCDD70693, 0x54DE5729, 0x23D967BF, 0xB3667A2E, 0xC4614AB8, 0x5D681B02, 0x2A6F2B94, 0xB40BBE37,
0xC30C8EA1, 0x5A05DF1B, 0x2D02EF8D
};
crc = ~crc;
for (usize i = 0; i < length; i++) {
const u8 byte = data[i];
crc = crcTable[(crc ^ byte) & 0xFF] ^ (crc >> 8);
}
return ~crc;
}
} // namespace Detail
namespace IPS {
static constexpr usize headerSize = 5;
// Need at least 5 (header) + 3 (EOF) bytes to be a valid IPS patch
static constexpr usize minimumPatchSize = headerSize + 3;
// "EOF" magic string
static constexpr usize endOfFile = 0x454F46;
template <typename T = u64, usize size>
T read(const u8* data, usize& offset, usize patchSize) {
return Detail::readBE<T, size>(data, offset, patchSize);
}
// The size isn't even encoded in the file properly, so we need to parse the file one time first to figure it out...
static usize getSize(const u8* patch, usize patchSize) {
usize outputSize = 0;
usize patchOffset = headerSize; // Skip header bytes
while (patchOffset < patchSize) {
const usize fileOffset = read<usize, 3>(patch, patchOffset, patchSize);
if (fileOffset == endOfFile) {
break;
}
usize newSize = fileOffset;
const u16 size = read<u16, 2>(patch, patchOffset, patchSize);
// RLE encoding
if (size == 0) {
const u16 rleSize = read<u16, 2>(patch, patchOffset, patchSize);
newSize += rleSize;
patchOffset += 1; // Skip value field
} else {
patchOffset += size; // Skip data field
newSize += size;
}
outputSize = std::max<usize>(outputSize, newSize);
}
if (patchOffset + 3 == patchSize) {
// Apparently some IPS files have a 3 byte footer with the ROM size after EOF
const usize actualSize = read<usize, 3>(patch, patchOffset, patchSize);
outputSize = std::max<usize>(outputSize, actualSize);
}
return outputSize;
}
}; // namespace IPS
static std::pair<std::vector<u8>, Result> patchIPS(const u8* data, usize dataSize, const u8* patch, usize patchSize) {
if (patch == nullptr || patchSize < IPS::minimumPatchSize) [[unlikely]] {
return {{}, Result::InvalidPatch};
}
// Header magic does not match, so the patch is invalid
if (patch[0] != 'P' || patch[1] != 'A' || patch[2] != 'T' || patch[3] != 'C' || patch[4] != 'H') [[unlikely]] {
return {{}, Result::InvalidPatch};
}
// Copy file to be patched in output buffer
std::vector<u8> output(IPS::getSize(patch, patchSize));
std::memcpy(output.data(), data, std::min<u64>(output.size(), dataSize));
// Skip header
usize offset = IPS::headerSize;
while (offset < patchSize) {
// Read the next record, starting from the 3-byte offset where the patch will be placed in the file to patch
const usize fileOffset = IPS::read<usize, 3>(patch, offset, patchSize);
if (fileOffset == IPS::endOfFile) {
// If we detect EOF, we are done applying the patch
break;
}
// Size of data to copy
u16 size = IPS::read<u16, 2>(patch, offset, patchSize);
if (size == 0) {
// RLE encoding
u16 rleSize = IPS::read<u16, 2>(patch, offset, patchSize);
u8 value = IPS::read<u8, 1>(patch, offset, patchSize);
for (int i = 0; i < rleSize; i++) {
// Going out of ROM bounds
if (fileOffset + i >= output.size()) {
break;
}
output[fileOffset + i] = value;
}
} else {
for (int i = 0; i < size; i++) {
// Going out of ROM bounds
if (fileOffset + i >= output.size()) {
break;
}
output[fileOffset + i] = IPS::read<u8, 1>(patch, offset, patchSize);
}
}
}
return {output, Result::Success};
}
namespace UPS {
static constexpr usize headerSize = 4;
// Need at least 4 (header) + 2 (minimum size for input/output sizes) + crc32s for input file, output file and patch
static constexpr usize minimumPatchSize = 18;
template <typename T = u64, usize size>
T read(const u8* data, usize& offset, usize patchSize) {
return Detail::readLE<T, size>(data, offset, patchSize);
}
template <typename T = u64>
T readRunLength(const u8* data, usize& offset, usize patchSize) {
return Detail::readRunLength<T>(data, offset, patchSize);
}
} // namespace UPS
static std::pair<std::vector<u8>, Result> patchUPS(const u8* data, usize dataSize, const u8* patch, usize patchSize) {
if (patch == nullptr || patchSize < IPS::minimumPatchSize) [[unlikely]] {
return {{}, Result::InvalidPatch};
}
// Header magic does not match, so the patch is invalid
if (patch[0] != 'U' || patch[1] != 'P' || patch[2] != 'S' || patch[3] != '1') [[unlikely]] {
return {{}, Result::InvalidPatch};
}
usize patchOffset = UPS::headerSize;
const u64 inputSize = UPS::readRunLength<u64>(patch, patchOffset, patchSize);
const u64 outputSize = UPS::readRunLength<u64>(patch, patchOffset, patchSize);
// The file we're trying to patch is smaller than the input is meant to be, reject it
if (dataSize < inputSize) {
return {{}, Result::SizeMismatch};
}
std::vector<u8> output(outputSize);
usize sourceOffset = 0;
usize outputOffset = 0;
while (patchOffset < patchSize - 12) {
u64 length = UPS::readRunLength<u64>(patch, patchOffset, patchSize);
// Copy length bytes as-is
while (length > 0 && outputOffset < outputSize) {
output[outputOffset++] = UPS::read<u8, 1>(data, sourceOffset, dataSize);
length -= 1;
}
while (outputOffset < outputSize) {
// Patch with XOR until we find the terminating patch value (0x00)
// Patching with XOR means patches are reversible, by simply applying the patch again
const u8 sourceValue = UPS::read<u8, 1>(data, sourceOffset, dataSize);
const u8 patchValue = UPS::read<u8, 1>(patch, patchOffset, patchSize);
output[outputOffset++] = sourceValue ^ patchValue;
// We need to check for this after applying the xor patch, otherwise the offset values will be wrong
if (patchValue == 0) {
break;
}
}
}
while (outputOffset < outputSize && sourceOffset < dataSize) {
// Copy the rest of the bytes
output[outputOffset++] = UPS::read<u8, 1>(data, sourceOffset, dataSize);
}
// Pad rest of the output with 0s
while (outputOffset < outputSize) {
output[outputOffset++] = 0;
}
const u32 inputCRC = UPS::read<u32, 4>(patch, patchOffset, patchSize);
const u32 outputCRC = UPS::read<u32, 4>(patch, patchOffset, patchSize);
const u32 patchCRC = UPS::read<u32, 4>(patch, patchOffset, patchSize);
if (outputCRC != Detail::crc32(output.data(), output.size())) {
return {output, Result::ChecksumMismatch};
}
return {output, Result::Success};
}
namespace BPS {
static constexpr usize headerSize = 4;
// Need at least 5 (header) + 3 (source/target/metadata size) + 9 (checksums) bytes to be a valid BPS patch
static constexpr usize minimumPatchSize = headerSize + 3 + 9;
template <typename T = u64, usize size>
T read(const u8* data, usize& offset, usize patchSize) {
return Detail::readLE<T, size>(data, offset, patchSize);
}
template <typename T = u64>
T readRunLength(const u8* data, usize& offset, usize patchSize) {
return Detail::readRunLength<T>(data, offset, patchSize);
}
namespace Action {
enum : u32 {
SourceRead = 0,
TargetRead = 1,
SourceCopy = 2,
TargetCopy = 3,
};
}
} // namespace BPS
static std::pair<std::vector<u8>, Result> patchBPS(const u8* data, usize dataSize, const u8* patch, usize patchSize) {
if (patch == nullptr || patchSize < BPS::minimumPatchSize) [[unlikely]] {
return {{}, Result::InvalidPatch};
}
// Header magic does not match, so the patch is invalid
if (patch[0] != 'B' || patch[1] != 'P' || patch[2] != 'S' || patch[3] != '1') [[unlikely]] {
return {{}, Result::InvalidPatch};
}
usize patchOffset = BPS::headerSize;
const u64 inputSize = BPS::readRunLength<u64>(patch, patchOffset, patchSize);
const u64 outputSize = BPS::readRunLength<u64>(patch, patchOffset, patchSize);
const u64 metadataSize = BPS::readRunLength<u64>(patch, patchOffset, patchSize);
// The file we're trying to patch is smaller than the input is meant to be, reject it
if (dataSize < inputSize) {
return {{}, Result::SizeMismatch};
}
// Copy file to be patched in output buffer
std::vector<u8> output(outputSize);
usize sourceOffset = 0;
usize outputOffset = 0;
usize outputOffset2 = 0; // Offset used for TargetCopy commands
while (patchOffset < patchSize - 12) {
// Each "record" in a BPS patch consists of a VLE word, whose bottom 2 bits are a patching "action" to perform
// And the top bits are the length of memory to operate on
const u64 word = BPS::readRunLength<u64>(patch, patchOffset, patchSize);
const u64 action = (word & 3);
u64 length = (word >> 2) + 1;
switch (action) {
case BPS::Action::SourceRead: {
while (length > 0 && outputOffset < outputSize && outputOffset < dataSize) {
output[outputOffset] = data[outputOffset];
outputOffset += 1;
length -= 1;
}
break;
}
case BPS::Action::TargetRead: {
while (length > 0 && outputOffset < outputSize) {
output[outputOffset++] = BPS::read<u8, 1>(patch, patchOffset, patchSize);
length -= 1;
}
break;
}
case BPS::Action::SourceCopy: {
const u64 word = BPS::readRunLength<u64>(patch, patchOffset, patchSize);
const s64 offset = s64(word >> 1);
sourceOffset += (word & 1) ? -offset : +offset;
while (length > 0) {
output[outputOffset++] = data[sourceOffset++];
length -= 1;
}
break;
}
case BPS::Action::TargetCopy: {
const u64 data = BPS::readRunLength<u64>(patch, patchOffset, patchSize);
const s64 offset = s64(data >> 1);
outputOffset2 += (data & 1) ? -offset : +offset;
while (length > 0) {
output[outputOffset++] = output[outputOffset2++];
length -= 1;
}
break;
}
}
}
// Pad rest of the output with 0s
while (outputOffset < outputSize) {
output[outputOffset++] = 0;
}
const u32 inputCRC = BPS::read<u32, 4>(patch, patchOffset, patchSize);
const u32 outputCRC = BPS::read<u32, 4>(patch, patchOffset, patchSize);
const u32 patchCRC = BPS::read<u32, 4>(patch, patchOffset, patchSize);
if (outputCRC != Detail::crc32(output.data(), output.size())) {
return {output, Result::ChecksumMismatch};
}
return {output, Result::Success};
}
static std::pair<std::vector<u8>, Result> patch(const u8* data, usize dataSize, const u8* patch, usize patchSize, PatchType type) {
switch (type) {
case PatchType::IPS: return patchIPS(data, dataSize, patch, patchSize);
case PatchType::UPS: return patchUPS(data, dataSize, patch, patchSize);
case PatchType::BPS: return patchBPS(data, dataSize, patch, patchSize);
default: return {{}, Result::UnknownFormat}; // Unknown patch format
}
}
} // namespace Hips
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment