wc3re/TreFile.cc

#include <cstdio>
#include <cstdint>

#include "common.hh"

#include "TreFile.hh"

struct TreHeader { // little endian
  char magic[4]; // == "XTRE"
  uint32_t unknown;
  uint32_t table1Ofs;
  uint32_t table2Ofs;
  uint32_t table3Ofs;
  uint32_t dataOfs;
} __attribute__((__packed__));

struct Table1Entry { // little endian
  uint32_t crc;
  uint32_t table3Ptr;
} __attribute__((__packed__));

TreFile::TreFile(char const* base, size_t length)
  : base_(base), length_(length)
{
  if (length_ < sizeof(TreHeader))
    throw FormatException{"Size < header size"};

  TreHeader const* header = reinterpret_cast<TreHeader const*>(base);

  if (memcmp(header->magic, "XTRE", 4) != 0)
    throw FormatException{"Wrong magic, not a TRE?"};

  if ((header->table1Ofs > length_) ||
      (header->table2Ofs > length_) ||
      (header->table3Ofs > length_) ||
      (header->dataOfs > length_))
    throw FormatException{"Table offset exceeds length"};

  if ((header->table1Ofs > header->table2Ofs) ||
      (header->table2Ofs > header->table3Ofs) ||
      (header->table3Ofs > header->dataOfs))
    throw FormatException{"Table offsets not in ascending order"};

  const size_t numTable1Entries = (header->table2Ofs-header->table1Ofs)/sizeof(Table1Entry);
  table3Size_ = (header->dataOfs-header->table3Ofs)/sizeof(Table3Entry);

#ifndef NDEBUG
  printf("Table 1: start %u, length %u (%lu entries)\n",
	 header->table1Ofs, header->table2Ofs-header->table1Ofs,
	 numTable1Entries);
  printf("Table 2: start %u, length %u\n",
	 header->table2Ofs, header->table3Ofs-header->table2Ofs);
  printf("Table 3: start %u, length %u (%lu entries)\n",
	 header->table3Ofs, header->dataOfs-header->table3Ofs,
	 table3Size_);
  printf("Data: start %u, length %lu\n",
	 header->dataOfs, length_ - header->dataOfs);
#endif

  // Check Table 3
  table3_ = reinterpret_cast<Table3Entry const*>(base+header->table3Ofs);
  for (size_t i = 0;i < table3Size_;++i) {
    if ((table3_[i].dataPtr > length_) ||
	(table3_[i].dataPtr < header->dataOfs))
      throw FormatException{"Data pointer out of range"};

    if ((table3_[i].dataPtr + table3_[i].length) > length_) {
      fprintf(stderr, "Data length exceeds file length: %.8x %.8x\n", table3_[i].dataPtr, table3_[i].length);
    }
  }

  // Read Table 2
  std::map<size_t, uint32_t> table2Pos;
    
  size_t pos = header->table2Ofs;
  while((pos+5) < header->table3Ofs) {
    uint8_t const* nameLen = reinterpret_cast<uint8_t const*>(base+pos);

    if (pos+*nameLen+5 > header->table3Ofs)
      throw FormatException{"Table 2 entry exceeds table "s + std::to_string(*nameLen)};

    std::string nameStr(base+pos+1, *nameLen);

    uint32_t const* table3Ptr = reinterpret_cast<uint32_t const*>(base+pos+*nameLen+1);
    
    if ((*table3Ptr < header->table3Ofs) ||
	(*table3Ptr >= header->dataOfs)) {
      throw FormatException{"Table3 pointer out of range"};
    }

    const size_t table3Index = (*table3Ptr - header->table3Ofs)/sizeof(Table3Entry);

    auto ins = table2_.emplace(nameStr, table3Index);
    if (!ins.second) {
      if (ins.first->second == table3Index)
	printf("Duplicate entry for name %s\n", nameStr.c_str());
      else
	printf("Collision for name %s: prev %lu, new %lu\n", nameStr.c_str(),
	       ins.first->second, table3Index);
    }
    // Store an position-indexed copy of Table2 to resolve Table1->Table2 ptr's
    table2Pos.emplace(pos, *table3Ptr);

    pos += 1+*nameLen+4;
  }

  // Read Table 1
  Table1Entry const* table1 = reinterpret_cast<Table1Entry const*>(base+header->table1Ofs);
  for (size_t i = 0;i < numTable1Entries;++i) {
    if ((table1[i].crc == 0x00000000) &&
	(table1[i].table3Ptr == 0xffffffff))
      continue; // Unused entry
      
    if ((table1[i].table3Ptr < header->table2Ofs) ||
	(table1[i].table3Ptr >= header->dataOfs)) {
      throw FormatException{"Table3 pointer out of range"};
    }

    uint32_t table3Ptr = table1[i].table3Ptr;
    if (table3Ptr < header->table3Ofs) {
      auto it = table2Pos.find(table3Ptr);
      if (it == table2Pos.end()) {
	fprintf(stderr, "Table1->Table2 pointer: Table 2 entry not found: crc %.8x ptr %.8x\n",
		table1[i].crc, table3Ptr);
	continue;
      }
      printf("Table1->Table2 pointer resolved: Table 2 %.8x -> Table 3 %.8x\n",
	     table3Ptr, it->second);
      table3Ptr = it->second;
    }

    const size_t table3Index = (table3Ptr - header->table3Ofs)/sizeof(Table3Entry);
    auto ins = table1_.emplace(static_cast<uint32_t>(table1[i].crc), table3Index);
    if (!ins.second) {
      if (ins.first->second == table3Index)
	printf("Duplicate entry for CRC %.8x\n", table1[i].crc);
      else
	printf("Collision for CRC %.8x: prev %lu, new %lu\n", table1[i].crc, ins.first->second, table3Index);
    }
  }
}


TreFile::~TreFile()
{
}

std::vector<std::string> TreFile::getNames() const
{
  std::vector<std::string> ret;
  for (auto ent : table2_) {
    ret.push_back(ent.first);
  }

  return ret;
}

std::vector<uint32_t> TreFile::getCRCs() const
{
  std::vector<uint32_t> ret;
  for (auto ent : table1_) {
    ret.push_back(ent.first);
  }

  return ret;
}

TreFile::Object TreFile::openName(std::string const& name) const
{
  return openIdx_(findName_(name));
}

TreFile::Object TreFile::openCRC(uint32_t crc) const
{
  return openIdx_(findCRC_(crc));
}
  
TreFile::Stat TreFile::statName(std::string const& name) const
{
  return statIdx_(findName_(name));
}

TreFile::Stat TreFile::statCRC(uint32_t crc) const
{
  return statIdx_(findCRC_(crc));
}

void TreFile::dumpName(std::string path, std::string const& name) const
{
  auto idx = findName_(name);

  if (path.back() != '/')
    path.push_back('/');
  path.append(name);
  
  dumpIdx_(path, idx);
}

void TreFile::dumpCRC(std::string path, uint32_t crc) const
{
  auto idx = findCRC_(crc);
  char crcStr[9];
  snprintf(crcStr, 9, "%.8X", crc);
  
  if (path.back() != '/')
    path.push_back('/');
  path.append(crcStr);
  
  dumpIdx_(path, idx);
}

void TreFile::dumpAll(std::string path) const
{
  if (path.back() != '/')
    path.push_back('/');
  
  for (auto ent : table1_) {
    char crcStr[9];
    snprintf(crcStr, 9, "%.8X", ent.first);
    std::string name = path;
    name.append(crcStr);

    dumpIdx_(name, ent.second);
  }

  for (auto ent : table2_) {
    std::string name = path;
    name.append(ent.first);

    dumpIdx_(name, ent.second);
  } 
}
  
void TreFile::printStructure()
{
  printf("Table 1: %lu entries\n", table1_.size());
  printf("Table 2: %lu entries\n", table2_.size());
  printf("Table 3: %lu entries\n", table3Size_);

  printf("Files by CRC:\n");
  for(auto const& ent : table1_) {
    printf("\t%.8x -> (ofs %.8x, len %.8x, flags %.2hhx)\n", ent.first,
	   table3_[ent.second].dataPtr,
	   table3_[ent.second].length,
	   table3_[ent.second].flags);
  }

  printf("Files by Name:\n");
  for(auto const& ent : table2_) {
    printf("\t%s -> (ofs %.8x, len %.8x, flags %.2hhx)\n", ent.first.c_str(),
	   table3_[ent.second].dataPtr,
	   table3_[ent.second].length,
	   table3_[ent.second].flags);
  }
}

size_t TreFile::findName_(std::string const& name) const
{
  auto it = table2_.find(name);
  if (it == table2_.end())
    throw Exception{name + " not found"};

  return it->second;
}

size_t TreFile::findCRC_(uint32_t crc) const
{
  auto it = table1_.find(crc);
  if (it == table1_.end()) {
    char crcStr[9];
    snprintf(crcStr, 9, "%.8X", crc);
    throw Exception{"CRC "s + crcStr + " not found"s};
  }

  return it->second;
}

void TreFile::dumpIdx_(std::string const& name, size_t table3Idx) const
{
  if(table3_[table3Idx].flags&0x80)
    throw Exception{"Compressed TRE objects NYI"};

  FILEUPtr outFile{fopen(name.c_str(), "wb")};
  if (!outFile)
    throw POSIXException{errno, "Could not open " + name};

  if (fwrite(base_+table3_[table3Idx].dataPtr, table3_[table3Idx].length, 1, outFile.get()) != 1)
    throw POSIXException{errno, "Could not write"};
}

TreFile::Object TreFile::openIdx_(size_t table3Idx) const
{
  if(table3_[table3Idx].flags&0x80)
    throw Exception{"Compressed TRE objects NYI"};

  if ((table3_[table3Idx].dataPtr + table3_[table3Idx].length) > length_)
    throw FormatException{"length exceeds file size"};
  
  return Object(base_+table3_[table3Idx].dataPtr,
	      table3_[table3Idx].length);
}

TreFile::Stat TreFile::statIdx_(size_t table3Idx) const
{
  Stat ret;
  ret.size = table3_[table3Idx].length;
  ret.flags = table3_[table3Idx].flags;
				       

  return ret;
}