Treat data files as uint8_t, not char

IffFile.cc

@@ -9,7 +9,7 @@ struct ChunkHeader {
   uint32_t length;
 };
 
-IffFile::IffFile(char const* base, size_t length)
+IffFile::IffFile(uint8_t const* base, size_t length)
   : base_(base), length_(length)
 {
   size_t pos = 0;
@@ -50,7 +50,7 @@ void IffFile::printStructure(unsigned level) const
 }
 
 
-std::unique_ptr<IffFile::Object> IffFile::parseObject(char const* base, size_t length)
+std::unique_ptr<IffFile::Object> IffFile::parseObject(uint8_t const* base, size_t length)
 {    
   if (length < 8)
     throw FormatException{"length < header size"};
@@ -72,7 +72,7 @@ std::unique_ptr<IffFile::Object> IffFile::parseObject(char const* base, size_t l
     return std::make_unique<Object>(std::string(header.typeID, 4), base+8, static_cast<size_t>(header.length));
 }
 
-IffFile::Object::Object(std::string type, char const* base, size_t length)
+IffFile::Object::Object(std::string type, uint8_t const* base, size_t length)
   : base_(base), length_(length), type_(std::move(type))
 {
 }
@@ -105,20 +105,23 @@ IffFile::Object::operator std::string() const
   }
     
   if (state == State::NULLS)
-    return std::string(base_);
+    return std::string(reinterpret_cast<char const*>(base_));
   else if (state == State::STARTASCII)
-    return std::string(base_, length_);
+    return std::string(reinterpret_cast<char const*>(base_), length_);
   else
     throw FormatException{"BLOB not string"};
 }
 
-IffFile::Form::Form(std::string type, char const* base, size_t length)
+IffFile::Form::Form(std::string type, uint8_t const* base, size_t length)
   : Object(std::move(type), base, length)
 {
   if (length < 4)
     throw FormatException{"length < subtype id length"};
 
-  subtype_ = std::string(base, 4);
+  for (unsigned i = 0;i < 4;++i)
+    if (!isprint(base[i]))
+      throw FormatException{"Subtype not printable"};
+  subtype_ = std::string(reinterpret_cast<char const*>(base), 4);
 
   size_t pos = 4;
   while (pos+8 < length) {

IffFile.hh

@@ -9,7 +9,7 @@
 
 class IffFile {
 public:
-  IffFile(char const* base, size_t length);
+  IffFile(uint8_t const* base, size_t length);
 
   ~IffFile();
 
@@ -111,7 +111,7 @@ public:
   
   class Object {
   public:
-    Object(std::string type, char const* base, size_t length);
+    Object(std::string type, uint8_t const* base, size_t length);
     Object(Object const& copy) = delete;
    
     virtual ~Object() {
@@ -129,25 +129,25 @@ public:
       return length_;
     }
 
-    char const* begin() const {
+    uint8_t const* begin() const {
       return base_;
     }
     
-    char const* end() const {
+    uint8_t const* end() const {
       return base_+length_;
     }
 
     operator std::string() const;
     
   protected:
-    char const* base_;
+    uint8_t const* base_;
     const size_t length_;
     std::string const type_;
   };
   
   class Form final : public Object {
   public:
-    Form(std::string type, char const* base, size_t length);
+    Form(std::string type, uint8_t const* base, size_t length);
     
     ~Form() {}
     
@@ -186,9 +186,9 @@ public:
   void printStructure(unsigned level = 0) const;
   
 private:
-  static std::unique_ptr<Object> parseObject(char const* base, size_t length);
+  static std::unique_ptr<Object> parseObject(uint8_t const* base, size_t length);
   
-  char const* base_;
+  uint8_t const* base_;
   const size_t length_;
   std::vector<std::unique_ptr<Object> > roots_;
 };

MveDecoder.cc

@@ -7,7 +7,7 @@
 #include "decompress.hh"
 #include "MveDecoder.hh"
 
-MveDecoder::MveDecoder(char const* base, size_t length)
+MveDecoder::MveDecoder(uint8_t const* base, size_t length)
   : iff_(base, length), width_(320), height_(165)
 {
   //iff_.printStructure();
@@ -130,7 +130,7 @@ MveDecoder::MveDecoder(char const* base, size_t length)
   }
 }
 
-std::vector<char> parseHuff_(char const* data, size_t len)
+std::vector<char> parseHuff_(uint8_t const* data, size_t len)
 {
   uint8_t numHuffVals;
   memcpy(&numHuffVals, data, 1);
@@ -196,7 +196,7 @@ std::vector<char> parseHuff_(char const* data, size_t len)
   return commands;
 }
 
-std::vector<uint8_t> parsePixels_(char const* data, size_t len)
+std::vector<uint8_t> parsePixels_(uint8_t const* data, size_t len)
 {
   if (*data == 0x02) {
     return decompressLZ(data+1, len-1);

MveDecoder.hh

@@ -8,7 +8,7 @@
 
 class MveDecoder {
 public:
-  MveDecoder(char const* base, size_t length);
+  MveDecoder(uint8_t const* base, size_t length);
 
   size_t numBranches() const {
     return branches_.size();

TreFile.cc

@@ -20,7 +20,7 @@ static const size_t headerSize = 24;
 static const size_t table1EntrySize = 8;
 static const size_t table3EntrySize = 8;
 
-TreFile::TreFile(char const* base, size_t length)
+TreFile::TreFile(uint8_t const* base, size_t length)
   : base_(base), length_(length)
 {
   if (length_ < 6*4)
@@ -66,7 +66,7 @@ TreFile::TreFile(char const* base, size_t length)
 
   // Read table 3
   for (size_t i = 0;i < numTable3Entries;++i) {
-    char const* const entryBase = base_+header.table3Ofs+i*table3EntrySize;
+    uint8_t const* const entryBase = base_+header.table3Ofs+i*table3EntrySize;
     
     const uint32_t dataPtr = readU32LE(entryBase);
     const uint32_t length = readU32LE(entryBase+4)&0x0fffffffu;
@@ -119,7 +119,11 @@ TreFile::TreFile(char const* base, size_t length)
     if (pos+nameLen+5 > header.table3Ofs)
       throw FormatException{"Table 2 entry exceeds table " + std::to_string(nameLen)};
 
-    std::string nameStr(base+pos+1, nameLen);
+    for (unsigned i = 0;i < nameLen;++i)
+      if (!isprint(base[pos+1+i]))
+	throw FormatException{"Filename not printable"};
+    
+    std::string nameStr(reinterpret_cast<char const*>(base)+pos+1, nameLen);
 
     const uint32_t table3Ptr = readU32LE(base+pos+nameLen+1);
     
@@ -146,7 +150,7 @@ TreFile::TreFile(char const* base, size_t length)
 
   // Read Table 1
   for (size_t i = 0;i < numTable1Entries;++i) {
-    char const* const entryBase = base+header.table1Ofs+i*table1EntrySize;
+    uint8_t const* const entryBase = base+header.table1Ofs+i*table1EntrySize;
 
     const uint32_t crc = readU32LE(entryBase);
     uint32_t table3Ptr = readU32LE(entryBase+4);

TreFile.hh

@@ -10,7 +10,7 @@
 
 class TreFile {
 public:
-  TreFile(char const* base, size_t length);
+  TreFile(uint8_t const* base, size_t length);
  
   ~TreFile();
 
@@ -47,9 +47,9 @@ public:
       return *this;
     }
     
-    char const* data() const {
+    uint8_t const* data() const {
       if (data_)
-	return (char const*)(data_->data());
+	return data_->data();
       else
 	return base_;
     }
@@ -66,10 +66,10 @@ public:
     // Object either owns data (when decompressed) ...
     std::experimental::optional<std::vector<uint8_t> > data_;
     // .. or points to area in underlying data (when not compressed)
-    char const* base_;
+    uint8_t const* base_;
     size_t length_;
 
-    Object(char const* base, size_t length)
+    Object(uint8_t const* base, size_t length)
       : data_(), base_(base), length_(length) {
     }
 
@@ -112,7 +112,7 @@ private:
   // base, size, comp. size, flags
   std::vector<std::tuple<uint32_t, uint32_t, uint32_t, uint8_t> > table3_;
 
-  char const* base_;
+  uint8_t const* base_;
   size_t length_;
 };
 

decompress.cc

@@ -2,7 +2,7 @@
 
 #include "decompress.hh"
 
-std::vector<uint8_t> decompressLZ(char const* data, size_t len)
+std::vector<uint8_t> decompressLZ(uint8_t const* data, size_t len)
 {
   std::vector<uint8_t> ret;
   

decompress.hh

@@ -9,6 +9,6 @@
 
 /* LZ77-like format consisting of codes specifing copies from the input stream
    and/or replication of previously output data */
-std::vector<uint8_t> decompressLZ(char const* data, size_t len);
+std::vector<uint8_t> decompressLZ(uint8_t const* data, size_t len);
 
 #endif

util.cc

@@ -74,11 +74,11 @@ size_t MmapFile::size() const
   return size_;
 }
 
-char const* MmapFile::data() const
+uint8_t const* MmapFile::data() const
 {
   if (!*this)
     throw std::logic_error("data() called on invalid MmapFile");
-  return static_cast<char const*>(base_);
+  return static_cast<uint8_t const*>(base_);
 }
 
 MmapFile::operator bool() const
@@ -87,7 +87,7 @@ MmapFile::operator bool() const
 }
 
 
-uint16_t readU16BE(char const* data)
+uint16_t readU16BE(uint8_t const* data)
 {
   if(!data)
     throw std::logic_error("readU16BE() called with nullptr");
@@ -96,7 +96,7 @@ uint16_t readU16BE(char const* data)
     static_cast<uint8_t>(*(data+1));
 }
   
-uint32_t readU24BE(char const* data)
+uint32_t readU24BE(uint8_t const* data)
 {
   if(!data)
     throw std::logic_error("readU24BE() called with nullptr");
@@ -106,7 +106,7 @@ uint32_t readU24BE(char const* data)
     static_cast<uint8_t>(*(data+2));
 }
 
-uint32_t readU32BE(char const* data)
+uint32_t readU32BE(uint8_t const* data)
 {
   if(!data)
     throw std::logic_error("readU32BE() called with nullptr");
@@ -117,7 +117,7 @@ uint32_t readU32BE(char const* data)
     static_cast<uint8_t>(*(data+3));
 }
 
-uint16_t readU16LE(char const* data)
+uint16_t readU16LE(uint8_t const* data)
 {
   if(!data)
     throw std::logic_error("readU16LE() called with nullptr");
@@ -126,7 +126,7 @@ uint16_t readU16LE(char const* data)
     static_cast<uint8_t>(*data);
 }
 
-uint32_t readU24LE(char const* data)
+uint32_t readU24LE(uint8_t const* data)
 {
   if(!data)
     throw std::logic_error("readU24LE() called with nullptr");
@@ -136,7 +136,7 @@ uint32_t readU24LE(char const* data)
     static_cast<uint8_t>(*data);
 }
 
-uint32_t readU32LE(char const* data)
+uint32_t readU32LE(uint8_t const* data)
 {
   if(!data)
     throw std::logic_error("readU32LE() called with nullptr");

util.hh

@@ -19,7 +19,7 @@ public:
 
   std::string const& name() const;
   size_t size() const;
-  char const* data() const;
+  uint8_t const* data() const;
 
   operator bool() const;
 private:
@@ -30,14 +30,14 @@ private:
 };
 
 // Helper functions to read big endian numbers
-uint16_t readU16BE(char const* data);
+uint16_t readU16BE(uint8_t const* data);
-uint32_t readU24BE(char const* data);
+uint32_t readU24BE(uint8_t const* data);
-uint32_t readU32BE(char const* data);
+uint32_t readU32BE(uint8_t const* data);
 
 // Helper functions to read little endian numbers
-uint16_t readU16LE(char const* data);
+uint16_t readU16LE(uint8_t const* data);
-uint32_t readU24LE(char const* data);
+uint32_t readU24LE(uint8_t const* data);
-uint32_t readU32LE(char const* data);
+uint32_t readU32LE(uint8_t const* data);
 
 // Sign-extend b starting at msb
 int sextend(unsigned b, unsigned msb);