If data is larger than the buffer size, then what happens is that the first set of buffered data (2048 bytes) is written over and over again until the entire length of input data has been processed.
For instance, 500KB input data would result in an output file that is around a gigabytes in size.
These changes to class fbuf should fix the issue:
int_type overflow(int_type c = traits_type::eof()) override
{
// Write buffer
if (flush() < 0)
{
return traits_type::eof();
}
setp(buffer, buffer + bufferSize); // reset the put pointers
if (c != traits_type::eof())
{
*pptr() = (char)c; // Add the overflow character to the put buffer
pbump(1); // increment the put pointer by one
}
return c;
}
int sync() override
{
const auto result = flush();
setp(buffer, buffer + bufferSize); // reset the put pointers
return result;
}
int flush() const
{
const auto len = pptr() - pbase();
if (len == 0)
{
return 0; // nothing to do
}
if (PHYSFS_writeBytes(file, pbase(), len) < len)
{
return -1;
}
return 0;
}
...
~fbuf() override
{
flush();
delete[] buffer;
}
If data is larger than the buffer size, then what happens is that the first set of buffered data (2048 bytes) is written over and over again until the entire length of input data has been processed.
For instance, 500KB input data would result in an output file that is around a gigabytes in size.
These changes to
class fbufshould fix the issue: