DCMTK

Using an int instead of size_t for computing the size of a buffer can make the int overflow and therefore create a much smaller buffer than actually needed.

Fix: Use a size_t cast (not int) for the malloc size so large ValueLength values can no longer truncate/sign-flip into an undersized buffer. Also run memset/memcpy only after the NULL check, avoiding a crash when the allocation fails.

Thanks to Dominik Blain for the report.