| 1 | #include <iostream> |
|---|---|
| 2 | #include <iomanip> |
| 3 | |
| 4 | #if __SSE2__ |
| 5 | #include <emmintrin.h> |
| 6 | #endif |
| 7 | |
| 8 | #include <common/Types.h> |
| 9 | #include <IO/ReadHelpers.h> |
| 10 | #include <IO/ReadBufferFromFileDescriptor.h> |
| 11 | |
| 12 | /** Counts number of 0 in a file. |
| 13 | * Outputs "fullness" of file - ration of non-0 bytes to the expected number of non-0 bytes in file with random bytes. |
| 14 | */ |
| 15 | |
| 16 | int main(int argc, char ** argv) |
| 17 | { |
| 18 | #if __SSE2__ |
| 19 | try |
| 20 | { |
| 21 | DB::ReadBufferFromFileDescriptor in(STDIN_FILENO); |
| 22 | size_t zeros = 0; |
| 23 | UInt64 limit = 0; |
| 24 | |
| 25 | if (argc == 2) |
| 26 | limit = DB::parse<UInt64>(argv[1]); |
| 27 | |
| 28 | while (!in.eof()) |
| 29 | { |
| 30 | const __m128i zero = {0}; |
| 31 | for (; in.position() + 15 < in.buffer().end(); in.position() += 16) |
| 32 | { |
| 33 | __m128i bytes = *reinterpret_cast<const __m128i *>(in.position()); |
| 34 | __m128i byte_mask = _mm_cmpeq_epi8(bytes, zero); |
| 35 | UInt16 bit_mask = _mm_movemask_epi8(byte_mask); |
| 36 | zeros += __builtin_popcount(bit_mask); |
| 37 | } |
| 38 | |
| 39 | for (; in.position() < in.buffer().end(); ++in.position()) |
| 40 | if (*in.position() == 0) |
| 41 | ++zeros; |
| 42 | |
| 43 | if (limit && in.count() >= limit) |
| 44 | break; |
| 45 | } |
| 46 | |
| 47 | std::cout << std::fixed |
| 48 | << 1 - std::max(0.0, static_cast<double>(zeros) / in.count() - 1.0 / 256) |
| 49 | << std::endl; |
| 50 | } |
| 51 | catch (const DB::Exception & e) |
| 52 | { |
| 53 | std::cerr << e.what() << ", "<< e.message() << std::endl |
| 54 | << std::endl |
| 55 | << "Stack trace:"<< std::endl |
| 56 | << e.getStackTrace().toString() |
| 57 | << std::endl; |
| 58 | throw; |
| 59 | } |
| 60 | #else |
| 61 | std::cerr << "Only for x86_64 arch "<< std::endl; |
| 62 | #endif |
| 63 | |
| 64 | return 0; |
| 65 | } |
| 66 |