Question

我正在尝试制作文件比较程序，我想要实现的功能之一是计算所选两个文件的相似性和差异性。我希望这个比较在大文件上快速（如果可能）。我不确定应该使用什么方法，但最后我想要一个百分比。

请参阅this gif以获得直观的想法。

Answer 1

你可能想要像二进制diff实用程序所看到的类似的东西 - 而不是逐字节的比较。但是，嘿，只是为了好玩......

unsafe static long DumbDifference(string file1Path, string file2Path)
{
    // completely untested! also, add some using()s here.
    // also, map views in chunks if you plan to use it on large files.

    MemoryMappedFile file1 = MemoryMappedFile.CreateFromFile(
             file1Path, System.IO.FileMode.Open,
             null, 0, MemoryMappedFileAccess.Read);
    MemoryMappedFile file2 = MemoryMappedFile.CreateFromFile(
             file2Path, System.IO.FileMode.Open,
             null, 0, MemoryMappedFileAccess.Read);
    MemoryMappedViewAccessor view1 = file1.CreateViewAccessor();
    MemoryMappedViewAccessor view2 = file2.CreateViewAccessor();

    long length1 = checked((long)view1.SafeMemoryMappedViewHandle.ByteLength);
    long length2 = checked((long)view2.SafeMemoryMappedViewHandle.ByteLength);
    long minLength = Math.Min(length1, length2);

    byte* ptr1 = null, ptr2 = null;
    view1.SafeMemoryMappedViewHandle.AcquirePointer(ref ptr1);
    view2.SafeMemoryMappedViewHandle.AcquirePointer(ref ptr2);

    ulong differences = (ulong)Math.Abs(length1 - length2);

    for (long i = 0; i < minLength; ++i)
    {
        // if you expect your files to be pretty similar,
        // you could optimize this by comparing long-sized chunks.
        differences += ptr1[i] != ptr2[i] ? 1u : 0u;
    }

    return checked((long)differences);
}

太糟糕了.NET没有内置的SIMD支持。

Answer 2

如果你可以使用linq这应该没问题。

var results = your1stEnumerable.Intersect(your2ndEnumerable);

.NET最快的方法来查找两个文件的相似性和差异性？

2 个答案: