简体繁体 English

具有恒定时间寻找特定字节的压缩算法？

[英]Compression Algorithms with Constant-Time Seek to Specific Byte?

原文 2022-01-10 00:35:51 6 1 compression/ lzo

I'm experimenting with building a data-structure optimized for a very specific use-case.我正在尝试构建针对特定用例优化的数据结构。 Essentially, I am trying to build a compressed bitset of a constant size, and obviously for that use-case, two operations exist: read the value of a bit or write the value of a bit.本质上，我正在尝试构建一个恒定大小的压缩位集，显然对于该用例，存在两个操作：读取位的值或写入位的值。

The best case scenario would be to be able to read a byte and write a byte in-place in constant time, but I can't imagine that it would be possible to write to an arbitrary byte without making changes to the rest of the compressed chunk of memory.最好的情况是能够在恒定时间内读取一个字节并就地写入一个字节，但我无法想象在不更改压缩文件的 rest 的情况下可以写入任意字节memory 块。 However, it might be possible to read an arbitrary byte in an amount of time that tends toward O(1).但是，有可能在接近 O(1) 的时间内读取任意字节。

I have been reading Wikipedia articles, and I'm familiar with LZO, but is there a table somewhere which describes the various features and tradeoffs that various compression systems provide?我一直在阅读 Wikipedia 文章，并且我对 LZO 很熟悉，但是否有一个表格描述了各种压缩系统提供的各种功能和权衡？ I'd like a moderate level of compression, and I'm mainly wanting to optimize around memory holes, eg large gaps of bytes which are zeroes.我想要中等程度的压缩，我主要想围绕 memory 孔进行优化，例如零字节的大间隙。

1 个解决方案

Assuming that you are doing many of these random accesses, you can build an index (once) to a compressed stream to get O(1).假设您正在执行许多这些随机访问，您可以为压缩的 stream 构建索引（一次）以获得 O(1)。 Here is an example for gzip streams .这是gzip 流的示例。