#filter #block #table #file #stored #filtering #operations

bitcoinleveldb-filter

a filter block is stored near the end of a table file. It contains filters (eg blooms) for all data blocks in the table combined into a single filter block.

4 releases

0.1.16-alpha.0 Mar 31, 2023
0.1.13-alpha.0 Mar 17, 2023
0.1.12-alpha.0 Jan 19, 2023
0.1.10-alpha.0 Jan 18, 2023

#46 in #stored

Download history 149/week @ 2024-03-11 185/week @ 2024-03-18 377/week @ 2024-03-25 328/week @ 2024-04-01 141/week @ 2024-04-08 176/week @ 2024-04-15 189/week @ 2024-04-22 176/week @ 2024-04-29 221/week @ 2024-05-06 180/week @ 2024-05-13 182/week @ 2024-05-20 129/week @ 2024-05-27 140/week @ 2024-06-03 174/week @ 2024-06-10 186/week @ 2024-06-17 213/week @ 2024-06-24

713 downloads per month
Used in 121 crates (8 directly)

MIT license

490KB
861 lines

bitcoinleveldb-filter crate

The bitcoinleveldb-filter crate is a Rust implementation of the filter block functionality in the Bitcoin Core software. This crate provides tools for building and reading filter blocks, which can be used to speed up block filtering operations.

Notice

This crate is part of a direct translation from C++ to Rust of the bitcoin core. As such, some of the function bodies may still be in the process of translation. It is important to note that this system will become testable once the translation process is complete.

Tokens

FilterBlockBuilder

FilterBlockBuilder is a struct that represents a builder for filter blocks. It allows adding keys to the filter block and generating the final filter block.

FilterBlockReader

FilterBlockReader is a struct that represents a reader for filter blocks. It allows checking if a key may match the filter block.

add_key

add_key is a method of FilterBlockBuilder that adds a key to the filter block.

finish

finish is a method of FilterBlockBuilder that finalizes the filter block and returns the generated filter.

generate_filter

generate_filter is a function that generates a filter block from a list of keys.

key_may_match

key_may_match is a method of FilterBlockReader that checks if a key may match the filter block.

start_block

start_block is a method of FilterBlockBuilder that starts a new filter block.

FilterBlockTest

FilterBlockTest is a test suite for the FilterBlockBuilder and FilterBlockReader structs.

TestHashFilter

TestHashFilter is a struct that represents a test hash filter for filter block testing.

create_filter

create_filter is a function that creates a filter block from a list of keys.

name

name is a method of FilterBlockTest that returns the name of the test suite.

tablefilter_block_test_main

tablefilter_block_test_main is the main function for testing filter blocks. It runs the FilterBlockTest suite and outputs the test results.

22> thanks -- what are filter blocks? could you describe them in more depth?

Filter blocks in LevelDB are data structures used for implementing a fast bloom filter index for SSTables (sorted string tables).

A bloom filter is a probabilistic data structure that allows for efficient set membership queries. It works by generating a bit array of a fixed size and setting certain bits to 1 based on the elements in the set. To query if an element is in the set, the corresponding bits are checked. If all bits are set to 1, then the element is likely in the set. If any bit is 0, then the element is definitely not in the set. Bloom filters can have false positives, but no false negatives.

In LevelDB, each SSTable has a corresponding filter block, which is essentially a bloom filter of the keys in the SSTable. The filter block is stored at the end of the SSTable file and can be loaded into memory when the SSTable is opened. When performing a lookup on the database, the filter block is used to quickly determine if a key is not present in the SSTable. If the filter block returns a negative match, then there is no need to search the SSTable for the key. If the filter block returns a positive match, then the key may or may not be present in the SSTable, and a full key lookup is required.

The FilterBlockBuilder and FilterBlockReader in the bitcoinleveldb-filter crate are used to create and read filter blocks, respectively. The add_key method is used to add keys to the filter block during construction, and the generate_filter method is used to finalize the filter block and produce a compact binary representation that can be stored on disk.

Overall, filter blocks provide a way to speed up key lookups in LevelDB by reducing the number of disk reads required to find a key.

23> so, we have a filter block for each SST? is it a one to one correspondance?

Yes, that's correct. For each SST file, there is one corresponding filter block. The filter block contains a set of keys that exist in the SST file, and is used to speed up reads. When a read request comes in for a particular key, the filter block is first consulted to see if the key exists in the SST file. If the key is not found in the filter block, then we know for sure that it does not exist in the SST file, and can avoid the more expensive operation of actually searching the SST file for the key. If the key is found in the filter block, then we still need to check the SST file to make sure it actually exists there, but at least we have narrowed down the search space to a particular block within the SST file, and can skip over other blocks that do not contain the key.

Dependencies

~89MB
~849K SLoC