bsearch

bsearch is a go library providing binary search functionality for line-ordered byte streams (such as LC_ALL=C sorted text files). It allows very fast lookups by key, in the spirit of the venerable look(1) unix utility. A line's key is its first field, and keys match as a whole: searching for foo finds foo,1 but not foobar,1.

bsearch can also be used via a DB object with a simple db-like interface, allowing sorted CSV files to be used as a key-value store, with excellent performance.

bsearch currently only supports bytewise key comparisons (not UTF-8 collations). This restriction may be removed in the future.

Usage

    import "github.com/ProfoundNetworks/bsearch"

    // Create the index (also see cmd/bsearch_index for a CLI tool)
    idx, err := bsearch.NewIndex(filepath)

    // Alternatively, if you need to tweak the options
    idx, err := bsearch.NewIndexOptions(filepath, bsearch.IndexOptions{Delimiter: []byte("|")})

    // Persist the created index to disk, so that the searcher may use it
    err = idx.Write()

    // Instantiate searcher from a file
    bss, err := bsearch.NewSearcher(filepath)
    defer bss.Close()

    // Find first line whose key is searchStr
    line, err := bss.Line([]byte(searchStr))

    // Distinguish not found from other errors
    if err != nil && err == bsearch.ErrNotFound {
        // do something
    } else if err != nil {
        log.Fatal(err)
    }


    // Or via the `DB` interface
    db, err := bsearch.NewDB(filepath)
    defer db.Close()

    // Lookup values via byteslices or strings
    valbytes, err := db.Get(keybytes)
    valstr, err := db.GetString(keystring)

Status

bsearch is alpha software. Interfaces may break and change until an official version 1.0.0 is released.

Copyright and Licence

Copyright 2020 Profound Networks LLC.

This project is licensed under the terms of the MIT license.