HORIZON HASKELLDocslts/ghc-9.10.x248f8f02026-10-05Search names, modules, packages, or :: a typeCtrl K

GHC 9.10.3 · lts/ghc-9.10.x · 248f8f0 · 2026-10-05

Modulestringsearch-0.3.6.6Haskell98

Data.ByteString.Search

Fast overlapping Boyer-Moore search of strict ByteString values. Breaking, splitting and replacing using the Boyer-Moore algorithm.

Descriptions of the algorithm can be found at http://www-igm.univ-mlv.fr/~lecroq/string/node14.html#SECTION00140 and http://en.wikipedia.org/wiki/Boyer-Moore_string_search_algorithm

Original authors: Daniel Fischer (daniel.is.fischer at googlemail.com) and Chris Kuklewicz (haskell at list.mightyreason.com).

  • 8 values

Overview

0 declarations

This module provides functions related to searching a substring within a string, using the Boyer-Moore algorithm with minor modifications to improve the overall performance and avoid the worst case performance degradation of the original Boyer-Moore algorithm for periodic patterns.

When searching a pattern in a UTF-8-encoded ByteString, be aware that these functions work on bytes, not characters, so the indices are byte-offsets, not character offsets.

Performance

In general, the Boyer-Moore algorithm is the most efficient method to search for a pattern inside a string. The advantage over other algorithms (e.g. Naïve, Knuth-Morris-Pratt, Horspool, Sunday) can be made arbitrarily large for specially selected patterns and targets, but usually, it's a factor of 2–3 versus Knuth-Morris-Pratt and of 6–10 versus the naïve algorithm. The Horspool and Sunday algorithms, which are simplified variants of the Boyer-Moore algorithm, typically have performance between Boyer-Moore and Knuth-Morris-Pratt, mostly closer to Boyer-Moore. The advantage of the Boyer-moore variants over other algorithms generally becomes larger for longer patterns. For very short patterns (or patterns with a very short period), other algorithms, e.g. Data.ByteString.Search.DFA can be faster (my tests suggest that "very short" means two, maybe three bytes).

In general, searching in a strict ByteString is slightly faster than searching in a lazy ByteString, but for long targets, the smaller memory footprint of lazy ByteStrings can make searching those (sometimes much) faster. On the other hand, there are cases where searching in a strict target is much faster, even for long targets.

Complexity

Preprocessing the pattern is O(patternLength + σ) in time and space (σ is the alphabet size, 256 here) for all functions. The time complexity of the searching phase for indices is O(targetLength / patternLength) in the best case. For non-periodic patterns, the worst case complexity is O(targetLength), but for periodic patterns, the worst case complexity is O(targetLength * patternLength) for the original Boyer-Moore algorithm.

The searching functions in this module contain a modification which drastically improves the performance for periodic patterns. I believe that for strict target strings, the worst case is now O(targetLength) also for periodic patterns. I may be wrong, though.

The other functions don't have to deal with possible overlapping patterns, hence the worst case complexity for the processing phase is O(targetLength) (respectively O(firstIndex + patternLength) for the breaking functions if the pattern occurs).

Partial application

All functions can usefully be partially applied. Given only a pattern, the pattern is preprocessed only once, allowing efficient re-use.

Finding substrings

2 declarations
valueindices
  1. :: ByteString

    Pattern to find

  2. -> ByteString

    String to search

  3. -> [Int]

    Offsets of matches

#

indices finds the starting indices of all possibly overlapping occurrences of the pattern in the target string. If the pattern is empty, the result is [0 .. length target].

In general, not . null $ indices pat target is a much more efficient version of isInfixOf.

Breaking on substrings

2 declarations
valuebreakOn
  1. :: ByteString

    String to search for

  2. -> ByteString

    String to search in

  3. -> (ByteString, ByteString)

    Head and tail of string broken at substring

#

breakOn pattern target splits target at the first occurrence of pattern. If the pattern does not occur in the target, the second component of the result is empty, otherwise it starts with pattern. If the pattern is empty, the first component is empty.

  uncurry append . breakOn pattern = id
valuebreakAfter
  1. :: ByteString

    String to search for

  2. -> ByteString

    String to search in

  3. -> (ByteString, ByteString)

    Head and tail of string broken after substring

#

breakAfter pattern target splits target behind the first occurrence of pattern. An empty second component means that either the pattern does not occur in the target or the first occurrence of pattern is at the very end of target. To discriminate between those cases, use e.g. isSuffixOf.

  uncurry append . breakAfter pattern = id

Replacing

1 declaration
valuereplace
  1. :: Substitution rep
  2. => ByteString

    Substring to replace

  3. -> rep

    Replacement string

  4. -> ByteString

    String to modify

  5. -> ByteString

    Lazy result

#

replace pat sub text replaces all (non-overlapping) occurrences of pat in text with sub. If occurrences of pat overlap, the first occurrence that does not overlap with a replaced previous occurrence is substituted. Occurrences of pat arising from a substitution will not be substituted. For example:

  replace "ana" "olog" "banana" = "bologna"
  replace "ana" "o" "bananana" = "bono"
  replace "aab" "abaa" "aaabb" = "aabaab"

The result is a lazy ByteString, which is lazily produced, without copying. Equality of pattern and substitution is not checked, but

  (concat . toChunks $ replace pat pat text) == text

holds. If the pattern is empty but not the substitution, the result is equivalent to (were they Strings) cycle sub.

For non-empty pat and sub a strict ByteString,

  fromChunks . intersperse sub . split pat = replace pat sub

and analogous relations hold for other types of sub.

Splitting

3 declarations
valuesplit
  1. :: ByteString

    Pattern to split on

  2. -> ByteString

    String to split

  3. -> [ByteString]

    Fragments of string

#

split pattern target splits target at each (non-overlapping) occurrence of pattern, removing pattern. If pattern is empty, the result is an infinite list of empty ByteStrings, if target is empty but not pattern, the result is an empty list, otherwise the following relations hold:

  concat . intersperse pat . split pat = id,
  length (split pattern target) ==
              length (nonOverlappingIndices pattern target) + 1,

no fragment in the result contains an occurrence of pattern.

valuesplitKeepEnd
  1. :: ByteString

    Pattern to split on

  2. -> ByteString

    String to split

  3. -> [ByteString]

    Fragments of string

#

splitKeepEnd pattern target splits target after each (non-overlapping) occurrence of pattern. If pattern is empty, the result is an infinite list of empty ByteStrings, otherwise the following relations hold:

  concat . splitKeepEnd pattern = id,

all fragments in the result except possibly the last end with pattern, no fragment contains more than one occurrence of pattern.

valuesplitKeepFront
  1. :: ByteString

    Pattern to split on

  2. -> ByteString

    String to split

  3. -> [ByteString]

    Fragments of string

#

splitKeepFront is like splitKeepEnd, except that target is split before each occurrence of pattern and hence all fragments with the possible exception of the first begin with pattern. No fragment contains more than one non-overlapping occurrence of pattern.