Fixed

• add missing symbol to pure Python version which prevented the usage of the fallback implementation

Fixed

• fix version number

Added

• add as_matching_blocks to Editops/Opcodes
• add support for deletions from Editops
• add Editops.apply/Opcodes.apply
• add Editops.remove_subsequence

Changed

• merge adjacent similar blocks in Opcodes

Fixed

• fix usage of eval(repr(Editop)), eval(repr(Editops)), eval(repr(Opcode)) and eval(repr(Opcodes))
• fix opcode conversion for empty source sequence
• fix validation for empty Opcode list passed into Opcodes.__init__

Changed

• added in-tree build backend to install cmake and ninja only when it is not installed yet and only when wheels are available

Changed

• changed internal implementation of cdist to remove build dependency to numpy

Added

• added wheels for musllinux and manylinux ppc64le, s390x

Fixed

• fix missing type stubs

Changed

• change src layout to make package import from root directory possible

Changed

• allow installation without the C++ extension if it fails to compile
• allow selection of implementation via the environment variable RAPIDFUZZ_IMPLEMENTATION which can be set to "cpp" or "python"

Added

• added pure python fallback for all implementations with the following exceptions:
  • no support for sequences of hashables. Only strings supported so far
  • *.editops / *.opcodes functions not implemented yet
  • process.cdist does not support multithreading

Fixed

• fuzz.partial_ratio_alignment ignored the score_cutoff
• fix implementation of Hamming.normalized_similarity
• fix default score_cutoff of Hamming.similarity
• fix implementation of LCSseq.distance when used in the process module
• treat hash for -1 and -2 as different

Fixed

• fix integer wraparound in partial_ratio/partial_ratio_alignment (#231)

Fixed

• fix unlimited recursion in LCSseq when used in combination with the process module

Changed

• add fallback implementations of taskflow, rapidfuzz-cpp and jarowinkler-cpp back to wheel, since some package building systems like piwheels can't clone sources

Changed

• use system version of cmake on arm platforms, since the cmake package fails to compile

Changed

• add tests to sdist
• remove cython dependency for sdist

Changed

• relax version requirements of dependencies to simplify packaging

Fixed

• Do not include installations of jaro_winkler in wheels (regression from 2.0.7)

Changed

• Allow installation from system installed versions of rapidfuzz-cpp, jarowinkler-cpp and taskflow

Added

• Added PyPy3.9 wheels on Linux

Fixed

• Add missing Cython code in sdist
• consider float imprecision in score_cutoff (see #210)

Fixed

• fix incorrect score_cutoff handling in token_set_ratio and token_ratio

Added

• add longest common subsequence

Fixed

• Do not include installations of jaro_winkler and taskflow in wheels

Fixed

• fix incorrect population of sys.modules which lead to submodules overshadowing other imports

Changed

• moved JaroWinkler and Jaro into a separate package

Fixed

• fix signed integer overflow inside hashmap implementation

Fixed

• fix binary size increase due to debug symbols
• fix segmentation fault in Levenshtein.editops

Added

• Added fuzz.partial_ratio_alignment, which returns the result of fuzz.partial_ratio combined with the alignment this result stems from

Fixed

• Fix Indel distance returning incorrect result when using score_cutoff=1, when the strings are not equal. This affected other scorers like fuzz.WRatio, which use the Indel distance as well.

Fixed

• fix type hints
• Add back transpiled cython files to the sdist to simplify builds in package builders like FreeBSD port build or conda-forge

Fixed

• fix type hints
• Indel.normalized_similarity mistakenly used the implementation of Indel.normalized_distance

Added

• added C-Api which can be used to extend RapidFuzz from different Python modules using any programming language which allows the usage of C-Apis (C/C++/Rust)
• added new scorers in rapidfuzz.distance.*
  • port existing distances to this new api
  • add Indel distance along with the corresponding editops function

Changed

• when the result of string_metric.levenshtein or string_metric.hamming is below max they do now return max + 1 instead of -1
• Build system moved from setuptools to scikit-build
• Stop including all modules in __init__.py, since they significantly slowed down import time

Removed

• remove the rapidfuzz.levenshtein module which was deprecated in v1.0.0 and scheduled for removal in v2.0.0
• dropped support for Python2.7 and Python3.5

Deprecated

• deprecate support to specify processor in form of a boolean (will be removed in v3.0.0)
  • new functions will not get support for this in the first place
• deprecate rapidfuzz.string_metric (will be removed in v3.0.0). Similar scorers are available in rapidfuzz.distance.*

Fixed

• process.cdist did raise an exception when used with a pure python scorer

Performance

• improve performance and memory usage of rapidfuzz.string_metric.levenshtein_editops
  • memory usage is reduced by 33%
  • performance is improved by around 10%-20%
• significantly improve performance of rapidfuzz.string_metric.levenshtein for max <= 31 using a banded implementation

Fixed

• fix bug in new editops implementation, causing it to SegFault on some inputs (see https://github.com/qurator-spk/dinglehopper/issues/64)

Fixed

• Fix some issues in the type annotations (see #163)

Performance

• improve performance and memory usage of rapidfuzz.string_metric.levenshtein_editops
  • memory usage is reduced by 10x
  • performance is improved from O(N * M) to O([N / 64] * M)

Added

• Added missing wheels for Python3.6 on MacOs and Windows (see #159)

Added

• Add wheels for Python 3.10 on MacOs

Fixed

• Fix incorrect editops results (See #148)

Changed

• Add Wheels for Python3.10 on all platforms except MacOs (see #141)
• Improve performance of string_metric.jaro_similarity and string_metric.jaro_winkler_similarity for strings with a length <= 64

Fixed

• fixed incorrect results of fuzz.partial_ratio for long needles (see #138)

Changed

• Added typing for process.cdist
• Added multithreading support to cdist using the argument process.cdist
• Add dtype argument to process.cdist to set the dtype of the result numpy array (see #132)
• Use a better hash collision strategy in the internal hashmap, which improves the worst case performance

Changed

• improved performance of fuzz.ratio
• only import process.cdist when numpy is available

Changed

• Add back wheels for Python2.7

Changed

• fuzz.partial_ratio uses a new implementation for short needles (<= 64). This implementation is
  • more accurate than the current implementation (it is guaranteed to find the optimal alignment)
  • it is significantly faster
• Add process.cdist to compare all elements of two lists (see #51)

Fixed

• Fix out of bounds access in levenshtein_editops

Changed

• all scorers do now support similarity/distance calculations between any sequence of hashables. So it is possible to calculate e.g. the WER as:

>>> string_metric.levenshtein(["word1", "word2"], ["word1", "word3"])
1

Added

• Added type stub files for all functions
• added jaro similarity in string_metric.jaro_similarity
• added jaro winkler similarity in string_metric.jaro_winkler_similarity
• added Levenshtein editops in string_metric.levenshtein_editops

Fixed

• Fixed support for set objects in process.extract
• Fixed inconsistent handling of empty strings

Performance

• improved performance of result creation in process.extract

Fixed

• Cython ABI stability issue (#95)
• fix missing decref in case of exceptions in process.extract

Changed

• added processor support to levenshtein and hamming
• added distance support to extract/extractOne/extract_iter

Fixed

• incorrect results of normalized_hamming and normalized_levenshtein when used with utils.default_process as processor

Fixed

• Fix a bug in the mbleven implementation of the uniform Levenshtein distance and cover it with fuzz tests

Fixed

• some of the newly activated warnings caused build failures in the conda-forge build

Fixed

• Fixed issue in LCS calculation for partial_ratio (see #90)
• Fixed incorrect results for normalized_hamming and normalized_levenshtein when the processor utils.default_process is used
• Fix many compiler warnings

Changed

• add wheels for a lot of new platforms
• drop support for Python 2.7

Performance

• use is instead of == to compare functions directly by address

Fixed

• Fix another ref counting issue
• Fix some issues in the Levenshtein distance algorithm (see #92)

Performance

• further improve bitparallel implementation of uniform Levenshtein distance for strings with a length > 64 (in many cases more than 50% faster)

Changed

• add more benchmarks to documentation

Performance

• add bitparallel implementation to InDel Distance (Levenshtein with the weights 1,1,2) for strings with a length > 64
• improve bitparallel implementation of uniform Levenshtein distance for strings with a length > 64
• use the InDel Distance and uniform Levenshtein distance in more cases instead of the generic implementation
• Directly use the Levenshtein implementation in C++ instead of using it through Python in process.*

Fixed

• Fix reference counting in process.extract (see #81)

Fixed

• Fix result conversion in process.extract (see #79)

Changed

• string_metric.normalized_levenshtein supports now all weights
• when different weights are used for Insertion and Deletion the strings are not swapped inside the Levenshtein implementation anymore. So different weights for Insertion and Deletion are now supported.
• replace C++ implementation with a Cython implementation. This has the following advantages:
  • The implementation is less error prone, since a lot of the complex things are done by Cython
  • slighly faster than the current implementation (up to 10% for some parts)
  • about 33% smaller binary size
  • reduced compile time
• Added **kwargs argument to process.extract/extractOne/extract_iter that is passed to the scorer
• Add max argument to hamming distance
• Add support for whole Unicode range to utils.default_process

Performance

• replaced Wagner Fischer usage in the normal Levenshtein distance with a bitparallel implementation

Fixed

• The bitparallel LCS algorithm in fuzz.partial_ratio did not find the longest common substring properly in some cases. The old algorithm is used again until this bug is fixed.

Changed

• string_metric.normalized_levenshtein supports now the weights (1, 1, N) with N >= 1

Performance

• The Levenshtein distance with the weights (1, 1, >2) do now use the same implementation as the weight (1, 1, 2), since Substitution > Insertion + Deletion has no effect

Fixed

• fix uninitialized variable in bitparallel Levenshtein distance with the weight (1, 1, 1)

Changed

• all normalized string_metrics can now be used as scorer for process.extract/extractOne
• Implementation of the C++ Wrapper completely refactored to make it easier to add more scorers, processors and string matching algorithms in the future.
• increased test coverage, that already helped to fix some bugs and help to prevent regressions in the future
• improved docstrings of functions

Performance

• Added bit-parallel implementation of the Levenshtein distance for the weights (1,1,1) and (1,1,2).
• Added specialized implementation of the Levenshtein distance for cases with a small maximum edit distance, that is even faster, than the bit-parallel implementation.
• Improved performance of fuzz.partial_ratio -> Since fuzz.ratio and fuzz.partial_ratio are used in most scorers, this improves the overall performance.
• Improved performance of process.extract and process.extractOne

Deprecated

• the rapidfuzz.levenshtein module is now deprecated and will be removed in v2.0.0 These functions are now placed in rapidfuzz.string_metric. distance, normalized_distance, weighted_distance and weighted_normalized_distance are combined into levenshtein and normalized_levenshtein.

Added

• added normalized version of the hamming distance in string_metric.normalized_hamming
• process.extract_iter as a generator, that yields the similarity of all elements, that have a similarity >= score_cutoff

Fixed

• multiple bugs in extractOne when used with a scorer, that's not from RapidFuzz
• fixed bug in token_ratio
• fixed bug in result normalization causing zero division

Fixed

• utf8 usage in the copyright header caused problems with python2.7 on some platforms (see #70)

Fixed

• when a custom processor like lambda s: s was used with any of the methods inside fuzz.* it always returned a score of 100. This release fixes this and adds a better test coverage to prevent this bug in the future.

Added

• added hamming distance metric in the levenshtein module

Performance

• improved performance of default_process by using lookup table

Fixed

• Add missing virtual destructor that caused a segmentation fault on Mac Os

Added

• C++11 Support
• manylinux