stringmatch is a small, lightweight string matching library written in Python, based on the Levenshtein distance.
Inspired by libraries like seatgeek/thefuzz, which did not quite fit my needs. And so I am building this library for myself, primarily.
Disclaimer: This library is still in an alpha development phase! Changes may be frequent and breaking changes can occur! It is recommended to update frequently to minimise bugs and maximise features.
- 🎯 Key Features
- 📋 Requirements
- ⚙️ Installation
- 🔨 Basic Usage
- 🛠️ Advanced Usage
- 🌟 Contributing
- 🔗 Links
⚠️ License
This library matches compares and strings to each other based mainly on, among others, the Levenshtein distance.
What makes stringmatch special compared to other libraries with similar functions:
- 💨 Lightweight, straightforward and easy to use
- ⚡ High speed - at least ~12x faster than thefuzz and up to 70x
- 🧰 Allows for highly customisable searches, that yield better results
- 📚 Lots of utility functions to make your life easier
- 📝 Statically typed with mypy, compiled with mypyc
- 🌍 Handles special unicode characters, like emojis or characters from other languages, like ジャパニーズ
- Python 3.9 or later.
- The packages in
requirements.txt, pip will handle these for you.
Install the latest stable version with pip:
pip install -U stringmatch
Or install the newest version via git (Might be unstable or unfinished):
pip install -U git+https://github.com/atomflunder/stringmatch
Below are some basic examples on how to use this library.
For a more detailed explanation head over to the Documentation.
For examples on how to use this library, head over to the examples directory.
The match functions allow you to compare 2 strings and check if they are "similar enough" to each other, or get the best match(es) from a list of strings:
from stringmatch import Match
match = Match()
# Checks if the strings are similar:
match.match("stringmatch", "strngmach") # returns True
match.match("stringmatch", "something else") # returns False
# Returns the best match(es) found in the list:
searches = ["stringmat", "strinma", "strings", "mtch", "whatever", "s"]
match.get_best_match("stringmatch", searches) # returns "stringmat"
match.get_best_matches("stringmatch", searches) # returns ["stringmat", "strinma"]The "ratio of similarity" describes how similar the strings are to each other. It ranges from 100 being an exact match to 0 being something completely different.
You can get the ratio between strings like this:
from stringmatch import Ratio
ratio = Ratio()
# Getting the ratio between the two strings:
ratio.ratio("stringmatch", "stringmatch") # returns 100
ratio.ratio("stringmatch", "strngmach") # returns 90
ratio.ratio("stringmatch", "eh") # returns 15
# Getting the ratio between the first string and the list of strings at once:
searches = ["stringmatch", "strngmach", "eh"]
ratio.ratio_list("stringmatch", searches) # returns [100, 90, 15]
# Searching for partial ratios with substrings:
ratio.partial_ratio("a string", "a string longer") # returns 80You can also get both the match and the ratio together in a tuple using these functions:
from stringmatch import Match
match = Match()
match.match_with_ratio("stringmatch", "strngmach") # returns (True, 90)
searches = ["test", "nope", "tset"]
match.get_best_match_with_ratio("test", searches) # returns ("test", 100)
match.get_best_matches_with_ratio("test", searches) # returns [("test", 100), ("tset", 75)]Instead of the ratio, you can also get the Levenshtein distance between strings directly. The bigger the distance, the more different the strings:
from stringmatch import Distance
distance = Distance()
distance.distance("kitten", "sitting") # returns 3
searches = ["sitting", "kitten"]
distance.distance_list("kitten", searches) # returns [3, 0]This is primarily meant for internal usage, but you can also use this library to modify strings:
from stringmatch import Strings
strings = Strings()
strings.latinise("Héllö, world!") # returns "Hello, world!"
strings.remove_punctuation("wh'at;, ever") # returns "what ever"
strings.alphanumeric("Héllö, world!") # returns "Hll world"
strings.ignore_case("test test!", lower=False) # returns "TEST TEST!"There are some optional arguments available for a few functions.
| Type | Default | Description | Available for: |
|---|---|---|---|
| Integer | 70 | The score cutoff for matching. If the score is below the threshold it will not get returned. | All functions from the Match() class. |
# Example:
from stringmatch import Match
match = Match()
match.match("stringmatch", "strngmach", score=95) # returns False
match.match("stringmatch", "strngmach", score=70) # returns True| Type | Default | Description | Available for: |
|---|---|---|---|
| Integer | 5 | The limit of how many matches to return. If you want to return every match set this to 0 or None. | get_best_matches(), get_best_matches_with_ratio() |
# Example:
from stringmatch import Match
match = Match()
searches = ["limit 5", "limit 4", "limit 3", "limit 2", "limit 1", "limit 0", "something else"]
# returns ["limit 5", "limit 4"]
match.get_best_matches("limit 5", searches, limit=2)
# returns ["limit 5"]
match.get_best_matches("limit 5", searches, limit=1)
# returns ["limit 5", "limit 4", "limit 3", "limit 2", "limit 1", "limit 0"]
match.get_best_matches("limit 5", searches, limit=None) You can also pass in on or more of these optional arguments when initialising the Match() and Ratio() classes to customize your search even further.
Of course you can use multiple of these keyword arguments at once, to customise the search to do exactly what you intend to do.
| Type | Default | Description |
|---|---|---|
| BaseScorer | LevenshteinScorer | Different scoring algorithms to use. The available options are: LevenshteinScorer, JaroScorer, JaroWinklerScorer. |
Click on the links above for detailed information about these, but speaking generally the Jaro Scorer will be the fastest, focussing on the characters the strings have in common.
The Jaro-Winkler Scorer slightly modified the Jaro Scorer to prioritise characters at the start of the string.
The Levenshtein Scorer will, most likely, produce the best results, focussing on the number of edits needed to get from one string to the other.
# Example:
from stringmatch import Match, LevenshteinScorer, JaroWinklerScorer
lev_matcher = Match(scorer=LevenshteinScorer)
lev_matcher.match_with_ratio("test", "th test") # returns (True, 73)
jw_matcher = Match(scorer=JaroWinklerScorer)
jw_matcher.match_with_ratio("test", "th test") # returns (False, 60)| Type | Default | Description |
|---|---|---|
| Boolean | False | Replaces special unicode characters with their latin alphabet equivalents. Examples: Ǽ -> AE, ノース -> nosu |
# Example:
from stringmatch import Match
lat_match = Match(latinise=True)
lat_match.match("séärçh", "search") # returns True
def_match = Match(latinise=False)
def_match.match("séärçh", "search") # returns False| Type | Default | Description |
|---|---|---|
| Boolean | True | If you want to ignore case sensitivity while searching. |
# Example:
from stringmatch import Match
def_match = Match(ignore_case=True)
def_match.match("test", "TEST") # returns True
case_match = Match(ignore_case=False)
case_match.match("test", "TEST") # returns False| Type | Default | Description |
|---|---|---|
| Boolean | False | Removes commonly used punctuation symbols from the strings, like .,;:!? and so on. |
# Example:
from stringmatch import Match
punc_match = Match(remove_punctuation=True)
punc_match.match("test,---....", "test") # returns True
def_match = Match(remove_punctuation=False)
def_match.match("test,---....", "test") # returns False| Type | Default | Description |
|---|---|---|
| Boolean | False | Removes every character that is not a number or in the latin alphabet, a more extreme version of remove_punctuation. |
# Example:
from stringmatch import Match
let_match = Match(alphanumeric=True)
let_match.match("»»ᅳtestᅳ►", "test") # returns True
def_match = Match(alphanumeric=False)
def_match.match("»»ᅳtestᅳ►", "test") # returns False| Type | Default | Description |
|---|---|---|
| Boolean | False | If set to true, also searches for partial substring matches. This may lead to more desirable results but is a bit slower. This will return a score of 65-95 depending on how far apart the sizes of the strings are to ensure only identical matches provide a score of 100. It will start matching at a length of 2, or 1 if it is the first letter of the string. |
# Example:
from stringmatch import Match
part_match = Match(include_partial=True)
# returns (True, 65)
part_match.match_with_ratio("A string", "A string thats like really really long", score=60)
def_match = Match(include_partial=False)
# returns (False, 35)
def_match.match_with_ratio("A string", "A string thats like really really long", score=60)If you are unhappy with the scoring algorithms provided, you can of course construct your own scorer class. Make sure it inherits from BaseScorer and has a score() method that takes 2 strings and returns a float between 0 and 100.
# Example:
from stringmatch import BaseScorer, Match
class MyOwnScorer(BaseScorer):
def score(self, string1: str, string2: str) -> float:
# Highly advanced technology
return 100
my_matcher = Match(scorer=MyOwnScorer)
my_matcher.match_with_ratio("anything", "whatever") # returns (True, 100)Contributions to this library are always appreciated! If you have any sort of feedback, or are interested in contributing, head on over to the Contributing Guidelines.
Additionally, if you like this library, leaving a star and spreading the word would be appreciated a lot!
Thanks in advance for taking the time to do so.
Packages used:
This project is licensed under the MIT License.