larzfuzzy

Fuzzy string matching in pure Python. Zero dependencies.

Compare strings that aren't exactly equal — edit distance, a 0-100 similarity ratio, Jaro-Winkler (great for names), and, most usefully, pick the best match for a query from a list of choices. The fuzzywuzzy/thefuzz surface, standard library only (no python-Levenshtein C build).

from larzfuzzy import ratio, extract_one, jaro_winkler

ratio("apple", "aple")                                   # 89
extract_one("new yrok", ["New York", "New Jersey", "Newark"])   # ('New York', 88)
jaro_winkler("MARTHA", "MARHTA")                         # 0.961

Why

• The functions you reach for. ratio, partial_ratio, token_sort_ratio, levenshtein, jaro, jaro_winkler, and extract/extract_one.
• Best-match extraction ranks a list of choices against a query — the actual job most fuzzy matching is for (autocomplete, dedupe, "did you mean…").
• Sensible defaults. extract lowercases before scoring (override with a processor), and any scorer can be plugged in.
• Zero dependencies. Pure stdlib — no C extension to compile.

Install

pip install larzfuzzy

Usage

from larzfuzzy import ratio, partial_ratio, token_sort_ratio, extract, extract_one, jaro_winkler

ratio("hello", "hallo")                 # 80
partial_ratio("york", "new york city")  # 100
token_sort_ratio("new york", "york new")# 100

extract("aple", ["apple", "maple", "grape"], limit=2)   # [('apple', 89), ('maple', 67)]
extract_one("marhta", names, scorer=lambda a, b: int(jaro_winkler(a, b) * 100))

Tests

python -m unittest discover -s tests -v   # 13 tests

The Larz stack

One of 30+ pure-Python, zero-dependency libraries at github.com/larz-scripter — pairs with larzsearch.

License

MIT © larz-scripter