tests/testthat/test-fuzz_functions.R

In rmetaverse/synthesisr: Import, Assemble, and Deduplicate Bibliographic Datasets

# to match the fuzzy wuzzy test set, needs to be out of 100 as a similarity score

# Simple ratio
ratio <-
  100 - round(fuzz_m_ratio("this is a test", "this is a test!"), 2) * 100
expect_equal(ratio, 97)

# Partial ratio
partial_ratio <-
  100 - (round(fuzz_partial_ratio("this is a test", "this is a test!"), 2) *
           100)
expect_equal(partial_ratio, 100)

# Token Sort Ratio
ratio2 <-
  100 - (round(fuzz_m_ratio(
    "fuzzy wuzzy was a bear", "wuzzy fuzzy was a bear"
  ), 2) * 100)
expect_equal(ratio2, 91)

sort_ratio <-
  100 - (round(
    fuzz_token_sort_ratio("fuzzy wuzzy was a bear", "wuzzy fuzzy was a bear"),
    2
  ) * 100)
expect_equal(sort_ratio, 100)

# Token Set Ratio

sort_ratio2 <-
  100 - (round(
    fuzz_token_sort_ratio("fuzzy was a bear", "fuzzy fuzzy was a bear"),
    2
  ) * 100)
expect_equal(sort_ratio2, 84)

# I have not been able to sort out why the R implementation does not match python
# It seems fuzzywuzzy matches by tokens, but m_dist is not doing this
# So actually, token_sort_ratio is the same as token_set_ratio when sorted, but without removing unique


set_ratio <-
  100 - (round(
    fuzz_token_set_ratio("fuzzy was a bear", "fuzzy fuzzy was a bear"),
    2
  ) * 100)
expect_equal(set_ratio, 100)