Merge pull request #106 from mammothb/feat-change_distance_comparer

Feat: change distance comparer

symspellpy/editdistance.py

@@ -65,7 +65,7 @@ def __init__(self, algorithm: DistanceAlgorithm) -> None:
  elif algorithm == DistanceAlgorithm.DAMERAU_OSA_FAST:
  self._distance_comparer = DamerauOsaFast()
  else:
- raise ValueError("Unknown distance algorithm")
+ raise ValueError("unknown distance algorithm")
 
  def compare(self, string_1: str, string_2: str, max_distance: int) -> int:
  """Compares a string to the base string to determine the edit distance,

symspellpy/helpers.py

@@ -40,8 +40,8 @@ def case_transfer_matching(cased_text: str, uncased_text: str) -> str:
  """
  if len(cased_text) != len(uncased_text):
  raise ValueError(
- "'cased_text' and 'uncased_text' don't have the same length. Use "
- "case_transfer_similar() instead."
+ "'cased_text' and 'uncased_text' don't have the same length, use "
+ "case_transfer_similar() instead"
  )
 
  return "".join(
@@ -84,7 +84,7 @@ def case_transfer_similar(cased_text: str, uncased_text: str) -> str:
  return uncased_text
 
  if not cased_text:
- raise ValueError("'cased_text' cannot be empty!")
+ raise ValueError("'cased_text' cannot be empty")
 
  matcher = SequenceMatcher(a=cased_text.lower(), b=uncased_text)
  result = ""

symspellpy/symspellpy.py

@@ -125,6 +125,19 @@ def deletes(self) -> Dict[str, List[str]]:
  """
  return self._deletes
 
+ @property
+ def distance_algorithm(self) -> DistanceAlgorithm:
+ """The current distance algorithm."""
+ return self._distance_algorithm
+
+ @distance_algorithm.setter
+ def distance_algorithm(self, value: DistanceAlgorithm) -> None:
+ if not isinstance(value, DistanceAlgorithm):
+ raise TypeError(
+ "can only assign DistanceAlgorithm type values to distance_algorithm"
+ )
+ self._distance_algorithm = value
+
  @property
  def entry_count(self) -> int:
  """Number of unique correct spelling words."""
@@ -365,7 +378,7 @@ def lookup(
  if max_edit_distance is None:
  max_edit_distance = self._max_dictionary_edit_distance
  if max_edit_distance > self._max_dictionary_edit_distance:
- raise ValueError("Distance too large")
+ raise ValueError("distance too large")
  suggestions: List[SuggestItem] = []
  phrase_len = len(phrase)
 

tests/test_editdistance.py

@@ -139,7 +139,7 @@ class TestEditDistance:
  def test_unknown_distance_algorithm(self):
  with pytest.raises(ValueError) as excinfo:
  _ = EditDistance(2)
- assert "Unknown distance algorithm" == str(excinfo.value)
+ assert "unknown distance algorithm" == str(excinfo.value)
 
  def test_abstract_distance_comparer(self):
  with pytest.raises(TypeError) as excinfo:

tests/test_helpers.py

@@ -55,8 +55,8 @@ def test_case_transfer_matching_diff_lengths(self):
  with pytest.raises(ValueError) as excinfo:
  case_transfer_matching("abc", "abcd")
  assert (
- "'cased_text' and 'uncased_text' don't have the same length. Use "
- "case_transfer_similar() instead."
+ "'cased_text' and 'uncased_text' don't have the same length, use "
+ "case_transfer_similar() instead"
  ) == str(excinfo.value)
 
  def test_case_transfer_matching(self):
@@ -78,7 +78,7 @@ def test_case_transfer_similar_empty_wo_casing(self):
  def test_case_transfer_similar_empty_w_casing(self):
  with pytest.raises(ValueError) as excinfo:
  case_transfer_similar("", "abcd")
- assert "'cased_text' cannot be empty!" == str(excinfo.value)
+ assert "'cased_text' cannot be empty" == str(excinfo.value)
 
  def test_case_transfer_similar(self, get_similar_texts):
  for cased_text, uncased_text, expected in get_similar_texts:

tests/test_symspellpy.py

@@ -3,6 +3,7 @@
 import pytest
 
 from symspellpy import SymSpell, Verbosity
+from symspellpy.editdistance import DistanceAlgorithm
 from symspellpy.helpers import DictIO
 
 FORTESTS_DIR = Path(__file__).resolve().parent / "fortests"
@@ -61,6 +62,27 @@ def test_negative_count_threshold(self):
  _ = SymSpell(1, 3, -1)
  assert "count_threshold cannot be negative" == str(excinfo.value)
 
+ @pytest.mark.parametrize(
+ "algorithm",
+ [
+ DistanceAlgorithm.LEVENSHTEIN,
+ DistanceAlgorithm.DAMERAU_OSA,
+ DistanceAlgorithm.LEVENSHTEIN_FAST,
+ DistanceAlgorithm.DAMERAU_OSA_FAST,
+ ],
+ )
+ def test_set_distance_algorithm(self, symspell_default, algorithm):
+ symspell_default.distance_algorithm = algorithm
+ assert algorithm == symspell_default.distance_algorithm
+
+ def test_set_invalid_distance_algorithm(self, symspell_default):
+ with pytest.raises(TypeError) as excinfo:
+ symspell_default.distance_algorithm = 1
+ assert (
+ "can only assign DistanceAlgorithm type values to distance_algorithm"
+ == str(excinfo.value)
+ )
+
  @pytest.mark.parametrize("symspell_short", [None, 0], indirect=True)
  def test_create_dictionary_entry_negative_count(self, symspell_short):
  assert (

tests/test_symspellpy_lookup.py

@@ -121,7 +121,7 @@ def test_should_not_return_low_count_word_that_are_also_delete_word(
  def test_max_edit_distance_too_large(self, symspell_high_thres_flame):
  with pytest.raises(ValueError) as excinfo:
  _ = symspell_high_thres_flame.lookup("flam", Verbosity.TOP, 3)
- assert "Distance too large" == str(excinfo.value)
+ assert "distance too large" == str(excinfo.value)
 
  def test_include_unknown(self, symspell_high_thres_flame):
  result = symspell_high_thres_flame.lookup("flam", Verbosity.TOP, 0, True)