Merge pull request #83 from yutanagano/move-to-logging

Migrate from warnings to logging

src/tidytcells/_query_engine/hla_query_engine.py

@@ -1,15 +1,17 @@
+import logging
 from typing import FrozenSet
-import warnings
-
 from tidytcells._query_engine import QueryEngine
 from tidytcells._resources import VALID_HOMOSAPIENS_MH
 
 
+logger = logging.getLogger(__name__)
+
+
 class HlaQueryEngine(QueryEngine):
     @classmethod
     def query(cls, precision: str, functionality: str) -> FrozenSet[str]:
         if precision == "allele":
-            warnings.warn(
+            logger.warning(
                 "tidytcells is not fully aware of all HLA alleles, and the "
                 "highest resolution it can provide is up to the level of the "
                 "protein (two allele designations)."

src/tidytcells/_query_engine/mus_musculus_mh_query_engine.py

@@ -1,15 +1,17 @@
+import logging
 from typing import FrozenSet
-import warnings
-
 from tidytcells._query_engine import QueryEngine
 from tidytcells._resources import VALID_MUSMUSCULUS_MH
 
 
+logger = logging.getLogger(__name__)
+
+
 class MusMusculusMhQueryEngine(QueryEngine):
     @classmethod
     def query(cls, precision: str, functionality: str) -> FrozenSet[str]:
         if precision == "allele":
-            warnings.warn(
+            logger.warning(
                 "tidytcells is not aware of Mus musculus MH alleles at all, "
                 "and can only provide up to the level of the gene."
             )

src/tidytcells/_standardized_gene_symbol/standardized_hla_symbol.py

@@ -137,7 +137,7 @@ def get_reason_why_invalid(self, enforce_functional: bool = False) -> Optional[s
             return None
 
         if not self._gene_name in VALID_HOMOSAPIENS_MH:
-            return "unrecognised gene name"
+            return "unrecognized gene name"
 
         # Verify allele designators up to the level of the protein (or G/P)
         allele_designation = self._allele_designation.copy()
@@ -149,7 +149,7 @@ def get_reason_why_invalid(self, enforce_functional: bool = False) -> Optional[s
             try:
                 current_root = current_root[allele_designation.pop(0)]
             except KeyError:
-                return "nonexistent allele for recognised gene"
+                return "nonexistent allele for recognized gene"
 
         # If there are designator fields past the protein level, just make sure
         # they look like legitimate designator field values

src/tidytcells/_standardized_gene_symbol/standardized_mus_musculus_mh_symbol.py

@@ -50,7 +50,7 @@ def _is_synonym(self) -> bool:
 
     def get_reason_why_invalid(self, enforce_functional: bool = False) -> Optional[str]:
         if not self._gene_name in VALID_MUSMUSCULUS_MH:
-            return "unrecognised gene name"
+            return "unrecognized gene name"
 
         return None
 

src/tidytcells/_standardized_gene_symbol/standardized_tr_symbol.py

@@ -126,15 +126,15 @@ def _try_adding_or_removing_dash1_to_gene_name(self) -> None:
 
     def get_reason_why_invalid(self, enforce_functional: bool = False) -> Optional[str]:
         if not self._gene_name in self._valid_tr_dictionary:
-            return "unrecognised gene name"
+            return "unrecognized gene name"
 
         if self._allele_designation:
             allele_valid = (
                 self._allele_designation in self._valid_tr_dictionary[self._gene_name]
             )
 
             if not allele_valid:
-                return "nonexistent allele for recognised gene"
+                return "nonexistent allele for recognized gene"
 
             if (
                 enforce_functional

src/tidytcells/_utils/warnings.py

@@ -1,16 +1,22 @@
-from warnings import warn
+from logging import Logger
 
 
 def warn_failure(
-    reason_for_failure: str, original_input: str, attempted_fix: str, species: str
+    reason_for_failure: str,
+    original_input: str,
+    attempted_fix: str,
+    species: str,
+    logger: Logger,
 ):
     warning_message = f'Failed to standardize "{original_input}" for species {species}: {reason_for_failure}.'
 
     if original_input != attempted_fix:
         warning_message += f' (best attempted fix: "{attempted_fix}").'
 
-    warn(warning_message)
+    logger.warning(warning_message)
 
 
-def warn_unsupported_species(species: str, gene_type: str):
-    warn(f'Unsupported species: "{species}". ' f"Skipping {gene_type} standardisation.")
+def warn_unsupported_species(species: str, gene_type: str, logger: Logger):
+    logger.warning(
+        f'Unsupported species: "{species}". ' f"Skipping {gene_type} standardisation."
+    )

src/tidytcells/aa/_standardize.py

@@ -1,9 +1,11 @@
-import warnings
-
+import logging
 from tidytcells._resources import AMINO_ACIDS
 from tidytcells._utils import Parameter
 
 
+logger = logging.getLogger(__name__)
+
+
 def standardize(seq: str, on_fail: str = "reject", suppress_warnings: bool = False):
     """
     Ensures that a string value looks like a valid amino acid sequence.
@@ -77,7 +79,7 @@ def standardize(seq: str, on_fail: str = "reject", suppress_warnings: bool = Fal
     for char in seq:
         if not char in AMINO_ACIDS:
             if not suppress_warnings:
-                warnings.warn(
+                logger.warning(
                     f"Failed to standardize {original_input}: not a valid amino acid sequence."
                 )
             if on_fail == "reject":

src/tidytcells/junction/_standardize.py

@@ -1,9 +1,11 @@
+import logging
 import re
-import warnings
-
 from tidytcells import aa
 
 
+logger = logging.getLogger(__name__)
+
+
 JUNCTION_MATCHING_REGEX = re.compile(f"^C[A-Z]*[FW]$")
 
 
@@ -112,7 +114,7 @@ def standardize(
     if not JUNCTION_MATCHING_REGEX.match(seq):
         if strict:
             if not suppress_warnings:
-                warnings.warn(
+                logger.warning(
                     f"Failed to standardize {original_input}: not a valid junction sequence."
                 )
             if on_fail == "reject":

src/tidytcells/mh/_get_chain.py

@@ -1,11 +1,13 @@
+import logging
 import re
 from typing import Optional
-import warnings
-
 from tidytcells._resources import VALID_HOMOSAPIENS_MH
 from tidytcells._utils import Parameter
 
 
+logger = logging.getLogger(__name__)
+
+
 ALPHA_MATCHING_REGEX = re.compile(r"HLA-([ABCEFG]|D[PQR]A)")
 BETA_MATCHING_REGEX = re.compile(r"HLA-D[PQR]B|B2M")
 
@@ -51,7 +53,7 @@ def get_chain(
 
     if not gene in (*VALID_HOMOSAPIENS_MH, "B2M"):
         if not suppress_warnings:
-            warnings.warn(f"Unrecognised gene {gene}. Is this standardized?")
+            logger.warning(f"Unrecognized gene {gene}. Is this standardized?")
         return None
 
     if ALPHA_MATCHING_REGEX.match(gene):
@@ -61,5 +63,5 @@ def get_chain(
         return "beta"
 
     if not suppress_warnings:
-        warnings.warn(f"Chain for {gene} unknown.")
+        logger.warning(f"Chain for {gene} unknown.")
     return None

src/tidytcells/mh/_get_class.py

@@ -1,11 +1,13 @@
+import logging
 import re
 from typing import Optional
-import warnings
-
 from tidytcells._resources import VALID_HOMOSAPIENS_MH
 from tidytcells._utils import Parameter
 
 
+logger = logging.getLogger(__name__)
+
+
 CLASS_1_MATCHING_REGEX = re.compile(r"HLA-[ABCEFG]|B2M")
 CLASS_2_MATCHING_REGEX = re.compile(r"HLA-D[PQR][AB]")
 
@@ -51,7 +53,7 @@ def get_class(
 
     if not gene in (*VALID_HOMOSAPIENS_MH, "B2M"):
         if not suppress_warnings:
-            warnings.warn(f"Unrecognised gene {gene}. Is this standardized?")
+            logger.warning(f"Unrecognized gene {gene}. Is this standardized?")
         return None
 
     if CLASS_1_MATCHING_REGEX.match(gene):
@@ -61,5 +63,5 @@ def get_class(
         return 2
 
     if not suppress_warnings:
-        warnings.warn(f"Class for {gene} unknown.")
+        logger.warning(f"Class for {gene} unknown.")
     return None

src/tidytcells/mh/_standardize.py

@@ -1,5 +1,5 @@
+import logging
 from typing import Dict, Optional, Type
-
 from tidytcells import _utils
 from tidytcells._utils import Parameter
 from tidytcells._standardized_gene_symbol import (
@@ -9,6 +9,9 @@
 )
 
 
+logger = logging.getLogger(__name__)
+
+
 SUPPORTED_SPECIES_AND_THEIR_STANDARDIZERS: Dict[str, Type[StandardizedGeneSymbol]] = {
     "homosapiens": StandardizedHlaSymbol,
     "musmusculus": StandardizedMusMusculusMhSymbol,
@@ -150,7 +153,7 @@ def standardize(
     species_is_supported = species in SUPPORTED_SPECIES_AND_THEIR_STANDARDIZERS
     if not species_is_supported:
         if not suppress_warnings:
-            _utils.warn_unsupported_species(species, "MH")
+            _utils.warn_unsupported_species(species, "MH", logger)
         return gene
 
     StandardizedMhSymbolClass = SUPPORTED_SPECIES_AND_THEIR_STANDARDIZERS[species]
@@ -164,6 +167,7 @@ def standardize(
                 original_input=gene,
                 attempted_fix=standardized_mh_symbol.compile("allele"),
                 species=species,
+                logger=logger,
             )
         if on_fail == "reject":
             return None

src/tidytcells/tr/_standardize.py

@@ -1,5 +1,5 @@
+import logging
 from typing import Dict, Optional, Type
-
 from tidytcells import _utils
 from tidytcells._utils import Parameter
 from tidytcells._standardized_gene_symbol import (
@@ -9,6 +9,9 @@
 )
 
 
+logger = logging.getLogger(__name__)
+
+
 SUPPORTED_SPECIES_AND_THEIR_STANDARDIZERS: Dict[str, Type[StandardizedGeneSymbol]] = {
     "homosapiens": StandardizedHomoSapiensTrSymbol,
     "musmusculus": StandardizedMusMusculusTrSymbol,
@@ -167,7 +170,7 @@ def standardize(
     species_is_supported = species in SUPPORTED_SPECIES_AND_THEIR_STANDARDIZERS
     if not species_is_supported:
         if not suppress_warnings:
-            _utils.warn_unsupported_species(species, "TR")
+            _utils.warn_unsupported_species(species, "TR", logger)
         return gene
 
     StandardizedTrSymbolClass = SUPPORTED_SPECIES_AND_THEIR_STANDARDIZERS[species]
@@ -181,6 +184,7 @@ def standardize(
                 original_input=gene,
                 attempted_fix=standardized_tr_symbol.compile("allele"),
                 species=species,
+                logger=logger,
             )
         if on_fail == "reject":
             return None

tests/test_aa.py

@@ -1,6 +1,5 @@
 import pytest
 from tidytcells import aa
-import warnings
 
 
 class TestStandardise:
@@ -11,10 +10,9 @@ def test_alreadY_correct(self, seq):
         assert result == seq
 
     @pytest.mark.parametrize("seq", ("123456", "ASDFGHJKL", "A?AAAA", "AAAXAA"))
-    def test_various_rejections(self, seq):
-        with pytest.warns(UserWarning, match="not a valid amino acid"):
-            result = aa.standardize(seq=seq)
-
+    def test_various_rejections(self, seq, caplog):
+        result = aa.standardize(seq=seq)
+        assert "not a valid amino acid" in caplog.text
         assert result == None
 
     @pytest.mark.parametrize(("seq", "expected"), (("klgak", "KLGAK"),))
@@ -33,13 +31,11 @@ def test_standardise(self):
 
         assert result == "KLGAK"
 
-    def test_suppress_warnings(self):
-        with warnings.catch_warnings():
-            warnings.simplefilter("error")
-            aa.standardize(seq="123456", suppress_warnings=True)
-
-    def test_on_fail(self):
-        with pytest.warns(UserWarning):
-            result = aa.standardize("foobarbaz", on_fail="keep")
+    def test_suppress_warnings(self, caplog):
+        aa.standardize(seq="123456", suppress_warnings=True)
+        assert len(caplog.records) == 0
 
+    def test_on_fail(self, caplog):
+        result = aa.standardize("foobarbaz", on_fail="keep")
+        assert "Failed to standardize" in caplog.text
         assert result == "foobarbaz"

tests/test_junction.py

@@ -1,6 +1,5 @@
 import pytest
 from tidytcells import junction
-import warnings
 
 
 class TestStandardise:
@@ -21,10 +20,9 @@ def test_already_correct(self, seq):
         assert result == seq
 
     @pytest.mark.parametrize("seq", ("123456", "ASDFGHJKL", "A?AAAA", "AAAXAA"))
-    def test_various_rejections(self, seq):
-        with pytest.warns(UserWarning, match="not a valid amino acid sequence"):
-            result = junction.standardise(seq=seq)
-
+    def test_various_rejections(self, seq, caplog):
+        result = junction.standardise(seq=seq)
+        assert "not a valid amino acid sequence" in caplog.text
         assert result is None
 
     @pytest.mark.parametrize(
@@ -47,24 +45,21 @@ def test_bad_input_type(self, seq):
             junction.standardise(seq=seq)
 
     @pytest.mark.parametrize("seq", ("ASQY", "CASQY", "ASQYF", "ASQYW"))
-    def test_strict(self, seq):
-        with pytest.warns(UserWarning, match="not a valid junction"):
-            result = junction.standardise(seq=seq, strict=True)
-
+    def test_strict(self, seq, caplog):
+        result = junction.standardise(seq=seq, strict=True)
+        assert "not a valid junction" in caplog.text
         assert result is None
 
     def test_standardize(self):
         result = junction.standardize(seq="CASSPGGADRRIDGYTF")
 
         assert result == "CASSPGGADRRIDGYTF"
 
-    def test_suppress_warnings(self):
-        with warnings.catch_warnings():
-            warnings.simplefilter("error")
-            junction.standardise(seq="123456", suppress_warnings=True)
-
-    def test_on_fail(self):
-        with pytest.warns(UserWarning):
-            result = junction.standardize("foobarbaz", on_fail="keep")
+    def test_suppress_warnings(self, caplog):
+        junction.standardise(seq="123456", suppress_warnings=True)
+        assert len(caplog.records) == 0
 
+    def test_on_fail(self, caplog):
+        result = junction.standardize("foobarbaz", on_fail="keep")
+        assert "Failed to standardize foobarbaz" in caplog.text
         assert result == "foobarbaz"