Merge pull request #695 from pyathena-dev/refactor/converter-cleanup

Author: laughingman7743

pyathena/arrow/converter.py

@@ -115,4 +115,5 @@ def __init__(self) -> None:
         )
 
     def convert(self, type_: str, value: str | None, type_hint: str | None = None) -> Any | None:
-        pass
+        converter = self.get(type_)
+        return converter(value)

pyathena/converter.py

@@ -3,20 +3,20 @@
 import binascii
 import json
 import logging
+import re
 from abc import ABCMeta, abstractmethod
 from collections.abc import Callable
 from copy import deepcopy
 from datetime import date, datetime, time
 from decimal import Decimal
-from typing import Any
+from typing import Any, ClassVar
 
 from dateutil.tz import gettz
 
 from pyathena.parser import (
     TypedValueConverter,
     TypeNode,
     TypeSignatureParser,
-    _normalize_hive_syntax,
     _split_array_items,
 )
 from pyathena.util import strtobool
@@ -551,6 +551,9 @@ class DefaultTypeConverter(Converter):
         ['1', '2', '3']
     """
 
+    _HIVE_SYNTAX_RE: ClassVar[re.Pattern[str]] = re.compile(r"[<>:]")
+    _HIVE_REPLACEMENTS: ClassVar[dict[str, str]] = {"<": "(", ">": ")", ":": " "}
+
     def __init__(self) -> None:
         super().__init__(mappings=deepcopy(_DEFAULT_CONVERTERS), default=_to_default)
         self._parser = TypeSignatureParser()
@@ -561,6 +564,25 @@ def __init__(self) -> None:
         )
         self._parsed_hints: dict[str, TypeNode] = {}
 
+    @staticmethod
+    def _normalize_hive_syntax(type_str: str) -> str:
+        """Normalize Hive-style DDL syntax to Trino-style.
+
+        Converts angle-bracket notation (``array<struct<a:int>>``) to
+        parenthesized notation (``array(struct(a int))``).
+
+        Args:
+            type_str: Type signature string, possibly using Hive syntax.
+
+        Returns:
+            Normalized type signature using Trino-style parenthesized notation.
+        """
+        if "<" not in type_str:
+            return type_str
+        return DefaultTypeConverter._HIVE_SYNTAX_RE.sub(
+            lambda m: DefaultTypeConverter._HIVE_REPLACEMENTS[m.group()], type_str
+        )
+
     def convert(self, type_: str, value: str | None, type_hint: str | None = None) -> Any | None:
         """Convert a string value to the appropriate Python type.
 
@@ -605,7 +627,7 @@ def _parse_type_hint(self, type_hint: str) -> TypeNode:
         Returns:
             Parsed TypeNode.
         """
-        normalized = _normalize_hive_syntax(type_hint)
+        normalized = self._normalize_hive_syntax(type_hint)
         if normalized not in self._parsed_hints:
             self._parsed_hints[normalized] = self._parser.parse(normalized)
         return self._parsed_hints[normalized]

pyathena/pandas/converter.py

@@ -81,7 +81,8 @@ def _dtypes(self) -> dict[str, type[Any]]:
         return self.__dtypes
 
     def convert(self, type_: str, value: str | None, type_hint: str | None = None) -> Any | None:
-        pass
+        converter = self.get(type_)
+        return converter(value)
 
 
 class DefaultPandasUnloadTypeConverter(Converter):
@@ -104,4 +105,5 @@ def __init__(self) -> None:
         )
 
     def convert(self, type_: str, value: str | None, type_hint: str | None = None) -> Any | None:
-        pass
+        converter = self.get(type_)
+        return converter(value)

pyathena/parser.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 import json
-import re
 from collections.abc import Callable
 from dataclasses import dataclass, field
 from typing import Any
@@ -11,28 +10,6 @@
     "int": "integer",
 }
 
-# Pattern for normalizing Hive-style type signatures to Trino-style.
-# Matches angle brackets and colons used in Hive DDL (e.g., array<struct<a:int>>).
-_HIVE_SYNTAX_RE: re.Pattern[str] = re.compile(r"[<>:]")
-_HIVE_REPLACEMENTS: dict[str, str] = {"<": "(", ">": ")", ":": " "}
-
-
-def _normalize_hive_syntax(type_str: str) -> str:
-    """Normalize Hive-style DDL syntax to Trino-style.
-
-    Converts angle-bracket notation (``array<struct<a:int>>``) to
-    parenthesized notation (``array(struct(a int))``).
-
-    Args:
-        type_str: Type signature string, possibly using Hive syntax.
-
-    Returns:
-        Normalized type signature using Trino-style parenthesized notation.
-    """
-    if "<" not in type_str:
-        return type_str
-    return _HIVE_SYNTAX_RE.sub(lambda m: _HIVE_REPLACEMENTS[m.group()], type_str)
-
 
 def _split_array_items(inner: str) -> list[str]:
     """Split array items by comma, respecting brace and bracket groupings.

pyathena/polars/converter.py

@@ -128,4 +128,5 @@ def __init__(self) -> None:
         )
 
     def convert(self, type_: str, value: str | None, type_hint: str | None = None) -> Any | None:
-        pass
+        converter = self.get(type_)
+        return converter(value)

tests/pyathena/arrow/test_converter.py

@@ -0,0 +1,8 @@
+from pyathena.arrow.converter import DefaultArrowUnloadTypeConverter
+
+
+class TestDefaultArrowUnloadTypeConverter:
+    def test_convert_delegates_to_default(self):
+        """convert() dispatches through the default converter instead of returning None."""
+        converter = DefaultArrowUnloadTypeConverter()
+        assert converter.convert("varchar", "hello") == "hello"

tests/pyathena/pandas/test_converter.py

@@ -0,0 +1,25 @@
+from pyathena.pandas.converter import (
+    DefaultPandasTypeConverter,
+    DefaultPandasUnloadTypeConverter,
+)
+
+
+class TestDefaultPandasTypeConverter:
+    def test_convert_delegates_to_mapping(self):
+        """convert() dispatches through self.get(type_) instead of returning None.
+
+        Verifies both the explicit mapping path (boolean → _to_boolean)
+        and the default converter path (varchar → _to_default), plus
+        None passthrough.
+        """
+        converter = DefaultPandasTypeConverter()
+        assert converter.convert("boolean", "true") is True
+        assert converter.convert("varchar", "hello") == "hello"
+        assert converter.convert("varchar", None) is None
+
+
+class TestDefaultPandasUnloadTypeConverter:
+    def test_convert_delegates_to_default(self):
+        """convert() dispatches through the default converter instead of returning None."""
+        converter = DefaultPandasUnloadTypeConverter()
+        assert converter.convert("varchar", "hello") == "hello"

tests/pyathena/polars/test_converter.py

@@ -0,0 +1,8 @@
+from pyathena.polars.converter import DefaultPolarsUnloadTypeConverter
+
+
+class TestDefaultPolarsUnloadTypeConverter:
+    def test_convert_delegates_to_default(self):
+        """convert() dispatches through the default converter instead of returning None."""
+        converter = DefaultPolarsUnloadTypeConverter()
+        assert converter.convert("varchar", "hello") == "hello"

tests/pyathena/test_converter.py

@@ -408,3 +408,50 @@ def test_hive_syntax_caching(self):
         # Both should normalize to "array(integer)" in the cache
         assert "array(integer)" in converter._parsed_hints
         assert len(converter._parsed_hints) == 1
+
+    def test_normalize_hive_syntax_noop(self):
+        """Trino-style input passes through unchanged."""
+        assert DefaultTypeConverter._normalize_hive_syntax("array(integer)") == "array(integer)"
+
+    def test_normalize_hive_syntax_replaces(self):
+        assert (
+            DefaultTypeConverter._normalize_hive_syntax("array<struct<a:int>>")
+            == "array(struct(a int))"
+        )
+
+    def test_normalize_hive_syntax_struct(self):
+        converter = DefaultTypeConverter()
+        result = converter.convert(
+            "row",
+            "{name=Alice, age=25}",
+            type_hint="struct<name:varchar,age:int>",
+        )
+        assert result == {"name": "Alice", "age": 25}
+
+    def test_normalize_hive_syntax_nested(self):
+        converter = DefaultTypeConverter()
+        result = converter.convert(
+            "array",
+            "[{a=1, b=hello}, {a=2, b=world}]",
+            type_hint="array<struct<a:int,b:varchar>>",
+        )
+        assert result == [{"a": 1, "b": "hello"}, {"a": 2, "b": "world"}]
+
+    def test_normalize_hive_syntax_map(self):
+        converter = DefaultTypeConverter()
+        result = converter.convert(
+            "map",
+            '{"x": 1, "y": 2}',
+            type_hint="map<string,int>",
+        )
+        assert result == {"x": 1, "y": 2}
+
+    def test_normalize_hive_syntax_mixed(self):
+        """Hive angle brackets wrapping Trino-style parenthesized inner type."""
+        converter = DefaultTypeConverter()
+        result = converter.convert(
+            "array",
+            "[{a=1, b=hello}]",
+            type_hint="array<row(a int, b varchar)>",
+        )
+        assert result == [{"a": 1, "b": "hello"}]

tests/pyathena/test_parser.py

@@ -5,7 +5,6 @@
     TypedValueConverter,
     TypeNode,
     TypeSignatureParser,
-    _normalize_hive_syntax,
 )
 
 
@@ -109,55 +108,6 @@ def test_type_alias_in_complex_type(self):
         assert node.type_name == "array"
         assert node.children[0].type_name == "integer"
 
-    def test_hive_syntax_simple(self):
-        parser = TypeSignatureParser()
-        node = parser.parse(_normalize_hive_syntax("array<int>"))
-        assert node.type_name == "array"
-        assert node.children[0].type_name == "integer"
-
-    def test_hive_syntax_struct(self):
-        parser = TypeSignatureParser()
-        node = parser.parse(_normalize_hive_syntax("struct<a:int,b:varchar>"))
-        assert node.type_name == "struct"
-        assert node.field_names == ["a", "b"]
-        assert node.children[0].type_name == "integer"
-        assert node.children[1].type_name == "varchar"
-
-    def test_hive_syntax_nested(self):
-        parser = TypeSignatureParser()
-        node = parser.parse(_normalize_hive_syntax("array<struct<a:int,b:varchar>>"))
-        assert node.type_name == "array"
-        struct_node = node.children[0]
-        assert struct_node.type_name == "struct"
-        assert struct_node.field_names == ["a", "b"]
-        assert struct_node.children[0].type_name == "integer"
-        assert struct_node.children[1].type_name == "varchar"
-
-    def test_hive_syntax_map(self):
-        parser = TypeSignatureParser()
-        node = parser.parse(_normalize_hive_syntax("map<string,int>"))
-        assert node.type_name == "map"
-        assert node.children[0].type_name == "string"
-        assert node.children[1].type_name == "integer"
-
-    def test_mixed_syntax(self):
-        """Hive angle brackets wrapping Trino-style parenthesized inner type."""
-        parser = TypeSignatureParser()
-        node = parser.parse(_normalize_hive_syntax("array<row(a int, b varchar)>"))
-        assert node.type_name == "array"
-        row_node = node.children[0]
-        assert row_node.type_name == "row"
-        assert row_node.field_names == ["a", "b"]
-        assert row_node.children[0].type_name == "integer"
-        assert row_node.children[1].type_name == "varchar"
-
-    def test_normalize_hive_syntax_noop(self):
-        """Trino-style input passes through unchanged."""
-        assert _normalize_hive_syntax("array(integer)") == "array(integer)"
-
-    def test_normalize_hive_syntax_replaces(self):
-        assert _normalize_hive_syntax("array<struct<a:int>>") == "array(struct(a int))"
-
     def test_trailing_modifier_after_paren(self):
         """Type with content after closing paren should not break parsing."""
         parser = TypeSignatureParser()