⚡️ Speed up _display_error_loc() by 25% in pydantic/v1/error_wrappers.py

[misrasaurabh1]

Change Summary

📄 _display_error_loc() in pydantic/v1/error_wrappers.py

📈 Performance improved by 25% (0.25x faster)

⏱️ Runtime went down from 81.6 milliseconds to 65.2 milliseconds

Explanation and details

To optimize this code for better performance, we can avoid the generator expression in the join method by using a list comprehension directly. This small change reduces overhead.

string.join can't join generators because it does not know the length upfront, so it converts it to list internally and it ends up being slower than joining list comprehensions directly.

Correctness verification

The new optimized code was tested for correctness. The results are listed below.

🔘 (none found) − ⚙️ Existing Unit Tests

✅ 13 Passed − 🌀 Generated Regression Tests

(click to show generated tests)

# imports
import pytest  # used for our unit tests
from pydantic.v1.error_wrappers import _display_error_loc

# unit tests

def test_basic_single_element_location():
    # Single element location
    error = {'loc': ['field1']}
    assert _display_error_loc(error) == 'field1'

def test_basic_multiple_elements_location():
    # Multiple elements location
    error = {'loc': ['field1', 'subfield2']}
    assert _display_error_loc(error) == 'field1 -> subfield2'

    error = {'loc': ['field1', 'subfield2', 'element3']}
    assert _display_error_loc(error) == 'field1 -> subfield2 -> element3'

def test_edge_empty_location_list():
    # Empty location list
    error = {'loc': []}
    assert _display_error_loc(error) == ''

def test_edge_non_string_elements():
    # Non-string elements in location list
    error = {'loc': [1, 2, 3]}
    assert _display_error_loc(error) == '1 -> 2 -> 3'

    error = {'loc': [None, True, 42.0]}
    assert _display_error_loc(error) == 'None -> True -> 42.0'

def test_error_handling_missing_loc_key():
    # Missing 'loc' key
    error = {'not_loc': ['field1']}
    with pytest.raises(KeyError):
        _display_error_loc(error)

def test_error_handling_non_iterable_loc_value():
    # Non-iterable 'loc' value
    error = {'loc': 'field1'}
    with pytest.raises(TypeError):
        _display_error_loc(error)

def test_performance_large_location_list():
    # Large location list
    error = {'loc': ['field' + str(i) for i in range(1000)]}
    expected_output = ' -> '.join('field' + str(i) for i in range(1000))
    assert _display_error_loc(error) == expected_output

def test_mixed_data_types_in_location_list():
    # Mixed data types in location list
    error = {'loc': ['field1', 2, 'subfield3', None]}
    assert _display_error_loc(error) == 'field1 -> 2 -> subfield3 -> None'

def test_special_characters_in_elements():
    # Special characters in elements
    error = {'loc': ['field1, 'sub@field2', '#element3']}
    assert _display_error_loc(error) == 'field1$ -> sub@field2 -> #element3'

def test_nested_structures():
    # Nested lists
    error = {'loc': [['field1', 'subfield2'], 'element3']}
    with pytest.raises(TypeError):
        _display_error_loc(error)

def test_unicode_characters():
    # Unicode characters
    error = {'loc': ['字段1', '子字段2']}
    assert _display_error_loc(error) == '字段1 -> 子字段2'

def test_immutable_types():
    # Immutable types (tuple)
    error = {'loc': ('field1', 'subfield2')}
    assert _display_error_loc(error) == 'field1 -> subfield2'

def test_large_scale_very_large_location_list():
    # Very large location list for performance and scalability
    error = {'loc': ['field' + str(i) for i in range(1000000)]}
    expected_output = ' -> '.join('field' + str(i) for i in range(1000000))
    assert _display_error_loc(error) == expected_output

🔘 (none found) − ⏪ Replay Tests

Checklist

• The pull request title is a good summary of the changes - it will be used in the changelog
• Unit tests for the changes exist
• Tests pass on CI
• Documentation reflects the changes where applicable
• My PR is ready to review, please add a comment including the phrase "please review" to assign reviewers

[codspeed-hq]

CodSpeed Performance Report

Merging #9653 will not alter performance

_{Comparing misrasaurabh1:codeflash/optimize-_display_error_loc-2024-06-06T03.21.31 (00aff67) with main (9dac684)}

Summary

✅ 13 untouched benchmarks

[Olegt0rr]

@misrasaurabh1, how 25% was measured? CodSpeed prints untouched benchmarks only

[misrasaurabh1]

Hi @Olegt0rr , the 25% gain is the speedup for only the _display_error_loc() function. It was measured on the inputs/tests we generated that are attached in the description of the PR. I also manually verified the gains with this quick benchmark I created.

[sydney-runkle]

Yep, makes sense that a list comprehension is going to be faster here! Increased space usage doesn't really matter in this case.

[sydney-runkle]

Oops, this should have gone in v1.10.X-fixes.

@misrasaurabh1, let's skip opening any PRs against v1 files - we're not looking to spend time on performance improvements there.

[sydney-runkle]

I already released our v1.10.17 patch, so this won't be included. Happy to consider other PRs that focus on V2 stuff though :).