fix(serialization/msgspec_hooks): avoid dict allocation in default_serializer when no custom type_encoders

[Peopl3s]

Description

Avoid dict allocation in default_serializer when no custom type_encoders

Closes #4701

---

📚 Documentation preview 📚: https://litestar-org.github.io/litestar-docs-preview/4710

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 67.38%. Comparing base (eb86bda) to head (f2bae4d).
⚠️ Report is 58 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4710      +/-   ##
==========================================
- Coverage   67.38%   67.38%   -0.01%     
==========================================
  Files         292      292              
  Lines       14995    14992       -3     
  Branches     1684     1685       +1     
==========================================
- Hits        10105    10102       -3     
  Misses       4755     4755              
  Partials      135      135              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[sobolevn]

This is a hot path indeed, do you have a benchmark to prove your optimization? :)

[Peopl3s]

@sobolevn Yeah, sure (I provided an example benchmark in the issue)

from __future__ import annotations

import timeit
from datetime import datetime
from typing import Any

from litestar.serialization.msgspec_hooks import DEFAULT_TYPE_ENCODERS

def default_serializer_before(value: Any, type_encoders=None) -> Any:
    type_encoders = {**DEFAULT_TYPE_ENCODERS, **(type_encoders or {})}
    for base in value.__class__.__mro__[:-1]:
        try:
            encoder = type_encoders[base]
        except KeyError:
            continue
        else:
            return encoder(value)
    raise TypeError(f"Unsupported type: {type(value)!r}")


def default_serializer_after(value: Any, type_encoders=None) -> Any:
    encoders = DEFAULT_TYPE_ENCODERS if not type_encoders else {**DEFAULT_TYPE_ENCODERS, **type_encoders}
    for base in value.__class__.__mro__[:-1]:
        if (encoder := encoders.get(base)) is not None:
            return encoder(value)
    raise TypeError(f"Unsupported type: {type(value)!r}")


from dataclasses import dataclass

@dataclass
class Record:
    id: int
    created_at: datetime
    name: str

records = [Record(id=i, created_at=datetime.now(), name=f"item-{i}") for i in range(10_000)]
payload = [{"id": r.id, "created_at": r.created_at, "name": r.name} for r in records]

ITERATIONS = 100

def run_before():
    for row in payload:
        default_serializer_before(row["created_at"])

def run_after():
    for row in payload:
        default_serializer_after(row["created_at"])

RUNS = 5

results_before = [timeit.timeit(run_before, number=ITERATIONS) for _ in range(RUNS)]
results_after  = [timeit.timeit(run_after,  number=ITERATIONS) for _ in range(RUNS)]

def stats(label, times):
    mn, mx, med = min(times), max(times), sorted(times)[len(times) // 2]
    ms = lambda t: f"{t / ITERATIONS * 1000:.2f} ms/iter"
    print(f"{label}")
    print(f"  min={mn:.3f}s ({ms(mn)})  max={mx:.3f}s ({ms(mx)})  median={med:.3f}s ({ms(med)})")

stats("BEFORE", results_before)
stats("AFTER ", results_after)
med_b = sorted(results_before)[RUNS // 2]
med_a = sorted(results_after)[RUNS // 2]
print(f"speedup (median): {med_b / med_a:.2f}x")

  ┌─────────┬─────────┬─────────┬─────────┐                                                                                                                                                                 
  │         │   min   │ median  │   max   │
  ├─────────┼─────────┼─────────┼─────────┤                                                                                                                                                                 
  │ BEFORE  │ 5.40 ms │ 5.46 ms │ 5.49 ms │                                                                                                                                                               
  ├─────────┼─────────┼─────────┼─────────┤                                                                                                                                                                 
  │ AFTER   │ 4.34 ms │ 4.39 ms │ 4.48 ms │                                                                                                                                                               
  ├─────────┼─────────┼─────────┼─────────┤                                                                                                                                                                 
  │ speedup │ 1.24x   │ 1.25x   │ 1.22x   │
  └─────────┴─────────┴─────────┴─────────┘   

[sobolevn]

LGTM :)

[Peopl3s]

@sobolevn ty for review :)