⚡ Bolt: Cache section header alias lookups

.jules/bolt.md

@@ -0,0 +1,3 @@
+## 2024-05-24 - [Cache Unnecessary Python Loop Variables]
+**Learning:** In highly recursive or looping NLP functions such as line-by-line regex scanning, rebuilding configurations or alias lookups dynamically creates enormous overhead in Python. The openmed package's `_alias_lookups` recreated language alias dicts on every single line, leading to severe slowdowns.
+**Action:** Use `@functools.lru_cache(maxsize=None)` on deterministic configuration generators within loops to cut down redundant operations. Ensure inputs are hashable and small in variance (like language codes).

openmed/openmed/clinical/sections/detect.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import functools
 from collections.abc import Iterable, Mapping
 from dataclasses import dataclass
 from typing import Any
@@ -244,6 +245,9 @@ def _is_underline(text: str) -> bool:
     return len(stripped) >= 3 and set(stripped) <= _UNDERLINE_CHARS
 
 
+# Bolt ⚡: Cache alias lookups as they are static per language to prevent
+# redundant tuple/dict allocations on every line during clinical section detection.
+@functools.lru_cache(maxsize=None)
 def _alias_lookups(language: str | None) -> tuple[tuple[str, Mapping[str, str]], ...]:
     languages = (
         tuple(dict.fromkeys((get_section_lexicon(language).language, "en")))