From de75698354447f9975c8e5cb0e92f1b13ab94b96 Mon Sep 17 00:00:00 2001
From: "google-labs-jules[bot]"
 <161369871+google-labs-jules[bot]@users.noreply.github.com>
Date: Fri, 28 Aug 2026 17:23:22 +0000
Subject: [PATCH] =?UTF-8?q?=E2=9A=A1=20Bolt:=20Cache=20context=20lexicon?=
 =?UTF-8?q?=20compilation=20for=20performance?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Co-authored-by: zrt219 <199104500+zrt219@users.noreply.github.com>
---
 .jules/bolt.md                      | 3 +++
 openmed/openmed/clinical/context.py | 2 ++
 2 files changed, 5 insertions(+)
 create mode 100644 .jules/bolt.md

diff --git a/.jules/bolt.md b/.jules/bolt.md
new file mode 100644
index 0000000..3bd58a5
--- /dev/null
+++ b/.jules/bolt.md
@@ -0,0 +1,3 @@
+## 2024-05-24 - Memoize deterministic regex compilations and lexicon generation in text-processing pipelines
+**Learning:** In NLP or text-processing pipelines (e.g., within `openmed.clinical`), deterministic operations like regex compilations (`re.compile`) and lexicon generation are extremely expensive if called repeatedly per string evaluation. Without memoization, functions like `_compiled_context_lexicon` cause severe performance bottlenecks during span evaluations.
+**Action:** Always memoize deterministic regex compilations and lexicon generation (e.g., using `@functools.lru_cache`) when they are repeatedly called per string or span evaluation, provided the return type is immutable (e.g., a frozen dataclass).
diff --git a/openmed/openmed/clinical/context.py b/openmed/openmed/clinical/context.py
index 9fd11df..e78f5ad 100644
--- a/openmed/openmed/clinical/context.py
+++ b/openmed/openmed/clinical/context.py
@@ -35,6 +35,7 @@
 
 from __future__ import annotations
 
+import functools
 import re
 from collections.abc import Iterable, Iterator, Mapping, Sequence
 from dataclasses import dataclass, replace
@@ -154,6 +155,7 @@ class _CompiledContextLexicon:
     backward_context_cues: frozenset[str]
 
 
+@functools.lru_cache
 def _compiled_context_lexicon(language: str | None = None) -> _CompiledContextLexicon:
     lexicon = get_clinical_cue_lexicon(language)
     token_boundaries = lexicon.token_boundaries