From 49d764cc85b17bca7e8b0863b9806f2080e525c3 Mon Sep 17 00:00:00 2001
From: Chris Grady <17553614+CGFixIT@users.noreply.github.com>
Date: Sun, 6 Sep 2026 11:22:20 -0400
Subject: [PATCH] perf: avoid copying chunked matches before sorting

---
 src/insight_extractor/stemmer.py |  3 ++-
 tests/unit/test_stemmer.py       | 29 ++++++++++++++++++++++++++++-
 2 files changed, 30 insertions(+), 2 deletions(-)

diff --git a/src/insight_extractor/stemmer.py b/src/insight_extractor/stemmer.py
index c99d889..19a69a1 100644
--- a/src/insight_extractor/stemmer.py
+++ b/src/insight_extractor/stemmer.py
@@ -35,7 +35,8 @@ def finditer(self, text: str) -> Iterator[re.Match[str]]:
         matches: list[re.Match[str]] = []
         for pattern in self.patterns:
             matches.extend(pattern.finditer(text))
-        yield from sorted(matches, key=lambda match: (match.start(), match.end()))
+        matches.sort(key=lambda match: (match.start(), match.end()))
+        yield from matches
 
 
 type KeywordPattern = re.Pattern[str] | ChunkedKeywordPattern
diff --git a/tests/unit/test_stemmer.py b/tests/unit/test_stemmer.py
index aefc94b..03c22c7 100644
--- a/tests/unit/test_stemmer.py
+++ b/tests/unit/test_stemmer.py
@@ -7,11 +7,38 @@
 import pytest
 
 from insight_extractor.config import StemMode
-from insight_extractor.stemmer import DynamicKeywordStemmer, KeywordPatternRegistry
+from insight_extractor.stemmer import (
+    ChunkedKeywordPattern,
+    DynamicKeywordStemmer,
+    KeywordPatternRegistry,
+)
 
 # ── DynamicKeywordStemmer tests ──────────────────────────────────────────────
 
 
+@pytest.mark.parametrize("text", ["", "no matches", "alpha beta alpha"])
+def test_chunked_finditer_preserves_span_order_and_tie_priority(text: str) -> None:
+    patterns = [
+        re.compile(pattern) for pattern in ("alpha beta", "alpha", "(?:alpha)", "(?=alpha)")
+    ]
+    matches = list(ChunkedKeywordPattern(patterns).finditer(text))
+
+    expected = (
+        [
+            ((0, 0), "", "(?=alpha)"),
+            ((0, 5), "alpha", "alpha"),
+            ((0, 5), "alpha", "(?:alpha)"),
+            ((0, 10), "alpha beta", "alpha beta"),
+            ((11, 11), "", "(?=alpha)"),
+            ((11, 16), "alpha", "alpha"),
+            ((11, 16), "alpha", "(?:alpha)"),
+        ]
+        if text == "alpha beta alpha"
+        else []
+    )
+    assert [(match.span(), match.group(), match.re.pattern) for match in matches] == expected
+
+
 class TestInit:
     """Construction and default attributes."""