From e73f8dfe054f32ff8987351c2f2bd89012be52f2 Mon Sep 17 00:00:00 2001
From: Andrew Paseltiner <apaseltiner@gmail.com>
Date: Mon, 15 Jul 2024 11:35:31 -0400
Subject: [PATCH] Remove duplicate matches for s-103

---
 se/se_epub_lint.py                             |  2 +-
 tests/lint/semantic/s-103/golden/s-103-out.txt | 18 ++++++------------
 .../s-103/in/src/epub/text/chapter-1.xhtml     |  2 --
 3 files changed, 7 insertions(+), 15 deletions(-)

diff --git a/se/se_epub_lint.py b/se/se_epub_lint.py
index 89b66c8e..88990e31 100644
--- a/se/se_epub_lint.py
+++ b/se/se_epub_lint.py
@@ -2248,7 +2248,7 @@ def _lint_xhtml_syntax_checks(self, filename: Path, dom: se.easy_xml.EasyXmlTree
 
 	# Check for common missing roman semantics for “I”
 	regent_regex = r"(?:Charles|Edward|George|Henry|James|William) I\b"
-	matches = regex.findall(fr"King {regent_regex}", file_contents) + regex.findall(fr"{regent_regex}’s", file_contents)
+	matches = regex.findall(fr"King {regent_regex}|{regent_regex}’s", file_contents)
 	if matches:
 		messages.append(LintMessage("s-103", "Probable missing semantics for a roman I numeral.", se.MESSAGE_TYPE_WARNING, filename, matches))
 
diff --git a/tests/lint/semantic/s-103/golden/s-103-out.txt b/tests/lint/semantic/s-103/golden/s-103-out.txt
index 6e316594..9d6cc84f 100644
--- a/tests/lint/semantic/s-103/golden/s-103-out.txt
+++ b/tests/lint/semantic/s-103/golden/s-103-out.txt
@@ -1,5 +1,11 @@
 s-103 [Manual Review] chapter-1.xhtml Probable missing semantics for a roman I 
 numeral.
+        Charles I’s
+        Edward I’s
+        George I’s
+        Henry I’s
+        James I’s
+        William I’s
         King Charles I
         King Edward I
         King George I
@@ -12,15 +18,3 @@ numeral.
         King Henry I
         King James I
         King William I
-        Charles I’s
-        Edward I’s
-        George I’s
-        Henry I’s
-        James I’s
-        William I’s
-        Charles I’s
-        Edward I’s
-        George I’s
-        Henry I’s
-        James I’s
-        William I’s
diff --git a/tests/lint/semantic/s-103/in/src/epub/text/chapter-1.xhtml b/tests/lint/semantic/s-103/in/src/epub/text/chapter-1.xhtml
index f5875457..f0384943 100644
--- a/tests/lint/semantic/s-103/in/src/epub/text/chapter-1.xhtml
+++ b/tests/lint/semantic/s-103/in/src/epub/text/chapter-1.xhtml
@@ -18,8 +18,6 @@
 			<p>King Henry I.</p>
 			<p>King James I.</p>
 			<p>King William I.</p>
-			<!-- these erroneously result in two matches each
-			due to the separate searches for the King prefix and possessive suffix -->
 			<p>King Charles I’s.</p>
 			<p>King Edward I’s.</p>
 			<p>King George I’s.</p>