From c08bb6f6211831199e4bc2473ba8fde7b6975aeb Mon Sep 17 00:00:00 2001
From: Alex-Wengg <hanweng9@gmail.com>
Date: Thu, 9 Jul 2026 12:37:30 -0400
Subject: [PATCH] =?UTF-8?q?feat(tn/en/roman):=20name=20+=20numeral=20reads?=
 =?UTF-8?q?=20as=20ordinal=20(3=E2=86=924,=20complete)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

A capitalized name followed by a two-or-more-character roman numeral reads as
an ordinal: "Sam II" → "Sam second". Section keywords still read as cardinals
("Chapter IV" → "Chapter four"). The 2+ character requirement avoids grabbing
a stray "I"/"V" after an ordinary capitalized word.

en TN roman parity: 3/4 → 4/4. No regressions.
---
 src/tn/en/roman.rs         | 34 +++++++++++++++++++++++++++++-----
 tests/data/en/tn_roman.txt |  1 +
 tests/parity_baseline.tsv  |  2 +-
 3 files changed, 31 insertions(+), 6 deletions(-)

diff --git a/src/tn/en/roman.rs b/src/tn/en/roman.rs
index fb1344c..006ea43 100644
--- a/src/tn/en/roman.rs
+++ b/src/tn/en/roman.rs
@@ -26,11 +26,35 @@ const KEYWORDS: &[&str] = &[
 pub fn parse(input: &str) -> Option<String> {
     let trimmed = input.trim();
     let (prefix, roman) = trimmed.split_once(' ')?;
-    if !KEYWORDS.contains(&prefix.to_lowercase().as_str()) {
-        return None;
+    let roman = roman.trim();
+    let value = roman_to_int(roman)?;
+
+    // Section keyword → cardinal ("Chapter IV" → "Chapter four").
+    if KEYWORDS.contains(&prefix.to_lowercase().as_str()) {
+        return Some(format!("{} {}", prefix, number_to_words(value)));
+    }
+
+    // A capitalized name followed by a multi-character numeral → ordinal
+    // ("Sam II" → "Sam second"). Requiring two or more numeral characters
+    // avoids grabbing a stray "I"/"V" after an ordinary capitalized word.
+    if roman.len() >= 2 && is_name(prefix) {
+        return Some(format!(
+            "{} {}",
+            prefix,
+            super::ordinal::number_to_ordinal_words(value)
+        ));
     }
-    let value = roman_to_int(roman.trim())?;
-    Some(format!("{} {}", prefix, number_to_words(value)))
+
+    None
+}
+
+/// A plausible personal name: an initial capital followed by lower-case
+/// letters ("Sam", "Henry").
+fn is_name(word: &str) -> bool {
+    let mut chars = word.chars();
+    word.len() >= 2
+        && matches!(chars.next(), Some(c) if c.is_ascii_uppercase())
+        && chars.all(|c| c.is_ascii_lowercase())
 }
 
 /// Convert a roman numeral to its value, or `None` if it is not a valid
@@ -77,7 +101,7 @@ mod tests {
     #[test]
     fn test_not_roman() {
         assert_eq!(parse("Chapter Five"), None); // not a numeral
-        assert_eq!(parse("Sam II"), None); // name path not handled
+        assert_eq!(parse("Sam II"), Some("Sam second".to_string())); // name → ordinal
         assert_eq!(parse("hello world"), None);
         assert_eq!(parse("Chapter"), None);
     }
diff --git a/tests/data/en/tn_roman.txt b/tests/data/en/tn_roman.txt
index 604b3c6..eb45fe3 100644
--- a/tests/data/en/tn_roman.txt
+++ b/tests/data/en/tn_roman.txt
@@ -5,3 +5,4 @@
 Chapter IV~Chapter four
 PART XL~PART forty
 Sam and I~Sam and I
+Sam II~Sam second
diff --git a/tests/parity_baseline.tsv b/tests/parity_baseline.tsv
index abc8420..8943b03 100644
--- a/tests/parity_baseline.tsv
+++ b/tests/parity_baseline.tsv
@@ -61,7 +61,7 @@ en	tn	ordinal	27	27
 en	tn	punctuation	34	63
 en	tn	punctuation_match_input	4	13
 en	tn	range	19	20
-en	tn	roman	3	4
+en	tn	roman	4	4
 en	tn	serial	8	32
 en	tn	special_text	9	10
 en	tn	telephone	17	20