From da2722b12c69977a02cf7eb1c838e0f2afdf127d Mon Sep 17 00:00:00 2001
From: niksedk <nikse.dk@gmail.com>
Date: Tue, 1 Sep 2026 19:10:31 +0200
Subject: [PATCH] Stop British to American from rewriting standard American
 words

The SE5 British to American plugin runs the shared WordList.xml in
reverse, which turned one-way vocabulary entries into corruptions of
text that is already correct American English: car -> automobile,
'flat tire' -> 'apartment tire', 'lift the box' -> 'elevator the box',
'Egyptian mummy' -> 'Egyptian mommy', 'disc jockey' -> 'disk jockey',
'the analyses' -> 'the analyzes', and more.

Add an optional only="UsToBr"/"BrToUs" attribute to word list pairs
and honor it in EnglishVariantConverter. Tag the 21 pairs whose British
side is also standard (often polysemous) American English so they only
run American -> British. Remove two pairs that are wrong in both
directions: 'pulled up'/'drew up' and 'luggage'/'baggage' (all four
words are standard in both dialects).

American -> British output is unchanged.

Fixes SubtitleEdit/subtitleedit#14374

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---
 se5/BritishToAmerican/README.md              |  5 +++
 se5/Plugin-Shared/EnglishVariantConverter.cs |  9 ++++
 se5/Plugin-Shared/WordList.xml               | 44 ++++++++++----------
 3 files changed, 35 insertions(+), 23 deletions(-)

diff --git a/se5/BritishToAmerican/README.md b/se5/BritishToAmerican/README.md
index 3c4aee7..59944dd 100644
--- a/se5/BritishToAmerican/README.md
+++ b/se5/BritishToAmerican/README.md
@@ -5,6 +5,11 @@ opposite direction. Reuses the bundled word list (~1850 pairs) and the shared
 `EnglishVariantConverter` from [`Plugin-Shared`](../Plugin-Shared/) with
 `Direction = BrToUs`.
 
+Word list pairs marked `only="UsToBr"` are skipped in this direction: their
+British-side word is also standard American English (car, film, flat, lift,
+torch...), so "converting" it would rewrite text that is already correct
+American (issue [#14374](https://github.com/SubtitleEdit/subtitleedit/issues/14374)).
+
 ## Build
 
 See `.github/workflows/british-to-american.yml`.
diff --git a/se5/Plugin-Shared/EnglishVariantConverter.cs b/se5/Plugin-Shared/EnglishVariantConverter.cs
index f02a468..b13e1c2 100644
--- a/se5/Plugin-Shared/EnglishVariantConverter.cs
+++ b/se5/Plugin-Shared/EnglishVariantConverter.cs
@@ -81,6 +81,15 @@ private void LoadBuiltInWordList()
                 continue;
             }
 
+            // only="UsToBr" / only="BrToUs" restricts a pair to one direction. Used for
+            // vocabulary pairs whose "source" word is also standard in the target dialect
+            // (car, film, flat, lift...) - converting those would corrupt correct text.
+            var only = element.Attribute("only")?.Value;
+            if (only != null && !only.Equals(_direction.ToString(), StringComparison.OrdinalIgnoreCase))
+            {
+                continue;
+            }
+
             var (from, to) = _direction == EnglishVariantDirection.UsToBr ? (us, br) : (br, us);
 
             AddRule(from, to);
diff --git a/se5/Plugin-Shared/WordList.xml b/se5/Plugin-Shared/WordList.xml
index b469cd6..f347487 100644
--- a/se5/Plugin-Shared/WordList.xml
+++ b/se5/Plugin-Shared/WordList.xml
@@ -14,7 +14,7 @@
   <Word us="analogs" br="analogues" />
   <Word us="analyze" br="analyse" />
   <Word us="analyzed" br="analysed" />
-  <Word us="analyzes" br="analyses" />
+  <Word us="analyzes" br="analyses" only="UsToBr" />
   <Word us="analyzing" br="analysing" />
   <Word us="anemia" br="anaemia" />
   <Word us="anemic" br="anaemic" />
@@ -61,8 +61,8 @@
   <Word us="armory" br="armoury" />
   <Word us="artifact" br="artefact" />
   <Word us="artifacts" br="artefacts" />
-  <Word us="automobile" br="car" />
-  <Word us="automobiles" br="cars" />
+  <Word us="automobile" br="car" only="UsToBr" />
+  <Word us="automobiles" br="cars" only="UsToBr" />
   <Word us="ax" br="axe" />
   <Word us="backpedaled" br="backpedalled" />
   <Word us="backpedaling" br="backpedalling" />
@@ -246,8 +246,8 @@
   <Word us="dishonored" br="dishonoured" />
   <Word us="dishonoring" br="dishonouring" />
   <Word us="dishonors" br="dishonours" />
-  <Word us="disk" br="disc" />
-  <Word us="disks" br="discs" />
+  <Word us="disk" br="disc" only="UsToBr" />
+  <Word us="disks" br="discs" only="UsToBr" />
   <Word us="disorganization" br="disorganisation" />
   <Word us="disorganizations" br="disorganisations" />
   <Word us="disorganize" br="disorganise" />
@@ -401,7 +401,7 @@
   <Word us="gamboling" br="gambolling" />
   <Word us="garbage can" br="dustbin" />
   <Word us="garbage cans" br="dustbins" />
-  <Word us="gasses" br="gases" />
+  <Word us="gasses" br="gases" only="UsToBr" />
   <Word us="glamor" br="glamour" />
   <Word us="gluing" br="glueing" />
   <Word us="goiter" br="goitre" />
@@ -548,7 +548,6 @@
   <Word us="louver" br="louvre" />
   <Word us="louvered" br="louvred" />
   <Word us="louvers" br="louvres" />
-  <Word us="luggage" br="baggage" />
   <Word us="luster" br="lustre" />
   <Word us="mailman" br="postman" />
   <Word us="mailmen" br="postmen" />
@@ -621,12 +620,12 @@
   <Word us="molting" br="moulting" />
   <Word us="molts" br="moults" />
   <Word us="mom" br="mum" />
-  <Word us="mommy" br="mummy" />
+  <Word us="mommy" br="mummy" only="UsToBr" />
   <Word us="moms" br="mums" />
   <Word us="monolog" br="monologue" />
   <Word us="monologs" br="monologues" />
-  <Word us="movie" br="film" />
-  <Word us="movies" br="films" />
+  <Word us="movie" br="film" only="UsToBr" />
+  <Word us="movies" br="films" only="UsToBr" />
   <Word us="multicolored" br="multicoloured" />
   <Word us="mustache" br="moustache" />
   <Word us="mustached" br="moustached" />
@@ -688,7 +687,7 @@
   <Word us="paralleled" br="parallelled" />
   <Word us="paralyze" br="paralyse" />
   <Word us="paralyzed" br="paralysed" />
-  <Word us="paralyzes" br="paralyses" />
+  <Word us="paralyzes" br="paralyses" only="UsToBr" />
   <Word us="paralyzing" br="paralysing" />
   <Word us="parceled" br="parcelled" />
   <Word us="parceling" br="parcelling" />
@@ -757,9 +756,8 @@
   <Word us="prologs" br="prologues" />
   <Word us="psychoanalyze" br="psychoanalyse" />
   <Word us="psychoanalyzed" br="psychoanalysed" />
-  <Word us="psychoanalyzes" br="psychoanalyses" />
+  <Word us="psychoanalyzes" br="psychoanalyses" only="UsToBr" />
   <Word us="psychoanalyzing" br="psychoanalysing" />
-  <Word us="pulled up" br="drew up" />
   <Word us="pummeled" br="pummelled" />
   <Word us="pummeling" br="pummelling" />
   <Word us="quarreled" br="quarrelled" />
@@ -878,8 +876,8 @@
   <Word us="smoldering" br="smouldering" />
   <Word us="smolderingly" br="smoulderingly" />
   <Word us="smolders" br="smoulders" />
-  <Word us="sneaker" br="trainer" />
-  <Word us="sneakers" br="trainers" />
+  <Word us="sneaker" br="trainer" only="UsToBr" />
+  <Word us="sneakers" br="trainers" only="UsToBr" />
   <Word us="sniveled" br="snivelled" />
   <Word us="sniveler" br="sniveller" />
   <Word us="snivelers" br="snivellers" />
@@ -1231,14 +1229,14 @@
   <!-- Additions: unambiguous vocabulary -->
   <Word us="truck" br="lorry" />
   <Word us="trucks" br="lorries" />
-  <Word us="elevator" br="lift" />
-  <Word us="elevators" br="lifts" />
-  <Word us="apartment" br="flat" />
-  <Word us="apartments" br="flats" />
-  <Word us="sidewalk" br="pavement" />
-  <Word us="sidewalks" br="pavements" />
-  <Word us="flashlight" br="torch" />
-  <Word us="flashlights" br="torches" />
+  <Word us="elevator" br="lift" only="UsToBr" />
+  <Word us="elevators" br="lifts" only="UsToBr" />
+  <Word us="apartment" br="flat" only="UsToBr" />
+  <Word us="apartments" br="flats" only="UsToBr" />
+  <Word us="sidewalk" br="pavement" only="UsToBr" />
+  <Word us="sidewalks" br="pavements" only="UsToBr" />
+  <Word us="flashlight" br="torch" only="UsToBr" />
+  <Word us="flashlights" br="torches" only="UsToBr" />
   <Word us="gasoline" br="petrol" />
   <Word us="eggplant" br="aubergine" />
   <Word us="eggplants" br="aubergines" />