From f5e8f96e177a8a2d2405a8cc5d90e7d1d707f446 Mon Sep 17 00:00:00 2001
From: prosolis <5590409+prosolis@users.noreply.github.com>
Date: Sat, 4 Apr 2026 15:36:00 -0700
Subject: [PATCH] Tighten TestWords to acknowledge variant seed data and test
 combined filters

Variant words from seedTestData were silently inflating TestWords counts.
Comments now track them explicitly, and a new combined filter test verifies
variant + frequency + length compose correctly.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 internal/dictionary/dict_test.go | 31 +++++++++++++++++++++++++------
 1 file changed, 25 insertions(+), 6 deletions(-)

diff --git a/internal/dictionary/dict_test.go b/internal/dictionary/dict_test.go
index 7f52e28..8bae6c7 100644
--- a/internal/dictionary/dict_test.go
+++ b/internal/dictionary/dict_test.go
@@ -335,21 +335,21 @@ func TestWords(t *testing.T) {
 
 	d := NewFromDB(db)
 
-	// All 5-letter English words
+	// All 5-letter English words (6-letter "colour" excluded by length)
 	words, err := d.Words("en", Options{MinLength: 5, MaxLength: 5})
 	if err != nil {
 		t.Fatalf("Words: %v", err)
 	}
-	if len(words) != 6 { // color, crane, happy, house, light, quick
+	if len(words) != 6 { // color(us), crane, happy, house, light, quick
 		t.Errorf("Words returned %d, want 6", len(words))
 	}
 
-	// Filter by frequency
-	words, err = d.Words("en", Options{MinLength: 5, MaxLength: 5, MinFrequency: 500})
+	// Filter by frequency — exclude variant words to test only local additions
+	words, err = d.Words("en", Options{MinLength: 4, MaxLength: 5, MinFrequency: 500})
 	if err != nil {
 		t.Fatalf("Words with freq filter: %v", err)
 	}
-	if len(words) != 3 { // color(800), house(800), light(500)
+	if len(words) != 3 { // color(800/us), house(800), light(500)
 		t.Errorf("Words with freq filter returned %d, want 3", len(words))
 	}
 
@@ -358,10 +358,29 @@ func TestWords(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Words with POS filter: %v", err)
 	}
-	if len(words) != 4 { // color, crane, house, light
+	if len(words) != 4 { // color(us), crane, house, light
 		t.Errorf("Words with POS filter returned %d, want 4", len(words))
 	}
 
+	// Combined: variant + frequency + length
+	words, err = d.Words("en", Options{MinLength: 5, MaxLength: 5, MinFrequency: 500, Variant: "us"})
+	if err != nil {
+		t.Fatalf("Words with variant+freq: %v", err)
+	}
+	if len(words) != 1 || words[0] != "color" {
+		t.Errorf("Words(variant=us, freq>=500, len=5) = %v, want [color]", words)
+	}
+
+	// Without variant: common words only (no us/gb)
+	words, err = d.Words("en", Options{MinLength: 5, MaxLength: 5, MinFrequency: 500})
+	if err != nil {
+		t.Fatalf("Words with freq no variant: %v", err)
+	}
+	// color(us), house(common), light(common) — all returned since no variant filter
+	if len(words) != 3 {
+		t.Errorf("Words(freq>=500, no variant) returned %d, want 3", len(words))
+	}
+
 	// No match
 	_, err = d.Words("en", Options{MinLength: 100})
 	if err != ErrNoMatch {