[LLM] Generate and package seed database asset with dictionary and compaction

app/BUILD.bazel

@@ -1 +1,10 @@
 package(default_visibility = ["//visibility:public"])
+
+exports_files([
+    "src/main/assets/databases/seed.db",
+])
+
+filegroup(
+    name = "seed_db_asset",
+    srcs = ["src/main/assets/databases/seed.db"],
+)

core/data/BUILD.bazel

@@ -13,6 +13,7 @@ kt_jvm_library(
 
 kt_jvm_test(
     name = "seed_database_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/seed/SeedDatabaseTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.seed.SeedDatabaseTest",
     deps = [
@@ -23,6 +24,7 @@ kt_jvm_test(
 
 kt_jvm_test(
     name = "seed_dictionary_repository_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/seed/SeedDictionaryRepositoryTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.seed.SeedDictionaryRepositoryTest",
     deps = [
@@ -34,6 +36,7 @@ kt_jvm_test(
 
 kt_jvm_test(
     name = "seed_false_cognates_repository_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/seed/SeedFalseCognatesRepositoryTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.seed.SeedFalseCognatesRepositoryTest",
     deps = [
@@ -45,6 +48,7 @@ kt_jvm_test(
 
 kt_jvm_test(
     name = "seed_curated_lessons_repository_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/seed/SeedCuratedLessonsRepositoryTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.seed.SeedCuratedLessonsRepositoryTest",
     deps = [
@@ -55,6 +59,7 @@ kt_jvm_test(
 
 kt_jvm_test(
     name = "seed_cognate_rules_repository_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/seed/SeedCognateRulesRepositoryTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.seed.SeedCognateRulesRepositoryTest",
     deps = [
@@ -63,8 +68,23 @@ kt_jvm_test(
     ],
 )
 
+kt_jvm_test(
+    name = "packaged_seed_asset_test",
+    size = "small",
+    srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/seed/PackagedSeedAssetTest.kt"],
+    data = [
+        "//app:seed_db_asset",
+    ],
+    test_class = "com.anysoftkeyboard.calque.data.seed.PackagedSeedAssetTest",
+    deps = [
+        ":data",
+        "@maven//:junit_junit",
+    ],
+)
+
 kt_jvm_test(
     name = "user_database_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/user/UserDatabaseTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.user.UserDatabaseTest",
     deps = [
@@ -75,6 +95,7 @@ kt_jvm_test(
 
 kt_jvm_test(
     name = "user_word_state_repository_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/user/UserWordStateRepositoryTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.user.UserWordStateRepositoryTest",
     deps = [
@@ -86,6 +107,7 @@ kt_jvm_test(
 
 kt_jvm_test(
     name = "user_custom_text_repository_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/user/UserCustomTextRepositoryTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.user.UserCustomTextRepositoryTest",
     deps = [
@@ -96,6 +118,7 @@ kt_jvm_test(
 
 kt_jvm_test(
     name = "calque_word_lookup_provider_test",
+    size = "small",
     srcs = ["src/test/kotlin/com/anysoftkeyboard/calque/data/CalqueWordLookupProviderTest.kt"],
     test_class = "com.anysoftkeyboard.calque.data.CalqueWordLookupProviderTest",
     deps = [

core/data/src/test/kotlin/com/anysoftkeyboard/calque/data/seed/PackagedSeedAssetTest.kt

@@ -0,0 +1,72 @@
+package com.anysoftkeyboard.calque.data.seed
+
+import java.io.File
+import org.junit.Assert.assertEquals
+import org.junit.Assert.assertNotNull
+import org.junit.Assert.assertTrue
+import org.junit.Test
+
+class PackagedSeedAssetTest {
+
+    @Test
+    fun packagedSeedAsset_existsAndHasAllSeedTablesPopulated() {
+        val assetFile = File("app/src/main/assets/databases/seed.db")
+        assertTrue("Expected packaged seed database at ${assetFile.path}", assetFile.exists())
+        assertTrue(
+            "Expected database size > 10KB, got ${assetFile.length()}",
+            assetFile.length() > 10_000
+        )
+
+        val database = SeedDatabase.open("jdbc:sqlite:${assetFile.path}")
+        database.use { db ->
+            val dictRepo = SqliteDictionaryRepository(db.connection)
+            val fcRepo = SqliteFalseCognatesRepository(db.connection)
+            val lessonRepo = SqliteCuratedLessonsRepository(db.connection)
+            val rulesRepo = SqliteCognateRulesRepository(db.connection)
+
+            // 1. Verify dictionary entries
+            val hotelWord = dictRepo.getBySurfaceForm("hotel")
+            assertNotNull("Dictionary should contain 'hotel'", hotelWord)
+            assertEquals("hotel", hotelWord?.lemma)
+            assertTrue("hotel should be marked as cognate", hotelWord?.isCognate == true)
+
+            val esWord = dictRepo.getBySurfaceForm("es")
+            assertNotNull("Dictionary should contain 'es'", esWord)
+            assertEquals("ser", esWord?.lemma)
+
+            // 2. Verify false cognates
+            val exitoCard = fcRepo.getBySurfaceForm("éxito")
+            assertNotNull("False cognates should contain 'éxito'", exitoCard)
+            assertEquals("exit", exitoCard?.apparentEnglish)
+            assertEquals("salida", exitoCard?.realSpanishForApparent)
+
+            val carpetasPlural = fcRepo.getBySurfaceForm("carpetas")
+            assertNotNull(
+                "False cognates should expand inflections like 'carpetas'",
+                carpetasPlural
+            )
+            assertEquals("carpeta", carpetasPlural?.lemma)
+
+            // 3. Verify curated lessons
+            val lessons = lessonRepo.getAll()
+            assertEquals("Should contain 8 curated lessons", 8, lessons.size)
+            val levels = lessons.map { it.level }.toSet()
+            assertEquals(setOf(0, 1, 2), levels)
+
+            val firstLesson = lessonRepo.getById("lvl0_00_cognate_unlock")
+            assertNotNull(firstLesson)
+            assertEquals("The Cognate Unlock & Vowel Soundboard", firstLesson?.title)
+            assertEquals(
+                "El hotel moderno es excelente.",
+                firstLesson?.parseSentences()?.first()?.sentenceEs,
+            )
+
+            // 4. Verify cognate rules
+            val rules = rulesRepo.getAll()
+            assertEquals("Should contain 12 cognate rules", 12, rules.size)
+            val matchingRule = rulesRepo.findRuleForSpanishWord("situación")
+            assertNotNull(matchingRule)
+            assertEquals("tion_to_cion", matchingRule?.ruleId)
+        }
+    }
+}

data/seeds/BUILD.bazel

@@ -3,6 +3,7 @@ package(default_visibility = ["//visibility:public"])
 exports_files([
     "cognate_rules_seed.json",
     "curated_lessons_seed.json",
+    "dictionary_seed.json",
     "false_cognates_seed.json",
 ])
 
@@ -11,6 +12,7 @@ filegroup(
     srcs = [
         "cognate_rules_seed.json",
         "curated_lessons_seed.json",
+        "dictionary_seed.json",
         "false_cognates_seed.json",
     ],
 )