feat(android): add reference parsing, SettingsManager bounds, BPE token cleaning with TDD tests

Author: github-actions[bot]

mobile/app/src/main/java/com/bytecats/metanoia/bible/UniversalBibleSearch.kt

@@ -161,8 +161,53 @@ class UniversalBibleSearch {
             bySection = bySection.mapKeys { it.key.name }.mapValues { it.value.size }
         )
     }
+
+    /**
+     * Map common Bible book abbreviations to their canonical full name.
+     */
+    fun resolveBookAbbreviation(input: String): String? {
+        val clean = input.lowercase().replace(".", "").trim()
+        val abbreviations = mapOf(
+            "gen" to "Genesis", "ex" to "Exodus", "exod" to "Exodus", "lev" to "Leviticus",
+            "num" to "Numbers", "deut" to "Deuteronomy", "dt" to "Deuteronomy", "josh" to "Joshua",
+            "judg" to "Judges", "ruth" to "Ruth", "1sam" to "1 Samuel", "2sam" to "2 Samuel",
+            "1kings" to "1 Kings", "2kings" to "2 Kings", "1chron" to "1 Chronicles", "2chron" to "2 Chronicles",
+            "ezra" to "Ezra", "neh" to "Nehemiah", "esth" to "Esther", "job" to "Job",
+            "ps" to "Psalms", "pss" to "Psalms", "psalm" to "Psalms", "psalms" to "Psalms",
+            "prov" to "Proverbs", "eccl" to "Ecclesiastes", "song" to "Song of Solomon", "isa" to "Isaiah",
+            "jer" to "Jeremiah", "lam" to "Lamentations", "ezek" to "Ezekiel", "dan" to "Daniel",
+            "hos" to "Hosea", "joel" to "Joel", "amos" to "Amos", "obad" to "Obadiah",
+            "jonah" to "Jonah", "mic" to "Micah", "nah" to "Nahum", "hab" to "Habakkuk",
+            "zeph" to "Zephaniah", "hag" to "Haggai", "zech" to "Zechariah", "mal" to "Malachi",
+            "matt" to "Matthew", "mt" to "Matthew", "mark" to "Mark", "mk" to "Mark",
+            "luke" to "Luke", "lk" to "Luke", "john" to "John", "jn" to "John",
+            "acts" to "Acts", "rom" to "Romans", "1cor" to "1 Corinthians", "2cor" to "2 Corinthians",
+            "gal" to "Galatians", "eph" to "Ephesians", "phil" to "Philippians", "col" to "Colossians",
+            "1thess" to "1 Thessalonians", "2thess" to "2 Thessalonians", "1tim" to "1 Timothy", "2tim" to "2 Timothy",
+            "titus" to "Titus", "philem" to "Philemon", "heb" to "Hebrews", "jas" to "James",
+            "1pet" to "1 Peter", "2pet" to "2 Peter", "1jn" to "1 John", "2jn" to "2 John", "3jn" to "3 John",
+            "jude" to "Jude", "rev" to "Revelation", "enoch" to "1 Enoch", "1enoch" to "1 Enoch"
+        )
+        return abbreviations[clean] ?: BOOKS.find { it.name.equals(input, ignoreCase = true) }?.name
+    }
+
+    /**
+     * Parse structured reference string like "John 3:16", "1 John 3:16", or "jn 3:16" into components.
+     */
+    fun parseReference(input: String): ParsedReference? {
+        val regex = Regex("""^((?:\d\s+)?[A-Za-z\s]+)\s+(\d+)(?::(\d+))?$""")
+        val match = regex.find(input.trim()) ?: return null
+        val bookRaw = match.groupValues[1].trim()
+        val chapter = match.groupValues[2].toIntOrNull() ?: return null
+        val verse = match.groupValues[3].toIntOrNull()
+
+        val resolvedBook = resolveBookAbbreviation(bookRaw) ?: bookRaw
+        return ParsedReference(resolvedBook, chapter, verse)
+    }
 }
 
+data class ParsedReference(val book: String, val chapter: Int, val verse: Int?)
+
 /**
  * Statistics about the complete biblical corpus.
  */

mobile/app/src/main/java/com/bytecats/metanoia/settings/SettingsManager.kt

@@ -55,13 +55,17 @@ class SettingsManager(context: Context) {
         set(value) = prefs.edit().putBoolean("speak_defs_on_tap", value).apply()
 
     // --- Reader UI ---
+    var themeMode: String
+        get() = prefs.getString("theme_mode", "system") ?: "system"
+        set(value) = prefs.edit().putString("theme_mode", value).apply()
+
     var englishFontSize: Int
-        get() = prefs.getInt("english_font_size", 20)
-        set(value) = prefs.edit().putInt("english_font_size", value).apply()
+        get() = prefs.getInt("english_font_size", 20).coerceIn(10, 48)
+        set(value) = prefs.edit().putInt("english_font_size", value.coerceIn(10, 48)).apply()
 
     var ancientFontSize: Int
-        get() = prefs.getInt("ancient_font_size", 22)
-        set(value) = prefs.edit().putInt("ancient_font_size", value).apply()
+        get() = prefs.getInt("ancient_font_size", 22).coerceIn(10, 48)
+        set(value) = prefs.edit().putInt("ancient_font_size", value.coerceIn(10, 48)).apply()
 
     var hapticFeedbackEnabled: Boolean
         get() = prefs.getBoolean("haptic_enabled", true)

mobile/app/src/main/java/com/bytecats/metanoia/tts/BPETokenizer.kt

@@ -43,10 +43,15 @@ class BPETokenizer(
         val cacheKey = text
         cache[cacheKey]?.let { return it }
 
+        // Fallback text cleaning: remove non-printable characters, normalize whitespace
+        val cleaned = text.replace(Regex("[\\p{C}&&[^\\r\\n\\t]]"), "")
+            .replace(Regex("\\s+"), " ")
+            .trim()
+            
         // Convert to lowercase (common for TTS)
-        val normalized = text.lowercase()
+        val normalized = cleaned.lowercase()
 
-        // Basic tokenization (character-based with bigram merging)
+        // Basic tokenization (character-based with bigram matching)
         val tokens = mutableListOf<Int>()
         var i = 0
 
@@ -58,7 +63,8 @@ class BPETokenizer(
                 val substr = normalized.substring(i, i + len)
                 val tokenId = vocab[substr]
 
-                if (tokenId != null) {
+                // Vocabulary boundary checks - ensure tokenId is valid
+                if (tokenId != null && tokenId >= 0) {
                     tokens.add(tokenId)
                     i += len
                     matched = true
@@ -68,7 +74,7 @@ class BPETokenizer(
 
             if (!matched) {
                 // Unknown character, use UNK token
-                vocab[UNK_TOKEN]?.let { tokens.add(it) }
+                vocab[UNK_TOKEN]?.let { if (it >= 0) tokens.add(it) }
                 i++
             }
         }

mobile/app/src/main/java/com/bytecats/metanoia/tts/Qwen3TTSEngine.kt

@@ -119,8 +119,17 @@ class Qwen3TTSEngine(
         temperature: Float = 0.5f,
         cfgScale: Float = 2.0f
     ): FloatArray {
-        val tokenIds = text.map { it.code % config.vocabSize }
-        val maxTokens = config.maxTokens(text.length)
+        // Fallback text cleaning
+        val cleanedText = text.replace(Regex("[\\p{C}&&[^\\r\\n\\t]]"), "")
+            .replace(Regex("\\s+"), " ")
+            .trim()
+            
+        // Resilient tokenization with boundary checks
+        val tokenIds = cleanedText.map { char ->
+            val code = char.code
+            if (code >= 0) code % config.vocabSize else 0
+        }
+        val maxTokens = config.maxTokens(cleanedText.length)
 
         // Create embeddings (matching Zig's embedding lookup)
         val embeddings = createEmbeddings(tokenIds)