diff options
| -rw-r--r-- | engine/src/main/kotlin/dev/m1sk9/lunaticChat/engine/converter/KanaConverter.kt | 69 | ||||
| -rw-r--r-- | platform-paper/src/main/kotlin/dev/m1sk9/lunaticChat/paper/converter/RomanjiConverter.kt | 13 |
2 files changed, 79 insertions, 3 deletions
diff --git a/engine/src/main/kotlin/dev/m1sk9/lunaticChat/engine/converter/KanaConverter.kt b/engine/src/main/kotlin/dev/m1sk9/lunaticChat/engine/converter/KanaConverter.kt index 1b3c67d..24e3560 100644 --- a/engine/src/main/kotlin/dev/m1sk9/lunaticChat/engine/converter/KanaConverter.kt +++ b/engine/src/main/kotlin/dev/m1sk9/lunaticChat/engine/converter/KanaConverter.kt @@ -254,6 +254,75 @@ object KanaConverter { } /** + * Checks if the input can be fully converted to hiragana without any remaining alphabetic characters. + * This validates that the input is valid romaji before attempting conversion. + * + * @param input The text to validate + * @return true if the input is valid romaji, false otherwise + */ + fun isValidRomaji(input: String): Boolean { + val lowerInput = input.lowercase() + var i = 0 + + while (i < lowerInput.length) { + val char = lowerInput[i] + + // Non-alphabetic characters are allowed (spaces, numbers, symbols) + if (char !in 'a'..'z') { + i++ + continue + } + + // Check for double consonant (valid in romaji for っ) + if (i + 1 < lowerInput.length) { + val next = lowerInput[i + 1] + if (char == next && char in "bcdfghjklmpqrstvwxyz") { + i++ + continue + } + } + + // Try to find the longest match in the trie + var node: TrieNode = romanjiTrie + var matchLength = 0 + var j = i + + while (j < lowerInput.length && lowerInput[j] in 'a'..'z') { + node = + when (node) { + is TrieNode.Branch -> { + if (node.value != null) { + matchLength = j - i + } + node.children[lowerInput[j]] ?: break + } + is TrieNode.Leaf -> { + matchLength = j - i + break + } + } + j++ + } + + // Check for terminal match + if (node is TrieNode.Leaf) { + matchLength = j - i + } else if (node is TrieNode.Branch && node.value != null) { + matchLength = j - i + } + + // If no match found, this character cannot be converted - not valid romaji + if (matchLength == 0) { + return false + } + + i += matchLength + } + + return true + } + + /** * Converts romanji text to hiragana. * * @param input The romanji text to convert diff --git a/platform-paper/src/main/kotlin/dev/m1sk9/lunaticChat/paper/converter/RomanjiConverter.kt b/platform-paper/src/main/kotlin/dev/m1sk9/lunaticChat/paper/converter/RomanjiConverter.kt index 9e54f24..c961af8 100644 --- a/platform-paper/src/main/kotlin/dev/m1sk9/lunaticChat/paper/converter/RomanjiConverter.kt +++ b/platform-paper/src/main/kotlin/dev/m1sk9/lunaticChat/paper/converter/RomanjiConverter.kt @@ -48,11 +48,18 @@ class RomanjiConverter( continue } + // Pre-validate: Check if the word is valid romaji before attempting conversion + // This prevents partial conversion of English words (e.g., "This" -> "てぃs") + if (!KanaConverter.isValidRomaji(word)) { + if (debugMode) { + logger.info("Word is not valid romaji, keeping original: $word") + } + results.add(word) + continue + } + // Step 1: Romanji -> Hiragana val hiragana = KanaConverter.toHiragana(word) - if (debugMode) { - logger.info("Romanji -> Hiragana: $word -> $hiragana") - } // Step 2: Hiragana -> Kanji/Kana val converted = |