// TranslationPrompt.swift // OSGKeyboard · Shared // // Builds the system prompt the LLM sees when the user has the // translation toggle on. Re-uses the same per-provider "primary // language" split the polish prompt uses (`AppGroupStore.defaultSystemPrompt`) // so Chinese-native LLMs (DeepSeek, Qwen, GLM, Moonshot) get a Chinese // prompt and English-native LLMs (OpenAI) get an English one — the LLM // is most reliable when the instructions are written in its strongest // language. // // The "translate AND polish" blend is intentional: ASR transcripts are // noisy (homophone errors, broken segmentation, dropped particles), so // the prompt asks the model to clean the noise while translating. // Scenario output format (`ScenarioStyleDirective`) is appended so // translate-and-polish honours the user's polish scenario choice. import Foundation public enum TranslationPrompt { /// Build the translate-and-polish system prompt. /// /// - Parameters: /// - target: target language entry resolved via `TranslationLanguageCatalog`. /// - providerId: provider preset id (e.g. `"deepseek"`, `"openai"`); /// drives the language the prompt is written in. /// - scenarioId: active polish scenario; format rules are shared /// with the polish-only path via `ScenarioStyleDirective`. /// - uiLanguage: host-app UI language (platform label variants). public static func make( target: TranslationLanguage, providerId: String, scenarioId: String = PolishScenarioCatalog.defaultId, uiLanguage: AppUILanguage? = nil ) -> String { let isChineseNative = ["zhipu", "moonshot", "qwen", "deepseek"].contains(providerId) let directive = ScenarioStyleDirective.make( scenarioId: scenarioId, providerId: providerId, uiLanguage: uiLanguage ) return isChineseNative ? chinesePrompt(target: target, directive: directive) : englishPrompt(target: target, directive: directive) } // MARK: - Chinese prompt (for DeepSeek / Qwen / GLM / Moonshot) private static func chinesePrompt(target: TranslationLanguage, directive: String) -> String { """ 你是一位语音输入翻译与润色助手。用户用 ASR 转写了一段可能含噪声的口述: 1) 先识别原话的主要语言(若不确定则按用户给定的方向处理); 2) 将内容翻译为「\(target.promptLanguageName)」,保留原意,不增删事实、不臆测; 3) 顺带修复 ASR 噪声(同音错字、漏字、断句错乱),让译文读起来自然; 4) 简洁;若下方场景未要求列表/分段,不超过原文 1.5 倍;去掉无意义的口头禅(嗯、啊、那个); 5) 若场景格式要求列表或分段,允许按格式组织译文;总长度不超过原长 2 倍; 6) 只输出译文正文,不要解释、不要加引号、不要前缀"以下是翻译"。 \(directive) """ } // MARK: - English prompt (for OpenAI / OpenAI-compatible non-Chinese) private static func englishPrompt(target: TranslationLanguage, directive: String) -> String { """ You are a voice-input translation and polishing assistant. The user has spoken informally and the transcript may contain ASR noise: 1) Identify the input language; if unclear, assume the user wants translation INTO \(target.promptLanguageName); 2) Translate the content INTO \(target.promptLanguageName), preserving meaning; do not invent facts or omit content; 3) Fix ASR noise (homophone errors, missing characters, broken segmentation) so the translation reads naturally; 4) Stay concise; if the scenario below does not require lists/sections, do not exceed 1.5x the spoken length; drop filler words (um, uh, like); 5) If the scenario requires bullets or paragraph breaks, use that layout in the translation; total length may be up to 2x when listing; 6) Output ONLY the translation. No quotes, no preamble, no explanation. \(directive) """ } }