markdown_translator/

engine.rs

//! 统一的翻译引擎模块
//!
//! 这个模块合并了原来的翻译引擎和服务，提供了一个统一的、
//! 函数式风格的翻译处理系统。
//!
//! ## 设计原则
//!
//! - **函数式组合**: 使用函数组合和管道操作
//! - **不可变性**: 优先使用不可变数据结构
//! - **错误处理**: 统一的Result类型和错误传播
//! - **异步友好**: 原生支持async/await模式

use std::sync::Arc;
use std::sync::atomic::{AtomicU64, AtomicUsize, Ordering};
use std::time::{Instant, Duration};

use crate::{
    TranslationService as BaseTranslationService,
    functional::{TextItem, TextFilter, BatchManager, Batch},
    collector::{DomNode, TextCollector},
    simple_config::SimpleTranslationConfig,
    error::{TranslationError, Result as TranslationResult},
};

// ============================================================================
// 统计和监控
// ============================================================================

/// 翻译引擎统计信息
#[derive(Debug, Default)]
pub struct EngineStats {
    /// 处理的文本总数
    pub texts_processed: AtomicUsize,
    /// 翻译的批次总数
    pub batches_processed: AtomicUsize,
    /// 缓存命中次数
    pub cache_hits: AtomicUsize,
    /// 缓存未命中次数
    pub cache_misses: AtomicUsize,
    /// 翻译API调用次数
    pub api_calls: AtomicUsize,
    /// 总翻译时间（毫秒）
    pub total_translation_time_ms: AtomicU64,
    /// 错误次数
    pub error_count: AtomicUsize,
}

impl EngineStats {
    pub fn new() -> Self {
        Self::default()
    }

    pub fn texts_processed(&self) -> usize {
        self.texts_processed.load(Ordering::Relaxed)
    }

    pub fn batches_processed(&self) -> usize {
        self.batches_processed.load(Ordering::Relaxed)
    }

    pub fn cache_hit_rate(&self) -> f64 {
        let hits = self.cache_hits.load(Ordering::Relaxed);
        let misses = self.cache_misses.load(Ordering::Relaxed);
        let total = hits + misses;
        if total == 0 {
            0.0
        } else {
            hits as f64 / total as f64
        }
    }

    pub fn average_translation_time_ms(&self) -> f64 {
        let total_time = self.total_translation_time_ms.load(Ordering::Relaxed);
        let api_calls = self.api_calls.load(Ordering::Relaxed);
        if api_calls == 0 {
            0.0
        } else {
            total_time as f64 / api_calls as f64
        }
    }

    pub fn error_rate(&self) -> f64 {
        let errors = self.error_count.load(Ordering::Relaxed);
        let total = self.api_calls.load(Ordering::Relaxed);
        if total == 0 {
            0.0
        } else {
            errors as f64 / total as f64
        }
    }

    fn increment_texts_processed(&self, count: usize) {
        self.texts_processed.fetch_add(count, Ordering::Relaxed);
    }

    fn increment_batches_processed(&self) {
        self.batches_processed.fetch_add(1, Ordering::Relaxed);
    }

    fn increment_cache_hits(&self) {
        self.cache_hits.fetch_add(1, Ordering::Relaxed);
    }

    fn increment_cache_misses(&self) {
        self.cache_misses.fetch_add(1, Ordering::Relaxed);
    }

    fn increment_api_calls(&self) {
        self.api_calls.fetch_add(1, Ordering::Relaxed);
    }

    fn add_translation_time(&self, duration: Duration) {
        self.total_translation_time_ms.fetch_add(
            duration.as_millis() as u64, 
            Ordering::Relaxed
        );
    }

    fn increment_errors(&self) {
        self.error_count.fetch_add(1, Ordering::Relaxed);
    }
}

// ============================================================================
// 简单缓存实现
// ============================================================================

use std::collections::HashMap;
use std::sync::Mutex;

/// 简单的内存缓存
struct SimpleCache {
    cache: Mutex<HashMap<String, (String, Instant)>>,
    ttl: Duration,
}

impl SimpleCache {
    fn new(ttl: Duration) -> Self {
        Self {
            cache: Mutex::new(HashMap::new()),
            ttl,
        }
    }

    fn get(&self, key: &str) -> Option<String> {
        let mut cache = self.cache.lock().ok()?;
        
        if let Some((value, timestamp)) = cache.get(key) {
            if timestamp.elapsed() < self.ttl {
                return Some(value.clone());
            } else {
                cache.remove(key);
            }
        }
        
        None
    }

    fn set(&self, key: String, value: String) {
        if let Ok(mut cache) = self.cache.lock() {
            cache.insert(key, (value, Instant::now()));
        }
    }

    fn clear_expired(&self) {
        if let Ok(mut cache) = self.cache.lock() {
            let now = Instant::now();
            cache.retain(|_, (_, timestamp)| now.duration_since(*timestamp) < self.ttl);
        }
    }
}

// ============================================================================
// 统一翻译引擎
// ============================================================================

/// 统一的翻译引擎
/// 
/// 合并了原来的翻译引擎和服务功能，提供简化的API和更好的性能。
pub struct UnifiedTranslationEngine {
    /// 基础翻译服务
    base_service: BaseTranslationService,
    /// 配置
    config: SimpleTranslationConfig,
    /// 文本过滤器
    filter: TextFilter,
    /// 文本收集器
    collector: TextCollector,
    /// 批次管理器
    batch_manager: BatchManager,
    /// 简单缓存
    cache: Option<SimpleCache>,
    /// 统计信息
    stats: Arc<EngineStats>,
}

impl UnifiedTranslationEngine {
    /// 创建新的翻译引擎
    pub fn new(config: SimpleTranslationConfig) -> TranslationResult<Self> {
        let base_service = BaseTranslationService::new(crate::types::TranslationConfig {
            enabled: config.enabled,
            source_lang: config.source_lang.clone(),
            target_lang: config.target_lang.clone(),
            deeplx_api_url: config.api_url.clone(),
            max_requests_per_second: config.requests_per_second,
            max_text_length: config.max_text_length,
            max_paragraphs_per_request: 10, // 固定值
        });

        let cache = if config.cache_enabled {
            Some(SimpleCache::new(config.cache_ttl()))
        } else {
            None
        };

        Ok(Self {
            base_service,
            config,
            filter: TextFilter::new(),
            collector: TextCollector::new(),
            batch_manager: BatchManager::new(),
            cache,
            stats: Arc::new(EngineStats::new()),
        })
    }

    /// 快速创建引擎
    pub fn quick(target_lang: &str, api_url: Option<&str>) -> TranslationResult<Self> {
        let config = crate::simple_config::quick_config(target_lang, api_url);
        Self::new(config)
    }

    /// 翻译单个文本
    pub async fn translate_text(&self, text: &str) -> TranslationResult<String> {
        if !self.config.enabled {
            return Ok(text.to_string());
        }

        if !self.filter.should_translate(text) {
            return Ok(text.to_string());
        }

        // 检查缓存
        let cache_key = format!("{}:{}", text, self.config.target_lang);
        if let Some(ref cache) = self.cache {
            if let Some(cached) = cache.get(&cache_key) {
                self.stats.increment_cache_hits();
                return Ok(cached);
            }
            self.stats.increment_cache_misses();
        }

        // 执行翻译
        let start = Instant::now();
        let result = self.base_service.translate(text).await
            .map_err(|e| TranslationError::ApiError { code: 500, message: e.to_string() })?;
        
        let duration = start.elapsed();
        self.stats.add_translation_time(duration);
        self.stats.increment_api_calls();
        self.stats.increment_texts_processed(1);

        // 存储到缓存
        if let Some(ref cache) = self.cache {
            cache.set(cache_key, result.clone());
        }

        Ok(result)
    }

    /// 翻译文本列表
    pub async fn translate_texts(&self, texts: &[String]) -> TranslationResult<Vec<String>> {
        if !self.config.enabled {
            return Ok(texts.to_vec());
        }

        // 过滤可翻译文本
        let translatable_items: Vec<TextItem> = texts
            .iter()
            .enumerate()
            .filter_map(|(i, text)| {
                if self.filter.should_translate(text) {
                    Some(crate::functional::create_text_item(
                        text.clone(), 
                        format!("text[{}]", i)
                    ))
                } else {
                    None
                }
            })
            .collect();

        // 创建批次
        let batches = self.batch_manager.create_batches(translatable_items);
        
        // 翻译每个批次
        let mut results = HashMap::new();
        for batch in batches {
            let batch_result = self.translate_batch(&batch).await?;
            for (item, translated) in batch.items.iter().zip(batch_result.iter()) {
                results.insert(item.location.clone(), translated.clone());
            }
        }

        // 构建结果向量
        let translated: Vec<String> = texts
            .iter()
            .enumerate()
            .map(|(i, original)| {
                let location = format!("text[{}]", i);
                results.get(&location).cloned().unwrap_or_else(|| original.clone())
            })
            .collect();

        Ok(translated)
    }

    /// 翻译批次
    async fn translate_batch(&self, batch: &Batch) -> TranslationResult<Vec<String>> {
        if batch.items.is_empty() {
            return Ok(Vec::new());
        }

        let texts: Vec<&str> = batch.items.iter().map(|item| item.text.as_str()).collect();
        let combined_text = texts.join("\n\n");

        let start = Instant::now();
        let translated = self.base_service.translate(&combined_text).await
            .map_err(|e| {
                self.stats.increment_errors();
                TranslationError::ApiError { code: 500, message: e.to_string() }
            })?;

        let duration = start.elapsed();
        self.stats.add_translation_time(duration);
        self.stats.increment_api_calls();
        self.stats.increment_batches_processed();
        self.stats.increment_texts_processed(batch.items.len());

        // 简单的分割逻辑（实际应用中需要更复杂的处理）
        let translated_parts: Vec<String> = translated
            .split("\n\n")
            .map(|s| s.trim().to_string())
            .collect();

        // 确保结果数量匹配
        if translated_parts.len() == texts.len() {
            Ok(translated_parts)
        } else {
            // 如果分割失败，返回原始文本
            Ok(texts.iter().map(|s| s.to_string()).collect())
        }
    }

    /// 从DOM节点收集并翻译文本
    pub async fn translate_dom_texts(&self, root: &dyn DomNode) -> TranslationResult<Vec<(String, String)>> {
        if !self.config.enabled {
            return Ok(Vec::new());
        }

        // 收集文本
        let text_items = self.collector.collect_texts(root);
        
        // 创建批次
        let batches = self.batch_manager.create_batches(text_items);
        
        // 翻译所有批次
        let mut results = Vec::new();
        for batch in batches {
            let translations = self.translate_batch(&batch).await?;
            for (item, translation) in batch.items.iter().zip(translations.iter()) {
                results.push((item.location.clone(), translation.clone()));
            }
        }

        Ok(results)
    }

    /// 获取统计信息
    pub fn stats(&self) -> Arc<EngineStats> {
        Arc::clone(&self.stats)
    }

    /// 获取配置
    pub fn config(&self) -> &SimpleTranslationConfig {
        &self.config
    }

    /// 清理过期缓存
    pub fn cleanup_cache(&self) {
        if let Some(ref cache) = self.cache {
            cache.clear_expired();
        }
    }

    /// 检查引擎健康状态
    pub fn health_check(&self) -> EngineHealth {
        let stats = &self.stats;
        
        let error_rate = stats.error_rate();
        let avg_time = stats.average_translation_time_ms();
        
        let status = if error_rate > 0.5 {
            HealthStatus::Critical
        } else if error_rate > 0.2 || avg_time > 5000.0 {
            HealthStatus::Warning
        } else {
            HealthStatus::Healthy
        };

        EngineHealth {
            status,
            error_rate,
            average_response_time_ms: avg_time,
            cache_hit_rate: stats.cache_hit_rate(),
            total_requests: stats.api_calls.load(Ordering::Relaxed),
        }
    }
}

// ============================================================================
// 健康检查
// ============================================================================

/// 引擎健康状态
#[derive(Debug, Clone)]
pub struct EngineHealth {
    pub status: HealthStatus,
    pub error_rate: f64,
    pub average_response_time_ms: f64,
    pub cache_hit_rate: f64,
    pub total_requests: usize,
}

/// 健康状态枚举
#[derive(Debug, Clone, PartialEq)]
pub enum HealthStatus {
    Healthy,
    Warning,
    Critical,
}

// ============================================================================
// 便利函数
// ============================================================================

/// 创建快速翻译引擎
pub fn create_engine(target_lang: &str, api_url: Option<&str>) -> TranslationResult<UnifiedTranslationEngine> {
    UnifiedTranslationEngine::quick(target_lang, api_url)
}

/// 创建开发环境引擎
pub fn create_dev_engine() -> TranslationResult<UnifiedTranslationEngine> {
    let config = crate::simple_config::presets::development();
    UnifiedTranslationEngine::new(config)
}

/// 创建生产环境引擎
pub fn create_prod_engine() -> TranslationResult<UnifiedTranslationEngine> {
    let config = crate::simple_config::presets::production();
    UnifiedTranslationEngine::new(config)
}

// ============================================================================
// 测试
// ============================================================================

#[cfg(test)]
mod tests {
    use super::*;
    use crate::collector::TestDomNode;

    #[tokio::test]
    async fn test_engine_creation() {
        let engine = UnifiedTranslationEngine::quick("zh", Some("http://localhost:1188/translate"));
        assert!(engine.is_ok());
    }

    #[tokio::test]
    async fn test_engine_stats() {
        let engine = UnifiedTranslationEngine::quick("zh", Some("http://localhost:1188/translate"))
            .expect("Failed to create engine");
        
        let stats = engine.stats();
        assert_eq!(stats.texts_processed(), 0);
        assert_eq!(stats.batches_processed(), 0);
    }

    #[tokio::test]
    async fn test_health_check() {
        let engine = UnifiedTranslationEngine::quick("zh", Some("http://localhost:1188/translate"))
            .expect("Failed to create engine");
        
        let health = engine.health_check();
        assert_eq!(health.status, HealthStatus::Healthy);
    }

    #[tokio::test]
    async fn test_disabled_engine() {
        let mut config = crate::simple_config::quick_config("zh", Some("http://localhost:1188/translate"));
        config.enabled = false;
        
        let engine = UnifiedTranslationEngine::new(config)
            .expect("Failed to create engine");
        
        let result = engine.translate_text("Hello World").await.unwrap();
        assert_eq!(result, "Hello World");
    }

    #[tokio::test]
    async fn test_dom_text_collection() {
        // 创建禁用翻译的引擎来避免网络请求
        let mut config = crate::simple_config::quick_config("zh", Some("http://localhost:1188/translate"));
        config.enabled = false;
        
        let engine = UnifiedTranslationEngine::new(config)
            .expect("Failed to create engine");
        
        let root = TestDomNode::new_element("div")
            .with_child(TestDomNode::new_text("Hello World"));
        
        let results = engine.translate_dom_texts(&root).await;
        assert!(results.is_ok());
    }

    #[test]
    fn test_convenience_functions() {
        let engine = create_engine("zh", Some("http://localhost:1188/translate"));
        assert!(engine.is_ok());

        let dev_engine = create_dev_engine();
        assert!(dev_engine.is_ok());

        let prod_engine = create_prod_engine();
        assert!(prod_engine.is_ok());
    }
}