postfix-log-parser 0.2.0

//! # Bounce 退信处理组件解析器
//!
//! Bounce 是 Postfix 的退信处理组件，负责：
//! - 生成和发送投递失败通知（NDN）
//! - 处理发送者通知和管理员通知
//! - 管理退信邮件的格式和内容
//! - 处理延迟投递通知和最终失败通知
//!
//! ## 支持的事件类型
//!
//! - **发送者通知**: 向原始发送者发送的投递失败通知
//! - **管理员通知**: 向邮件管理员发送的系统通知
//! - **警告事件**: 格式错误、配置问题、资源限制等警告
//!
//! ## 示例日志格式
//!
//! ```text
//! # 发送者投递失败通知
//! queue_id: sender non-delivery notification: new_queue_id
//!
//! # 管理员通知
//! queue_id: postmaster non-delivery notification: new_queue_id
//!
//! # 警告事件
//! malformed request
//! ```

use regex::Regex;

use crate::error::ParseError;
use crate::events::bounce::{BounceEvent, BounceWarningType};
use crate::events::{base::BaseEvent, ComponentEvent};

use super::ComponentParser;

/// BOUNCE组件解析器
///
/// 基于Postfix BOUNCE守护进程的真实日志格式开发
/// BOUNCE守护进程的特点：
/// - 处理投递失败通知（non-delivery notification）
/// - 生成发送者通知邮件
/// - 处理退信和延迟通知
/// - 管理通知邮件的生成和发送
pub struct BounceParser {
    /// 发送者投递失败通知事件解析 - 主要模式（95%+频率）
    /// 格式: "queue_id: sender non-delivery notification: new_queue_id"
    sender_notification_regex: Regex,

    /// 收件人投递失败通知事件解析
    /// 格式: "queue_id: postmaster non-delivery notification: new_queue_id"  
    postmaster_notification_regex: Regex,

    /// 警告事件解析 - malformed request
    /// 格式: "malformed request" (注意：MasterParser已剥离"warning:"前缀)
    warning_malformed_regex: Regex,

    /// 其他警告事件解析
    /// 格式: 各种警告消息 (注意：MasterParser已剥离"warning:"前缀)
    warning_general_regex: Regex,
}

impl BounceParser {
    pub fn new() -> Self {
        Self {
            // 主要的发送者投递失败通知事件模式
            sender_notification_regex: Regex::new(
                r"^([A-F0-9]+):\s+sender non-delivery notification:\s+([A-F0-9]+)$",
            )
            .expect("BOUNCE发送者通知正则表达式编译失败"),

            // 邮件管理员投递失败通知事件模式
            postmaster_notification_regex: Regex::new(
                r"^([A-F0-9]+):\s+postmaster non-delivery notification:\s+([A-F0-9]+)$",
            )
            .expect("BOUNCE邮件管理员通知正则表达式编译失败"),

            // 格式错误请求警告模式 (MasterParser已剥离"warning:"前缀)
            warning_malformed_regex: Regex::new(r"^malformed request$")
                .expect("BOUNCE格式错误警告正则表达式编译失败"),

            // 一般警告事件模式 (MasterParser已剥离"warning:"前缀)
            warning_general_regex: Regex::new(r"^(.+)$").expect("BOUNCE一般警告正则表达式编译失败"),
        }
    }

    /// 解析BOUNCE日志行
    pub fn parse_line(&self, line: &str, base_event: BaseEvent) -> Option<BounceEvent> {
        // 尝试解析发送者投递失败通知事件（主要模式）
        if let Some(captures) = self.sender_notification_regex.captures(line) {
            return self.parse_sender_notification(captures, base_event);
        }

        // 尝试解析邮件管理员投递失败通知事件
        if let Some(captures) = self.postmaster_notification_regex.captures(line) {
            return self.parse_postmaster_notification(captures, base_event);
        }

        // 尝试解析格式错误警告
        if self.warning_malformed_regex.is_match(line) {
            return self.parse_malformed_warning(base_event);
        }

        // 尝试解析其他警告事件 (只有在确认是警告消息时才处理)
        if self.is_bounce_warning(line) {
            if let Some(captures) = self.warning_general_regex.captures(line) {
                return self.parse_general_warning(captures, base_event);
            }
        }

        None
    }

    /// 解析发送者投递失败通知事件
    fn parse_sender_notification(
        &self,
        captures: regex::Captures,
        base_event: BaseEvent,
    ) -> Option<BounceEvent> {
        let original_queue_id = captures.get(1)?.as_str().to_string();
        let bounce_queue_id = captures.get(2)?.as_str().to_string();

        Some(BounceEvent::SenderNotification {
            base: base_event,
            original_queue_id,
            bounce_queue_id,
        })
    }

    /// 解析邮件管理员投递失败通知事件
    fn parse_postmaster_notification(
        &self,
        captures: regex::Captures,
        base_event: BaseEvent,
    ) -> Option<BounceEvent> {
        let original_queue_id = captures.get(1)?.as_str().to_string();
        let bounce_queue_id = captures.get(2)?.as_str().to_string();

        Some(BounceEvent::PostmasterNotification {
            base: base_event,
            original_queue_id,
            bounce_queue_id,
        })
    }

    /// 解析格式错误警告事件
    fn parse_malformed_warning(&self, base_event: BaseEvent) -> Option<BounceEvent> {
        Some(BounceEvent::Warning {
            base: base_event,
            warning_type: BounceWarningType::MalformedRequest,
            details: "malformed request".to_string(),
        })
    }

    /// 解析其他警告事件
    /// 处理BOUNCE服务的其他警告类型
    fn parse_general_warning(
        &self,
        captures: regex::Captures,
        base_event: BaseEvent,
    ) -> Option<BounceEvent> {
        let warning_message = captures.get(1)?.as_str();

        // 根据警告内容确定警告类型
        let warning_type = if warning_message.contains("malformed") {
            BounceWarningType::MalformedRequest
        } else if warning_message.contains("configuration") || warning_message.contains("config") {
            BounceWarningType::Configuration
        } else if warning_message.contains("resource")
            || warning_message.contains("memory")
            || warning_message.contains("disk")
        {
            BounceWarningType::ResourceLimit
        } else {
            BounceWarningType::Other
        };

        Some(BounceEvent::Warning {
            base: base_event,
            warning_type,
            details: warning_message.to_string(),
        })
    }

    /// 判断是否是BOUNCE组件的警告消息
    fn is_bounce_warning(&self, message: &str) -> bool {
        // 只有在日志级别为Warning时才考虑作为警告处理
        // 这个判断会在实际使用中由MasterParser的日志级别提取来处理
        message.contains("malformed")
            || message.contains("configuration")
            || message.contains("resource")
            || message.contains("memory")
            || message.contains("disk")
            || message.contains("queue")
            || message.contains("bounce")
    }
}

impl ComponentParser for BounceParser {
    fn parse(&self, message: &str) -> Result<ComponentEvent, ParseError> {
        // 创建一个临时的BaseEvent，用于解析
        // 在实际使用中，这些字段会被MasterParser正确填充
        let base_event = BaseEvent {
            timestamp: chrono::Utc::now(),
            hostname: "temp".to_string(),
            component: "bounce".to_string(),
            process_id: 0,
            log_level: crate::events::base::PostfixLogLevel::Info,
            raw_message: message.to_string(),
        };

        if let Some(bounce_event) = self.parse_line(message, base_event) {
            Ok(ComponentEvent::Bounce(bounce_event))
        } else {
            Err(ParseError::ComponentParseError {
                component: "bounce".to_string(),
                reason: "无法识别的bounce日志格式".to_string(),
            })
        }
    }

    fn component_name(&self) -> &'static str {
        "bounce"
    }

    fn can_parse(&self, message: &str) -> bool {
        // 检查是否包含bounce特征
        self.sender_notification_regex.is_match(message)
            || self.postmaster_notification_regex.is_match(message)
            || self.warning_malformed_regex.is_match(message)
            // 对于一般警告，我们需要更精确的判断，而不是匹配所有内容
            || self.is_bounce_warning(message)
    }
}

impl Default for BounceParser {
    fn default() -> Self {
        Self::new()
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::events::base::BaseEvent;
    use chrono::{DateTime, Utc};

    fn create_test_base_event() -> BaseEvent {
        BaseEvent {
            timestamp: DateTime::parse_from_rfc3339("2024-04-27T16:20:48+00:00")
                .unwrap()
                .with_timezone(&Utc),
            hostname: "m01".to_string(),
            component: "bounce".to_string(),
            process_id: 133,
            log_level: crate::events::base::PostfixLogLevel::Info,
            raw_message: "test message".to_string(),
        }
    }

    #[test]
    fn test_parse_sender_notification() {
        let parser = BounceParser::new();
        let base_event = create_test_base_event();

        let message = "5FC392A20996: sender non-delivery notification: 732B92A209A3";

        let result = parser.parse_line(message, base_event);
        assert!(result.is_some());

        if let Some(BounceEvent::SenderNotification {
            original_queue_id,
            bounce_queue_id,
            ..
        }) = result
        {
            assert_eq!(original_queue_id, "5FC392A20996");
            assert_eq!(bounce_queue_id, "732B92A209A3");
        } else {
            panic!("解析结果类型不正确");
        }
    }

    #[test]
    fn test_parse_postmaster_notification() {
        let parser = BounceParser::new();
        let base_event = create_test_base_event();

        let message = "633F488423: postmaster non-delivery notification: 6DFA788422";

        let result = parser.parse_line(message, base_event);
        assert!(result.is_some());

        if let Some(BounceEvent::PostmasterNotification {
            original_queue_id,
            bounce_queue_id,
            ..
        }) = result
        {
            assert_eq!(original_queue_id, "633F488423");
            assert_eq!(bounce_queue_id, "6DFA788422");
        } else {
            panic!("解析结果类型不正确");
        }
    }

    #[test]
    fn test_parse_malformed_warning() {
        let parser = BounceParser::new();
        let base_event = create_test_base_event();

        // 注意：MasterParser已剥离"warning:"前缀
        let message = "malformed request";

        let result = parser.parse_line(message, base_event);
        assert!(result.is_some());

        if let Some(BounceEvent::Warning {
            warning_type,
            details,
            ..
        }) = result
        {
            assert!(matches!(warning_type, BounceWarningType::MalformedRequest));
            assert_eq!(details, "malformed request");
        } else {
            panic!("解析结果类型不正确");
        }
    }

    #[test]
    fn test_parse_general_warning() {
        let parser = BounceParser::new();
        let base_event = create_test_base_event();

        // 注意：MasterParser已剥离"warning:"前缀
        let message = "configuration file error";

        let result = parser.parse_line(message, base_event);
        assert!(result.is_some());

        if let Some(BounceEvent::Warning {
            warning_type,
            details,
            ..
        }) = result
        {
            assert!(matches!(warning_type, BounceWarningType::Configuration));
            assert_eq!(details, "configuration file error");
        } else {
            panic!("解析结果类型不正确");
        }
    }

    #[test]
    fn test_can_parse() {
        let parser = BounceParser::new();

        // 应该能解析的消息
        assert!(parser.can_parse("5FC392A20996: sender non-delivery notification: 732B92A209A3"));
        assert!(parser.can_parse("633F488423: postmaster non-delivery notification: 6DFA788422"));
        // 注意：MasterParser已剥离"warning:"前缀
        assert!(parser.can_parse("malformed request"));
        assert!(parser.can_parse("configuration error"));

        // 不应该解析的消息
        assert!(!parser.can_parse("some random message"));
        assert!(!parser.can_parse("info: normal operation"));
        assert!(!parser.can_parse("5FC392A20996: to=<user@domain.com>, status=sent"));
    }

    #[test]
    fn test_component_parser_interface() {
        let parser = BounceParser::new();

        // 测试component_name
        assert_eq!(parser.component_name(), "bounce");

        // 测试parse方法成功案例
        let result = parser.parse("5FC392A20996: sender non-delivery notification: 732B92A209A3");
        assert!(result.is_ok());

        if let Ok(ComponentEvent::Bounce(bounce_event)) = result {
            match bounce_event {
                BounceEvent::SenderNotification {
                    original_queue_id,
                    bounce_queue_id,
                    ..
                } => {
                    assert_eq!(original_queue_id, "5FC392A20996");
                    assert_eq!(bounce_queue_id, "732B92A209A3");
                }
                _ => panic!("解析结果类型不正确"),
            }
        } else {
            panic!("解析失败");
        }

        // 测试parse方法失败案例
        let result = parser.parse("some random text that should not match");
        assert!(result.is_err());

        if let Err(ParseError::ComponentParseError { component, reason }) = result {
            assert_eq!(component, "bounce");
            assert!(reason.contains("无法识别的bounce日志格式"));
        } else {
            panic!("错误类型不正确");
        }
    }
}