idns : Fast DNS Record Parsing with Racing Query

DNS record parsing library with racing query support. Works with idoq (DNS over QUIC) and idoh (DNS over HTTPS).

Features
Installation
Usage
API Reference
Architecture
Project Structure
Tech Stack
History

Features

Racing DNS queries across multiple servers for optimal latency
Automatic server weight adjustment based on response time
Built-in caching with TTL support
Parse A, AAAA, MX, TXT records
SPF record verification (IPv4/IPv6)
Async/await support

Installation

[dependencies]
idns = { version = "0.2", features = ["spf"] }

Available features:

a - A record parsing
aaaa - AAAA record parsing
mx - MX record parsing
spf - SPF verification (includes a, aaaa, mx)
cache - Response caching (enabled by default)

Usage

Query DNS Records

use idns::{DnsRace, Query, Spf};

let dns = DnsRace::new(dns_servers);

// Query SPF record
if let Some(records) = dns.query::<Spf>("gmail.com").await? {
  for spf in records {
    println!("{spf:?}");
  }
}

SPF Verification

use std::net::IpAddr;
use idns::{Spf, spf::Status};

let ip: IpAddr = "209.85.220.41".parse()?;
let status = Spf::verify(&dns, "gmail.com", ip).await?;

match status {
  Status::Pass => println!("Accept"),
  Status::Fail => println!("Reject"),
  Status::SoftFail => println!("Mark as suspicious"),
  Status::Neutral => println!("No policy"),
  Status::None => println!("No SPF record"),
}

API Reference

Structs

Name	Description
`DnsRace<Q>`	Racing DNS client that queries multiple servers concurrently
`Cache<P>`	TTL-based cache for DNS records
`Answer`	Raw DNS answer containing name, value, type_id, ttl
`A`	Parsed A record (IPv4)
`Aaaa`	Parsed AAAA record (IPv6)
`Mx`	Parsed MX record with priority and server
`Spf`	Parsed SPF record with IP ranges and mechanisms

Enums

Name	Description
`QType`	DNS record types (A, AAAA, MX, TXT, etc.)
`spf::Status`	SPF verification result (Pass, Fail, SoftFail, Neutral, None)
`spf::Act`	SPF mechanism action

Traits

Name	Description
`Query`	DNS query interface
`Parse`	DNS record parsing interface

Architecture

graph TD
  A[User Code] --> B[DnsRace]
  B --> C[DNS Server 1]
  B --> D[DNS Server 2]
  B --> E[DNS Server N]
  C --> F[Race Result]
  D --> F
  E --> F
  F --> G[Cache]
  G --> H[Parse Trait]
  H --> I[A / AAAA / MX / Spf]

SPF Verification Flow

Spf::verify performs multi-stage verification:

Query domain's SPF record (TXT)
Check if IP matches ip4/ip6 ranges in record
Process host mechanisms (include, redirect, a, mx, exists)
If no match, fallback to MX record check - query domain's MX servers and verify if IP belongs to any MX host
Return final status

graph TD
  A[Spf::verify] --> B[Query SPF TXT]
  B --> C{Has Record?}
  C -->|No| D[Status::None]
  C -->|Yes| E[verify_deep]
  E --> F[Check ip4_pass/ip4_fail HashSet]
  F --> G[Check ip4/ip6 CIDR ranges]
  G --> H{Match?}
  H -->|Yes| I[Return Act]
  H -->|No| J[Process Host Mechanisms]
  J --> K[include: recursive verify]
  J --> L[redirect: jump to other SPF]
  J --> M[a: check A/AAAA record]
  J --> N[mx: check MX hosts IP]
  J --> O[exists: check A record exists]
  K --> P{Any Match?}
  L --> P
  M --> P
  N --> P
  O --> P
  P -->|Yes| Q[Return Act]
  P -->|No| R[Return all or Neutral]
  R --> S{Status != Pass?}
  I --> S
  Q --> S
  S -->|Yes| T[MX Fallback]
  S -->|No| U[Return Status]
  T --> V[Query MX Records]
  V --> W[For each MX server]
  W --> X[Query A/AAAA of MX host]
  X --> Y{IP Match?}
  Y -->|Yes| Z[Status::Pass]
  Y -->|No| U
  Z --> U

MX fallback ensures emails from legitimate mail servers pass verification even without explicit SPF authorization.

Caching

Each record type has independent cache with configurable TTL:

Record	TTL	Description
A	300s	IPv4 address
AAAA	300s	IPv6 address
MX	600s	Mail exchange
SPF	600s	SPF record

During SPF verification, A/AAAA/MX queries are cached. When processing include or redirect mechanisms that reference the same domain, cached results are reused. This avoids redundant DNS queries and parsing.

Cache uses papaya HashMap for concurrent read/write access, suitable for high-concurrency scenarios.

Project Structure

src/
├── lib.rs        # Public exports
├── error.rs      # Error definitions
├── qtype.rs      # DNS record types
├── dns_race.rs   # Racing DNS client
├── cache.rs      # TTL cache
└── parse/
    ├── mod.rs    # Parse trait
    ├── a.rs      # A record
    ├── aaaa.rs   # AAAA record
    ├── mx.rs     # MX record
    └── spf.rs    # SPF record & verification

Tech Stack

thiserror - Error handling
papaya - Concurrent hashmap
static_init - Lazy static initialization

History

SPF (Sender Policy Framework) was first proposed in 2003 by Meng Weng Wong to combat email spoofing. The protocol allows domain owners to specify which mail servers are authorized to send email on their behalf.

Before SPF, anyone could send email claiming to be from any domain. This made phishing attacks trivially easy. SPF introduced a simple TXT record format that lists authorized IP addresses and mechanisms.

The "v=spf1" prefix you see in SPF records stands for "version SPF 1". Despite being over 20 years old, this remains the only version in widespread use. The protocol was standardized as RFC 7208 in 2014.

Racing DNS queries, as implemented in this library, is a technique popularized by Google's "Happy Eyeballs" algorithm (RFC 6555). The idea is simple: send queries to multiple servers simultaneously and use the first response. This dramatically reduces latency in unreliable network conditions.

About

This project is an open-source component of js0.site ⋅ Refactoring the Internet Plan.

We are redefining the development paradigm of the Internet in a componentized way. Welcome to follow us:

idns : 竞速查询的 DNS 记录解析库

支持竞速查询的 DNS 记录解析库。可配合 idoq（DNS over QUIC）和 idoh（DNS over HTTPS）使用。

特性

多服务器竞速查询，获取最优延迟
根据响应时间自动调整服务器权重
内置 TTL 缓存
解析 A、AAAA、MX、TXT 记录
SPF 记录验证（支持 IPv4/IPv6）
异步支持

安装

[dependencies]
idns = { version = "0.2", features = ["spf"] }

可用特性：

a - A 记录解析
aaaa - AAAA 记录解析
mx - MX 记录解析
spf - SPF 验证（包含 a、aaaa、mx）
cache - 响应缓存（默认启用）

使用

查询 DNS 记录

use idns::{DnsRace, Query, Spf};

let dns = DnsRace::new(dns_servers);

// 查询 SPF 记录
if let Some(records) = dns.query::<Spf>("gmail.com").await? {
  for spf in records {
    println!("{spf:?}");
  }
}

SPF 验证

use std::net::IpAddr;
use idns::{Spf, spf::Status};

let ip: IpAddr = "209.85.220.41".parse()?;
let status = Spf::verify(&dns, "gmail.com", ip).await?;

match status {
  Status::Pass => println!("接受"),
  Status::Fail => println!("拒绝"),
  Status::SoftFail => println!("标记为可疑"),
  Status::Neutral => println!("无策略"),
  Status::None => println!("无 SPF 记录"),
}

API 参考

结构体

名称	说明
`DnsRace<Q>`	竞速 DNS 客户端，并发查询多服务器
`Cache<P>`	基于 TTL 的 DNS 记录缓存
`Answer`	原始 DNS 应答，包含 name、value、type_id、ttl
`A`	解析后的 A 记录（IPv4）
`Aaaa`	解析后的 AAAA 记录（IPv6）
`Mx`	解析后的 MX 记录，含优先级和服务器
`Spf`	解析后的 SPF 记录，含 IP 范围和机制

枚举

名称	说明
`QType`	DNS 记录类型（A、AAAA、MX、TXT 等）
`spf::Status`	SPF 验证结果（Pass、Fail、SoftFail、Neutral、None）
`spf::Act`	SPF 机制动作

Trait

名称	说明
`Query`	DNS 查询接口
`Parse`	DNS 记录解析接口

架构

graph TD
  A[用户代码] --> B[DnsRace]
  B --> C[DNS 服务器 1]
  B --> D[DNS 服务器 2]
  B --> E[DNS 服务器 N]
  C --> F[竞速结果]
  D --> F
  E --> F
  F --> G[Cache]
  G --> H[Parse Trait]
  H --> I[A / AAAA / MX / Spf]

SPF 验证流程

Spf::verify 执行多阶段验证：

查询域名的 SPF 记录（TXT）
检查 IP 是否匹配记录中的 ip4/ip6 范围
处理 host 机制（include、redirect、a、mx、exists）
若无匹配，回退到 MX 记录检查 - 查询域名的 MX 服务器，验证 IP 是否属于任意 MX 主机
返回最终状态

graph TD
  A[Spf::verify] --> B[查询 SPF TXT]
  B --> C{有记录?}
  C -->|否| D[Status::None]
  C -->|是| E[verify_deep]
  E --> F[检查 ip4_pass/ip4_fail HashSet]
  F --> G[检查 ip4/ip6 CIDR 范围]
  G --> H{匹配?}
  H -->|是| I[返回 Act]
  H -->|否| J[处理 Host 机制]
  J --> K[include: 递归验证]
  J --> L[redirect: 跳转其他 SPF]
  J --> M[a: 检查 A/AAAA 记录]
  J --> N[mx: 检查 MX 主机 IP]
  J --> O[exists: 检查 A 记录存在]
  K --> P{任意匹配?}
  L --> P
  M --> P
  N --> P
  O --> P
  P -->|是| Q[返回 Act]
  P -->|否| R[返回 all 或 Neutral]
  R --> S{Status != Pass?}
  I --> S
  Q --> S
  S -->|是| T[MX 回退]
  S -->|否| U[返回 Status]
  T --> V[查询 MX 记录]
  V --> W[遍历 MX 服务器]
  W --> X[查询 MX 主机的 A/AAAA]
  X --> Y{IP 匹配?}
  Y -->|是| Z[Status::Pass]
  Y -->|否| U
  Z --> U

MX 回退机制确保来自合法邮件服务器的邮件即使没有显式 SPF 授权也能通过验证。

缓存机制

每种记录类型有独立缓存，TTL 可配置：

记录	TTL	说明
A	300s	IPv4 地址
AAAA	300s	IPv6 地址
MX	600s	邮件交换
SPF	600s	SPF 记录

SPF 验证过程中，A/AAAA/MX 查询结果会被缓存。处理 include 或 redirect 机制时若引用相同域名，直接复用缓存结果，避免重复 DNS 查询和解析。

缓存使用 papaya HashMap 实现并发读写，适用于高并发场景。

项目结构

src/
├── lib.rs        # 公开导出
├── error.rs      # 错误定义
├── qtype.rs      # DNS 记录类型
├── dns_race.rs   # 竞速 DNS 客户端
├── cache.rs      # TTL 缓存
└── parse/
    ├── mod.rs    # Parse trait
    ├── a.rs      # A 记录
    ├── aaaa.rs   # AAAA 记录
    ├── mx.rs     # MX 记录
    └── spf.rs    # SPF 记录与验证

技术栈

thiserror - 错误处理
papaya - 并发哈希表
static_init - 延迟静态初始化

历史

SPF（发件人策略框架）由 Meng Weng Wong 于 2003 年提出，用于对抗邮件伪造。该协议允许域名所有者指定哪些邮件服务器有权代表其发送邮件。

在 SPF 出现之前，任何人都可以发送声称来自任意域名的邮件，这使得钓鱼攻击极为容易。SPF 引入了简单的 TXT 记录格式，列出授权的 IP 地址和机制。

SPF 记录中的 "v=spf1" 前缀表示 "SPF 版本 1"。尽管已有 20 多年历史，这仍是唯一广泛使用的版本。该协议于 2014 年标准化为 RFC 7208。

本库实现的竞速 DNS 查询技术源自 Google 的 "Happy Eyeballs" 算法（RFC 6555）。原理很简单：同时向多个服务器发送查询，使用最先返回的响应。这在不稳定的网络环境下能显著降低延迟。

关于

本项目为 js0.site ⋅ 重构互联网计划的开源组件。

我们正在以组件化的方式重新定义互联网的开发范式，欢迎关注：

idns 0.2.7