国际化域名 (IDN)
包含非ASCII Unicode字符的域名,内部以Punycode(xn--...)存储,但向用户显示为Unicode,安全隐患:同形字攻击。
What Are Internationalized Domain Names?
Internationalized Domain Names (IDNs) allow domain names to contain non-ASCII characters — letters from Arabic, Chinese, Cyrillic, Devanagari, Hebrew, Japanese, Korean, Thai, and hundreds of other scripts. Before IDNs, domain names were restricted to ASCII letters, digits, and hyphens (LDH characters).
IDNs let users register and navigate to domain names written entirely in their native scripts: 例え.jp, مثال.إختبار, пример.испытание. This dramatically improves accessibility for the billions of internet users whose primary scripts are not Latin.
How IDNs Work Technically
The DNS infrastructure only understands ASCII. IDNs bridge this gap using Punycode encoding: non-ASCII domain labels are converted to an ASCII-compatible encoding (ACE) prefixed with xn--. The Punycode-encoded forms travel through DNS; the original Unicode forms are displayed to users.
User sees: 例え.jp
DNS query: xn--r8jz45g.jp
The conversion is handled by the operating system's DNS resolver or the browser's IDN processing layer. Applications interact with the human-readable Unicode form; the network sees only ASCII.
IDNA Standards
Two versions of the IDNA (Internationalizing Domain Names in Applications) protocol exist:
- IDNA2003 (RFC 3490): First standard. Uses NAMEPREP profile of Stringprep for normalization.
- IDNA2008 (RFC 5891/5892): Stricter. Removes some characters IDNA2003 allowed (like
ß→ssmapping). More conservative and consistent.
Most modern systems use IDNA2008. Some compatibility issues exist between the versions for a small set of characters.
Security Considerations
IDNs introduce homograph attacks: characters from different scripts can look identical to Latin letters. For example, Cyrillic а (U+0430) looks like Latin a (U+0061). A malicious domain pаypal.com might use Cyrillic а to impersonate paypal.com.
Browsers defend against this by:
- Displaying the Punycode form (xn--...) when a domain mixes scripts or contains confusable characters.
- Restricting which labels show as Unicode (typically: all characters from a single script, using registered scripts for the TLD).
# This URL shows as Punycode in Chrome (mixed scripts)
http://xn--80ak6aa92e.com/
Using IDNs in Python
import idna # pip install idna
# Encode: Unicode → Punycode
idna.encode("例え.jp") # b"xn--r8jz45g.jp"
idna.encode("münchen.de") # b"xn--mnchen-3ya.de"
idna.encode("пример.испытание") # b"xn--e1afmapc.xn--80akhbyknj4f"
# Decode: Punycode → Unicode
idna.decode("xn--r8jz45g.jp") # "例え.jp"
idna.decode("xn--mnchen-3ya.de") # "münchen.de"
# Standard library (limited to IDNA2003)
"例え.jp".encode("idna") # b"xn--r8jz45g.jp"
b"xn--r8jz45g.jp".decode("idna") # "例え.jp"
IDN Email Addresses
Email also supports international addresses (EAI — Email Address Internationalization, RFC 6531). A full internationalized email address can use Unicode in both the local part and the domain:
用户@例子.广告 (Chinese)
उपयोगकर्ता@उदाहरण.भारत (Hindi)
Support for EAI in mail clients and servers is still growing.
Quick Facts
| Property | Value |
|---|---|
| Full name | Internationalized Domain Names in Applications (IDNA) |
| DNS representation | Punycode-encoded ASCII (xn--...) |
| User-visible form | Unicode characters in native script |
| Current standard | IDNA2008 (RFC 5891/5892) |
| Max label length (encoded) | 63 ASCII characters |
| Homograph attacks | Mitigated by browser mixed-script detection |
| Python library | idna (pip install idna) for IDNA2008 |
相关术语
网页与 HTML 中的更多内容
声明响应字符编码的HTTP头参数(Content-Type: text/html; charset=utf-8),优先级高于文档内的编码声明。
通过::before和::after伪元素使用Unicode转义插入生成内容的CSS属性:content: '\2713'可插入✓。
CSS properties (direction, writing-mode, unicode-bidi) controlling text layout direction. Works with Unicode …
HTML中字符的文本表示方式,有三种形式:命名(&)、十进制(&)、十六进制(&),对于与HTML语法冲突的字符是必需的。
ECMAScript Internationalization API providing locale-aware string comparison (Collator), number formatting (NumberFormat), date …
将Unicode域名转换为xn--前缀ASCII字符串的ASCII兼容编码,例如münchen.de → xn--mnchen-3ya.de。
CSS supports Unicode via escape sequences (\2713 for ✓), the content property …
XML版本的数字字符引用:✓或✓,XML只有5个命名实体(& < > " '),而HTML5有2,231个。
选择特定字形变体的字符(U+FE00–U+FE0F、U+E0100–U+E01EF),VS15(U+FE0E)表示文本呈现,VS16(U+FE0F)表示表情符号呈现。
使用人类可读名称的HTML实体:© → ©,— → —。HTML5定义了2,231个命名引用,且区分大小写。