[Unicode]  Technical Reports
 

Unicode Technical Standard #46

Unicode IDNA Compatibility Processing

Version 6.0.0
Editors Mark Davis (markdavis@google.com),
Michel Suignard (michel@suignard.com)
Date 2010-10-28
This Version http://www.unicode.org/reports/tr46/tr46-5.html
Previous Version http://www.unicode.org/reports/tr46/tr46-3.html
Latest Version http://www.unicode.org/reports/tr46/
Latest Proposed Update http://www.unicode.org/reports/tr46/proposed.html
Revision 5

Summary

Client software, such as browsers and emailers, faces a difficult transition from the version of international domain names approved in 2003 (IDNA2003), to the revision approved in 2010 (IDNA2008). The specification in this document provides a mechanism that minimizes the impact of this transition for client software, allowing client software to access domains that are valid under either system.

The specification provides two main features: One is a comprehensive mapping to support current user expectations for casing and other variants of domain names. Such a mapping is allowed by IDNA2008. The second is a compatibility mechanism that supports the existing domain names that were allowed under IDNA2003. This second feature is intended to improve client behavior during the transitional period.

Status

This document has been reviewed by Unicode members and other interested parties, and has been approved for publication by the Unicode Consortium. This is a stable document and may be used as reference material or cited as a normative reference by other specifications.

A Unicode Technical Standard (UTS) is an independent specification. Conformance to the Unicode Standard does not imply conformance to any UTS.

Please submit corrigenda and other comments with the online reporting form [Feedback]. Related information that is useful in understanding this document is found in the References. For the latest version of the Unicode Standard see [Unicode]. For a list of current Unicode Technical Reports see [Reports]. For more information about versions of the Unicode Standard, see [Versions].

Contents




1 Introduction

One of the great strengths of domain names is universality. The URL http://Apple.com goes to Apple's website from anywhere in the world, using any browser. The email address markdavis@google.com can be used to send email to an editor of this specification from anywhere in the world, using any emailer.

Initially, domain names were restricted to ASCII characters. This was a significant burden on people using other characters. Suppose, for example, that the domain name system had been invented by Greeks, and one could only use Greek characters in URLs. Rather than apple.com, one would have to write something like αππλε.κομ. An English speaker would not only have to be acquainted with Greek characters, but would also have to pick those Greek letters that would correspond to the desired English letters. One would have to guess at the spelling of particular words, because there are not exact matches between scripts.

Most of the world’s population faced this situation until recently, because their languages use non-ASCII characters. A system was introduced in 2003 for internationalized domain names (IDN). This system is called Internationalizing Domain Names for Applications, or IDNA2003 for short. This mechanism supports IDNs by means of a client software transformation into a format known as Punycode. A revision of IDNA was approved in 2010 (IDNA2008). This revision has a number of incompatibilities with IDNA2003.

The incompatibilities force implementers of client software, such as browsers and emailers, to face difficult choices during the transition period as registries shift from IDNA2003 to IDNA2008. This document specifies a mechanism that minimizes the impact of this transition for client software, allowing client software to access domains that are valid under either system.

The specification provides two main features. The first is a comprehensive mapping to support current user expectations for casing and other variants of domain names. Such a mapping is allowed by IDNA2008. The second feature is a compatibility mechanism that supports the existing domain names that were allowed under IDNA2003. This second feature is intended to improve client behavior during the transitional period. This specification contains both normative and informative material. Only the conformance clauses and the text that they directly or indirectly reference are considered normative.

1.1 IDNA2003

The series of RFCs collectively known as IDNA2003 [IDNA2003] allows domain names to contain non-ASCII Unicode characters, which includes not only the characters needed for Latin-script languages other than English (such as Å, Ħ, or Þ), but also different scripts, such as Greek, Cyrillic, Tamil, or Korean. An internationalized domain name such as Bücher.de can then be used in an "internationalized" URL, called an IRI, such as http://Bücher.de#titel.

The IDNA mechanism for allowing non-ASCII Unicode characters in domain names involves applying the following steps to each label in the domain name that contains Unicode characters:

  1. Transforming (mapping) a Unicode string to remove case and other variant differences.
  2. Checking the resulting string for validity, according to certain rules.
  3. Transforming the Unicode characters into a DNS-compatible ASCII string using a specialized encoding called Punycode [RFC3492].

For example, typing the IRI http://Bücher.de into the address bar of any modern browser goes to a corresponding site, even though the "ü" is not an ASCII character. This works because the IDN in that IRI resolves to the Punycode string which is actually stored by the DNS for that site. Similarly, when a browser interprets a web page containing a link such as <a href="http://Bücher.de">, the appropriate site is reached. (In this document, phrases such as "a browser interprets" refer to domain names parsed out of IRIs entered in an address bar as well as to those contained in links internal to HTML text.)

In the case of IDN Bücher.de, the Punycode value actually used for the domain names on the wire is xn--bcher-kva.de. The Punycode version is also typically transformed back into Unicode form for display. The resulting display string will be a string which has already been mapped according to the IDNA2003 rules. This example results in a display string for the IRI that has been casefolded to lowercase:

http://Bücher.de http://xn--bcher-kva.de http://bücher.de

A major limitation of IDNA2003 is its restriction to the repertoire of characters in Unicode 3.2, which means that some modern languages are excluded or not fully supported. Furthermore, within the constraints of IDNA2003, there is no simple way to extend the repertoire. IDNA2003 also does not make it clear to users of registries exactly which string they are registering for a domain name (between Bücher.de and bücher.de, for example).

1.2 IDNA2008

In early 2010, a new version of IDNA was approved. Like IDNA2003, this version consists of a collection of RFCs and is called IDNA2008 [IDNA2008]. IDNA2008 is intended to solve the major problems in IDNA2003. It extends the valid repertoire of characters in domain names, and establishes an automatic process for updating to future versions of the Unicode Standard. Furthermore, it defines the concept of a valid domain name clearly, so that registrants understand exactly what domain name string is being registered.

Processing in IDNA2008 is identical to IDNA2003 for many common domain names. Both IDNA2003 and IDNA2008 transform a Unicode domain name in an IRI (like http://öbb.at) to the Punycode version (like http://xn--bb-eka.at). However, IDNA2008 does not maintain strict backward compatibility with IDNA2003. The main differences are:

For more details, see Section 7, IDNA Comparison.

1.3 Transition Considerations

The differences between IDNA2008 and IDNA2003 may cause interoperability and security problems. They affect extremely common characters, such as all uppercase characters, all halfwidth or fullwidth characters (commonly used in Japan, China, and Korea), and certain other characters like the German eszett (U+00DF ß LATIN SMALL LETTER SHARP S) and Greek final sigma (U+03C2 ς GREEK SMALL LETTER FINAL SIGMA).

1.3.1 Mapping

IDNA2003 requires a mapping phase, which maps ÖBB.at to öbb.at, for example. Mapping typically involves mapping uppercase characters to their lowercase pairs, but it also involves other types of mappings between equivalent characters, such as mapping halfwidth katakana characters to normal katakana characters in Japanese. The mapping phase in IDNA2003 was included to match the insensitivity of ASCII domain names. Users are accustomed to having both CNN.com and cnn.com work identically. They expect domain names with accents to have the same casing behavior, so that ÖBB.at is the same as öbb.at. There are variations similar to case differences in other scripts. The IDNA2003 mapping is based on data specified in the Unicode Standard, Version 3.2; this mapping was later formalized as the Unicode property [NFKC_Casefold].

IDNA2008 does not require a mapping phase, but does permit one (called "Local Mapping" or "Custom Mapping"). For more information on the permitted mappings, see the Protocol document of [IDNA2008], Section 4.2, Permitted Character and Label Validation and Section 5.2, Conversion to Unicode.

The UTS #46 specification defines a mapping consistent with the normative requirements of the IDNA2008 protocol, and which is as compatible as possible with IDNA2003. For client software, this provides behavior that is the most consistent with user expectations about the handling of domain names with existing data—namely, that domain names will map consistently both on clients supporting IDNA2003 and on clients supporting IDNA2008 with the UTS #46 mapping.

1.3.2 Deviations

There are a few situations where the use of IDNA2008 without compatibility mapping will result in the resolution of IDNs to different IP addresses than in IDNA2003, unless the registry or registrant takes special action. This affects a very small number of characters, but because these characters are very common in particular languages, a significant number of domain names in those languages are affected. This set of characters is referred to as "Deviations" and is shown in Table 1, Deviation Characters, illustrated in the context of IRIs.

Table 1. Deviation Characters
Char
Example
IDNA2003 Result
IDNA2008 Result
ß
00DF
href="http://faß.de"
http://fass.de
= http://fass.de
http://faß.de
= http://xn--fa-hia.de
ς
03C2
href="http://βόλος.com"
http://βόλοσ.com
= http://xn--nxasmq6b.com
http://βόλος.com
= http://xn--nxasmm1c.com
ZWJ
200D
href="http://ශ්‍රී.com"
http://ශ්රී.com
= http://xn--10cl1a0b.com
http://ශ්‍රී.com
= http://xn--10cl1a0b660p.com
ZWNJ
200C
href="http://نامه‌ای.com"
http://نامهای.com
= http://xn--mgba3gch31f.com
http://نامه‌ای.com
= http://xn--mgba3gch31f060k.com

For more information on the rationale for the occurrence of these Deviations in IDNA2008, see the [IDN FAQ].

The differences in interpretation of Deviation characters result in potential for security exploits. Consider a scenario involving http://www.sparkasse-gießen.de, a German IRI containing an IDN for "Gießen Savings and Loan".

  1. Alice's browser supports IDNA2003. Under those rules, http://www.sparkasse-gießen.de is mapped to http://www.sparkasse-giessen.de, which leads to a site with the IP address 01.23.45.67.
  2. She visits her friend Bob, and checks her bank statement on his browser. His browser supports IDNA2008. Under those rules, http://www.sparkasse-gießen.de is also valid, but converts to a different Punycode domain name in http://www.xn--sparkasse-gieen-2ib.de. This can lead to a different site with the IP address 101.123.145.167, a spoof site.

Alice ends up at the phishing site, supplies her bank password, and her money is stolen. While the .DE registar (DENIC) might have a policy about bundling all of the variants of ß together (so that they all have the same owner) it is not required of registries. It is unlikely that all registries will have and enforce such a bundling policy in all such cases.

There are two Deviations of particular concern. IDNA2008 allows the joiner characters (ZWJ and ZWNJ) in labels. By contrast, these are removed by the mapping in IDNA2003. When used in the intended contexts in particular scripts, the joiner characters produce a noticeable change in displayed text. However, when used between any other characters in those scripts, or in any other scripts, they are invisible. For example, when used between the Latin characters "a" and "b" there is no visible different: the sequence "a<ZWJ>b" looks just like "ab".

Because of the visual confusability introduced by the joiner characters, IDNA2008 provides a special category for them called CONTEXTJ, and only permits CONTEXTJ characters in limited contexts: certain sequences of Arabic or Indic characters. However, applications that perform IDNA2008 lookup are not required to check for these contexts, so overall security is dependent on registries having correct implementations. Moreover, the IDNA2008 context restrictions do not catch most cases where distinct domain names have visually confusable appearances because of ZWJ and ZWNJ.

2 Unicode IDNA Compatibility Processing

To satisfy user expectations for mapping, and provide maximal compatibility with IDNA2003, this document specifies a mapping for use with IDNA2008. In addition, to transition more smoothly to IDNA2008, this document provides a Unicode algorithm for a standardized processing that allows conformant implementations to minimize the security and interoperability problems caused by the differences between IDNA2003 and IDNA2008. This Unicode IDNA Compatibility Processing is structured according to IDNA2003 principles, but extends those principles to Unicode 5.2 and later. It also incorporates the repertoire extensions provided by IDNA2008.

Where the transition processing is not needed, UTS #46 can be used purely as a preprocessing (local mapping) for IDNA2008 by claiming conformance specifically to Conformance Clause C3.

By using this Compatibility Processing, a domain name such as ÖBB.at will be mapped to the valid domain name öbb.at, thus matching user expectation for case behavior in domain names. For transitional use, the Compatibility Processing also allows domain names containing symbols and punctuation that were valid in IDNA2003, such as √.com (which has an associated web page). Such domain names containing symbols will gradually disappear as registries shift to IDNA2008.

Implementations may also restrict or flag (in a UI) domain names that include symbols and punctuation. For more information, see Unicode Technical Report # 36, "Unicode Security Considerations" [UTR36].

Using the Unicode IDNA Compatibility Processing to transform an IDN into a form suitable for DNS lookup is similar to the tactic of "try IDNA2008 then try IDNA2003". However, this approach avoids a potentially problematic dual lookup. It allows browsers and other clients, such as search engines, to have a single processing step, without the burden of maintaining two different implementations and multiple tables. It accounts for a number of edge cases that would cause problems, and provides a stable definition with predictable results.

The Unicode IDNA Compatibility Processing also provides alternate mappings for the Deviation characters. This facilitates the transition from IDNA2003 to IDNA2008. It is up to the registeries to decide how to handle the transition, for example, by either bundling or blocking the Deviation characters that they support. The course of the transition will also depend on how soon the IDNA2003 client software is retired.

The term "registries" includes far more than top-level registries, such as for .de or .com. For example, .blogspot.com has more domain names registered than most top-level registries. There may be different policies in place for a registry and any of its subregistries. Thus millions of registries need to be considered in a transition strategy, not just hundreds.

The retirement of IDNA2003 client software may also take considerable time. IE6 was superseded in October 2006, yet as of April 2010, it still has a usage share of over 20%—higher than all other browsers except IE8! In lookup software, transitions may be fine-grained: for example, it may be possible to transition to IDNA2008 rules regarding Deviations for .blogspot.com at a given point but not for .com, or vice versa. If .de bundles or blocks the Deviation characters, then clients could transition Deviations for .de, but not for (say) .blogspot.de. Moreover, client software with a UI, such as the address bar in a browser, could provide more options for the transition. A full discussion of such transition strategies is outside of the scope of this document.

During the interim, authors of documents, such as HTML documents, can unambiguously refer to the IDNA2008 interpretation of characters by explicitly using the Punycode form of the domain name label.

There are two slightly different compatibility mechanisms for domain names during a transition and afterward. UTS #46 therefore specifies two specific types of processing: Transitional Processing (Conformance Clause C1) and Nontransitional Processing (Conformance Clause C2). The only difference between them is the handling of the four Deviation characters.

Summarized briefly, UTS #46 builds upon IDNA2008 in three areas:

For a demonstration of differences between IDNA2003, IDNA2008, and the Unicode IDNA Compatibility Processing, see the [IDN_Demo]. For more detail on the differences, see Section 7, IDNA Comparison. UTS #46 does not change any of the terms defined in IDNA2008, such as A-Label or U-Label.

Neither the Unicode IDNA Compatibility Processing nor IDNA2008 address security problems associated with confusables (the so-called "paypal.com" problem). IDNA2008 disallows certain symbols and punctuation characters that can be used for spoofing, such as spoofs of the slash character ("/"). However, these are an extremely small fraction of the confusable characters used for spoofing. Moreover, confusable characters themselves account for a small proportion of phishing problems: most are cases like "secure-wellsfargo.com". For more information, see [Bortzmeyer] and the [IDN FAQ]. It is strongly recommended that Unicode Technical Report #36, "Unicode Security Considerations" [UTR36] and Unicode Technical Standard #39, "Unicode Security Mechanisms" [UTS39] be consulted for information on dealing with confusables, both for client software and registries. In particular, [UTS39] provides information that can be used to drastically reduce the number of confusables when dealing with international domain names, much beyond what IDNA2008 does. See also the [DemoConf].

2.1 Display of Internationalized Domain Names

IDNA2003 applications customarily display the processed string to the user. This improves security by reducing the opportunity for visual confusability. Thus, for example, the URL http://googIe.com (with a capital I in place of the L) is revealed as http://googie.com. However, for Deviations the distinction between the original and processed form is especially important for users. Thus the Nontransitional Processing should be used for displaying domain names. This is the same as Transitional Processing, except that it excludes the Deviations: ß and ς and the joiners. It is thus fully compatible with IDNA2008 for these Deviation characters.

Except for direct DNS lookup during the transitional period, the Nontransitional Processing should always be used, preserving the Deviation characters in the original string as per IDNA2008. Once the transition for Deviation characters is over, Nontransitional Processing can be used exclusively.

2.2 Registries

This specification is primarily targeted at applications doing lookup of IDNs. There is, however, one strong recommendation for registries: do not allow the registration of labels that are invalid according to Nontransitional Processing, and for a transitional period, bundle or block labels containing Deviation characters.

These tactics can be described as follows:

The label that is actually registered and inserted into a registry has always been processed. For example, xn--bcher-kva corresponds to bücher. However, it may be useful for a registry to also ask for "unprocessed" labels, such as Bücher, as part of the registration process, so that they are aware of the registrant's intent. However, such unprocessed labels must be handled carefully:

2.3 Notation

Sets of code points are defined using properties and the syntax of Unicode Technical Standard #18, "Unicode Regular Expressions" [UTS18]. For example, the set of combining marks is represented by the syntax \p{gc=M}. Additionally, the "+" indicates the addition of elements to a set, for clarity.

In this document, a label is a substring of a domain name. That substring is bounded on both sides by either the start or the end of the string, or any of the following characters, called label-separators:

  1. U+002E ( . ) FULL STOP
  2. U+FF0E ( . ) FULLWIDTH FULL STOP
  3. U+3002 ( 。 ) IDEOGRAPHIC FULL STOP
  4. U+FF61 ( 。 ) HALFWIDTH IDEOGRAPHIC FULL STOP

Many people use the terms "domain names" and "host names" interchangeably. This document follows [RFC3490] in use of the term "domain name".

3 Conformance

The requirements for conformance on implementations of the Unicode IDNA Compatibility Processing algorithm are stated in the following clauses. An implementation can claim conformance to any or all of these clauses independently.

C1 Given a version of Unicode and a Unicode String, a conformant implementation of Transitional Processing shall replicate the results given by applying the Transitional Processing algorithm specified by Section 4, Processing.
C2 Given a version of Unicode and a Unicode String, a conformant implementation of Nontransitional Processing shall replicate the results given by applying the Nontransitional Processing algorithm specified by Section 4, Processing.
C3 Given a version of Unicode and a Unicode String, a conformant implementation of Preprocessing for IDNA2008 shall replicate the results specified by Section 4.4, Preprocessing for IDNA2008.

These specifications are logical ones, designed to be straightforward to describe. An actual implementation is free to use different methods as long the result is the same as that specified by the logical algorithm.

Any conformant implementation may also have tighter validity criteria than those imposed by Section 4.1, Validity Criteria. For example, an application could disallow or warn of domain name labels with certain characteristics, such as:

For more information, see Unicode Technical Report #36, "Unicode Security Considerations" [UTR36] and Unicode Technical Standard #39, "Unicode Security Mechanisms" [UTS39].

3.1 STD3 Rules

IDNA2003 provides for a flag, UseSTD3ASCIIRules, that allows for implementations to choose whether or not to abide by the rules in [STD3]. These rules exclude ASCII characters outside the set consisting of A-Z, a-z, 0-9, and U+002D ( - ) HYPHEN-MINUS. For example, some browsers also allow characters such as U+005F ( _ ) LOW LINE (underbar) in domain names, and thus use UseSTD3ASCIIRules=false, plus their own validity checks for the other ASCII characters.

While UseSTD3ASCIIRules=true is strongly recommended, Section 5, IDNA Mapping Table provides data to allow implementations to support UseSTD3ASCIIRules=false for compatibility with IDNA2003 implementations where necessary. The mapping table does this: providing the status values and mapping values for both UseSTD3ASCIIRules=true and UseSTD3ASCIIRules=false. Implementations that use UseSTD3ASCIIRules=false will need to apply their own validation to the mapped values as indicated in Section 4.1, Validity Criteria.

4 Processing

The input to Unicode IDNA Compatibility Processing is a prospective domain_name string expressed in Unicode, and a choice of Transitional or Nontransitional Processing. The domain name consists of a sequence of labels with dot separators, such as "Bücher.de". For more information about the composition of a URL, see Section 3.5 of [STD13].

The following steps, performed in order, successively alter the input domain_name string and then output it as a converted Unicode string, plus a flag to indicate whether there was an error. Even if an error occurs, the conversion of the string is performed as much as is possible.

  1. Map. For each code point in the domain_name string, look up the status value in Section 5, IDNA Mapping Table, and take the following actions:
    • disallowed: Leave the code point unchanged in the string, and record that there was an error.
    • ignored: Remove the code point from the string. This is equivalent to mapping the code point to an empty string.
    • mapped: Replace the code point in the string by the value for the mapping in Section 5, IDNA Mapping Table.
    • deviation:
      • For Transitional Processing, replace the code point in the string by the value for the mapping in Section 5, IDNA Mapping Table.
      • For Nontransitional Processing, leave the code point unchanged in the string.
    • valid: Leave the code point unchanged in the string.
  2. Normalize. Normalize the domain_name string to Unicode Normalization Form C.
  3. Break. Break the string into labels at U+002E ( . ) FULL STOP.
  4. Convert/Validate. For each label in the domain_name string:
    • If the label starts with "xn--":
      1. Attempt to convert the rest of the label to Unicode according to Punycode [RFC3492]. If that conversion fails, record that there was an error, and continue with the next label. Otherwise replace the original label in the string by the results of the conversion.
      2. Verify that the label meets the validity criteria in Section 4.1, Validity Criteria for Nontransitional Processing. If any of the validity criteria are not satisfied, record that there was an error.
    • If the label does not start with "xn--":
      • Verify that the label meets the validity criteria in Section 4.1, Validity Criteria for the input Processing choice (Transitional or Nontransitional). If any of the validity criteria are not satisfied, record that there was an error.

Any input domain_name string that does not record an error has been successfully processed according to this specification. Conversely, if an input domain_name string causes an error, then the processing of the input domain_name string fails. The processing is idempotent—reapplying the processing to the output will make no further changes. For examples, see Table 2, Examples of Transitional Processing.

Implementations may make further modifications to the resulting Unicode string when showing it to the user. For example, it is recommended that disallowed characters be replaced by a U+FFFD to make them visible to the user. Similarly, labels that fail processing during steps 4 or 5 may be marked by the insertion of a U+FFFD or other visual device.

With either Transitional or Nontransitional Processing, sources already in Punycode are validated without mapping. In particular, Punycode containing Deviation characters, such as href="xn--fu-hia.de" (for fuß.de) is not remapped. This provides a mechanism allowing explicit use of Deviation characters even during a transition period.

4.1 Validity Criteria

Each of the following criteria must be satisfied for a label:

  1. The label must be in Unicode Normalization Form NFC.
  2. The label must not contain a U+002D HYPHEN-MINUS character in both the third position and fourth positions.
  3. The label must neither begin nor end with a U+002D HYPHEN-MINUS character.
  4. The label must not contain a U+002E ( . ) FULL STOP.
  5. The label must not begin with a combining mark, that is: General_Category=Mark.
  6. Each code point in the label must only have certain status values according to Section 5, IDNA Mapping Table:
    1. For Transitional Processing, each value must be valid.
    2. For Nontransitional Processing, each value must be either valid or deviation.

Any particular application may have tighter validity criteria, as discussed in Section 3, Conformance.

UseSTD3ASCIIRules

If UseSTD3ASCIIRules=false, then the validity tests for ASCII characters are not provided by the table status values, but are implementation-dependent. For example, if an implementation allows the characters [\u002Da-zA-Z0-9] and also the underbar (_), then it needs to use the table values for UseSTD3ASCIIRules=false, and test for any other ASCII characters as part of its validity criteria. These ASCII characters may have resulted from a mapping: for example, a U+005F ( _ ) LOW LINE (underbar) may have originally been a U+FF3F ( _ ) FULLWIDTH LOW LINE.

There are a very small number of non-ASCII characters with the data file status disallowed_STD3_valid:

U+2260 ( ≠ ) NOT EQUAL TO
U+226E ( ≮ ) NOT LESS-THAN
U+226F ( ≯ ) NOT GREATER-THAN

Those characters are disallowed with UseSTD3ASCIIRules=true because the set of characters in their canonical decompositions are not
entirely in the valid set (Step 7 of the Table Derivation). However, they are allowed with UseSTD3ASCIIRules=false,
because the base characters of their canonical decompositions, U+003D ( = ) EQUALS SIGN, U+003C ( < ) LESS-THAN SIGN, and U+003E ( > ) GREATER-THAN SIGN, are each valid under that option. If an implementation uses UseSTD3ASCIIRules=false but disallows any of these three ASCII characters, then it must also disallow the corresponding precomposed character for its negation.

Right-to-Left Scripts

In addition, the label should meet the requirements for right-to-left characters specified in the Right-to-Left Scripts document of [IDNA2008], and for the CONTEXTJ requirements in the Protocol document of [IDNA2008]. It is strongly recommended that Unicode Technical Report #36, "Unicode Security Considerations" [UTR36] and Unicode Technical Standard #39, "Unicode Security Mechanisms" [UTS39] be consulted for information on dealing with confusables, and for characters that should be excluded from identifiers. Note that the recommended exclusions are a superset of those in [IDNA2008].

4.2 ToASCII

The operation corresponding to ToASCII of [RFC3490] is defined by the following steps:

  1. Apply the appropriate processing. This may record an error. The appropriate processing is either:
    • Transitional Processing for transitional handling of Deviation characters, or
    • Nontransitional Processing otherwise
  2. Break the result into labels at U+002E FULL STOP.
  3. Convert each label with non-ASCII characters into Punycode [RFC3492]. This may record an error.
  4. Verify DNS length restrictions. This may record an error. For more information, see [STD13] and [STD3].
    1. The length of the domain name, excluding the root label and its dot, is from 1 to 253.
    2. The length of each label is from 1 to 63.
  5. If an error was recorded, then the operation failed, and no DNS lookup should be done.

Implementations are advised to apply additional tests to these labels, such as those described in Unicode Technical Report #36, "Unicode Security Considerations"[UTR36] and Unicode Technical Standard #39, "Unicode Security Mechanisms" [UTS39], and take appropriate actions. For example, a label with mixed scripts or confusables may be called out in the UI.

4.3 ToUnicode

The operation corresponding to ToUnicode of [RFC3490] is defined by the following steps:

  1. Apply the Nontransitional Processing. Note that unlike RFC3490 ToASCII, this always signals whether or not there was an error.
  2. Like RFC3490, this will always produce a converted Unicode string, even if there was an error.

Implementations are advised to apply additional tests to these labels, such as those described in Unicode Technical Report #36, "Unicode Security Considerations" [UTR36] and Unicode Technical Standard #39, "Unicode Security Mechanisms" [UTS39], and take appropriate actions. For example, a label with mixed scripts or confusables may be called out in the UI.

4.4 Preprocessing for IDNA2008

The table specified in Section 5, IDNA Mapping Table may also be used for a pure preprocessing step for IDNA2008, mapping a Unicode string for input directly to the algorithm specified in IDNA2008.

Preprocessing for IDNA2008 is specified as follows:

Apply the Section 4.3, ToUnicode processing to the Unicode string.

Note that this preprocessing allows some characters that are invalid according to IDNA2008. However, the IDNA2008 processing will catch those characters. For example, a Unicode string containing a character listed as DISALLOWED in IDNA2008, such as U+2665 (♥) BLACK HEART SUIT, will pass the preprocessing step without an error, but subsequent application of the IDNA2008 processing will fail with an error, indicating that the string is not a valid IDN according to IDNA2008.

4.5 Implementation Notes

A number of optimizations can be applied to the Unicode IDNA Compatibility Processing. These optimizations can improve performance, reduce table size, make use of existing NFKC transform mechanisms, and so on. For example:

Note that the input domain_name string for the Unicode IDNA Compatibility Processing must have had all escaped Unicode code points converted to Unicode code points. For example, U+5341 ( 十 ) CJK UNIFIED IDEOGRAPH-5341 could have been escaped as any of the following:

Examples are shown in Table 2, Examples of Processing:

Table 2. Examples of Processing
Input
Validate
Comment
Bloß.de
bloss.de
=
n/a
ok
Transitional: maps uppercase and eszett
bloß.de
=
n/a
ok
Nontransitional: maps uppercase
xn--blo-7ka.de
=
=
bloß.com
ok
Punycode is not mapped, so ß never changes (whether transitional or not).
u¨.com
=
ü.com
n/a
ok
Normalize changes u + umlaut to ü
xn--tda.com
=
=
ü.com
ok
Punycode xn--tda changes to ü
xn--u-ccb.com
=
=
u¨.com
error
Punycode is not mapped, but is validated. Because u + umlaut is not NFC, it fails.
a⒈com
error
error
error
error
The character "" is disallowed, because it would produce a dot when mapped.
xn--a-ecp.ru
xn--a-ecp.ru
=
a⒈.ru
error
Punycode xn--a-ecp = a⒈, which fails validation.
xn--0.pt
xn--0.pt
=
error
error
Punycode xn--0 is invalid.
日本語。JP
日本語.jp
=
n/a
ok
Fullwidth characters are remapped, including 。
☕.us
=
=
n/a
ok
Post-Unicode 3.2 characters are allowed.

5 IDNA Mapping Table

For each code point in Unicode, the IDNA Mapping Table provides one of the following status values:

If this status value is mapped, disallowed_STD3_mapped or deviation, the table also supplies a mapping value for that code point.

A table is provided for each version of Unicode starting with Unicode 5.1, in versioned directories under [IDNA-Table]. Each table for a version of the Unicode Standard will always be backward compatible with previous versions of the table: only characters with the status value disallowed may change in status or mapping value. Unlike the IDNA2008 table, this table is designed to be applied to the entire domain name, not just to individual labels. That design provides for the IDNA2003 handling of label separators. In particular, the table is constructed to forbid problematic characters such as U+2488 ( ⒈ ) DIGIT ONE FULL STOP, whose decompositions contain a "dot".

The Unicode IDNA Compatibility Processing is based on the Unicode character mapping property [NFKC_Casefold]. Section 6, Mapping Table Derivation describes the derivation of these tables. Like derived properties in the Unicode Character Database, the description of the derivation is informative. Only the data in IDNA Mapping Table is normative for the application of this specification.

The files use a semicolon-delimited format similar to those in the Unicode Character Database. The first field is the code point; the second field is the status value; and the third field is the mapping value. Code points are expressed in hexadecimal. The status values are the following: valid, disallowed, ignored, mapped, and deviation.

Example:

0000..002C    ; disallowed                    #  NULL..COMMA
002D ; valid # HYPHEN-MINUS
...
0041 ; mapped ; 0061 # LATIN CAPITAL LETTER A ...
00AD ; ignored # SOFT HYPHEN ... 00DF ; deviation ; 0073 0073 # LATIN SMALL LETTER SHARP S
...

6 Mapping Table Derivation

The following describes the derivation of the mapping table. This description has nothing to do with the actual mapping of labels in Section 4, Processing. Instead, this section describes the derivation of the table in Section 5, IDNA Mapping Table. That table is then normatively used for mapping in Section 4, Processing.

The derivation is described as a series of steps. Step 1 defines a base mapping value; Steps 2-5 define three sets of characters. These are all used in Step 6 to produce the mapping and status values for the table. Step 7 removes characters whose NFD form would be invalid. Each numbered step may have substeps: for example, Step 1 consists of Step 1.1 followed by Step 1.2.

The computation is done twice, once with UseSTD3ASCIIRules=true, and once with UseSTD3ASCIIRules=false. Code points that are disallowed with UseSTD3ASCIIRules=true, but valid or mapped with UseSTD3ASCIIRules=false, are given the special values disallowed_STD3_valid and disallowed_STD3_mapped.

If a Unicode property changes in a future version in a way that would affect backward compatibility, a grandfathering clause will be added to Step 5 to maintain compatibility. For more information on compatibility, see Section 5, IDNA Mapping Table.

Step 1: Define a base mapping value for each character

  1. Map the following label separator characters to U+002E ( . ) FULL STOP:
    • U+FF0E ( . ) FULLWIDTH FULL STOP
    • U+3002 ( 。 ) IDEOGRAPHIC FULL STOP
    • U+FF61 ( 。 ) HALFWIDTH IDEOGRAPHIC FULL STOP
  2. Map each other character to its NFKC_Casefold value [NFKC_Casefold].

Step 2: Specify the base valid set

The base valid set is defined by the sequential list of additions and subtractions in Table 3, Base Valid Set. This definition is based on the principles of IDNA2003. When applied to the repertoire of Unicode 3.2 characters, this produces a set which is closely aligned with IDNA2003.

Table 3. Base Valid Set
Formal Sets Descriptions
  \P{Changes_When_NFKC_Casefolded}
Start with characters that are equal to their [NFKC_Casefold] value. This criterion excludes uppercase letters, for example, as well as characters that are unstable under NFKC normalization, and default ignorable code points.

Note that according to Perl/Java syntax, \P means the inverse of \p, so these are the characters that do not change when individually mapped according to [NFKC_Casefold].

- \p{c} - \p{z} Remove Unassigned, Controls, Private Use, Format, Surrogate, and Whitespace
- \p{Block=Ideographic_Description_Characters} Remove ideographic description characters
- \p{ascii} + [\u002Da-zA-Z0-9] if UseSTD3ASCIIRules = true Remove disallowed ASCII; '-' is valid
+ \p{ascii} - [\u002E] if UseSTD3ASCIIRules = false Add all ASCII except for "."

Step 3: Specify the base exclusion set

Form the base exclusion set in the following way:

  1. Start with the empty set.
  2. Add each code point C such that:
    1. In IDNA2003, C is neither prohibited nor unassigned nor a label separator (that is, it is either valid or mapped), and
    2. In IDNA2003, C has a different mapping than C's base mapping value specified in Step 1.
  3. Add each code point C such that:
    1. In IDNA2003, C is prohibited, and
    2. either C is in the base valid set, or every code point in C's base mapping value is in the base valid set.

For Unicode 5.2 and 6.0, the base exclusion set consists of list that follows. The subheads (like "Case Changes") are informational, and do not represent the principle for excluding the characters listed under them.

Characters that have a different mapping in IDNA2003 (Step 3.2 above)

Characters that are disallowed in IDNA2003 (Step 3.3 above)

Step 4: Specify the deviation set

This is the set of characters that deviate between IDNA2003 and IDNA2008.

Step 5: Specify grandfathered changes

This set is currently empty. Adjustments will be made in this section if the steps would cause an already existing character to change status or mapping under a future version of Unicode.

Step 6: Produce the initial status and mapping values for the table

For each code point:

  1. If the code point is in the deviation set
    • the status is deviation and the mapping value is the base mapping value for that code point.
  2. Otherwise, if the code point is in the base exclusion set or is unassigned
    • the status is disallowed and there is no mapping value in the table.
  3. Otherwise, if the code point is not a label separator and some code point in its base mapping value is not in the base valid set
    • the status is disallowed and there is no mapping value in the table.
  4. Otherwise, if the base mapping value is an empty string
    • the status is ignored and there is no mapping value in the table.
  5. Otherwise, if the base mapping value is the same as the code point
    • the status is valid and there is no mapping value in the table.
  6. Otherwise,
    • the status is mapped and the mapping value is the base mapping value for that code point.

Step 7: Produce the final status and mapping values

After processing all code points in Step 6:

  1. Iterate through the set of characters with a status of valid. Any whose canonical decompositions (NFD) are not wholly in the valid set, make disallowed.
  2. Iterate through the set of characters with a status of mapped. Any with mapping values whose canonical decompositions (NFD) are not wholly in the valid set, make disallowed.
  3. Recursively apply these actions until there are no more status changes.

For Unicode 5.2 and 6.0, the set of characters set to disallowed in Step 7 consists of the following:

Note that characters such as U+2488 ( ⒈ ) DIGIT ONE FULL STOP are disallowed by Step 6.3.

7 IDNA Comparison

Table 4, IDNA Comparisons illustrates the differences between the three specifications in terms of valid character repertoire. It omits the ASCII-repertoire code points, all code points unassigned in Unicode 6.0, as well as control characters, private-use characters, and surrogate code points. It also includes labels separators that are valid or mapped. The table separates the Unicode 3.2 characters from those encoded later, because they have a special status in IDNA2003. It also separates buckets where UTS #46 and IDNA2008 behave the same from those where they behave differently.

Each row in the table defines a bucket of code points that share a pattern of behavior across the three specifications. The columns provide the following information:

Table 4. IDNA Comparisons
  Count
IDNA2003
UTS46
IDNA2008
Comments and Samples
Unicode 3.2 (IDNA2003 = UTS46 = IDNA2008)
(a)
86,676
Valid
Valid
Valid
Valid in all three systems
U+00E0 ( à ) LATIN SMALL LETTER A WITH GRAVE
(b)
431
Disallowed
Disallowed
Disallowed
Disallowed in all three systems
U+FF01 ( ! ) FULLWIDTH EXCLAMATION MARK
Unicode 3.2 (IDNA2003 ≠ UTS46 = IDNA2008)
(c)
48
Valid
Disallowed
Disallowed
Mappings changed after Unicode 3.2
U+2132 ( Ⅎ ) TURNED CAPITAL F
(d)
8
Mapped
Disallowed
Disallowed
Mappings changed after Unicode 3.2
U+2F868 ( 㛼 ) CJK COMP.
Unicode 3.2 (IDNA2003 = UTS46 ≠ IDNA2008)
(e)
4,640
Mapped / Ignored
Mapped / Ignored
Disallowed
Case and compatibility variants, default ignorables
U+00C0 ( À ) LATIN CAPITAL LETTER A WITH GRAVE
(f)
3,254
Valid
Valid
Disallowed
Punctuation, Symbols, etc.
U+2665 ( ♥ ) BLACK HEART SUIT
(g)
4
Mapped / Ignored
Display: Valid;
Lookup: Mapped / Ignored
Valid
Deviations
U+200C ZERO WIDTH NON-JOINER
U+200D ZERO WIDTH JOINER
U+00DF ( ß ) LATIN SMALL LETTER SHARP S
U+03C2 ( ς ) GREEK SMALL LETTER FINAL SIGMA
Unicode 4.0 to 6.0 (UTS46 = IDNA2008)
(h)
10,729
LookupValid
Valid
Valid
U+0221 ( ȡ ) LATIN SMALL LETTER D WITH CURL
(i)
53
LookupValid
Disallowed
Disallowed
U+0602 ( ؂ ) ARABIC FOOTNOTE MARKER
Unicode 4.0 to 6.0 (UTS46 ≠ IDNA2008)
(j)
2,604
LookupValid
Valid
Disallowed
U+2615 ( ☕ ) HOT BEVERAGE
(k)
842
LookupValid
Mapped / Ignored
Disallowed
U+023A ( Ⱥ ) LATIN CAPITAL LETTER A WITH STROKE

A more detailed online listing of differences is found at [DemoIDNChars] and [DemoIDN]. The implications for confusability can be seen at [DemoConf].

7.1 Implications for Implementers

Table 4, IDNA Comparisons can also be used to categorize implications for implementers.

With the exception of Row (g), if any characters are Mapped/Ignored in any specification—Rows (d), (e), (k)—then in the other specifications they are either Mapped/Ignored in precisely the same way, or they are Disallowed. This prevents domain names from being mapped differently on different browsers: either the characters map to the same result, or they do not work. Row (k) is unproblematic in this regard, assuming that registries follow one of the specifications, because characters like U+023A ( Ⱥ ) will not be valid in registered labels.

The only exceptions are the four problematic Deviations in Row (g), which require more complex handling because of their treatment in IDNA2008, as discussed earlier. Transitions for Deviation characters will depend upon how registries handle IDNA2008 going forward, and how soon IDNA2003 clients are retired. Outside of the transition from IDNA2003 to IDNA2008, the UTS #46 Nontransitional Processing should be used, thus preserving Deviation characters.

This presumes that IDNA2008 implementations do not use custom, incompatible mappings: that is, that they do not take advantage of the fact that arbitrary mappings are allowed in IDNA2008, and choose a mapping that is incompatible with IDNA2003 or UTS #46. This pertains to any of Rows (e), (f), (j), (k). If custom mappings were used by any signficant client base, it would result in serious problems for security and interoperability. For more information, see the [IDN_FAQ].

With the exception of the above issues, implementation is straightforward:

8 Conformance Testing

A conformance testing file (IdnaTest.txt) is provided for each version of Unicode starting with Unicode 6.0, in versioned directories under [IDNA-Table]. It only provides test cases for UseSTD3ASCIIRules=true.

To test for conformance to UTS46, on each line, perform the toASCII and to Unicode operations on the source string, with the indicated type. The results must match what is given in the toUnicode and toASCII columns, except for errors. In the case of errors, an implementation only needs to record that there is an error; it need not reproduce the results in [...], which are only informative.

Format

The file is UTF8, with certain characters escaped using the \uXXXX convention for readability. Columns (c1, c2,...) are separated by semicolons. Leading and trailing spaces and tabs in each column are ignored. Comments are indicated with hash marks. The columns contain the following information:

Column 1: type - T for transitional, N for nontransitional, B for both
Column 2: source - the source string to be tested
Column 3: toUnicode - the result of applying toUnicode to the source, using the specified type
Column 4: toASCII - the result of applying toASCII to the source, using nontransitional

If the value of toUnicode or toASCII is the same as source, the column will be blank.

An error in toUnicode or toASCII is indicated by a value of the form [...]. In such a case, the contents is a list of error codes based on the step numbers in UTS46 and IDNA2008:

Pn for Section 4 Processing step n
Vn for 4.1 Validity Criteria step n
An for 4.2 ToASCII step n
Bn for Bidi (in IDNA2008)
Cn for ContextJ (in IDNA2008)

However, these particular error codes are only informative; the important feature is whether or not there is an error.

Acknowledgements

Mark Davis and Michel Suignard authored the bulk of the text of this document, under direction from the Unicode Technical Committee. For their contributions of ideas or text to this specification, the editors thank Julie Allen, Matitiahu Allouche, Peter Constable, Craig Cummings, Martin Dürst, Peter Edberg, Deborah Goldsmith, Laurentiu Iancu, Gervase Markham, Simon Montagu, Lisa Moore, Eric Muller, Murray Sargent, Markus Scherer, Jungshik Shin, Shawn Steele, Erik van der Poel, Chris Weber, and Ken Whistler. The specification builds upon [IDNA2008], developed in the IETF Idnabis working group, especially contributions from Matitiahu Allouche, Harald Alvestrand, Vint Cerf, Martin J. Dürst, Lisa Dusseault, Patrik Fältström, Paul Hoffman, Cary Karp, John Klensin, and Peter Resnick, and also upon [IDNA2003], authored by Marc Blanchet, Adam Costello, Patrik Fältström, and Paul Hoffman.

References

[Bortzmeyer] http://www.bortzmeyer.org/idn-et-phishing.html

The most interesting studies cited there (originally from Mike Beltzner of Mozilla) are:
[DemoConf] http://unicode.org/cldr/utility/confusables.jsp
[DemoIDN] http://unicode.org/cldr/utility/idna.jsp
[DemoIDNChars] http://unicode.org/cldr/utility/list-unicodeset.jsp?a=\p{age%3D3.2}-\p{cn}-\p{cs}-\p{co}&abb=on&g=uts46+idna+idna2008
[Feedback] Reporting Form
http://www.unicode.org/reporting.html
For reporting errors and requesting information online.
[IDNA2003] The IDNA2003 specification is defined by a cluster of IETF RFCs:
[IDNA2008] The IDNA2008 specification is defined by a cluster of IETF RFCs: There is also an informative document:
[IDNA-Table] http://www.unicode.org/Public/idna
[IDN-Demo] http://unicode.org/cldr/utility/idna.jsp
[IDN-FAQ] http://www.unicode.org/faq/idn.html
[NFKC_Casefold] The Unicode property specified in [UAX44], and defined by the data in DerivedNormalizationProps.txt (search for "NFKC_Casefold").

The Unicode character mapping property defined by the data in DerivedNormalizationProps.txt. (Search for "NFKC_Casefold".) See also the documentation for NFKC_Casefold in [UAX44].
[Reports] Unicode Technical Reports
http://www.unicode.org/reports/
For information on the status and development process for technical reports, and for a list of technical reports.
[RFC3454] P. Hoffman, M. Blanchet. "Preparation of Internationalized Strings ("stringprep")", RFC 3454, December 2002.
http://ietf.org/rfc/rfc3454.txt
[RFC3490] Faltstrom, P., Hoffman, P. and A. Costello, "Internationalizing Domain Names in Applications (IDNA)", RFC 3490, March 2003.
http://ietf.org/rfc/rfc3490.txt
[RFC3491] Hoffman, P. and M. Blanchet, "Nameprep: A Stringprep Profile for Internationalized Domain Names (IDN)", RFC 3491, March 2003.
http://ietf.org/rfc/rfc3491.txt
[RFC3492] Costello, A., "Punycode: A Bootstring encoding of Unicode for Internationalized Domain Names in Applications (IDNA)", RFC 3492, March 2003.
http://ietf.org/rfc/rfc3492.txt
[SafeBrowsing] http://code.google.com/apis/safebrowsing/
[Stability] Unicode Consortium Stability Policies
http://www.unicode.org/policies/stability_policy.html 
[STD3] Braden, R., "Requirements for Internet Hosts -- Communication Layers", STD 3, RFC 1122, and "Requirements for Internet Hosts -- Application and Support", STD 3, RFC 1123, October 1989.
http://www.rfc-editor.org/std/std3.txt
[STD13] Mockapetris, P., "Domain names - concepts and facilities", STD 13, RFC 1034 and "Domain names - implementation and specification", STD 13, RFC 1035, November 1987.
http://www.rfc-editor.org/std/std13.txt
[UAX29] UAX #29: Unicode Text Segmentation
http://www.unicode.org/reports/tr29/
[UAX31] UAX #31: Unicode Identifier and Pattern Syntax
http://www.unicode.org/reports/tr31/
[UAX44] UAX #44:Unicode Character Database
http://www.unicode.org/reports/tr44/
[Unicode] The Unicode Standard
For the latest version, see:
http://www.unicode.org/versions/latest/
For the 5.2.0 version, see:
http://www.unicode.org/versions/Unicode5.2.0/
For the 6.0.0 version, see:
http://www.unicode.org/versions/Unicode6.0.0/
[UTR36] UTR #36: Unicode Security Considerations
http://www.unicode.org/reports/tr36/
[UTS18] UTS #18: Unicode Regular Expressions
http://www.unicode.org/reports/tr18/
[UTS39] UTS #39: Unicode Security Mechanisms
http://www.unicode.org/reports/tr39/
[Versions] Versions of the Unicode Standard
http://www.unicode.org/versions/
For details on the precise contents of each version of the Unicode Standard, and how to cite them.

Modifications

The following summarizes modifications from the previous revisions of this document.

Revision 5

Revision 4 being a proposed update, only changes between revisions 5 and 3 are noted here.

Revision 3

Revision 2

Revision 1