Brandenburg InternetWorking

675 Spruce Dr. Sunnyvale CA 94086 US +1.408.246.8253 dcrocker@bbiw.net

THUS plc.

1/2 Berkeley Square, 99 Berkeley Street Glasgow G3 7HR UK paul.overell@thus.net

ABNF Augmented Backus-Naur Form electronic mail Internet technical specifications often need to define a formal syntax. Over the years, a modified version of Backus-Naur Form (BNF), called Augmented BNF (ABNF), has been popular among many Internet specifications. The current specification documents ABNF. It balances compactness and simplicity with reasonable representational power. The differences between standard BNF and ABNF involve naming rules, repetition, alternatives, order- independence, and value ranges. This specification also supplies additional rule definitions and encoding for a core lexical analyzer of the type common to several Internet specifications.

Internet technical specifications often need to define a formal syntax and are free to employ whatever notation their authors deem useful. Over the years, a modified version of Backus-Naur Form (BNF), called Augmented BNF (ABNF), has been popular among many Internet specifications. It balances compactness and simplicity with reasonable representational power. In the early days of the Arpanet, each specification contained its own definition of ABNF. This included the email specifications, and then , which came to be the common citations for defining ABNF. The current document separates those definitions to permit selective reference. Predictably, it also provides some modifications and enhancements. The differences between standard BNF and ABNF involve naming rules, repetition, alternatives, order-independence, and value ranges. supplies rule definitions and encoding for a core lexical analyzer of the type common to several Internet specifications. It is provided as a convenience and is otherwise separate from the meta language defined in the body of this document, and separate from its formal status.

The name of a rule is simply the name itself, that is, a sequence of characters, beginning with an alphabetic character, and followed by a combination of alphabetics, digits, and hyphens (dashes). Rule names are case insensitive. The names <rulename>, <Rulename>, <RULENAME>, and <rUlENamE> all refer to the same rule. Unlike original BNF, angle brackets ("<", ">") are not required. However, angle brackets may be used around a rule name whenever their presence facilitates in discerning the use of a rule name. This is typically restricted to rule name references in free-form prose, or to distinguish partial rules that combine into a string not separated by white space, such as shown in the discussion about repetition, below.

A rule is defined by the following sequence:

where <name> is the name of the rule, <elements> is one or more rule names or terminal specifications, and <crlf> is the end-of-line indicator (carriage return followed by line feed). The equal sign separates the name from the definition of the rule. The elements form a sequence of one or more rule names and/or value definitions, combined according to the various operators defined in this document, such as alternative and repetition. For visual ease, rule definitions are left aligned. When a rule requires multiple lines, the continuation lines are indented. The left alignment and indentation are relative to the first lines of the ABNF rules and need not match the left margin of the document.

Rules resolve into a string of terminal values, sometimes called characters. In ABNF, a character is merely a non-negative integer. In certain contexts, a specific mapping (encoding) of values into a character set (such as ASCII) will be specified.

Terminals are specified by one or more numeric characters, with the base interpretation of those characters indicated explicitly. The following bases are currently defined:

Hence: respectively specify the decimal and hexadecimal representation of for carriage return.

A concatenated string of such values is specified compactly, using a period (".") to indicate a separation of characters within that value. Hence:

ABNF permits the specification of literal text strings directly, enclosed in quotation marks. Hence:

Literal text strings are interpreted as a concatenated set of printable characters. ABNF strings are case insensitive and the character set for these strings is US-ASCII.

Hence:

and: will match "abc", "Abc", "aBc", "abC", "ABc", "aBC", "AbC", and "ABC".

To specify a rule that is case sensitive, specify the characters individually.

For example:

or will match only the string that comprises only the lowercase characters, abc.

External representations of terminal value characters will vary according to constraints in the storage or transmission environment. Hence, the same ABNF-based grammar may have multiple external encodings, such as one for a 7-bit US-ASCII environment, another for a binary octet environment, and still a different one when 16-bit Unicode is used. Encoding details are beyond the scope of ABNF, although provides definitions for a 7-bit US-ASCII environment as has been common to much of the Internet. By separating external encoding from the syntax, it is intended that alternate encoding environments can be used for the same syntax.

A rule can define a simple, ordered string of values (i.e., a concatenation of contiguous characters) by listing a sequence of rule names. For example:

So that the rule <mumble> matches the lowercase string "aba". Linear white space: Concatenation is at the core of the ABNF parsing model. A string of contiguous characters (values) is parsed according to the rules defined in ABNF. For Internet specifications, there is some history of permitting linear white space (space and horizontal tab) to be freely and implicitly interspersed around major constructs, such as delimiting special characters or atomic strings. NOTE: This specification for ABNF does not provide for implicit specification of linear white space. Any grammar that wishes to permit linear white space around delimiters or string segments must specify it explicitly. It is often useful to provide for such white space in "core" rules that are then used variously among higher-level rules. The "core" rules might be formed into a lexical analyzer or simply be part of the main ruleset.

Elements separated by a forward slash ("/") are alternatives. Therefore,

will accept <foo> or <bar>.

A quoted string containing alphabetic characters is a special form for specifying alternative characters and is interpreted as a non-terminal representing the set of combinatorial strings with the contained characters, in the specified order but with any mixture of upper- and lowercase.

It is sometimes convenient to specify a list of alternatives in fragments. That is, an initial rule may match one or more alternatives, with later rule definitions adding to the set of alternatives. This is particularly useful for otherwise independent specifications that derive from the same parent ruleset, such as often occurs with parameter lists. ABNF permits this incremental definition through the construct:

So that the ruleset

is the same as specifying

A range of alternative numeric values can be specified compactly, using a dash ("-") to indicate the range of alternative values. Hence:

is equivalent to:

Concatenated numeric values and numeric value ranges cannot be specified in the same string. A numeric value may use the dotted notation for concatenation or it may use the dash notation to specify one value range. Hence, to specify one printable character between end-of-line sequences, the specification could be:

Elements enclosed in parentheses are treated as a single element, whose contents are strictly ordered. Thus, matches (elem foo blat) or (elem bar blat), and

matches (elem foo) or (bar blat).

It is strongly advised that grouping notation be used, rather than relying on the proper reading of "bare" alternations, when alternatives consist of multiple rule names or literals.

Hence, it is recommended that the following form be used: It will avoid misinterpretation by casual readers.

The sequence group notation is also used within free text to set off an element sequence from the prose.

The operator "*" preceding an element indicates repetition. The full form is: *element ]]> where <a> and <b> are optional decimal values, indicating at least <a> and at most <b> occurrences of the element.

Default values are 0 and infinity so that *<element> allows any number, including zero; 1*<element> requires at least one; 3*3<element> allows exactly 3; and 1*2<element> allows one or two.

A rule of the form: element ]]>

is equivalent to *element ]]>

That is, exactly <n> occurrences of <element>. Thus, 2DIGIT is a 2-digit number, and 3ALPHA is a string of three alphabetic characters.

Square brackets enclose an optional element sequence:

is equivalent to

A semicolon starts a comment that continues to the end of line. This is a simple way of including useful notes in parallel with the specifications.

The various mechanisms described above have the following precedence, from highest (binding tightest) at the top, to lowest (loosest) at the bottom: Rule name, prose-val, Terminal value Comment Value range Repetition Grouping, Optional Concatenation Alternative Use of the alternative operator, freely mixed with concatenations, can be confusing. Again, it is recommended that the grouping operator be used to make explicit concatenation groups.

This syntax requires a formatting of rules that is relatively strict. Hence, the version of a ruleset included in a specification might need preprocessing to ensure that it can be interpreted by an ABNF parser. This syntax uses the rules provided in .

" ; bracketed string of SP and VCHAR ; without angles ; prose description, to be used as ; last resort ]]>

Security is truly believed to be irrelevant to this document.

American National Standards Institute The Rand Corporation, Information Sciences Department

1700 Main St Santa Monica CA 90406 US DCrocker@Rand-Unix

Bolt Beranek and Newman Inc. (BBN)

50 Moulton St. Cambridge MA 02138 US Vittal@BBN-TenexD

Massachusets Institute of Technology (MIT), Laboratory for Computer Science

545 Technology Square Cambridge MA 02139 US Pogran@MIT-Multics

Bolt Beranek and Newman Inc. (BBN)

50 Moulton St. Cambridge MA 02138 US Henderson@BBN-TenexD

University of Delaware, Dept. of Electrical Engineering

Newark DE 19711 US DCrocker@UDel-Relay

The syntax for ABNF was originally specified in RFC 733. Ken L. Harrenstien, of SRI International, was responsible for re-coding the BNF into an Augmented BNF that makes the representation smaller and easier to understand. This recent project began as a simple effort to cull out the portion of RFC 822 that has been repeatedly cited by non-email specification writers, namely the description of Augmented BNF. Rather than simply and blindly converting the existing text into a separate document, the working group chose to give careful consideration to the deficiencies, as well as benefits, of the existing specification and related specifications made available over the last 15 years, and therefore to pursue enhancement. This turned the project into something rather more ambitious than was first intended. Interestingly, the result is not massively different from that original, although decisions, such as removing the list notation, came as a surprise. This "separated" version of the specification was part of the DRUMS working group, with significant contributions from Jerome Abela, Harald Alvestrand, Robert Elz, Roger Fajman, Aviva Garrett, Tom Harsch, Dan Kohn, Bill McQuillan, Keith Moore, Chris Newman, Pete Resnick, and Henning Schulzrinne. Julian Reschke warrants a special thanks for converting the Draft Standard version to XML source form.

This appendix contains some basic rules that are in common use. Basic rules are in uppercase. Note that these rules are only valid for ABNF encoded in 7-bit ASCII or in characters sets that are a superset of 7-bit ASCII.

Certain basic rules are in uppercase, such as SP, HTAB, CRLF, DIGIT, ALPHA, etc.

Externally, data are represented as "network virtual ASCII" (namely, 7-bit US-ASCII in an 8-bit field), with the high (8th) bit set to zero. A string of values is in "network byte order", in which the higher-valued bytes are represented on the left-hand side and are sent over the network first.