In computer science, Extended Backus–Naur Form (EBNF) is a family of metasyntax notations used for expressing context-free grammars: that is, a formal way to describe computer programming languages and other formal languages. They are extensions of the basic Backus–Naur Form (BNF) metasyntax notation.
The earliest EBNF was originally developed by Niklaus Wirth. However, many variants of EBNF are in use. The International Organization for Standardization has adopted an EBNF standard (ISO/IEC 14977). This article uses EBNF as specified by the ISO for examples applying to all EBNF:s. Other EBNF variants use somewhat different syntactic conventions.
Contents |
EBNF is a code that expresses the grammar of a computer language. An EBNF consists of terminal symbol and non-terminal production rules which are the restrictions governing how terminal symbols can be combined into a legal sequence. Examples of terminal symbols include alphanumeric characters, punctuation marks, and white space characters.
The EBNF defines production rules where sequences of symbols are respectively assigned to a nonterminal:
digit excluding zero = "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" ; digit = "0" | digit excluding zero ;
This production rule defines the nonterminal digit which is on the left side of the assignment. The vertical bar represents an alternative and the terminal symbols are enclosed with quotation marks followed by a semicolon as terminating character. Hence a Digit is a 0 or a digit excluding zero that can be 1 or 2 or 3 and so forth until 9.
A production rule can also include a sequence of terminals or nonterminals, each separated by a comma:
twelve = "1" , "2" ; two hundred one = "2" , "0" , "1" ; three hundred twelve = "3" , twelve ; twelve thousand two hundred one = twelve , two hundred one ;
Expressions that may be omitted or repeated can be represented through curly braces { ... }:
natural number = digit excluding zero , { digit } ;
In this case, the strings 1, 2, ...,10,...,12345,... are correct expressions. To represent this, everything that is set within the curly braces may be repeated arbitrarily often, including not at all.
An option can be represented through squared brackets [ ... ]. That is, everything that is set within the square brackets may be present just once, or not at all:
integer = "0" | [ "-" ] , natural number ;
Therefore an integer is a zero (0) or a natural number that may be preceded by an optional minus sign.
EBNF also provides, among other things, the syntax to describe repetitions of a specified number of times, to exclude some part of a production, or to insert comments in an EBNF grammar.
The following represents a proposed standard.
| Usage | Notation |
|---|---|
| definition | = |
| concatenation | , |
| termination | ; |
| alternation | | |
| option | [ ... ] |
| repetition | { ... } |
| grouping | ( ... ) |
| terminal string | " ... " |
| terminal string | ' ... ' |
| comment | (* ... *) |
| special sequence | ? ... ? |
| exception | - |
A Pascal-like programming language that allows only assignments can be defined in EBNF as follows:
(* a simple program syntax in EBNF − Wikipedia *)
program = 'PROGRAM' , white space , identifier , white space ,
'BEGIN' , white space ,
{ assignment , ";" , white space } ,
'END.' ;
identifier = alphabetic character , { alphabetic character | digit } ;
number = [ "-" ] , digit , { digit } ;
string = '"' , { all characters − '"' } , '"' ;
assignment = identifier , ":=" , ( number | identifier | string ) ;
alphabetic character = "A" | "B" | "C" | "D" | "E" | "F" | "G"
| "H" | "I" | "J" | "K" | "L" | "M" | "N"
| "O" | "P" | "Q" | "R" | "S" | "T" | "U"
| "V" | "W" | "X" | "Y" | "Z" ;
digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" ;
white space = ? white space characters ? ;
all characters = ? all visible characters ? ;
A syntactically correct program then would be:
PROGRAM DEMO1 BEGIN A0:=3; B:=45; H:=-100023; C:=A; D123:=B34A; BABOON:=GIRAFFE; TEXT:="Hello world!"; END.
The language can easily be extended with control flows, arithmetical expressions, and Input/Output instructions. Then a small, usable programming language would be developed.
The BNF had the problem that options and repetitions could not be directly expressed. Instead, they needed the use of an intermediate rule or alternative production defined to be either nothing or the optional production for option, or either the repeated production or itself, recursively, for repetition. The same constructs can still be used in EBNF.
The BNF used the symbols (<, >, |, ::=) for itself, but did not include quotes around terminal strings. This prevented these characters from being used in the languages, and required a special symbol for the empty string. In EBNF, terminals are strictly enclosed within quotation marks ("..." or '...'). The angle brackets ("<...>") for nonterminals can be omitted.
BNF syntax can only represent a rule in one line, whereas in EBNF a terminating character, the semicolon, marks the end of a rule.
Furthermore, EBNF includes mechanisms for enhancements, defining the number of repetitions, excluding alternatives, comments, etc.
It should be noted that EBNF is not "more powerful" than the BNF: As a matter of principle, any grammar defined in EBNF can also be represented in BNF (though representations in the latter are generally lengthier).
1. The following conventions are used:
2. The normal character representing each operator of Extended BNF and its implied precedence is (highest precedence at the top):
* repetition-symbol - except-symbol , concatenate-symbol | definition-separator-symbol = defining-symbol ; terminator-symbol
3. The normal precedence is overridden by the following bracket pairs:
' first-quote-symbol first-quote-symbol '
" second-quote-symbol second-quote-symbol "
(* start-comment-symbol end-comment-symbol *)
( start-group-symbol end-group-symbol )
[ start-option-symbol end-option-symbol ]
{ start-repeat-symbol end-repeat-symbol }
? special-sequence-symbol special-sequence-symbol ?
The first-quote-symbol is the apostrophe as defined by ISO/IEC 646:1991, that is to say Unicode U+0027 ('); the font used in ISO/IEC 14977:1996(E) renders it very much like the acute, Unicode U+00B4 (´), so confusion sometimes arises. However, the ISO Extended BNF standard invokes ISO/IEC 646:1991, "ISO 7-bit coded character set for information interchange", as a normative reference and makes no mention of any other character sets, so formally, there is no confusion with Unicode characters outside the 7-bit ASCII range.
As examples, the following syntax rules illustrate the facilities for expressing repetition:
aa = "A";
bb = 3 * aa, "B";
cc = 3 * [aa], "C";
dd = {aa}, "D";
ee = aa, {aa}, "E";
ff = 3 * aa, 3 * [aa], "F";
gg = {3 * aa}, "G";
Terminal strings defined by these rules are as follows:
aa: A bb: AAAB cc: C AC AAC AAAC dd: D AD AAD AAAD AAAAD etc. ee: AE AAE AAAE AAAAE AAAAAE etc. ff: AAAF AAAAF AAAAAF AAAAAAF gg: G AAAG AAAAAAG etc.
According to the ISO 14977 standard EBNF is meant to be extensible, and two facilities are mentioned. The first is part of EBNF grammar, the special sequence, which is arbitrary text enclosed with question marks. The interpretation of the text inside a special sequence is beyond the scope of the EBNF standard. For example, the space character could be defined by the following rule:
space = ? US-ASCII character 32 ?;
The second facility for extension is using the fact that parentheses cannot in EBNF be placed next to identifiers (they must be concatenated with them). The following is valid EBNF:
something = foo, ( bar );
The following is not valid EBNF:
something = foo ( bar );
Therefore, an extension of EBNF could use that notation. For example, in a Lisp grammar, function application could be defined by the following rule:
function application = list( symbol , { expression } );
This article was originally based on material from the Free On-line Dictionary of Computing, which is licensed under the GFDL.