# This is the Pegex grammar for Pegex grammars!

%grammar pegex
%version 0.2.0
%include pegex-atoms


grammar:
  meta-section
  rule-section
  ( doc-ending | ERROR-rule-definition )

meta-section: ( meta-definition | + | ERROR-meta-definition )*

rule-section: ( rule-definition | + )*

meta-definition: / '%' meta-name BLANK+ meta-value /

rule-definition: rule-start rule-group ending

rule-start: / ( rule-name ) BLANK* ':' -/

rule-group: any-group

any-group: /- '|'? -/ all-group ( /- '|' -/ all-group )*

all-group: rule-part (- rule-part)*

rule-part: (rule-item)1-2 % /+ ( '%'{1,2} ) +/

rule-item:
  | bracketed-group
  | whitespace-token
  | rule-reference
  | quoted-regex
  | regular-expression
  | error-message

rule-reference:
  /
    ( rule-modifier? )      # [=!.-+]
      (:                    # foo | <foo>
        ( rule-name ) |
        (: '<' ( rule-name ) '>' )
      )
    ( rule-quantifier? )    # [?*+] 2+ 2-3
    (! BLANK* ':' )         # Avoid parsing 'foo:'
  /                        # as a rule reference.

quoted-regex:
  / TICK ( [^ TICK ]* ) TICK /

regular-expression:
  '/'
  whitespace-start?
  (
  | whitespace-must
  | whitespace-maybe
  | quoted-regex
  | regex-rule-reference
  | +
  | regex-raw
  )*
  '/'

whitespace-start: / ([ PLUS DASH]) (! [ DASH TILDE ]) /

whitespace-must: /+ (: PLUS | DASH DASH )  (= [ SPACE SLASH CR NL ]) /

whitespace-maybe: /- DASH (= [ SPACE SLASH CR NL ]) /

regex-rule-reference:
  /
    (:
      + ( rule-name ) |
      (: '<' ( rule-name ) '>' )
    )
    (! BLANK* ':' )
  /

regex-raw: / ([^ WS SLASH TICK LANGLE ]+) /

bracketed-group:
    / ( group-modifier? ) '(' -/
    rule-group
    /- ')' ( rule-quantifier? ) /

whitespace-token:
    / ( (: PLUS | DASH | DASH DASH | TILDE | TILDE TILDE ) ) (= + )/

error-message:
    / '`' ( [^ '`' DOS ]* ) '`' /

rule-modifier: / [ BANG EQUAL PLUS DASH DOT ] /

group-modifier: / [ DASH DOT ] /

rule-quantifier:
    / (:
        [ STAR PLUS QMARK ] |
        DIGIT+ (: DASH DIGIT+ | PLUS)?
    ) /

meta-name:
    / ( 'grammar' | 'extends' | 'include' | 'version' ) /

meta-value:
    /
        BLANK*
        ( [^ SEMI BREAK ]*? )
        BLANK*
        ending
    /

rule_name: /
  (:
    ALPHA ALNUM* (:[ DASH UNDER ] ALNUM+)* |
    DASH+ |
    UNDER+
  )
  (= [^ WORD DASH ])
/

ending: /
  ~?
  (:
    BREAK - SEMI? - |
    comment - SEMI? - |
    SEMI - |
    EOS
  )
/

ws: / (: WS | comment ) /

comment: / '#' ANY* (: BREAK | EOS ) /

###
# Pegex common error recognition and reporting:
###

doc-ending: /- EOS /

illegal-non-modifier-char: / [^
    WORD LPAREN RPAREN LANGLE SLASH TILDE PIPE GRAVE WS
] /

illegal-non-quantifier-char: / [^
    WORD LPAREN RPAREN LANGLE SLASH TILDE PIPE GRAVE WS
    STAR PLUS QMARK BANG EQUAL PLUS DASH DOT COLON SEMI
] /

ERROR-meta-definition:
    /(= PERCENT WORD+ )/
    `Illegal meta rule`

# Much of this is essentially a duplicate of the above rules, except with added
# error checking

ERROR-rule-definition: ERROR-rule-start ERROR-rule-group ( ending | `Rule ending syntax error` )

ERROR-rule-group: ERROR-any-group | ERROR-all-group

ERROR-all-group: ERROR-rule-part+ % -

ERROR-any-group: (ERROR-all-group)2+ % /- PIPE -/

ERROR-rule-part: (ERROR-rule-item)1-2 % /+ ( PERCENT{1,2} ) +/

ERROR-rule-start: / ( rule-name ) BLANK* COLON -/ | `Rule header syntax error`

ERROR-rule-item:
    rule-item |
    ERROR-rule-reference |
    ERROR-regular-expression |
    ERROR-bracketed-group |
    ERROR-error-message

# Errors - rule-reference
ERROR-rule-reference:
    /(= rule-modifier? LANGLE rule-name (! RANGLE ) )/
    `Missing > in rule reference`
|
    /(= rule-modifier? rule-name RANGLE )/
    `Missing < in rule reference`
|
    /(=
        rule-modifier? (: rule-name | LANGLE rule-name RANGLE )
        illegal-non-quantifier-char
    )/
    `Illegal character in rule quantifier`
|
    /(= rule-modifier? rule-name DASH )/
    `Unprotected rule name with numeric quantifier; please use <rule>#-# syntax!`
|
    !rule-modifier
    /(=
        illegal-non-modifier-char
        (: rule-name | LANGLE rule-name RANGLE )
        rule-quantifier?         # [?*+] 2+ 2-3
        (! BLANK* COLON )      # Avoid parsing 'foo:'
    )/                             # as a rule reference.
    `Illegal rule modifier (must be [=!.-+]?)`

# Errors - regular-expression
ERROR-regular-expression:
    /(= SLASH ( [^ SLASH ]* ) doc-ending )/
    `Runaway regular expression; no ending slash at EOF`

# Errors - bracketed-group
ERROR-bracketed-group:
    /(! group-modifier) (= illegal-non-modifier-char LPAREN )/
    `Illegal group rule modifier (can only use .)`
|
    / ( group-modifier? ) LPAREN -/
    rule-group
    (
        =doc-ending
        `Runaway rule group; no ending parens at EOF`
    |
        / (= - RPAREN illegal-non-quantifier-char ) /
        `Illegal character in group rule quantifier`
    )

# Errors - error-message
ERROR-error-message:
    /(= GRAVE [^ GRAVE DOS ]* [ DOS ] [^ GRAVE ]* GRAVE )/
    `Multi-line error messages not allowed!`
|
    /(= GRAVE [^ GRAVE ]* doc-ending )/
    `Runaway error message; no ending grave at EOF`

# Errors - separation
ERROR-separation:
    /(= - PERCENT{3} )/
    `Leading separator form (BOK) no longer supported`
|
    /(= - PERCENT{1,2} [^ WS ] )/
    `Illegal characters in separator indicator`

# vim: set lisp: