AWK.sublime-syntax

%YAML 1.2
---
# http://www.sublimetext.com/docs/3/syntax.html
name: AWK

scope: source.awk

file_extensions:
  - awk

# References
# http://pubs.opengroup.org/onlinepubs/009695399/utilities/awk.html
# http://www.gnu.org/software/gawk/manual/html_node/index.html
# http://www.grymoire.com/Unix/Awk.html
# http://www.cs.mun.ca/~yzchen/bib/nawkUS.pdf
variables:

  NAME: '[A-Za-z_][A-Za-z_0-9]*'

contexts:

  main:
    - include: comment
    - include: procedure
    - include: pattern

  comment:
    - match: '(?<!\S)(#)(?!\{).*$\n?'
      scope: comment.line.number-sign.awk
      captures:
        1: punctuation.definition.comment.awk

  procedure:
    - match: '{'
      scope: meta.brace.curly.awk
      push:
        - match: '}'
          scope: meta.brace.curly.awk
          pop: true
        - include: comment
        - include: procedure
        - include: keyword
        - include: expression

  pattern:
    - include: function-definition
    - include: builtin-pattern
    - include: expression

  #TODO: Resolve regex / operator ambiguity
  #Will probably have to split expression
  #into seperate unary/binary states
  #See https://github.com/sublimehq/Packages/blob/master/JavaScript/JavaScript.sublime-syntax
  expression:
    - include: comment
    - include: command
    - include: function
    - include: constant
    - include: variable
    - include: groupings
    - include: prefix-operator
    - include: regexp
    - match: '(?=[\S])'
      pop: true

  groupings:
    - match: \(
      scope: meta.brace.round.awk
    - match: \)(?=\s*{)
      scope: meta.brace.round.awk
    - match: \)
      scope: meta.brace.round.awk
      push: infix-operator
    - match: \,
      scope: punctuation.separator.parameters.awk

  builtin-pattern:
    - match: \b(BEGINFILE|BEGIN|ENDFILE|END)\b
      scope: constant.language.awk

  function-definition:
    - match: \b(function)\s+({{NAME}})(\()
      captures:
        1: storage.type.function.awk
        2: entity.name.function.awk
        3: punctuation.definition.parameters.begin.awk
      push:
        - match: \)
          captures:
            0: punctuation.definition.parameters.end.awk
          pop: true
        - match: \b(\w+)\b
          scope: variable.parameter.function.awk
        - match: \b(,)\b
          scope: punctuation.separator.parameters.awk

  constant:
    - include: numeric-constant
    - include: string-constant

  numeric-constant:
    - match: '\b[0-9]+(?:\.[0-9]+)?(?:e[+-][0-9]+)?\b'
      scope: constant.numeric.awk
      push: infix-operator

  string-constant:
    - match: '"'
      scope: punctuation.definition.string.begin.awk
      push:
        - meta_scope: string.quoted.double.awk
        - match: '"'
          scope: punctuation.definition.string.end.awk
          pop: true
        - include: escaped-char

  escaped-char:
    - match: '\\(?:[\\abfnrtv/"]|x[0-9A-Fa-f]{2}|[0-7]{3})'
      scope: constant.character.escape.awk

  # To avoid ambiguity with division operators regexps are only
  # valid in places were division is not.
  regexp:
    - match: /
      scope: punctuation.definition.regex.begin.awk
      push:
        - meta_scope: string.regexp
        - match: "(/)([gimy]*)"
          captures:
            1: punctuation.definition.regex.end.awk
            2: keyword.other.awk
          set: infix-operator
        - match: '(?=.|\n)'
          with_prototype:
            - match: '(?=/)'
              pop: true
          push:
            - include: scope:source.regexp

  variable:
    - match: '\$[0-9]+'
      scope: variable.language.awk
      push: infix-operator
    - match: \b(?:FILENAME|FS|NF|NR|OFMT|OFS|ORS|RS)\b
      scope: variable.language.awk
      push: infix-operator
    - match: \b(?:ARGC|ARGV|CONVFMT|ENVIRON|FNR|RLENGTH|RSTART|SUBSEP)\b
      scope: variable.language.nawk
      push: infix-operator
    - match: \b(?:ARGIND|ERRNO|FIELDWIDTHS|IGNORECASE|RT)\b
      scope: variable.language.gawk
      push: infix-operator
    - match: '\b{{NAME}}\b'
      scope: variable.other.readwrite.awk
      push: infix-operator

  keyword:
    - match: \b(?:break|continue|do|while|exit|for|if|else|return)\b
      scope: keyword.control.awk

  command:
    - match: \b(?:next|print|printf)\b
      scope: keyword.other.command.awk
    - match: \b(?:close|getline|delete|system)\b
      scope: keyword.other.command.nawk
    - match: \b(?:fflush|nextfile)\b
      scope: keyword.other.command.bell-awk

  function:
    - match: \b(?:exp|int|log|sqrt|index|length|split|sprintf|substr)\b
      scope: support.function.awk
    - match: \b(?:atan2|cos|rand|sin|srand|gsub|match|sub|tolower|toupper)\b
      scope: support.function.nawk
    - match: \b(?:gensub|strftime|systime)\b
      scope: support.function.gawk
    - match: \b{{NAME}}(?=\s*\()
      scope: entity.name.function.awk

  prefix-operator:
    - match: '[+-]'
      scope: keyword.operator.arithmetic.awk
      set: expression

  infix-operator:
    - match: "!?~|[=<>!]=|[<>]"
      scope: keyword.operator.comparison.awk
      set: expression
    - match: \b(in)\b
      scope: keyword.operator.comparison.awk
      set: expression
    - match: '[+\-*/%^]=|\+\+|--|>>|='
      scope: keyword.operator.assignment.awk
      set: expression
    - match: \|\||&&|!
      scope: keyword.operator.boolean.awk
      set: expression
    - match: '[+\-*/%^]'
      scope: keyword.operator.arithmetic.awk
      set: expression
    - match: '[?:]'
      scope: keyword.operator.trinary.awk
      set: expression
    - match: '\['
      scope: keyword.operator.index.awk
      push:
        - match: '\]'
          scope: keyword.operator.index.awk
          pop: true
        - match: ','
          scope: punctuation.separator.index.awk
        - include: expression
    - match: '(?=[\S])'
      pop: true