path: root/runtime/syntax/julia.vim

" Vim syntax file
" Language:	julia
" Maintainer:	Carlo Baldassi <carlobaldassi@gmail.com>
" Homepage:	https://github.com/JuliaEditorSupport/julia-vim
" Last Change:	2013 feb 11

if version < 600
  syntax clear
elseif exists("b:current_syntax")
  finish
endif

let s:cpo_save = &cpo
set cpo&vim

if version < 704
  " this is used to disable regex syntax like `\@3<='
  " on older vim versions
  function! s:d(x)
    return ''
  endfunction
else
  function! s:d(x)
    return string(a:x)
  endfunction
endif

scriptencoding utf-8

let s:julia_spellcheck_strings = get(g:, "julia_spellcheck_strings", 0)
let s:julia_spellcheck_docstrings = get(g:, "julia_spellcheck_docstrings", 1)
let s:julia_spellcheck_comments = get(g:, "julia_spellcheck_comments", 1)

let s:julia_highlight_operators = get(g:, "julia_highlight_operators", 1)

" List of characters, up to \UFF, which cannot be used in identifiers.
" (It includes operator characters; we don't consider them identifiers.)
" This is used mostly in lookbehinds with `\@<=`, e.g. when we need to check
" that that we're not in the middle of an identifier.
" It doesn't include a few characters (spaces and all closing parentheses)
" because those may or may not be valid in the lookbehind on a case-by-case
" basis.
let s:nonid_chars = '\U00-\U08' . '\U0A-\U1F'
      \             . '\U21-\U28' . '\U2A-\U2F' . '\U3A-\U40' . '\U5B-\U5E' . '\U60' . '\U7B\U7C'
      \             . '\U7E-\UA1' . '\UA7\UA8' . '\UAB-\UAD' . '\UAF\UB1\UB4' . '\UB6-\UB8' . '\UBB\UBF' . '\UD7\UF7'

" The complete list
let s:nonidS_chars = '[:space:])\U5D}' . s:nonid_chars


" List of all valid operator chars up to \UFF (NOTE: they must all be included
" in s:nonidS_chars, so that if we include that, then this is redundant)
" It does not include '!' since it can be used in an identifier.
" The list contains the following characters: '%&*+-/<=>\\^|~¬±×÷'
let s:op_chars = '\U25\U26\U2A\U2B\U2D\U2F\U3C-\U3E\U5C\U5E\U7C\U7E\UAC\UB1\UD7\UF7'

" List of all valid operator chars above \UFF
" Written with ranges for performance reasons
" The list contains the following characters: '…⁝⅋←↑→↓↔↚↛↜↝↞↠↢↣↤↦↩↪↫↬↮↶↷↺↻↼↽⇀⇁⇄⇆⇇⇉⇋⇌⇍⇎⇏⇐⇒⇔⇚⇛⇜⇝⇠⇢⇴⇵⇶⇷⇸⇹⇺⇻⇼⇽⇾⇿∈∉∊∋∌∍∓∔∗∘∙√∛∜∝∤∥∦∧∨∩∪∷∸∺∻∽∾≀≁≂≃≄≅≆≇≈≉≊≋≌≍≎≏≐≑≒≓≔≕≖≗≘≙≚≛≜≝≞≟≠≡≢≣≤≥≦≧≨≩≪≫≬≭≮≯≰≱≲≳≴≵≶≷≸≹≺≻≼≽≾≿⊀⊁⊂⊃⊄⊅⊆⊇⊈⊉⊊⊋⊍⊎⊏⊐⊑⊒⊓⊔⊕⊖⊗⊘⊙⊚⊛⊜⊞⊟⊠⊡⊢⊣⊩⊬⊮⊰⊱⊲⊳⊴⊵⊶⊷⊻⊼⊽⋄⋅⋆⋇⋉⋊⋋⋌⋍⋎⋏⋐⋑⋒⋓⋕⋖⋗⋘⋙⋚⋛⋜⋝⋞⋟⋠⋡⋢⋣⋤⋥⋦⋧⋨⋩⋪⋫⋬⋭⋮⋯⋰⋱⋲⋳⋴⋵⋶⋷⋸⋹⋺⋻⋼⋽⋾⋿⌿▷⟂⟈⟉⟑⟒⟕⟖⟗⟰⟱⟵⟶⟷⟹⟺⟻⟼⟽⟾⟿⤀⤁⤂⤃⤄⤅⤆⤇⤈⤉⤊⤋⤌⤍⤎⤏⤐⤑⤒⤓⤔⤕⤖⤗⤘⤝⤞⤟⤠⥄⥅⥆⥇⥈⥉⥊⥋⥌⥍⥎⥏⥐⥑⥒⥓⥔⥕⥖⥗⥘⥙⥚⥛⥜⥝⥞⥟⥠⥡⥢⥣⥤⥥⥦⥧⥨⥩⥪⥫⥬⥭⥮⥯⥰⦷⦸⦼⦾⦿⧀⧁⧡⧣⧤⧥⧴⧶⧷⧺⧻⨇⨈⨝⨟⨢⨣⨤⨥⨦⨧⨨⨩⨪⨫⨬⨭⨮⨰⨱⨲⨳⨴⨵⨶⨷⨸⨹⨺⨻⨼⨽⩀⩁⩂⩃⩄⩅⩊⩋⩌⩍⩎⩏⩐⩑⩒⩓⩔⩕⩖⩗⩘⩚⩛⩜⩝⩞⩟⩠⩡⩢⩣⩦⩧⩪⩫⩬⩭⩮⩯⩰⩱⩲⩳⩴⩵⩶⩷⩸⩹⩺⩻⩼⩽⩾⩿⪀⪁⪂⪃⪄⪅⪆⪇⪈⪉⪊⪋⪌⪍⪎⪏⪐⪑⪒⪓⪔⪕⪖⪗⪘⪙⪚⪛⪜⪝⪞⪟⪠⪡⪢⪣⪤⪥⪦⪧⪨⪩⪪⪫⪬⪭⪮⪯⪰⪱⪲⪳⪴⪵⪶⪷⪸⪹⪺⪻⪼⪽⪾⪿⫀⫁⫂⫃⫄⫅⫆⫇⫈⫉⫊⫋⫌⫍⫎⫏⫐⫑⫒⫓⫔⫕⫖⫗⫘⫙⫛⫷⫸⫹⫺⬰⬱⬲⬳⬴⬵⬶⬷⬸⬹⬺⬻⬼⬽⬾⬿⭀⭁⭂⭃⭄⭇⭈⭉⭊⭋⭌￩￪￫￬'
let s:op_chars_wc = '\U2026\U205D\U214B\U2190-\U2194\U219A-\U219E\U21A0\U21A2-\U21A4\U21A6\U21A9-\U21AC\U21AE\U21B6\U21B7\U21BA-\U21BD\U21C0\U21C1\U21C4\U21C6\U21C7\U21C9\U21CB-\U21D0\U21D2\U21D4\U21DA-\U21DD\U21E0\U21E2\U21F4-\U21FF\U2208-\U220D\U2213\U2214\U2217-\U221D\U2224-\U222A\U2237\U2238\U223A\U223B\U223D\U223E\U2240-\U228B\U228D-\U229C\U229E-\U22A3\U22A9\U22AC\U22AE\U22B0-\U22B7\U22BB-\U22BD\U22C4-\U22C7\U22C9-\U22D3\U22D5-\U22FF\U233F\U25B7\U27C2\U27C8\U27C9\U27D1\U27D2\U27D5-\U27D7\U27F0\U27F1\U27F5-\U27F7\U27F9-\U27FF\U2900-\U2918\U291D-\U2920\U2944-\U2970\U29B7\U29B8\U29BC\U29BE-\U29C1\U29E1\U29E3-\U29E5\U29F4\U29F6\U29F7\U29FA\U29FB\U2A07\U2A08\U2A1D\U2A1F\U2A22-\U2A2E\U2A30-\U2A3D\U2A40-\U2A45\U2A4A-\U2A58\U2A5A-\U2A63\U2A66\U2A67\U2A6A-\U2AD9\U2ADB\U2AF7-\U2AFA\U2B30-\U2B44\U2B47-\U2B4C\UFFE9-\UFFEC'

" Full operators regex
let s:operators = '\%(' . '\.\%([-+*/^÷%|&⊻]\|//\|\\\|>>\|>>>\?\)\?=' .
      \           '\|'  . '[:<>]=\|||\|&&\||>\|<|\|[<>:]:\|<<\|>>>\?\|//\|[-=]>\|\.\.\.\?' .
      \           '\|'  . '\.\?[!' . s:op_chars . s:op_chars_wc . ']' .
      \           '\)'


" Characters that can be used to start an identifier. Above \UBF we don't
" bother checking. (If a UTF8 operator is used, it will take precedence anyway.)
let s:id_charsH = '\%([A-Za-z_\UA2-\UA6\UA9\UAA\UAE\UB0\UB5\UBA]\|[^\U00-\UBF]\)'
" Characters that can appear in an identifier, starting in 2nd position. Above
" \UBF we check for operators since we need to stop the identifier if one
" appears. We don't check for invalid characters though.
let s:id_charsW = '\%([0-9A-Za-z_!\UA2-\UA6\UA9\UAA\UAE-\UB0\UB2-\UB5\UB8-\UBA\UBC-\UBE]\|[^\U00-\UBF]\@=[^' . s:op_chars_wc . ']\)'

" A valid julia identifier, more or less
let s:idregex = '\%(' . s:id_charsH . s:id_charsW . '*\)'



syn case match

syntax cluster juliaExpressions		contains=@juliaParItems,@juliaStringItems,@juliaKeywordItems,@juliaBlocksItems,@juliaTypesItems,@juliaConstItems,@juliaMacroItems,@juliaSymbolItems,@juliaOperatorItems,@juliaNumberItems,@juliaCommentItems,@juliaErrorItems,@juliaSyntaxRegions
syntax cluster juliaExprsPrintf		contains=@juliaExpressions,@juliaPrintfItems
syntax cluster juliaExprsNodot		contains=@juliaParItems,@juliaStringItems,@juliaMacroItems,@juliaSymbolItems,@juliaOperatorItems,@juliaCommentItems,juliaIdSymbol

syntax cluster juliaParItems		contains=juliaParBlock,juliaSqBraIdxBlock,juliaSqBraBlock,juliaCurBraBlock,juliaQuotedParBlock,juliaQuotedQMarkPar
syntax cluster juliaKeywordItems	contains=juliaKeyword,juliaWhereKeyword,juliaImportLine,juliaInfixKeyword,juliaRepKeyword
syntax cluster juliaBlocksItems		contains=juliaConditionalBlock,juliaWhileBlock,juliaForBlock,juliaBeginBlock,juliaFunctionBlock,juliaMacroBlock,juliaQuoteBlock,juliaTypeBlock,juliaImmutableBlock,juliaExceptionBlock,juliaLetBlock,juliaDoBlock,juliaModuleBlock,juliaStructBlock,juliaMutableStructBlock,juliaAbstractBlock,juliaPrimitiveBlock
syntax cluster juliaTypesItems		contains=juliaBaseTypeBasic,juliaBaseTypeNum,juliaBaseTypeC,juliaBaseTypeError,juliaBaseTypeIter,juliaBaseTypeString,juliaBaseTypeArray,juliaBaseTypeDict,juliaBaseTypeSet,juliaBaseTypeIO,juliaBaseTypeProcess,juliaBaseTypeRange,juliaBaseTypeRegex,juliaBaseTypeFact,juliaBaseTypeFact,juliaBaseTypeSort,juliaBaseTypeRound,juliaBaseTypeSpecial,juliaBaseTypeRandom,juliaBaseTypeDisplay,juliaBaseTypeTime,juliaBaseTypeOther

syntax cluster juliaConstItems  	contains=juliaConstNum,juliaConstBool,juliaConstEnv,juliaConstMMap,juliaConstC,juliaConstGeneric,juliaConstIO,juliaPossibleEuler

syntax cluster juliaMacroItems		contains=juliaPossibleMacro,juliaDollarVar,juliaDollarPar,juliaDollarSqBra
syntax cluster juliaSymbolItems		contains=juliaPossibleSymbol
syntax cluster juliaNumberItems		contains=juliaNumbers
syntax cluster juliaStringItems		contains=juliaChar,juliaString,juliabString,juliasString,juliaShellString,juliaDocString,juliaRegEx
syntax cluster juliaPrintfItems		contains=juliaPrintfParBlock,juliaPrintfString
syntax cluster juliaOperatorItems	contains=juliaOperator,juliaRangeOperator,juliaCTransOperator,juliaTernaryRegion,juliaColon,juliaSemicolon,juliaComma
syntax cluster juliaCommentItems	contains=juliaCommentL,juliaCommentM
syntax cluster juliaErrorItems		contains=juliaErrorPar,juliaErrorEnd,juliaErrorElse,juliaErrorCatch,juliaErrorFinally

syntax cluster juliaSyntaxRegions	contains=juliaIdSymbol,juliaTypeOperatorR2,juliaTypeOperatorR3,juliaWhereR,juliaDotted

syntax cluster juliaSpellcheckStrings		contains=@spell
syntax cluster juliaSpellcheckDocStrings	contains=@spell
syntax cluster juliaSpellcheckComments		contains=@spell

if !s:julia_spellcheck_docstrings
  syntax cluster juliaSpellcheckDocStrings	remove=@spell
endif
if !s:julia_spellcheck_strings
  syntax cluster juliaSpellcheckStrings		remove=@spell
endif
if !s:julia_spellcheck_comments
  syntax cluster juliaSpellcheckComments	remove=@spell
endif

syntax match   juliaSemicolon		display ";"
syntax match   juliaComma		display ","
syntax match   juliaColon		display ":"

" A dot can introduce a sort of 'environment' such that words after it are not
" recognized as keywords. This has low precedence so that it can be overridden
" by operators
syntax match   juliaDotted		transparent "\.\s*[^])}.]" contains=@juliaExprsNodot
syntax match   juliaDottedT		contained transparent "\.\s*[^])}.]" contains=@juliaExprsNodot,juliaType

syntax match   juliaErrorPar		display "[])}]"
syntax match   juliaErrorEnd		display "\<end\>"
syntax match   juliaErrorElse		display "\<\%(else\|elseif\)\>"
syntax match   juliaErrorCatch		display "\<catch\>"
syntax match   juliaErrorFinally	display "\<finally\>"
syntax match   juliaErrorSemicol	display contained ";"

syntax region  juliaParBlock		matchgroup=juliaParDelim start="(" end=")" contains=@juliaExpressions,juliaComprehensionFor
syntax region  juliaParBlockInRange	matchgroup=juliaParDelim contained start="(" end=")" contains=@juliaExpressions,juliaParBlockInRange,juliaRangeKeyword,juliaComprehensionFor
syntax region  juliaSqBraIdxBlock	matchgroup=juliaParDelim start="\[" end="\]" contains=@juliaExpressions,juliaParBlockInRange,juliaRangeKeyword,juliaComprehensionFor,juliaSymbolS,juliaQuotedParBlockS,juliaQuotedQMarkParS
exec 'syntax region  juliaSqBraBlock	matchgroup=juliaParDelim start="\%(^\|\s\|' . s:operators . '\)\@'.s:d(3).'<=\[" end="\]" contains=@juliaExpressions,juliaComprehensionFor,juliaSymbolS,juliaQuotedParBlockS,juliaQuotedQMarkParS'
syntax region  juliaCurBraBlock		matchgroup=juliaParDelim start="{" end="}" contains=juliaType,juliaDottedT,@juliaExpressions

exec 'syntax match   juliaType		contained "\%(' . s:idregex . '\.\)*\zs' . s:idregex . '"'

" This is a generic identifier followed by some symbol, either a type
" operator (<: or >:), or an open parenthesis, or an open curly bracket.
" It's used to recognize one of the contained regions looking for identifiers
" only once. Once recognized, those regions no longer need to use the
" expensive s:idregex.
exec 'syntax match   juliaIdSymbol	transparent "' . s:idregex . '\%(\s*[<>]:\|\.\?(\|{\|\"\)\@=" contains=juliaFunctionCall,juliaParamType,juliaStringPrefixed,juliaTypeOperatorR1'

syntax match  juliaFunctionCall		contained "[^{([:space:]<>\"]\+(\@=" nextgroup=juliaParBlock

exec 'syntax match   juliaFunctionDef	contained transparent "\%(\<\%(function\|macro\)\)\@'.s:d(8).'<=\s\+\zs' . s:idregex . '\%(\.' . s:idregex . '\)*\ze\s*\%((\|\send\>\|$\)" contains=juliaFunctionName'
exec 'syntax match   juliaFunctionName	contained "\%(\<\%(function\|macro\)\s\+\)\@'.s:d(20).'<=\%(' . s:idregex . '\.\)*\zs' . s:idregex . '"'

exec 'syntax match   juliaStructR	contained transparent "\%(\<\%(\%(mutable\s\+\)\?struct\|\%(abstract\|primitive\)\s\+type\)\s\+\)\@'.s:d(20).'<=\%(' . s:idregex . '\.\)*' . s:idregex . '\>\(\s*(\)\@!" contains=juliaType'

syntax match   juliaKeyword		display "\<\%(return\|local\|global\|const\)\>"
syntax match   juliaInfixKeyword	display "\%(=\s*\)\@<!\<\%(in\|isa\)\>\S\@!\%(\s*=\)\@!"

" The import/export/using keywords introduce a sort of special parsing
" environment with its own rules
exec 'syntax region  juliaImportLine		matchgroup=juliaKeyword excludenl start="\<\%(import\|using\|export\)\>" skip="\%(\%(\<\%(import\|using\|export\)\>\)\|^\)\@'.s:d(6).'<=$" end="$" end="\%([])}]\)\@=" contains=@juliaExpressions,juliaAsKeyword,@juliaContinuationItems,juliaMacroName'
syntax match   juliaAsKeyword		display contained "\<as\>"

syntax match   juliaRepKeyword		display "\<\%(break\|continue\)\>"
syntax region  juliaConditionalBlock	matchgroup=juliaConditional start="\<if\>" end="\<end\>" contains=@juliaExpressions,juliaConditionalEIBlock,juliaConditionalEBlock fold
syntax region  juliaConditionalEIBlock	matchgroup=juliaConditional transparent contained start="\<elseif\>" end="\<\%(end\|else\|elseif\)\>"me=s-1 contains=@juliaExpressions,juliaConditionalEIBlock,juliaConditionalEBlock
syntax region  juliaConditionalEBlock	matchgroup=juliaConditional transparent contained start="\<else\>" end="\<end\>"me=s-1 contains=@juliaExpressions
syntax region  juliaWhileBlock		matchgroup=juliaRepeat start="\<while\>" end="\<end\>" contains=@juliaExpressions fold
syntax region  juliaForBlock		matchgroup=juliaRepeat start="\<for\>" end="\<end\>" contains=@juliaExpressions,juliaOuter fold
syntax region  juliaBeginBlock		matchgroup=juliaBlKeyword start="\<begin\>" end="\<end\>" contains=@juliaExpressions fold
syntax region  juliaFunctionBlock	matchgroup=juliaBlKeyword start="\<function\>" end="\<end\>" contains=@juliaExpressions,juliaFunctionDef fold
syntax region  juliaMacroBlock		matchgroup=juliaBlKeyword start="\<macro\>" end="\<end\>" contains=@juliaExpressions,juliaFunctionDef fold
syntax region  juliaQuoteBlock		matchgroup=juliaBlKeyword start="\<quote\>" end="\<end\>" contains=@juliaExpressions fold
syntax region  juliaStructBlock		matchgroup=juliaBlKeyword start="\<struct\>" end="\<end\>" contains=@juliaExpressions,juliaStructR fold
syntax region  juliaMutableStructBlock	matchgroup=juliaBlKeyword start="\<mutable\s\+struct\>" end="\<end\>" contains=@juliaExpressions,juliaStructR fold
syntax region  juliaLetBlock		matchgroup=juliaBlKeyword start="\<let\>" end="\<end\>" contains=@juliaExpressions fold
syntax region  juliaDoBlock		matchgroup=juliaBlKeyword start="\<do\>" end="\<end\>" contains=@juliaExpressions fold
syntax region  juliaModuleBlock		matchgroup=juliaBlKeyword start="\<\%(bare\)\?module\>" end="\<end\>" contains=@juliaExpressions fold
syntax region  juliaExceptionBlock	matchgroup=juliaException start="\<try\>" end="\<end\>" contains=@juliaExpressions,juliaCatchBlock,juliaFinallyBlock fold
syntax region  juliaCatchBlock		matchgroup=juliaException transparent contained start="\<catch\>" end="\<end\>"me=s-1 contains=@juliaExpressions,juliaFinallyBlock
syntax region  juliaFinallyBlock	matchgroup=juliaException transparent contained start="\<finally\>" end="\<end\>"me=s-1 contains=@juliaExpressions
syntax region  juliaAbstractBlock	matchgroup=juliaBlKeyword start="\<abstract\s\+type\>" end="\<end\>" fold contains=@juliaExpressions,juliaStructR
syntax region  juliaPrimitiveBlock	matchgroup=juliaBlKeyword start="\<primitive\s\+type\>" end="\<end\>" fold contains=@juliaExpressions,juliaStructR

exec 'syntax region  juliaComprehensionFor	matchgroup=juliaComprehensionFor transparent contained start="\%([^[:space:],;:({[]\_s*\)\@'.s:d(80).'<=\<for\>" end="\ze[]);]" contains=@juliaExpressions,juliaComprehensionIf,juliaComprehensionFor'
syntax match   juliaComprehensionIf	contained "\<if\>"

exec 'syntax match   juliaOuter		contained "\<outer\ze\s\+' . s:idregex . '\>"'

syntax match   juliaRangeKeyword	contained "\<\%(begin\|end\)\>"

syntax match   juliaBaseTypeBasic	display "\<\%(\%(N\|Named\)\?Tuple\|Symbol\|Function\|Union\%(All\)\?\|Type\%(Name\|Var\)\?\|Any\|ANY\|Vararg\|Ptr\|Exception\|Module\|Expr\|DataType\|\%(LineNumber\|Quote\)Node\|\%(Weak\|Global\)\?Ref\|Method\|Pair\|Val\|Nothing\|Some\|Missing\)\>"
syntax match   juliaBaseTypeNum		display "\<\%(U\?Int\%(8\|16\|32\|64\|128\)\?\|Float\%(16\|32\|64\)\|Complex\|Bool\|Char\|Number\|Signed\|Unsigned\|Integer\|AbstractFloat\|Real\|Rational\|\%(Abstract\)\?Irrational\|Enum\|BigInt\|BigFloat\|MathConst\|ComplexF\%(16\|32\|64\)\)\>"
syntax match   juliaBaseTypeC		display "\<\%(FileOffset\|C\%(u\?\%(char\|short\|int\|long\(long\)\?\|w\?string\)\|float\|double\|\%(ptrdiff\|s\?size\|wchar\|off\|u\?intmax\)_t\|void\)\)\>"
syntax match   juliaBaseTypeError	display "\<\%(\%(Bounds\|Divide\|Domain\|\%(Stack\)\?Overflow\|EOF\|Undef\%(Ref\|Var\)\|System\|Type\|Parse\|Argument\|Key\|Load\|Method\|Inexact\|OutOfMemory\|Init\|Assertion\|ReadOnlyMemory\|StringIndex\)Error\|\%(Interrupt\|Error\|ProcessExited\|Captured\|Composite\|InvalidState\|Missing\|\%(Process\|Task\)Failed\)Exception\|DimensionMismatch\|SegmentationFault\)\>"
syntax match   juliaBaseTypeIter	display "\<\%(EachLine\|Enumerate\|Cartesian\%(Index\|Range\)\|LinSpace\|CartesianIndices\)\>"
syntax match   juliaBaseTypeString	display "\<\%(DirectIndex\|Sub\|Rep\|Rev\|Abstract\|Substitution\)\?String\>"
syntax match   juliaBaseTypeArray	display "\<\%(\%(Sub\)\?Array\|\%(Abstract\|Dense\|Strided\)\?\%(Array\|Matrix\|Vec\%(tor\|OrMat\)\)\|SparseMatrixCSC\|\%(AbstractSparse\|Bit\|Shared\)\%(Array\|Vector\|Matrix\)\|\%\(D\|Bid\|\%(Sym\)\?Trid\)iagonal\|Hermitian\|Symmetric\|UniformScaling\|\%(Lower\|Upper\)Triangular\|\%(Sparse\|Row\)Vector\|VecElement\|Conj\%(Array\|Matrix\|Vector\)\|Index\%(Cartesian\|Linear\|Style\)\|PermutedDimsArray\|Broadcasted\|Adjoint\|Transpose\|LinearIndices\)\>"
syntax match