path: root/runtime/syntax/ocaml.vim

" Vim syntax file
" Language:     OCaml
" Filenames:    *.ml *.mli *.mll *.mly
" Maintainers:  Markus Mottl      <markus.mottl@gmail.com>
"               Karl-Heinz Sylla  <Karl-Heinz.Sylla@gmd.de>
"               Issac Trotts      <ijtrotts@ucdavis.edu>
" URL:          https://github.com/ocaml/vim-ocaml
" Last Change:
"               2019 Nov 05 - Accurate type highlighting (Maëlan)
"               2018 Nov 08 - Improved highlighting of operators (Maëlan)
"               2018 Apr 22 - Improved support for PPX (Andrey Popp)
"               2018 Mar 16 - Remove raise, lnot and not from keywords (Étienne Millon, "copy")
"               2017 Apr 11 - Improved matching of negative numbers (MM)
"               2016 Mar 11 - Improved support for quoted strings (Glen Mével)
"               2015 Aug 13 - Allow apostrophes in identifiers (Jonathan Chan, Einar Lielmanis)
"               2015 Jun 17 - Added new "nonrec" keyword (MM)

" A minor patch was applied to the official version so that object/end
" can be distinguished from begin/end, which is used for indentation,
" and folding. (David Baelde)

" Quit when a syntax file was already loaded
if exists("b:current_syntax") && b:current_syntax == "ocaml"
  finish
endif

let s:keepcpo = &cpo
set cpo&vim

" ' can be used in OCaml identifiers
setlocal iskeyword+='

" ` is part of the name of polymorphic variants
setlocal iskeyword+=`

" OCaml is case sensitive.
syn case match

" Access to the method of an object
syn match    ocamlMethod       "#"

" Scripting directives
syn match    ocamlScript "^#\<\(quit\|labels\|warnings\|warn_error\|directory\|remove_directory\|cd\|load\|load_rec\|use\|mod_use\|install_printer\|remove_printer\|require\|list\|ppx\|principal\|predicates\|rectypes\|thread\|trace\|untrace\|untrace_all\|print_depth\|print_length\|camlp4o\|camlp4r\|topfind_log\|topfind_verbose\)\>"

" lowercase identifier - the standard way to match
syn match    ocamlLCIdentifier /\<\(\l\|_\)\(\w\|'\)*\>/

" Errors
syn match    ocamlBraceErr   "}"
syn match    ocamlBrackErr   "\]"
syn match    ocamlParenErr   ")"
syn match    ocamlArrErr     "|]"

syn match    ocamlCountErr   "\<downto\>"
syn match    ocamlCountErr   "\<to\>"

if !exists("ocaml_revised")
  syn match    ocamlDoErr      "\<do\>"
endif

syn match    ocamlDoneErr    "\<done\>"
syn match    ocamlThenErr    "\<then\>"

" Error-highlighting of "end" without synchronization:
" as keyword or as error (default)
if exists("ocaml_noend_error")
  syn match    ocamlKeyword    "\<end\>"
else
  syn match    ocamlEndErr     "\<end\>"
endif

" These keywords are only expected nested in constructions that are handled by
" the type linter, so outside of type contexts we highlight them as errors:
syn match    ocamlKwErr  "\<\(mutable\|nonrec\|of\|private\)\>"

" Some convenient clusters
syn cluster  ocamlAllErrs contains=@ocamlAENoParen,ocamlParenErr
syn cluster  ocamlAENoParen contains=ocamlBraceErr,ocamlBrackErr,ocamlCountErr,ocamlDoErr,ocamlDoneErr,ocamlEndErr,ocamlThenErr,ocamlKwErr

syn cluster  ocamlContained contains=ocamlTodo,ocamlPreDef,ocamlModParam,ocamlModParam1,ocamlModTypePre,ocamlModRHS,ocamlFuncWith,ocamlModTypeRestr,ocamlModTRWith,ocamlWith,ocamlWithRest,ocamlFullMod,ocamlVal


" Enclosing delimiters
syn region   ocamlNone transparent matchgroup=ocamlEncl start="(" matchgroup=ocamlEncl end=")" contains=ALLBUT,@ocamlContained,ocamlParenErr
syn region   ocamlNone transparent matchgroup=ocamlEncl start="{" matchgroup=ocamlEncl end="}"  contains=ALLBUT,@ocamlContained,ocamlBraceErr
syn region   ocamlNone transparent matchgroup=ocamlEncl start="\[" matchgroup=ocamlEncl end="\]" contains=ALLBUT,@ocamlContained,ocamlBrackErr
syn region   ocamlNone transparent matchgroup=ocamlEncl start="\[|" matchgroup=ocamlEncl end="|\]" contains=ALLBUT,@ocamlContained,ocamlArrErr


" Comments
syn region   ocamlComment start="(\*" end="\*)" contains=@Spell,ocamlComment,ocamlTodo
syn keyword  ocamlTodo contained TODO FIXME XXX NOTE


" Objects
syn region   ocamlEnd matchgroup=ocamlObject start="\<object\>" matchgroup=ocamlObject end="\<end\>" contains=ALLBUT,@ocamlContained,ocamlEndErr


" Blocks
if !exists("ocaml_revised")
  syn region   ocamlEnd matchgroup=ocamlKeyword start="\<begin\>" matchgroup=ocamlKeyword end="\<end\>" contains=ALLBUT,@ocamlContained,ocamlEndErr
endif


" "for"
syn region   ocamlNone matchgroup=ocamlKeyword start="\<for\>" matchgroup=ocamlKeyword end="\<\(to\|downto\)\>" contains=ALLBUT,@ocamlContained,ocamlCountErr


" "do"
if !exists("ocaml_revised")
  syn region   ocamlDo matchgroup=ocamlKeyword start="\<do\>" matchgroup=ocamlKeyword end="\<done\>" contains=ALLBUT,@ocamlContained,ocamlDoneErr
endif

" "if"
syn region   ocamlNone matchgroup=ocamlKeyword start="\<if\>" matchgroup=ocamlKeyword end="\<then\>" contains=ALLBUT,@ocamlContained,ocamlThenErr

"" PPX nodes

syn match ocamlPpxIdentifier /\(\[@\{1,3\}\)\@<=\w\+\(\.\w\+\)*/
syn region ocamlPpx matchgroup=ocamlPpxEncl start="\[@\{1,3\}" contains=TOP end="\]"

"" Modules

" "open"
syn match   ocamlKeyword "\<open\>" skipwhite skipempty nextgroup=ocamlFullMod

" "include"
syn match    ocamlKeyword "\<include\>" skipwhite skipempty nextgroup=ocamlModParam,ocamlFullMod

" "module" - somewhat complicated stuff ;-)
" 2022-10: please document it?
syn region   ocamlModule matchgroup=ocamlKeyword start="\<module\>" matchgroup=ocamlModule end="\<_\|\u\(\w\|'\)*\>" contains=@ocamlAllErrs,ocamlComment skipwhite skipempty nextgroup=ocamlPreDef
syn region   ocamlPreDef start="."me=e-1 end="[a-z:=)]\@=" contained contains=@ocamlAllErrs,ocamlComment,ocamlModParam,ocamlGenMod,ocamlModTypeRestr nextgroup=ocamlModTypePre,ocamlModPreRHS
syn region   ocamlModParam start="(\*\@!" end=")" contained contains=ocamlGenMod,ocamlModParam,ocamlModParam1,ocamlSig,ocamlVal
syn match    ocamlModParam1 "\<\u\(\w\|'\)*\>" contained skipwhite skipempty
syn match    ocamlGenMod "()" contained skipwhite skipempty

syn match    ocamlModTypePre ":" contained skipwhite skipempty nextgroup=ocamlModTRWith,ocamlSig,ocamlFunctor,ocamlModTypeRestr,ocamlModTypeOf
syn match    ocamlModTypeRestr "\<\w\(\w\|'\)*\( *\. *\w\(\w\|'\)*\)*\>" contained

syn match    ocamlModPreRHS "=" contained skipwhite skipempty nextgroup=ocamlModParam,ocamlFullMod
syn keyword  ocamlKeyword val
syn region   ocamlVal matchgroup=ocamlKeyword start="\<val\>" matchgroup=ocamlLCIdentifier end="\<\l\(\w\|'\)*\>" contains=@ocamlAllErrs,ocamlComment,ocamlFullMod skipwhite skipempty nextgroup=ocamlModTypePre
syn region   ocamlModRHS start="." end=". *\w\|([^*]"me=e-2 contained contains=ocamlComment skipwhite skipempty nextgroup=ocamlModParam,ocamlFullMod
syn match    ocamlFullMod "\<\u\(\w\|'\)*\( *\. *\u\(\w\|'\)*\)*" contained skipwhite skipempty nextgroup=ocamlFuncWith

syn region   ocamlFuncWith start="([*)]\@!" end=")" contained contains=ocamlComment,ocamlWith,ocamlStruct skipwhite skipempty nextgroup=ocamlFuncWith

syn region   ocamlModTRWith start="(\*\@!" end=")" contained contains=@ocamlAENoParen,ocamlWith
syn match    ocamlWith "\<\(\u\(\w\|'\)* *\. *\)*\w\(\w\|'\)*\>" contained skipwhite skipempty nextgroup=ocamlWithRest
syn region   ocamlWithRest start="[^)]" end=")"me=e-1 contained contains=ALLBUT,@ocamlContained

" "struct"
syn region   ocamlStruct matchgroup=ocamlStructEncl start="\<\(module\s\+\)\=struct\>" matchgroup=ocamlStructEncl end="\<end\>" contains=ALLBUT,@ocamlContained,ocamlEndErr

" "sig"
syn region   ocamlSig matchgroup=ocamlSigEncl start="\<sig\>" matchgroup=ocamlSigEncl end="\<end\>" contains=ALLBUT,@ocamlContained,ocamlEndErr

" "functor"
syn region   ocamlFunctor start="\<functor\>" matchgroup=ocamlKeyword end="->" contains=@ocamlAllErrs,ocamlComment,ocamlModParam,ocamlGenMod skipwhite skipempty nextgroup=ocamlStruct,ocamlSig,ocamlFuncWith,ocamlFunctor

" "module type"
syn region   ocamlModTypeOf start="\<module\s\+type\(\s\+of\)\=\>" matchgroup=ocamlModule end="\<\w\(\w\|'\)*\>" contains=ocamlComment skipwhite skipempty nextgroup=ocamlMTDef
syn match    ocamlMTDef "=\s*\w\(\w\|'\)*\>"hs=s+1,me=s+1 skipwhite skipempty nextgroup=ocamlFullMod

" Quoted strings
syn region ocamlString matchgroup=ocamlQuotedStringDelim start="{\z\([a-z_]*\)|" end="|\z1}" contains=@Spell
syn region ocamlString matchgroup=ocamlQuotedStringDelim start="{%[a-z_]\+\(\.[a-z_]\+\)\?\( \z\([a-z_]\+\)\)\?|" end="|\z1}" contains=@Spell

syn keyword  ocamlKeyword  and as assert class
syn keyword  ocamlKeyword  else
syn keyword  ocamlKeyword  external
syn keyword  ocamlKeyword  in inherit initializer
syn keyword  ocamlKeyword  lazy let match
syn keyword  ocamlKeyword  method new
syn keyword  ocamlKeyword  parser rec
syn keyword  ocamlKeyword  try
syn keyword  ocamlKeyword  virtual when while with

" Keywords which are handled by the type linter:
"     as (within a type equation)
"     constraint exception mutable nonrec of private type

" The `fun` keyword has special treatment because of the syntax `fun … : t -> e`
" where `->` ends the type context rather than being part of it; to handle that,
" we blacklist the ocamlTypeAnnot matchgroup, and we plug ocamlFunTypeAnnot
" instead (later in this file, by using containedin=ocamlFun):
syn region ocamlFun matchgroup=ocamlKeyword start='\<fun\>' matchgroup=ocamlArrow end='->'
\ contains=ALLBUT,@ocamlContained,ocamlArrow,ocamlInfixOp,ocamlTypeAnnot

if exists("ocaml_revised")
  syn keyword  ocamlKeyword  do value
  syn keyword  ocamlBoolean  True False
else
  syn keyword  ocamlKeyword  function
  syn keyword  ocamlBoolean  true false
endif

syn match    ocamlEmptyConstructor  "(\s*)"
syn match    ocamlEmptyConstructor  "\[\s*\]"
syn match    ocamlEmptyConstructor  "\[|\s*>|]"
syn match    ocamlEmptyConstructor  "\[<\s*>\]"
syn match    ocamlConstructor  "\u\(\w\|'\)*\>"

" Polymorphic variants
syn match    ocamlConstructor  "`\w\(\w\|'\)*\>"

" Module prefix
syn match    ocamlModPath      "\u\(\w\|'\)* *\."he=e-1

syn match    ocamlCharacter    "'\\\d\d\d'\|'\\[\'ntbr]'\|'.'"
syn match    ocamlCharacter    "'\\x\x\x'"
syn match    ocamlCharErr      "'\\\d\d'\|'\\\d'"
syn match    ocamlCharErr      "'\\[^\'ntbr]'"
syn region   ocamlString       start=+"+ skip=+\\\\\|\\"+ end=+"+ contains=@Spell

syn match    ocamlAnyVar       "\<_\>"
syn match    ocamlKeyChar      "|]\@!"
syn match    ocamlKeyChar      ";"
syn match    ocamlKeyChar      "\~"
syn match    ocamlKeyChar      "?"

" NOTE: for correct precedence, the rule for ";;" must come after that for ";"
syn match    ocamlTopStop      ";;"

"" Operators

" The grammar of operators is found there:
"     https://caml.inria.fr/pub/docs/manual-ocaml/names.html#operator-name
"     https://caml.inria.fr/pub/docs/manual-ocaml/extn.html#s:ext-ops
"     https://caml.inria.fr/pub/docs/manual-ocaml/extn.html#s:index-operators
" = is both an operator name and a keyword, we let the user choose how
" to display it (has to be declared before regular infix operators):
syn match    ocamlEqual        "="
" Custom indexing operators:
syn region   ocamlIndexing matchgroup=ocamlIndexingOp
  \ start="\.[~?!:|&$%=>@^/*+-][~?!.:|&$%<=>@^*/+-]*\_s*("
  \ end=")\(\_s*<-\)\?"
  \ contains=ALLBUT,@ocamlContained,ocamlParenErr
syn region   ocamlIndexing matchgroup=ocamlIndexingOp
  \ start="\.[~?!:|&$%=>@^/*+-][~?!.:|&$%<=>@^*/+-]*\_s*\["
  \ end="]\(\_s*<-\)\?"
  \ contains=ALLBUT,@ocamlContained,ocamlBrackErr
syn region   ocamlIndexing matchgroup=ocamlIndexingOp
  \ start="\.[~?!:|&$%=>@^/*+-][~?!.:|&$%<=>@^*/+-]*\_s*{"
  \ end="}\(\_s*<-\)\?"
  \ contains=ALLBUT,@ocamlContained,ocamlBraceErr
" Extension operators (has to be declared before regular infix operators):
syn match    ocamlExtensionOp          "#[#~?!.:|&$%<=>@^*/+-]\+"
" Infix and prefix operators:
syn match    ocamlPrefixOp              "![~?!.:|&$%<=>@^*/+-]*"
syn match    ocamlPrefixOp           "[~?][~?!.:|&$%<=>@^*/+-]\+"
syn match    ocamlInfixOp   "[&$%<>@^*/+-][~?!.:|&$%<=>@^*/+-]*"
syn match    ocamlInfixOp            "[|=][~?!.:|&$%<=>@^*/+-]\+"
syn match    ocamlInfixOp               "#[~?!.:|&$%<=>@^*/+-]\+#\@!"
syn match    ocamlInfixOp              "!=[~?!.:|&$%<=>@^*/+-]\@!"
syn keyword  ocamlInfixOpKeyword      asr land lor lsl lsr lxor mod or
" := is technically an infix operator, but we may want to show it as a keyword
" (somewhat analogously to = for let‐bindings and <- for assignations):
syn match    ocamlRefAssign    ":="
" :: is technically not an operator, but we may want to show it as such:
syn match    ocamlCons         "::"
" -> and <- are keywords, not operators (but can appear in longer operators):
syn match    ocamlArrow        "->[~?!.:|&$%<=>@^*/+-]\@!"
if exists("ocaml_revised")
  syn match    ocamlErr        "<-[~?!.:|&$%<=>@^*/+-]\@!"
else
  syn match    ocamlKeyChar    "<-[~?!.:|&$%<=>@^*/+-]\@!"
endif

" Script shebang (has to be declared after operators)
syn match    ocamlShebang       "\%1l^#!.*$"

syn match    ocamlNumber        "-\=\<\d\(_\|\d\)*[l|L|n]\?\>"
syn match    ocamlNumber        "-\=\<0[x|X]\(\x\|_\)\+[l|L|n]\?\>"
syn match    ocamlNumber        "-\=\<0[o|O]\(\o\|_\)\+[l|L|n]\?\>"
syn match    ocamlNumber        "-\=\<0[b|B]\([01]\|_\)\+[l|L|n]\?\>"
syn match    ocamlFloat         "-\=\<\d\(_\|\d\)*\.\?\(_\|\d\)*\([eE][-+]\=\d\(_\|\d\)*\)\=\>"

" Labels
syn match    ocamlLabel        "[~?]\(\l\|_\)\(\w\|'\)*:\?"
syn region   ocamlLabel transparent matchgroup=ocamlLabel start="[~?](\(\l\|_\)\(\w\|'\)*"lc=2 end=")"me=e-1 contains=ALLBUT,@ocamlContained,ocamlParenErr

""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""&q