Update tree-sitter-julia (#10031)

Update julia parser to latest version, along with: - updating the queries, - pulling changes from `nvim-treesitter`’s queries (as the maintainters of the parser update the queries there), - reversing the queries’ order to be compatible with upstream.
6 months ago · 951b454a33
parent 0364f7c879
commit 951b454a33
6 changed files with 478 additions and 197 deletions
--- a/languages.toml
+++ b/languages.toml
@ -1115,7 +1115,7 @@ indent = { tab-width = 4, unit = "    " }

 [[grammar]]
 name = "julia"
-source = { git = "https://github.com/tree-sitter/tree-sitter-julia", rev = "8fb38abff74652c4faddbf04d2d5bbbc6b4bae25" }
+source = { git = "https://github.com/tree-sitter/tree-sitter-julia", rev = "e84f10db8eeb8b9807786bfc658808edaa1b4fa2" }

 [[language]]
 name = "java"
--- a/runtime/queries/julia/folds.scm
+++ b/runtime/queries/julia/folds.scm
@ -1,11 +1,14 @@
 [
- (module_definition)
- (struct_definition)
- (macro_definition)
- (function_definition)
- (compound_expression) ; begin blocks
- (let_statement)
- (if_statement)
- (for_statement)
- (while_statement)
+  (module_definition)
+  (struct_definition)
+  (macro_definition)
+  (function_definition)
+  (if_statement)
+  (try_statement)
+  (for_statement)
+  (while_statement)
+  (let_statement)
+  (quote_statement)
+  (do_clause)
+  (compound_statement) ; begin block
 ] @fold
--- a/runtime/queries/julia/highlights.scm
+++ b/runtime/queries/julia/highlights.scm
@ -1,38 +1,47 @@
-; ----------
-; Primitives
-; ----------
+; ------------
+; Variables identifiers
+; ------------

-[
-  (line_comment)
-  (block_comment)
-] @comment
+(identifier) @variable

+; Remaining identifiers that start with capital letters should be types (PascalCase)
 (
-  ((identifier) @constant.builtin)
-  (#match? @constant.builtin "^(nothing|missing|undef)$"))
+  (identifier) @type
+  (match? @type "^[A-Z]"))

-[
-  (true)
-  (false)
-] @constant.builtin.boolean
+; SCREAMING_SNAKE_CASE
+(
+  (identifier) @constant
+  (match? @constant "^[A-Z][A-Z0-9_]*$"))

-(integer_literal) @constant.numeric.integer
-(float_literal) @constant.numeric.float
+(const_statement
+  (assignment
+    . (identifier) @constant))

-(
-  ((identifier) @constant.numeric.float)
-  (#match? @constant.numeric.float "^((Inf|NaN)(16|32|64)?)$"))
+; Field expressions are either module content or struct fields.
+; Module types and constants should already be captured, so this
+; assumes the remaining identifiers to be struct fields.
+(field_expression
+  (_)
+  (identifier) @variable.other.member)

-(character_literal) @constant.character
-(escape_sequence) @constant.character.escape
+(quote_expression
+  ":" @string.special.symbol
+  [
+    (identifier)
+    (operator)
+  ] @string.special.symbol)

-(string_literal) @string
+; ------
+; Macros
+; ------

-(prefixed_string_literal
-  prefix: (identifier) @function.macro) @string
+(macro_definition
+  name: (identifier) @function.macro)

-(quote_expression
-  (identifier) @string.special.symbol) 
+(macro_identifier
+  "@" @function.macro
+  (identifier) @function.macro)

 ; -------------------
 ; Modules and Imports
@ -50,49 +59,6 @@
 (scoped_identifier
  (identifier) @namespace)

-; -----
-; Types
-; -----
-
-(abstract_definition
-  name: (identifier) @type)
-  
-(primitive_definition
-  name: (identifier) @type)
-
-(struct_definition
-  name: (identifier) @type)
-
-(struct_definition
-  . (_)
-    (identifier) @variable.other.member)
-
-(struct_definition
-  . (_)
-  (typed_expression
-    . (identifier) @variable.other.member))
-    
-(type_parameter_list
-  (identifier) @type)
-
-(constrained_type_parameter
-  (identifier) @type)
-  
-(subtype_clause
-  (identifier) @type)
-
-(typed_expression
-  (identifier) @type . )
-
-(parameterized_identifier
-  (identifier) @type)
-  
-(type_argument_list
-  (identifier) @type)
-
-(where_clause
-  (identifier) @type)
-
 ; -------------------
 ; Function definition
 ; -------------------
@ -119,22 +85,6 @@
  ; prevent constructors (PascalCase) to be highlighted as functions
  (#match? @function "^[^A-Z]"))

-(parameter_list
-  (identifier) @variable.parameter)
-
-(typed_parameter
-  (identifier) @variable.parameter
-  (identifier)? @type)
-
-(optional_parameter
-  . (identifier) @variable.parameter)
-
-(slurp_parameter
-  (identifier) @variable.parameter)
-
-(function_expression
-  . (identifier) @variable.parameter)
-
 ; ---------------
 ; Functions calls
 ; ---------------
@ -146,13 +96,13 @@
  (#match? @function "^[^A-Z]"))

 (
-  (broadcast_call_expression
-    (identifier) @function)
+  (call_expression
+    (field_expression (identifier) @function .))
  (#match? @function "^[^A-Z]"))

 (
-  (call_expression
-    (field_expression (identifier) @function .))
+  (broadcast_call_expression
+    (identifier) @function)
  (#match? @function "^[^A-Z]"))

 (
@ -160,78 +110,279 @@
    (field_expression (identifier) @function .))
  (#match? @function "^[^A-Z]"))

-; ------
-; Macros
-; ------

-(macro_definition
-  name: (identifier) @function.macro)
+; -------------------
+; Functions builtins
+; -------------------
+
+((identifier) @function.builtin
+  (#any-of? @function.builtin
+    "_abstracttype" "_apply_iterate" "_apply_pure" "_call_in_world" "_call_in_world_total"
+    "_call_latest" "_equiv_typedef" "_expr" "_primitivetype" "_setsuper!" "_structtype" "_typebody!"
+    "_typevar" "applicable" "apply_type" "arrayref" "arrayset" "arraysize" "const_arrayref"
+    "donotdelete" "fieldtype" "get_binding_type" "getfield" "ifelse" "invoke" "isa" "isdefined"
+    "modifyfield!" "nfields" "replacefield!" "set_binding_type!" "setfield!" "sizeof" "svec"
+    "swapfield!" "throw" "tuple" "typeassert" "typeof"))
+
+; -----------
+; Parameters
+; -----------
+
+(parameter_list
+  (identifier) @variable.parameter)
+
+(optional_parameter
+  . (identifier) @variable.parameter)
+
+(slurp_parameter
+  (identifier) @variable.parameter)
+
+(typed_parameter
+  parameter: (identifier)? @variable.parameter
+  type: (_) @type)
+
+(function_expression
+  . (identifier) @variable.parameter) ; Single parameter arrow functions
+
+; -----
+; Types
+; -----
+
+; Definitions
+(abstract_definition
+  name: (identifier) @type.definition) @keyword
+
+(primitive_definition
+  name: (identifier) @type.definition) @keyword
+
+(struct_definition
+  name: (identifier) @type)
+
+(struct_definition
+  . (_)
+    (identifier) @variable.other.member)
+
+(struct_definition
+  . (_)
+  (typed_expression
+    . (identifier) @variable.other.member))
+
+(type_clause
+  [
+    (identifier) @type
+    (field_expression
+      (identifier) @type .)
+  ])
+
+; Annotations
+(parametrized_type_expression
+  (_) @type
+  (curly_expression
+    (_) @type))
+
+(type_parameter_list
+  (identifier) @type)
+
+(typed_expression
+  (identifier) @type . )
+
+(function_definition
+  return_type: (identifier) @type)
+
+(short_function_definition
+  return_type: (identifier) @type)
+
+(where_clause
+  (identifier) @type)
+
+(where_clause
+  (curly_expression
+    (_) @type))
+
+; ---------
+; Builtins
+; ---------
+
+; This list was generated with:
+;
+;  istype(x) = typeof(x) === DataType || typeof(x) === UnionAll
+;  get_types(m) = filter(x -> istype(Base.eval(m, x)), names(m))
+;  type_names = sort(union(get_types(Core), get_types(Base)))
+;
+((identifier) @type.builtin
+  (#any-of? @type.builtin
+    "AbstractArray" "AbstractChannel" "AbstractChar" "AbstractDict" "AbstractDisplay"
+    "AbstractFloat" "AbstractIrrational" "AbstractLock" "AbstractMatch" "AbstractMatrix"
+    "AbstractPattern" "AbstractRange" "AbstractSet" "AbstractSlices" "AbstractString"
+    "AbstractUnitRange" "AbstractVecOrMat" "AbstractVector" "Any" "ArgumentError" "Array"
+    "AssertionError" "Atomic" "BigFloat" "BigInt" "BitArray" "BitMatrix" "BitSet" "BitVector" "Bool"
+    "BoundsError" "By" "CanonicalIndexError" "CapturedException" "CartesianIndex" "CartesianIndices"
+    "Cchar" "Cdouble" "Cfloat" "Channel" "Char" "Cint" "Cintmax_t" "Clong" "Clonglong" "Cmd" "Colon"
+    "ColumnSlices" "Complex" "ComplexF16" "ComplexF32" "ComplexF64" "ComposedFunction"
+    "CompositeException" "ConcurrencyViolationError" "Condition" "Cptrdiff_t" "Cshort" "Csize_t"
+    "Cssize_t" "Cstring" "Cuchar" "Cuint" "Cuintmax_t" "Culong" "Culonglong" "Cushort" "Cvoid"
+    "Cwchar_t" "Cwstring" "DataType" "DenseArray" "DenseMatrix" "DenseVecOrMat" "DenseVector" "Dict"
+    "DimensionMismatch" "Dims" "DivideError" "DomainError" "EOFError" "Enum" "ErrorException"
+    "Exception" "ExponentialBackOff" "Expr" "Float16" "Float32" "Float64" "Function" "GlobalRef"
+    "HTML" "IO" "IOBuffer" "IOContext" "IOStream" "IdDict" "IndexCartesian" "IndexLinear"
+    "IndexStyle" "InexactError" "InitError" "Int" "Int128" "Int16" "Int32" "Int64" "Int8" "Integer"
+    "InterruptException" "InvalidStateException" "Irrational" "KeyError" "LazyString" "LinRange"
+    "LineNumberNode" "LinearIndices" "LoadError" "Lt" "MIME" "Matrix" "Method" "MethodError"
+    "Missing" "MissingException" "Module" "NTuple" "NamedTuple" "Nothing" "Number" "Ordering"
+    "OrdinalRange" "OutOfMemoryError" "OverflowError" "Pair" "ParseError" "PartialQuickSort" "Perm"
+    "PermutedDimsArray" "Pipe" "ProcessFailedException" "Ptr" "QuoteNode" "Rational" "RawFD"
+    "ReadOnlyMemoryError" "Real" "ReentrantLock" "Ref" "Regex" "RegexMatch" "Returns"
+    "ReverseOrdering" "RoundingMode" "RowSlices" "SegmentationFault" "Set" "Signed" "Slices" "Some"
+    "SpinLock" "StackFrame" "StackOverflowError" "StackTrace" "Stateful" "StepRange" "StepRangeLen"
+    "StridedArray" "StridedMatrix" "StridedVecOrMat" "StridedVector" "String" "StringIndexError"
+    "SubArray" "SubString" "SubstitutionString" "Symbol" "SystemError" "Task" "TaskFailedException"
+    "Text" "TextDisplay" "Timer" "Tmstruct" "Tuple" "Type" "TypeError" "TypeVar" "UInt" "UInt128"
+    "UInt16" "UInt32" "UInt64" "UInt8" "UndefInitializer" "UndefKeywordError" "UndefRefError"
+    "UndefVarError" "Union" "UnionAll" "UnitRange" "Unsigned" "Val" "VecElement" "VecOrMat" "Vector"
+    "VersionNumber" "WeakKeyDict" "WeakRef"))
+
+((identifier) @variable.builtin
+  (#any-of? @variable.builtin "begin" "end")
+  (#has-ancestor? @variable.builtin index_expression))
+
+((identifier) @variable.builtin
+  (#any-of? @variable.builtin "begin" "end")
+  (#has-ancestor? @variable.builtin range_expression))

-(macro_identifier
-  "@" @function.macro
-  (identifier) @function.macro)

 ; --------
 ; Keywords
 ; --------

-(function_definition 
-  ["function" "end"] @keyword.function)
+[
+  "global"
+  "local"
+] @keyword
+
+(compound_statement
+  [
+    "begin"
+    "end"
+  ] @keyword)
+
+(quote_statement
+  [
+    "quote"
+    "end"
+  ] @keyword)
+
+(let_statement
+  [
+    "let"
+    "end"
+  ] @keyword)

 (if_statement
-  ["if" "end"] @keyword.control.conditional)
+  [
+    "if"
+    "end"
+  ] @keyword.control.conditional)
+
 (elseif_clause
-  ["elseif"] @keyword.control.conditional)
+  "elseif" @keyword.control.conditional)
+
 (else_clause
-  ["else"] @keyword.control.conditional)
-(ternary_expression
-  ["?" ":"] @keyword.control.conditional)
+  "else" @keyword.control.conditional)

-(for_statement
-  ["for" "end"] @keyword.control.repeat)
-(while_statement
-  ["while" "end"] @keyword.control.repeat)
-(break_statement) @keyword.control.repeat
-(continue_statement) @keyword.control.repeat
-(for_binding
-  "in" @keyword.control.repeat)
-(for_clause
-  "for" @keyword.control.repeat)
+(if_clause
+  "if" @keyword.control.conditional) ; `if` clause in comprehensions
+
+(ternary_expression
+  [
+    "?"
+    ":"
+  ] @keyword.control.conditional)

 (try_statement
-  ["try" "end" ] @keyword.control.exception)
+  [
+    "try"
+    "end"
+  ] @keyword.control.exception)
+
 (finally_clause
  "finally" @keyword.control.exception)
+
 (catch_clause
  "catch" @keyword.control.exception)

+(for_statement
+  [
+    "for"
+    "end"
+  ] @keyword.control.repeat)
+
+(while_statement
+  [
+    "while"
+    "end"
+  ] @keyword.control.repeat)
+
+(for_clause
+  "for" @keyword.control.repeat)
+
 [
-  "export"
-  "import"
-  "using"
-] @keyword.control.import
+  (break_statement)
+  (continue_statement)
+] @keyword.control.repeat
+
+(module_definition
+  [
+    "module"
+    "baremodule"
+    "end"
+  ] @keyword.control.import)
+
+(import_statement
+  [
+    "import"
+    "using"
+  ] @keyword.control.import)
+
+(import_alias
+  "as" @keyword.control.import)
+
+(export_statement
+  "export" @keyword.control.import)
+
+(selected_import
+  ":" @punctuation.delimiter)
+
+(struct_definition
+  [
+    "struct"
+    "end"
+  ] @keyword)
+
+(macro_definition
+  [
+    "macro"
+    "end"
+  ] @keyword)
+
+(function_definition
+  [
+    "function"
+    "end"
+  ] @keyword.function)
+
+(do_clause
+  [
+    "do"
+    "end"
+  ] @keyword.function)
+
+(return_statement
+  "return" @keyword.control.return)

 [
-  "abstract"
-  "baremodule"
-  "begin"
  "const"
-  "do"
-  "end"
-  "let"
-  "macro"
-  "module"
  "mutable"
-  "primitive"
-  "quote"
-  "return"
-  "struct"
-  "type"
-  "where"
-] @keyword
-
-; TODO: fix this
-((identifier) @keyword (#match? @keyword "global|local"))
+] @keyword.storage.modifier

 ; ---------
 ; Operators
@ -239,14 +390,34 @@

 [
  (operator)
-  "::"
-  "<:"
-  ":"
-  "=>"
-  "..."
-  "$"
+  "="
+  "∈"
 ] @operator

+(adjoint_expression
+  "'" @operator)
+
+(range_expression
+  ":" @operator)
+
+((operator) @keyword.operator
+  (#any-of? @keyword.operator "in" "isa"))
+
+(for_binding
+  "in" @keyword.operator)
+
+(where_clause
+  "where" @keyword.operator)
+
+(where_expression
+  "where" @keyword.operator)
+
+(binary_expression
+  (_)
+  (operator) @operator
+  (identifier) @function
+  (#any-of? @operator "|>" ".|>"))
+
 ; ------------
 ; Punctuations
 ; ------------
@ -255,40 +426,58 @@
  "."
  "," 
  ";"
+  "::"
+  "->"
 ] @punctuation.delimiter

+"..." @punctuation.special
+
 [
-  "["
-  "]"
  "("
  ")" 
+  "["
+  "]"
  "{" 
  "}"
 ] @punctuation.bracket

-; ---------------------
-; Remaining identifiers
-; ---------------------
+; ---------
+; Literals
+; ---------

-(const_statement
-  (variable_declaration
-    . (identifier) @constant))
+(boolean_literal) @constant.builtin.boolean
+
+(integer_literal) @constant.numeric.integer
+
+(float_literal) @constant.numeric.float

-; SCREAMING_SNAKE_CASE
 (
-  (identifier) @constant
-  (#match? @constant "^[A-Z][A-Z0-9_]*$"))
+  ((identifier) @constant.numeric.float)
+  (#match? @constant.numeric.float "^((Inf|NaN)(16|32|64)?)$"))

-; remaining identifiers that start with capital letters should be types (PascalCase)
 (
-  (identifier) @type
-  (#match? @type "^[A-Z]"))
+  ((identifier) @constant.builtin)
+  (#match? @constant.builtin "^(nothing|missing|undef)$"))

-; Field expressions are either module content or struct fields.
-; Module types and constants should already be captured, so this
-; assumes the remaining identifiers to be struct fields.
-(field_expression
-  (_)
-  (identifier) @variable.other.member)
+(character_literal) @constant.character

-(identifier) @variable
+(escape_sequence) @constant.character.escape
+
+(string_literal) @string
+
+(prefixed_string_literal
+  prefix: (identifier) @function.macro) @string
+
+(command_literal) @string
+
+(prefixed_command_literal
+  prefix: (identifier) @function.macro) @string
+
+; ---------
+; Comments
+; ---------
+
+[
+  (line_comment)
+  (block_comment)
+] @comment
--- a/runtime/queries/julia/indents.scm
+++ b/runtime/queries/julia/indents.scm
@ -2,15 +2,39 @@
  (struct_definition)
  (macro_definition)
  (function_definition)
-  (compound_expression)
-  (let_statement)
+  (compound_statement)
  (if_statement)
+  (try_statement)
  (for_statement)
  (while_statement)
+  (let_statement)
+  (quote_statement)
  (do_clause)
-  (parameter_list)
+  (assignment)
+  (for_binding)
+  (call_expression)
+  (parenthesized_expression)
+  (tuple_expression)
+  (comprehension_expression)
+  (matrix_expression)
+  (vector_expression)
 ] @indent

 [
  "end"
+  ")"
+  "]"
+  "}"
 ] @outdent
+
+(argument_list
+  . (_) @anchor
+  (#set! "scope" "tail")) @align
+
+(parameter_list
+  . (_) @anchor
+  (#set! "scope" "tail")) @align
+
+(curly_expression
+  . (_) @anchor
+  (#set! "scope" "tail")) @align
--- a/runtime/queries/julia/injections.scm
+++ b/runtime/queries/julia/injections.scm
@ -9,7 +9,8 @@
      (primitive_definition)
      (abstract_definition)
      (struct_definition)
-      (assignment_expression)
+      (short_function_definition)
+      (assignment)
      (const_statement)
    ])
  (#set! injection.language "markdown"))
@ -21,10 +22,17 @@
  ] @injection.content
  (#set! injection.language "comment"))

+(
+  [
+    (command_literal)
+    (prefixed_command_literal)
+  ] @injection.content
+  (#set! injection.language "sh"))
+
 (
  (prefixed_string_literal
    prefix: (identifier) @function.macro) @injection.content
-  (#eq? @function.macro "re")
+  (#eq? @function.macro "r")
  (#set! injection.language "regex"))

 (
--- a/runtime/queries/julia/locals.scm
+++ b/runtime/queries/julia/locals.scm
@ -2,43 +2,100 @@
 ; Definitions
 ; -----------

-; Imports
-(import_statement
+; Variables
+(assignment
  (identifier) @local.definition)

+(assignment
+  (tuple_expression
+    (identifier) @local.definition))
+
 ; Constants
 (const_statement
-  (variable_declaration
+  (assignment
    . (identifier) @local.definition))

+; let/const bindings
+(let_binding
+  (identifier) @local.definition)
+
+(let_binding
+  (tuple_expression
+    (identifier) @local.definition))
+
+; For bindings
+(for_binding
+  (identifier) @local.definition)
+
+(for_binding
+  (tuple_expression
+    (identifier) @local.definition))
+
+; Types
+(struct_definition
+  name: (identifier) @local.definition)
+
+(abstract_definition
+  name: (identifier) @local.definition)
+
+(abstract_definition
+  name: (identifier) @local.definition)
+
+(type_parameter_list
+  (identifier) @local.definition)
+
+; Module imports
+(import_statement
+  (identifier) @local.definition)
+
 ; Parameters
 (parameter_list
  (identifier) @local.definition)

-(typed_parameter
-  . (identifier) @local.definition)
-
-(optional_parameter .
+(optional_parameter
+  .
  (identifier) @local.definition)

 (slurp_parameter
  (identifier) @local.definition)

+(typed_parameter
+  parameter: (identifier) @local.definition
+  (_))
+
+; Single parameter arrow function
 (function_expression
-  . (identifier) @local.definition)
+  .
+  (identifier) @local.definition)

-; ------
-; Scopes
-; ------
+; Function/macro definitions
+(function_definition
+  name: (identifier) @local.definition) @local.scope

-[
-  (function_definition)
-  (short_function_definition)
-  (macro_definition)
-] @local.scope
+(short_function_definition
+  name: (identifier) @local.definition) @local.scope
+
+(macro_definition
+  name: (identifier) @local.definition) @local.scope

 ; ----------
 ; References
 ; ----------

 (identifier) @local.reference
+ 
+; ------
+; Scopes
+; ------
+
+[
+  (for_statement)
+  (while_statement)
+  (try_statement)
+  (catch_clause)
+  (finally_clause)
+  (let_statement)
+  (quote_statement)
+  (do_clause)
+] @local.scope 
+