shrimp/src/parser/shrimp.grammar

@external propSource highlighting from "./highlight"

@context trackScope from "./parserScopeContext"

@skip { space | Comment }

@top Program { item* }

@external tokens operatorTokenizer from "./operatorTokenizer" { Star, Slash, Plus, Minus, And, Or, Eq, EqEq, Neq, Lt, Lte, Gt, Gte, Modulo, PlusEq, MinusEq, StarEq, SlashEq, ModuloEq }

@tokens {
  @precedence { Number Regex }

  StringFragment { !['\\$]+ }
  NamedArgPrefix { $[a-z-]+ "=" }
  Number {
    ("-" | "+")? "0x" $[0-9a-fA-F]+ |
    ("-" | "+")? "0b" $[01]+ |
    ("-" | "+")? $[0-9]+ ('.' $[0-9]+)?
  }
  Boolean { "true" | "false" }
  newlineOrSemicolon { "\n" | ";" }
  eof { @eof }
  space { " " | "\t" }
  Comment { "#" ![\n]* }
  leftParen { "(" }
  rightParen { ")" }
  colon[closedBy="end", @name="colon"] { ":" }
  Underscore { "_" }
  Regex { "//" (![/\\\n[] | "\\" ![\n] | "[" (![\n\\\]] | "\\" ![\n])* "]")+ ("//" $[gimsuy]*)? } // Stolen from the lezer JavaScript grammar
  "|"[@name=operator]
}

end { @specialize[@name=keyword]<Identifier, "end"> }
while { @specialize[@name=keyword]<Identifier, "while"> }
if { @specialize[@name=keyword]<Identifier, "if"> }
else { @specialize[@name=keyword]<Identifier, "else"> }
try { @specialize[@name=keyword]<Identifier, "try"> }
catch { @specialize[@name=keyword]<Identifier, "catch"> }
finally { @specialize[@name=keyword]<Identifier, "finally"> }
throw { @specialize[@name=keyword]<Identifier, "throw"> }
null { @specialize[@name=Null]<Identifier, "null"> }

@external tokens tokenizer from "./tokenizer" { Identifier, AssignableIdentifier, Word, IdentifierBeforeDot }
@external specialize {Identifier} specializeKeyword from "./tokenizer" { Do }

@precedence {
  pipe @left,
  or @left,
  and @left,
  comparison @left,
  multiplicative @left,
  additive @left,
  call
}

item {
  consumeToTerminator newlineOrSemicolon |
  consumeToTerminator eof |
  newlineOrSemicolon // allow blank lines
}

consumeToTerminator {
  PipeExpr |
  WhileExpr |
  FunctionCallWithBlock |
  ambiguousFunctionCall |
  TryExpr |
  Throw |
  IfExpr |
  FunctionDef |
  CompoundAssign |
  Assign |
  BinOp |
  ConditionalOp |
  expressionWithoutIdentifier
}

PipeExpr {
  pipeOperand (!pipe "|" pipeOperand)+
}

pipeOperand {
  consumeToTerminator
}

WhileExpr {
  while (ConditionalOp | expression) colon Block end
}

Block {
  consumeToTerminator | newlineOrSemicolon block
}

FunctionCallWithBlock {
  ambiguousFunctionCall colon Block CatchExpr? FinallyExpr? end
}

FunctionCallOrIdentifier {
  DotGet | Identifier
}

ambiguousFunctionCall {
  FunctionCall | FunctionCallOrIdentifier
}

FunctionCall {
  (DotGet | Identifier | ParenExpr) arg+
}

arg {
  PositionalArg | NamedArg
}

PositionalArg {
  expression | FunctionDef | Underscore
}

NamedArg {
  NamedArgPrefix (expression | FunctionDef | Underscore)
}

FunctionDef {
  Do Params colon (consumeToTerminator | newlineOrSemicolon block) CatchExpr? FinallyExpr? end
}

ifTest {
  ConditionalOp | expression | FunctionCall
}

IfExpr {
  if ifTest colon Block ElseIfExpr* ElseExpr? end
}

ElseIfExpr {
  else if ifTest colon Block
}

ElseExpr {
  else colon Block
}

TryExpr {
  try colon Block CatchExpr? FinallyExpr? end
}

CatchExpr {
  catch Identifier colon Block
}

FinallyExpr {
  finally colon Block
}

Throw {
  throw (BinOp | ConditionalOp | expression)
}

ConditionalOp {
  expression !comparison EqEq expression |
  expression !comparison Neq expression |
  expression !comparison Lt expression |
  expression !comparison Lte expression |
  expression !comparison Gt expression |
  expression !comparison Gte expression |
  (expression | ConditionalOp) !and And (expression | ConditionalOp) |
  (expression | ConditionalOp) !or Or (expression | ConditionalOp)
}

Params {
  Identifier* NamedParam*
}

NamedParam {
  NamedArgPrefix (String | Number | Boolean | null)
}

Assign {
  (AssignableIdentifier | Array) Eq consumeToTerminator
}

CompoundAssign {
  AssignableIdentifier (PlusEq | MinusEq | StarEq | SlashEq | ModuloEq) consumeToTerminator
}

BinOp {
  expression !multiplicative Modulo expression |
  (expression | BinOp) !multiplicative Star (expression | BinOp) |
  (expression | BinOp) !multiplicative Slash (expression | BinOp) |
  (expression | BinOp) !additive Plus (expression | BinOp) |
  (expression | BinOp) !additive Minus (expression | BinOp)
}

ParenExpr {
  leftParen (IfExpr | ambiguousFunctionCall | BinOp | expressionWithoutIdentifier | ConditionalOp | PipeExpr | FunctionDef) rightParen
}

expression {
  expressionWithoutIdentifier | DotGet | Identifier
}


@local tokens {
  dot { "." }
}

@skip {} {
  DotGet {
    IdentifierBeforeDot dot (Number | Identifier | ParenExpr)
  }

  String { "'" stringContent* "'" }
}

stringContent {
  StringFragment |
  Interpolation |
  EscapeSeq
}

Interpolation {
  "$" Identifier |
  "$" ParenExpr
}

EscapeSeq {
  "\\" ("$" | "n" | "t" | "r" | "\\" | "'")
}

Dict {
  "[=]" |
  "[" newlineOrSemicolon* NamedArg (newlineOrSemicolon | NamedArg)* "]"
}

Array {
  "[" newlineOrSemicolon* (expression (newlineOrSemicolon | expression)*)? "]"
}

// We need expressionWithoutIdentifier to avoid conflicts in consumeToTerminator.
// Without this, when parsing "my-var" at statement level, the parser can't decide:
//   - ambiguousFunctionCall → FunctionCallOrIdentifier → Identifier
//   - expression → Identifier
// Both want the same Identifier token! So we use expressionWithoutIdentifier
// to remove Identifier from the second path, forcing standalone identifiers
// to go through ambiguousFunctionCall (which is what we want semantically).
// Yes, it is annoying and I gave up trying to use GLR to fix it.
expressionWithoutIdentifier {
  ParenExpr | Word | String | Number | Boolean | Regex | Dict | Array | null
}

block {
  (consumeToTerminator? newlineOrSemicolon)*
}