carbon-lang/executable_semantics/syntax/parser.ypp

// Part of the Carbon Language project, under the Apache License v2.0 with LLVM
// Exceptions. See /LICENSE for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception

// -----------------------------------------------------------------------------
// Bison Configuration
// -----------------------------------------------------------------------------

%require "3.2"
%language "c++"

// We don't need a separate header for Bison locations.
%define api.location.file none

// Use a type-safe C++ variant for semantic values
%define api.value.type variant

// Have Bison generate the functions ‘make_TEXT’ and ‘make_NUMBER’, but also
// ‘make_YYEOF’, for the end of input.
%define api.token.constructor

// Make parse error messages more detailed
%define parse.error verbose

// Enable support for parser debugging
%define parse.trace true

//
// Parameters to the parser and lexer
//
// Parameters to the parser are stored therein as protected data members, and
// thus available to its methods.

// "out" parameter passed to the parser, where the AST is written.
%parse-param {std::optional<Carbon::AST>& parsed_program}

// "inout" parameter passed to both the parser and the lexer.
%param {Carbon::ParseAndLexContext& context}

// No shift-reduce conflicts are expected.
%expect 0

// -----------------------------------------------------------------------------

%code top {
#include <algorithm>
#include <cstdarg>
#include <cstdio>
#include <cstdlib>
#include <iostream>
#include <list>
#include <vector>

#include "executable_semantics/syntax/syntax_helpers.h"
#include "executable_semantics/syntax/parse_and_lex_context.h"
}  // %code top

%code requires {
#include <optional>

#include "executable_semantics/ast/abstract_syntax_tree.h"
#include "executable_semantics/ast/declaration.h"
#include "executable_semantics/ast/expression.h"
#include "executable_semantics/ast/function_definition.h"
#include "executable_semantics/ast/pattern.h"
#include "executable_semantics/common/arena.h"
#include "executable_semantics/syntax/paren_contents.h"

namespace Carbon {
class ParseAndLexContext;
}  // namespace Carbon

}  // %code requires

%code {

extern int yylineno;

void yy::parser::error(const location_type&, const std::string& message) {
  context.PrintDiagnostic(message, yylineno);
}

}  // %code

%token <int> integer_literal
%token <std::string> identifier
%type <std::string> designator
%type <Carbon::Declaration> declaration
%type <Carbon::FunctionDefinition> function_declaration
%type <Carbon::FunctionDefinition> function_definition
%type <std::list<Carbon::Declaration>> declaration_list
%type <const Carbon::Statement*> statement
%type <const Carbon::Statement*> if_statement
%type <const Carbon::Statement*> optional_else
%type <std::pair<const Carbon::Expression*, bool>> return_expression
%type <const Carbon::Statement*> block
%type <const Carbon::Statement*> statement_list
%type <const Carbon::Expression*> expression
%type <Carbon::GenericBinding> generic_binding
%type <std::vector<Carbon::GenericBinding>> deduced_params
%type <std::vector<Carbon::GenericBinding>> deduced_param_list
%type <const Carbon::Pattern*> pattern
%type <const Carbon::Pattern*> non_expression_pattern
%type <std::pair<const Carbon::Expression*, bool>> return_type
%type <const Carbon::Expression*> paren_expression
%type <const Carbon::Expression*> tuple
%type <std::optional<std::string>> binding_lhs
%type <const Carbon::BindingPattern*> variable_declaration
%type <Carbon::Member*> member
%type <std::list<Carbon::Member*>> member_list
%type <Carbon::ParenContents<Carbon::Expression>::Element> paren_expression_element
%type <Carbon::ParenContents<Carbon::Expression>> paren_expression_base
%type <Carbon::ParenContents<Carbon::Expression>> paren_expression_contents
%type <const Carbon::Pattern*> paren_pattern
%type <const Carbon::TuplePattern*> tuple_pattern
%type <const Carbon::TuplePattern*> maybe_empty_tuple_pattern
%type <Carbon::ParenContents<Carbon::Pattern>> paren_pattern_base
%type <Carbon::ParenContents<Carbon::Pattern>::Element> paren_pattern_element
%type <Carbon::ParenContents<Carbon::Pattern>> paren_pattern_contents
%type <std::pair<std::string, const Carbon::Expression*>> alternative
%type <std::list<std::pair<std::string, const Carbon::Expression*>>> alternative_list
%type <std::pair<const Carbon::Pattern*, const Carbon::Statement*>*> clause
%type <std::list<std::pair<const Carbon::Pattern*, const Carbon::Statement*>>*> clause_list
%token END_OF_FILE 0
%token AND
%token OR
%token NOT
%token INT
%token BOOL
%token TYPE
%token FN
%token FNTY
%token ARROW "->"
%token FNARROW "-> in return type"
%token VAR
%token EQUAL_EQUAL
%token IF
%token ELSE
%token WHILE
%token CONTINUATION_TYPE
%token CONTINUATION
%token RUN
%token AWAIT
%token BREAK
%token CONTINUE
%token RETURN
%token TRUE
%token FALSE
%token STRUCT
%token CHOICE
%token MATCH
%token CASE
%token DBLARROW "=>"
%token DEFAULT
%token AUTO
%token UNDERSCORE
%token
  EQUAL  "="
  MINUS  "-"
  PLUS   "+"
  // The lexer determines the arity and fixity of each `*` based on whitespace
  // and adjacent tokens. UNARY_STAR indicates that the operator is unary but
  // could be either prefix or postfix.
  UNARY_STAR "unary *"
  PREFIX_STAR "prefix *"
  POSTFIX_STAR "postfix *"
  BINARY_STAR "binary *"
  SLASH  "/"
  LEFT_PARENTHESIS "("
  RIGHT_PARENTHESIS ")"
  LEFT_CURLY_BRACE "{"
  RIGHT_CURLY_BRACE "}"
  LEFT_SQUARE_BRACKET "["
  RIGHT_SQUARE_BRACKET "]"
  PERIOD "."
  COMMA ","
  SEMICOLON ";"
  COLON_BANG ":!"
  COLON ":"
;

%precedence FNARROW
%precedence "{" "}"
%precedence ":!" ":" "," DBLARROW
%left OR AND
%nonassoc EQUAL_EQUAL
%left "+" "-"
%left BINARY_STAR
%precedence NOT UNARY_MINUS PREFIX_STAR
// We need to give the `UNARY_STAR` token a precedence, rather than overriding
// the precedence of the `expression UNARY_STAR` rule below, because bison
// compares the precedence of the final token (for a shift) to the precedence
// of the other rule (for a reduce) when attempting to resolve a shift-reduce
// conflict. See https://stackoverflow.com/a/26188429/1041090. When UNARY_STAR
// is the final token of a rule, it must be a postfix usage, so we give it the
// same precedence as POSTFIX_STAR.
%precedence POSTFIX_STAR UNARY_STAR
%left "." ARROW
%precedence "(" ")" "[" "]"

%start input
%locations
%%
input: declaration_list
    { parsed_program = $1; }
;
expression:
  identifier
    { $$ = Carbon::Expression::MakeIdentifierExpression(yylineno, $1); }
| expression designator
    { $$ = Carbon::Expression::MakeFieldAccessExpression(yylineno, $1, $2); }
| expression "[" expression "]"
    { $$ = Carbon::Expression::MakeIndexExpression(yylineno, $1, $3); }
| integer_literal
    { $$ = Carbon::Expression::MakeIntLiteral(yylineno, $1); }
| TRUE
    { $$ = Carbon::Expression::MakeBoolLiteral(yylineno, true); }
| FALSE
    { $$ = Carbon::Expression::MakeBoolLiteral(yylineno, false); }
| INT
    { $$ = Carbon::Expression::MakeIntTypeLiteral(yylineno); }
| BOOL
    { $$ = Carbon::Expression::MakeBoolTypeLiteral(yylineno); }
| TYPE
    { $$ = Carbon::Expression::MakeTypeTypeLiteral(yylineno); }
| CONTINUATION_TYPE
    { $$ = Carbon::Expression::MakeContinuationTypeLiteral(yylineno); }
| paren_expression { $$ = $1; }
| expression EQUAL_EQUAL expression
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Eq, {$1, $3}); }
| expression "+" expression
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Add, {$1, $3}); }
| expression "-" expression
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Sub, {$1, $3}); }
| expression BINARY_STAR expression
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Mul, {$1, $3}); }
| expression AND expression
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::And, {$1, $3}); }
| expression OR expression
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Or, {$1, $3}); }
| NOT expression
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Not, {$2}); }
| "-" expression %prec UNARY_MINUS
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Neg, {$2}); }
| PREFIX_STAR expression
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Deref, {$2}); }
| UNARY_STAR expression %prec PREFIX_STAR
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Deref, {$2}); }
| expression tuple
    { $$ = Carbon::Expression::MakeCallExpression(yylineno, $1, $2); }
| expression POSTFIX_STAR
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Ptr, {$1}); }
| expression UNARY_STAR
    { $$ = Carbon::Expression::MakePrimitiveOperatorExpression(
        yylineno, Carbon::Operator::Ptr, {$1}); }
| FNTY tuple return_type
    { $$ = Carbon::Expression::MakeFunctionTypeLiteral(
        yylineno, $2, $3.first, $3.second); }
;
designator: "." identifier { $$ = $2; }
;
paren_expression: paren_expression_base
    { $$ = Carbon::ExpressionFromParenContents(yylineno, $1); }
;
tuple: paren_expression_base
    { $$ = Carbon::TupleExpressionFromParenContents(yylineno, $1); }
;
paren_expression_element:
  expression
    { $$ = {.name = std::nullopt, .term = $1}; }
| designator "=" expression
    { $$ = {.name = $1, .term = $3}; }
;
paren_expression_base:
  "(" ")"
    { $$ = {.elements = {}, .has_trailing_comma = false}; }
| "(" paren_expression_contents ")"
    { $$ = $2; }
| "(" paren_expression_contents "," ")"
    {
      $$ = $2;
      $$.has_trailing_comma = true;
    }
;
paren_expression_contents:
  paren_expression_element
    { $$ = {.elements = {$1}, .has_trailing_comma = false}; }
| paren_expression_contents "," paren_expression_element
    {
      $$ = $1;
      $$.elements.push_back($3);
    }
;

// In many cases, using `pattern` recursively will result in ambiguities.
// When that happens, it's necessary to factor out two separate productions,
// one for when the sub-pattern is an expression, and one for when it is not.
// To facilitate this, non-terminals besides `pattern` whose names contain
// `pattern` are structured to be disjoint from `expression`, unless otherwise
// specified.
pattern:
  non_expression_pattern
    { $$ = $1; }
| expression
    { $$ = Carbon::global_arena->New<Carbon::ExpressionPattern>($1); }
;
non_expression_pattern:
  AUTO
    { $$ = Carbon::global_arena->New<Carbon::AutoPattern>(yylineno); }
| binding_lhs ":" pattern
    { $$ = Carbon::global_arena->New<Carbon::BindingPattern>(yylineno, $1, $3); }
| paren_pattern
    { $$ = $1; }
| expression tuple_pattern
    { $$ = Carbon::global_arena->New<Carbon::AlternativePattern>(yylineno, $1, $2); }
;
binding_lhs:
  identifier { $$ = $1; }
| UNDERSCORE { $$ = std::nullopt; }
;
paren_pattern: paren_pattern_base
    { $$ = Carbon::PatternFromParenContents(yylineno, $1); }
;
paren_pattern_base:
  "(" paren_pattern_contents ")"
    { $$ = $2; }
| "(" paren_pattern_contents "," ")"
    {
      $$ = $2;
      $$.has_trailing_comma = true;
    }
;
// paren_pattern is analogous to paren_expression, but in order to avoid
// ambiguities, it must be disjoint from paren_expression, meaning it must
// contain at least one non_expression_pattern. The structure of this rule
// is very different from the corresponding expression rule because is has to
// enforce that requirement.
paren_pattern_contents:
  paren_pattern_element
    { $$ = {.elements = {$1}, .has_trailing_comma = false }; }
| paren_expression_contents "," paren_pattern_element
    {
      $$ = Carbon::ParenExpressionToParenPattern($1);
      $$.elements.push_back($3);
    }
| paren_pattern_contents "," paren_expression_element
    {
      $$ = $1;
      $$.elements.push_back({.name = $3.name, .term = Carbon::global_arena->New<Carbon::ExpressionPattern>($3.term)});
    }
| paren_pattern_contents "," paren_pattern_element
    {
      $$ = $1;
      $$.elements.push_back($3);
    }
;
paren_pattern_element:
  non_expression_pattern
    { $$ = {.name = std::nullopt, .term = $1}; }
| designator "=" non_expression_pattern
    { $$ = {.name = $1, .term = $3}; }
;
tuple_pattern: paren_pattern_base
    { $$ = Carbon::TuplePatternFromParenContents(yylineno, $1); }
;
// Unlike most `pattern` nonterminals, this one overlaps with `expression`,
// so it should be used only when prior context (such as an introducer)
// rules out the possibility of an `expression` at this point.
maybe_empty_tuple_pattern:
  "(" ")"
    { $$ = Carbon::global_arena->New<Carbon::TuplePattern>(yylineno, std::vector<Carbon::TuplePattern::Field>()); }
| tuple_pattern
    { $$ = $1; }
;
clause:
  CASE pattern DBLARROW statement
    { $$ = Carbon::global_arena->New<std::pair<const Carbon::Pattern*, const Carbon::Statement*>>($2, $4); }
| DEFAULT DBLARROW statement
    {
      auto vp = Carbon::global_arena->New<Carbon::BindingPattern>(
          yylineno, std::nullopt, Carbon::global_arena->New<Carbon::AutoPattern>(yylineno));
      $$ = Carbon::global_arena->New<std::pair<const Carbon::Pattern*, const Carbon::Statement*>>(vp, $3);
    }
;
clause_list:
  // Empty
    {
      $$ = Carbon::global_arena->New<std::list<
          std::pair<const Carbon::Pattern*, const Carbon::Statement*>>>();
    }
| clause clause_list
    { $$ = $2; $$->push_front(*$1); }
;
statement:
  expression "=" expression ";"
    { $$ = Carbon::Statement::MakeAssign(yylineno, $1, $3); }
| VAR pattern "=" expression ";"
    { $$ = Carbon::Statement::MakeVariableDefinition(yylineno, $2, $4); }
| expression ";"
    { $$ = Carbon::Statement::MakeExpressionStatement(yylineno, $1); }
| if_statement
    { $$ = $1; }
| WHILE "(" expression ")" block
    { $$ = Carbon::Statement::MakeWhile(yylineno, $3, $5); }
| BREAK ";"
    { $$ = Carbon::Statement::MakeBreak(yylineno); }
| CONTINUE ";"
    { $$ = Carbon::Statement::MakeContinue(yylineno); }
| RETURN return_expression ";"
    { $$ = Carbon::Statement::MakeReturn(yylineno, $2.first, $2.second); }
| block
    { $$ = $1; }
| MATCH "(" expression ")" "{" clause_list "}"
    { $$ = Carbon::Statement::MakeMatch(yylineno, $3, $6); }
| CONTINUATION identifier statement
    { $$ = Carbon::Statement::MakeContinuation(yylineno, $2, $3); }
| RUN expression ";"
    { $$ = Carbon::Statement::MakeRun(yylineno, $2); }
| AWAIT ";"
    { $$ = Carbon::Statement::MakeAwait(yylineno); }
;
if_statement:
  IF "(" expression ")" block optional_else
    { $$ = Carbon::Statement::MakeIf(yylineno, $3, $5, $6); }
;
optional_else:
  // Empty
    { $$ = 0; }
| ELSE if_statement
    { $$ = $2; }
| ELSE block
    { $$ = $2; }
;
return_expression:
  // Empty
    { $$ = {Carbon::Expression::MakeTupleLiteral(yylineno, {}), true}; }
| expression
    { $$ = {$1, false}; }
;
statement_list:
  // Empty
    { $$ = 0; }
| statement statement_list
    { $$ = Carbon::Statement::MakeSequence(yylineno, $1, $2); }
;
block:
  "{" statement_list "}"
    { $$ = Carbon::Statement::MakeBlock(yylineno, $2); }
;
return_type:
  // Empty
    { $$ = {Carbon::Expression::MakeTupleLiteral(yylineno, {}), true}; }
| ARROW expression %prec FNARROW
    { $$ = {$2, false}; }
;
generic_binding:
  identifier ":!" expression
    {
      $$ = Carbon::GenericBinding({.name = std::move($1), .type = $3});
    }
;
deduced_param_list:
  // Empty
    { $$ = std::vector<Carbon::GenericBinding>(); }
| generic_binding
    {
      $$ = std::vector<Carbon::GenericBinding>();
      $$.push_back($1);
    }
| generic_binding "," deduced_param_list
    {
      $$ = $3;
      $$.push_back($1);
    }
;
deduced_params:
  // Empty
    { $$ = std::vector<Carbon::GenericBinding>(); }
| "[" deduced_param_list "]"
    { $$ = $2; }
;
function_definition:
  FN identifier deduced_params maybe_empty_tuple_pattern return_type block
    {
      $$ = Carbon::FunctionDefinition(
          yylineno, $2, $3, $4,
          Carbon::global_arena->New<Carbon::ExpressionPattern>($5.first),
          $5.second, $6);
    }
| FN identifier deduced_params maybe_empty_tuple_pattern DBLARROW expression ";"
    {
      $$ = Carbon::FunctionDefinition(
          yylineno, $2, $3, $4,
          Carbon::global_arena->New<Carbon::AutoPattern>(yylineno), true,
          Carbon::Statement::MakeReturn(yylineno, $6, false));
    }
;
function_declaration:
  FN identifier deduced_params maybe_empty_tuple_pattern return_type ";"
    {
      $$ = Carbon::FunctionDefinition(
          yylineno, $2, $3, $4,
          Carbon::global_arena->New<Carbon::ExpressionPattern>($5.first),
          $5.second, nullptr); }
;
variable_declaration: identifier ":" pattern
    { $$ = Carbon::global_arena->New<Carbon::BindingPattern>(yylineno, $1, $3); }
;
member: VAR variable_declaration ";"
    { $$ = Carbon::Member::MakeFieldMember(yylineno, $2); }
;
member_list:
  // Empty
    { $$ = std::list<Carbon::Member*>(); }
| member member_list
    { $$ = $2; $$.push_front($1); }
;
alternative:
  identifier tuple
    { $$ = std::pair<std::string, const Carbon::Expression*>($1, $2); }
| identifier
    {
      $$ = std::pair<std::string, const Carbon::Expression*>(
          $1, Carbon::Expression::MakeTupleLiteral(yylineno, {}));
    }
;
alternative_list:
  // Empty
    { $$ = std::list<std::pair<std::string, const Carbon::Expression*>>(); }
| alternative
    {
      $$ = std::list<std::pair<std::string, const Carbon::Expression*>>();
      $$.push_front($1);
    }
| alternative "," alternative_list
    { $$ = std::move($3); $$.push_front($1); }
;
declaration:
  function_definition
    { $$ = Carbon::Declaration::MakeFunctionDeclaration(std::move($1)); }
| function_declaration
    { $$ = Carbon::Declaration::MakeFunctionDeclaration(std::move($1)); }
| STRUCT identifier "{" member_list "}"
    {
      $$ = Carbon::Declaration::MakeStructDeclaration(yylineno, $2, $4);
    }
| CHOICE identifier "{" alternative_list "}"
    {
      $$ = Carbon::Declaration::MakeChoiceDeclaration(yylineno, $2, $4);
    }
| VAR variable_declaration "=" expression ";"
    {
      $$ = Carbon::Declaration::MakeVariableDeclaration(yylineno, $2, $4);
    }
;
declaration_list:
  // Empty
    { $$ = std::list<Carbon::Declaration>(); }
| declaration declaration_list
    {
      $$ = $2;
      $$.push_front($1);
    }
;
%%