muesli/parser.go

package muesli

import (
	_ "bufio"
	_ "bytes"
	_ "errors"
	_ "fmt"
	_ "io"
	_ "os"
	_ "reflect"
	_ "runtime"
	_ "strings"
	_ "unicode"
	// "gitlab.com/beoran/woe/graphviz"
	// _ "gitlab.com/beoran/woe/monolog"
)

/* Grammar:

Desrired syntax (verified LL(1) on smlweb.cpsc.ucalgary.ca)

PROGRAM -> STATEMENTS.
STATEMENTS -> STATEMENT STATEMENTS | .
STATEMENT -> EXPRESSION eos | BLOCK .
WORDOPS -> WORDOP WORDOPS | .
EXPRESSION -> SETTER | GETTER | COMMAND.
COMMAND -> WORDVALUE PARAMETERS.
PARAMETERS -> PARAMETER PARAMETERS | .
PARAMETER -> WORDVALUE | PARENTHESIS | GETTER | ARRARY | BLOCK .
PARENTHESIS -> '(' EXPRESSION ')' .
BLOCK -> '{' STATEMENTS '}' .
WORDVALUE -> word | VALUE.
VALUE -> string | integer | float | symbol.
SETTER -> set word PARAMETERS .
GETTER -> get word .

 *
 * program -> statements
 * statements -> statement+
 * statement -> get / set / command
 *
*/

type Parser struct {
	Lexer   *Lexer
	current *Token
	LoggerWrapper
}

func (parser *Parser) SetLogger(logger Logger) {
	parser.LoggerWrapper = LoggerWrapper{logger}
}

func (parser *Parser) Advance() {
	token := parser.Lexer.Lex()
	parser.current = &token
	parser.LogDebug("Next token: %s\n", token.String())
}

/* Looks at the current token and advances the lexer if the token is of any of
the token kinds given in kinds. In this case it will return the accepted
token and advance the parser. Otherwise, if no token kind matches, the lexer
does not advance and the current token remains the same, except if that was nil.*/
func (parser *Parser) Accept(kinds ...TokenKind) *Token {
	if parser.current == nil {
		parser.Advance()
	}

	for _, kind := range kinds {
		if kind == parser.current.TokenKind {
			accepted := parser.current
			parser.Advance()
			return accepted
		}
	}
	return nil
}

func (parser *Parser) ParseMany(astkind AstKind, parsefunc func(*Parser) *Ast) *Ast {
	ast := NewAst(astkind, nil, nil)
	for sub := parsefunc(parser); sub != nil && sub.AstKind != AstKindError; sub = parsefunc(parser) {
		ast.AppendChild(sub)
	}
	return ast
}

func (parser *Parser) NewAstError(message string) *Ast {
	sv := StringValue(message + " at token " + parser.current.String())
	pos := parser.current.Position
	tok := NewToken(TokenKindError, sv, pos)
	return NewAst(AstKindError, nil, &tok)
}

func (parser *Parser) ParseAny(astkind AstKind, parsefuncs ...(func(*Parser) *Ast)) *Ast {
	ast := NewAst(astkind, nil, nil)
	for _, parsefunc := range parsefuncs {
		sub := parsefunc(parser)
		if sub != nil {
			ast.AppendChild(sub)
			return ast
		}
	}
	return nil
}

func (parser *Parser) ParseRequireAny(astkind AstKind, parsefuncs ...(func(*Parser) *Ast)) *Ast {
	ast := parser.ParseAny(astkind, parsefuncs...)
	if ast == nil {
		err := parser.NewAstError("Unexpected token")
		return err
	}
	return ast
}

func (parser *Parser) ParseValue() *Ast {
	value := parser.Accept(TokenKindInteger, TokenKindString,
		TokenKindBoolean, TokenKindFloat, TokenKindSymbol)
	if value == nil {
		return nil
	}
	return NewAst(AstKindValue, nil, value)
}

func (parser *Parser) ParseWordValue() *Ast {
	value := parser.Accept(TokenKindInteger, TokenKindString,
		TokenKindBoolean, TokenKindFloat, TokenKindSymbol, TokenKindWord)
	if value == nil {
		return nil
	}
	return NewAst(AstKindWordValue, nil, value)
}

func (parser *Parser) ParseArgument() *Ast {
	return parser.ParseAny(AstKindArgument,
		(*Parser).ParseWordValue,
		(*Parser).ParseGet,
		(*Parser).ParseSet,
		(*Parser).ParseParenthesis,
		(*Parser).ParseList,
		(*Parser).ParseBlock)
}

func (parser *Parser) ParseArguments() *Ast {
	return parser.ParseMany(AstKindArguments, (*Parser).ParseArgument)
}

func (parser *Parser) ParseList() *Ast {
	op := parser.Accept(TokenKindOpenList)
	if op == nil {
		return nil
	}

	ast := NewAst(AstKindList, nil, op)
	args := parser.ParseArguments()
	if args.AstKind == AstKindError {
		return args
	}
	if cp := parser.Accept(TokenKindCloseList); cp == nil {
		return parser.NewAstError("expected closing brackets")
	}

	ast.AppendChild(args)
	return ast
}

func (parser *Parser) ParseParenthesis() *Ast {
	op := parser.Accept(TokenKindOpenParen)

	if op == nil {
		return nil
	}

	ast := NewAst(AstKindParenthesis, nil, op)
	expr := parser.ParseExpression()
	if expr == nil {
		return parser.NewAstError("expected expression")
	}
	if expr.AstKind == AstKindError {
		return expr
	}
	if cp := parser.Accept(TokenKindCloseParen); cp == nil {
		return parser.NewAstError("expected closing parenthesis")
	}

	ast.AppendChild(expr)
	return ast
}

func (parser *Parser) ParseBlock() *Ast {
	op := parser.Accept(TokenKindOpenBlock)
	if op == nil {
		return nil
	}

	ast := NewAst(AstKindBlock, nil, op)
	stats := parser.ParseStatements()
	if stats == nil {
		return parser.NewAstError("expected expression")
	}
	if stats.AstKind == AstKindError {
		return stats
	}
	if cp := parser.Accept(TokenKindCloseBlock); cp == nil {
		return parser.NewAstError("expected closing block")
	}

	ast.AppendChild(stats)
	return ast
}

/* Parses the target of a set or get expression */
func (parser *Parser) ParseTarget() *Ast {
	target := parser.Accept(TokenKindWord, TokenKindSymbol)
	ast := NewAst(AstKindTarget, nil, target)

	if target == nil {
		paren := parser.ParseParenthesis()
		if paren == nil {
			return parser.NewAstError("expected word, symbol or parenthesis")
		}
		ast.AppendChild(paren)
	}

	return ast
}

func (parser *Parser) ParseSet() *Ast {
	set := parser.Accept(TokenKindSet)
	if set == nil {
		return nil
	}

	ast := NewAst(AstKindGet, nil, set)
	target := parser.ParseTarget()
	ast.AppendChild(target)

	argument := parser.ParseArgument()
	if argument == nil {
		return parser.NewAstError("Expected argument to set")
	}
	ast.AppendChild(argument)
	return ast
}

func (parser *Parser) ParseGet() *Ast {
	get := parser.Accept(TokenKindGet)
	if get == nil {
		return nil
	}
	ast := NewAst(AstKindGet, nil, get)
	target := parser.ParseTarget()
	ast.AppendChild(target)
	return ast
}

func (parser *Parser) ParseCommand() *Ast {
	word := parser.Accept(TokenKindWord)
	if word == nil {
		return nil
	}
	arguments := parser.ParseArguments()
	command := NewAst(AstKindCommand, nil, word)
	command.AppendChild(arguments)
	return command
}

func (parser *Parser) ParseExpression() *Ast {
	exp := parser.ParseRequireAny(AstKindExpression, (*Parser).ParseSet,
		(*Parser).ParseGet, (*Parser).ParseCommand)
	return exp
}

func (parser *Parser) ParseEmptyStatement() *Ast {
	eox := parser.Accept(TokenKindEOX)
	if eox == nil {
		return nil
	}
	ast := NewAst(AstKindEnd, nil, eox)
	return ast
}

func (parser *Parser) ParseStatement() *Ast {
	ast := parser.ParseAny(AstKindStatement,
		(*Parser).ParseBlock,
		(*Parser).ParseSet,
		(*Parser).ParseGet,
		(*Parser).ParseCommand,
		(*Parser).ParseParenthesis)
	if eox := parser.Accept(TokenKindEOX); eox == nil {
		return parser.NewAstError("expected end of statement")
	}
	return ast
}

func (parser *Parser) ParseStatements() *Ast {
	return parser.ParseMany(AstKindStatements, (*Parser).ParseStatement)
}

func (parser *Parser) ParseProgram() *Ast {
	ast := NewAst(AstKindProgram, nil, nil)
	stats := parser.ParseStatements()
	eof := parser.Accept(TokenKindEOF)
	aeof := NewAst(AstKindEnd, nil, eof)
	// be lenient with missing EOF for now...
	ast.AppendChild(stats)
	ast.AppendChild(aeof)
	return ast
}

func (parser *Parser) Parse() *Ast {
	ast := parser.ParseProgram()
	return ast
}

func NewParser(lexer *Lexer) *Parser {
	parser := &Parser{lexer, nil, LoggerWrapper{nil}}
	return parser
}

func NewParserFromString(input string) *Parser {
	lexer := NewLexerFromString(input)
	return NewParser(lexer)
}

func NewParserFromFilename(filename string) (*Parser, error) {
	lexer, err := NewLexerFromFileName(filename)
	if err != nil {
		return nil, err
	}
	return NewParser(lexer), nil
}