path: root/main/parse.go

package main

import (
	"strings"
	"strconv"
	"fmt"
	"main/walk"
	"main/subex"
)

type parser struct {
	tokenStream chan Token
	rewinds []Token
}
func (p *parser) next() Token {
	var token Token
	if len(p.rewinds) == 0 {
		token = <- p.tokenStream
	} else {
		token = p.rewinds[len(p.rewinds)-1]
		p.rewinds = p.rewinds[:len(p.rewinds)-1]
	}
	if token.typ == TokenErr {
		fmt.Println(token)
		panic("Lexing error")
	}
	return token
}
func (p *parser) rewind(token Token) {
	p.rewinds = append(p.rewinds, token)
}
func (p *parser) peek() Token {
	token := p.next()
	p.rewind(token)
	return token
}

var segmentTokens map[TokenType]bool = map[TokenType]bool {
	TokenHash: true,
	TokenAt: true,
	TokenDot: true,
	TokenLBrack: true,
}

func (p *parser) parsePathPatternFilter(minPower int) PathFilterAST {
	var lhs PathFilterAST
	token := p.next()
	switch token.typ {
		case TokenHash:
			stringIndex := p.next()
			if stringIndex.typ != TokenPatternStringIndex {
				panic("Expected string index after # in pattern")
			}
			lhs = StringSegmentPathFilterAST{stringIndex.val}
		case TokenAt:
			intIndex := p.next()
			if intIndex.typ != TokenPatternIntegerIndex {
				panic("Expected integer index after @ in pattern")
			}
			index, err := strconv.Atoi(intIndex.val)
			if err != nil {
				panic("Expected integer index after @ in pattern")
			}
			lhs = IntegerSegmentPathFilterAST{index}
		case TokenDot:
			lhs = AnySegmentPathFilterAST{}
		case TokenLBrack:
			lhs = p.parsePathPatternFilter(0)
			if p.next().typ != TokenRBrack {
				panic("Expected ] in path filter")
			}
		default:
			panic("Expected path pattern filter segment")
	}
	loop: for {
		token = p.next()
		switch {
			case token.typ == TokenAst && 10 >= minPower:
				lhs = RepeatPathFilterAST {lhs}
			case token.typ == TokenQuestion && 10 >= minPower:
				lhs = OrPathFilterAST{lhs, NonePathFilterAST{}}
			case token.typ == TokenBar && 0 >= minPower:
				lhs = OrPathFilterAST{lhs, p.parsePathPatternFilter(1)}
			case segmentTokens[token.typ] && 2 >= minPower:
				p.rewind(token)
				lhs = SequencePathFilterAST {lhs, p.parsePathPatternFilter(3)}
			default:
				p.rewind(token)
				break loop
		}
	}
	return lhs
}

func (p *parser) parseFilter(minPower int) Filter {
	var lhs Filter
	token := p.next()
	switch token.typ {
		case TokenHash, TokenAt, TokenDot, TokenLBrack:
			p.rewind(token)
			filterAst := p.parsePathPatternFilter(0)
			lhs = compilePathFilterAST(filterAst)
		case TokenHat:
			lhs = BeginTerminalFilter{}
		case TokenDollar:
			lhs = EndTerminalFilter{}
		case TokenHatDollar:
			lhs = TerminalFilter{}
		case TokenTilde:
			lhs = RootFilter{}
		case TokenLParen:
			lhs = p.parseFilter(0)
			rParen := p.next()
			if rParen.typ != TokenRParen {
				panic("Missing ) in filter")
			}
		default:
			panic("Expected filter")
	}
	loop: for {
		token = p.next()
		switch {
			case token.typ == TokenAnd && 2 >= minPower:
				lhs = AndFilter {lhs, p.parseFilter(3)}
			case token.typ == TokenOr && 0 >= minPower:
				lhs = OrFilter {lhs, p.parseFilter(1)}
			default:
				p.rewind(token)
				break loop
		}
	}
	return lhs
}

func (p *parser) parseLiterals() (items []walk.WalkItem) {
	var path walk.Path
	var value walk.WalkValue
	loop: for {
		token := p.next()
		switch token.typ {
			case TokenSemicolon, TokenEOF:
				p.rewind(token)
				break loop
			case TokenComma:
			case TokenNullLiteral:
				value = walk.ValueNull{}
			case TokenTrueLiteral:
				value = walk.ValueBool(true)
			case TokenFalseLiteral:
				value = walk.ValueBool(false)
			case TokenNumberLiteral:
				numberLiteral, err := strconv.ParseFloat(token.val, 64)
				if err != nil {
					panic("Error parsing number literal to float64")
				}
				value = walk.ValueNumber(numberLiteral)
			case TokenDoubleQuote:
				stringToken := p.next()
				if stringToken.typ != TokenStringLiteral {
					panic("Expected string literal after \"")
				}
				// TODO: resolve escape characters
				stringLiteral := stringToken.val
				if p.next().typ != TokenDoubleQuote {
					panic("Expected \" after string literal")
				}
				colon := p.next()
				if colon.typ == TokenColon {
					if path != nil {
						panic("Expected value after path:")
					}
					path = walk.Path{stringLiteral}
				} else {
					p.rewind(colon)
					value = walk.ValueString(stringLiteral)
				}
			case TokenTerminalLiteral:
				switch token.val {
					case "{":
						value = walk.MapBegin
					case "}":
						value = walk.MapEnd
					case "[":
						value = walk.ArrayBegin
					case "]":
						value = walk.ArrayEnd
					default:
						panic("Invalid terminal token")
				}
		}
		if value != nil {
			items = append(items, walk.WalkItem {
				Path: path,
				Value: value,
			})
			path = nil
			value = nil
		}
	}
	if path != nil {
		panic("Expected value after path:")
	}
	return items
}

func (p *parser) parseSubex() subex.SubexState {
	delim := p.next()
	if delim.typ != TokenSubstituteDelimiter {
		panic("Missing substitute delimiter")
	}
	subexProgramToken := p.next()
	if subexProgramToken.typ != TokenSubex {
		panic("Missing subex from substitution")
	}
	reader := subex.NewStringRuneReader(subexProgramToken.val)
	subexAST := subex.Parse(reader)
	subex := subex.CompileTransducer(subexAST)
	delim = p.next()
	if delim.typ != TokenSubstituteDelimiter {
		panic("Missing end substitute delimiter")
	}
	return subex
}

func (p *parser) parseBasicCommand(commandChar rune) Command {
	switch commandChar {
		case 'p':
			return PrintValueCommand{}
		case 'd':
			return DeleteAllCommand{}
		case 'n':
			return NextCommand{}
		case 'N':
			return AppendNextCommand{}
		case 's':
			subex := p.parseSubex()
			var next Command
			token := p.peek()
			switch token.typ {
				case TokenEOF, TokenRBrace:
					next = NoopCommand{}
				default:
					next = p.parseCommand()
			}
			return SubstituteCommand {
				subex: subex,
				next: next,
			}
		case 'i':
			items := p.parseLiterals()
			return PrintLiteralsCommand {items: items}
		default:
			panic("Invalid command")
	}
}

func (p *parser) parseCommand() Command {
	token := p.next()
	switch token.typ {
		case TokenHash, TokenAt, TokenDot, TokenLParen, TokenHat, TokenDollar, TokenHatDollar, TokenTilde:
			p.rewind(token)
			filter := p.parseFilter(0)
			notToken := p.next()
			if notToken.typ == TokenExclamation {
				filter = NotFilter {filter}
			} else {
				p.rewind(notToken)
			}
			command := p.parseCommand()
			command = FilteredCommand {
				filter: filter,
				command: command,
			}
			return command
		case TokenLBrace:
			commands := p.parseCommands()
			if p.next().typ != TokenRBrace {
				panic("Missing matching }")
			}
			return SequenceCommand {commands}
		case TokenCommand:
			commandChar, _, err := strings.NewReader(token.val).ReadRune()
			if err != nil {
				panic("Error reading a command character!?")
			}
			return p.parseBasicCommand(commandChar)
		default:
			panic("Invalid token, expected command")
	}
}

func (p *parser) parseCommands() []Command {
	var commands []Command
	for {
		nextToken := p.peek()
		if nextToken.typ == TokenEOF || nextToken.typ == TokenRBrace {
			return commands
		}
		commands = append(commands, p.parseCommand())
		semicolon := p.next()
		if semicolon.typ == TokenEOF || semicolon.typ == TokenRBrace {
			p.rewind(semicolon)
			return commands
		}
		if semicolon.typ != TokenSemicolon {
			panic("Expected ; after command")
		}
	}
}

func Parse(tokens chan Token) []Command {
	p := parser {
		tokenStream: tokens,
	}
	return p.parseCommands()
}