str/src/vm/parser.ts

import {BehaviorSubject} from '../util/subject.js'
import {Lexer, LexInput, LexToken} from './lexer.js'
import {StreamLogger} from '../util/log.js'
import {log} from '../log.js'
import {Commands} from './commands/index.js'
import {Command, CommandData, ParseContext} from './commands/command.js'
import {Executable, InternalParseError, InvalidCommandError, IsNotKeywordError} from './parse.js'

export class Parser extends BehaviorSubject<Executable<CommandData>> {
    private logger: StreamLogger

    private parseCandidate?: Command<CommandData>
    private inputForCandidate: LexInput[] = []

    constructor(lexer: Lexer, private commands: Commands) {
        super()
        this.logger = log.getStreamLogger('parser')
        lexer.subscribe(token => this.handleToken(token))
    }

    async handleToken(token: LexToken) {
        // We are in between full commands, so try to identify a new parse candidate:
        if ( !this.parseCandidate ) {
            // Ignore duplicated terminators between commands
            if ( token.type === 'terminator' ) {
                return
            }

            this.logger.verbose({ identifyParseCandidate: token })
            if ( !this.isKeyword(token) ) {
                throw new IsNotKeywordError('Expected keyword, found: ' + this.displayToken(token))
            }

            this.parseCandidate = this.getParseCandidate(token)
            return
        }

        // We have already identified a parse candidate:
        // If this is normal input token, collect it so we can give it to the candidate to parse:
        if ( token.type === 'input' ) {
            this.inputForCandidate.push(token)
            return
        }

        // If we got a terminator, then ask the candidate to actually perform its parse:
        if ( token.type === 'terminator' ) {
            try {
                // Have the candidate attempt to parse itself from the collecte data:
                const context = new ParseContext(this.inputForCandidate)
                this.logger.verbose({ parsing: this.parseCandidate.getDisplayName(), context })
                const data = this.parseCandidate.attemptParse(context)

                // The candidate must consume every token in the context:
                context.assertEmpty()

                // Emit the parsed command:
                this.logger.debug({ parsed: this.parseCandidate.getDisplayName() })
                await this.next({
                    command: this.parseCandidate,
                    data,
                })
                return
            } finally {
                this.parseCandidate = undefined
                this.inputForCandidate = []
            }
        }

        throw new InternalParseError('Encountered invalid token.')
    }

    private isKeyword(token: LexToken): token is (LexInput & {literal: undefined}) {
        return token.type === 'input' && !token.literal
    }

    private getParseCandidate(token: LexInput): Command<CommandData> {
        for ( const command of this.commands ) {
            if ( command.isParseCandidate(token) ) {
                this.logger.debug({ foundParseCandidate: command.getDisplayName(), token })
                return command
            }
        }

        throw new InvalidCommandError('Could not find parser for: ' + this.displayToken(token))
    }

    private displayToken(token: LexToken) {
        if ( token.type === 'terminator' ) {
            return '(TERMINATOR)'
        }

        return `(${token.literal ? 'LITERAL' : 'INPUT'}) ${token.value}`
    }
}