prototype-3x5/src/parser.ts

import { escapeHtml } from "./helpers";
import {
  AtLeast,
  Choose,
  End,
  Pattern,
  Peek,
  Regex,
  Sequence,
  Use,
} from "./peg";
import {
  InterpolatedPiece,
  Script,
  ScriptPiece,
  SimplifyWord,
  TextWord,
  Word as WordType,
} from "./words";

const Comment = Regex(/#[^\n]*/y)
  .expects("#")
  .map(() => []);

const PreWordWhitespace = Regex(/[^\S\n;]+/y).expects("whitespace");

const BackslashEscape = Regex(/\\(.)/y)
  .expects("\\")
  .map(([, char]) => ({ text: char }));

const BARE_WORD_CHAR = /[^\s\\;\[]+/y;
const BARE_BRACKET_WORD_CHAR = /[^\s\\;\[\]]+/y;

let BracketScript: Pattern<Script>;
const Bracket: Pattern<ScriptPiece> = Sequence(
  Regex(/\[/y).expects("["),
  Use(() => BracketScript)
)
  .expects("[")
  .map(([, script]) => ({ script }));

function bareWordTmpl(charRegex: RegExp) {
  return Sequence(
    Regex(/(?!["\{])/y),
    AtLeast(
      1,
      Choose<InterpolatedPiece>(
        BackslashEscape,
        Bracket,
        Regex(charRegex)
          .expects("CHAR")
          .map(([text]) => ({ enchanted: text }))
      )
    )
  ).map(([, pieces]) => SimplifyWord(pieces));
}

const QuotedWord = Sequence(
  Regex(/"/y).expects('"'),
  AtLeast(
    0,
    Choose<InterpolatedPiece>(
      BackslashEscape,
      Bracket,
      Regex(/[^"\\\[]+/y)
        .expects("CHAR")
        .map(([text]) => ({ text }))
    )
  ),
  Regex(/"/y).expects('"')
).map(([, pieces]) => SimplifyWord(pieces));

const Brace: Pattern<string> = Sequence(
  Regex(/\{/y).expects("{"),
  AtLeast(
    0,
    Choose(
      Use(() => Brace)
        .expects("{")
        .map((text) => `{${text}}`),
      Regex(/\\./y)
        .expects("\\")
        .map(([escape]) => escape),
      Regex(/[^\\{}]+/y)
        .expects("text")
        .map(([text]) => text)
    )
  ),
  Regex(/\}/y).expects("}")
).map(([, fragments]) => fragments.join(""));

function wordTmpl(bareWordCharRegex: RegExp): Pattern<WordType> {
  return Choose<WordType>(
    Brace.map((text) => ({ text } as TextWord)),
    QuotedWord,
    bareWordTmpl(bareWordCharRegex)
  );
}

const CommandTerminator = Regex(/[\n;]/y)
  .expects("NEWLINE | ;")
  .map(() => true);

function commandTmpl(bareWordCharRegex: RegExp) {
  const word = wordTmpl(bareWordCharRegex);
  return Sequence(
    word,
    AtLeast(
      0,
      Sequence(PreWordWhitespace, word).map(([, word]) => word)
    ),
    AtLeast(0, PreWordWhitespace)
  ).map(([word, moreWords]) => [word].concat(moreWords));
}

function scriptTmpl(bareWordCharRegex: RegExp, endPattern: Pattern<unknown>) {
  return Sequence(
    AtLeast(
      0,
      Choose(
        PreWordWhitespace.map(() => []),
        CommandTerminator.map(() => []),
        Sequence(Comment, Choose(CommandTerminator, Peek(endPattern))).map(
          () => []
        ),
        Sequence(
          commandTmpl(bareWordCharRegex),
          Choose(CommandTerminator, Peek(endPattern))
        ).map(([words]) => words)
      )
    ),
    endPattern
  ).map(([commands]) => commands.filter((command) => command.length > 0));
}

const Script = scriptTmpl(BARE_WORD_CHAR, End());
BracketScript = scriptTmpl(BARE_BRACKET_WORD_CHAR, Regex(/\]/y).expects("]"));

const ERROR_CONTEXT = /(?<=([^\n]{0,50}))([^\n]{0,50})/y;

/**
 * Parse out a Notcl script into an easier-to-interpret representation.
 * No script is actually executed yet.
 *
 * @param - code to parse
 * @returns - parsed list of commands, or error message on failure
 */
export function parse(code: string): [true, Script] | [false, string] {
  /* Preprocess */
  // fold line endings
  code = code.replace(/(?<!\\)((\\\\)*)\\\n[ \t]*/g, "$1 ");

  /* Parse */
  const [commands, errorPos, expected] = Script(code, 0);

  if (commands) {
    return [true, commands[0]];
  } else {
    ERROR_CONTEXT.lastIndex = errorPos;
    const [, before, after] = ERROR_CONTEXT.exec(code)!;
    return [
      false,
      `<pre>Error at position ${errorPos}
${escapeHtml(before + "" + after)}
${"-".repeat(before.length)}^

Expected: ${escapeHtml(expected)}</pre>`,
    ];
  }
}