sql/src/lexer.ts

import Error, { isError } from "./error";
import Token, { TokenKind } from "./token";

export default class Lexer {
  public source: string;
  public position: number;
  public line: number;

  constructor(source: string) {
    this.source = source;
    this.position = 0;
    this.line = 1;
  }
  public scan(): Token[] | Error {
    const tokens = [];

    while (!this.atEnd()) {
      const result = this.getToken();
      if (isError(result)) {
        return result;
      } else if (result) {
        tokens.push(result);
      }
    }

    tokens.push(new Token(TokenKind.EOF, null, this.line));

    return tokens;
  }

  private getToken(): Token | Error | null {
    const source = this.source.slice(this.position);

    if (source.match(/^select/i)) {
      this.advance(6);
      return new Token(TokenKind.SELECT, null, this.line);
    } else if (source.match(/^where/i)) {
      this.advance(5);
      return new Token(TokenKind.WHERE, null, this.line);
    } else if (source.match(/^from/i)) {
      this.advance(4);
      return new Token(TokenKind.FROM, null, this.line);
    } else if (source.match(/^as/i)) {
      this.advance(2);
      return new Token(TokenKind.AS, null, this.line);
    } else if (source.match(/^\*/)) {
      this.advance();
      return new Token(TokenKind.STAR, null, this.line);
    } else if (source.match(/^=/)) {
      this.advance();
      return new Token(TokenKind.EQUALS, null, this.line);
    } else if (source.match(/^,/)) {
      this.advance();
      return new Token(TokenKind.COMMA, null, this.line);
    } else if (source.match(/^`/)) {
      this.advance();
      return new Token(TokenKind.BACKTICK, null, this.line);
    } else if (source.match(/^\.([^0-9]|$)/)) {
      this.advance();
      return new Token(TokenKind.DOT, null, this.line);
    } else if (source.match(/^;/)) {
      this.advance();
      return new Token(TokenKind.SEMICOLON, null, this.line);
    } else if (source.match(/^[0-9]+(\.[0-9]+)?/)) {
      const match = source.match(/^[0-9]+(\.[0-9]+)?/);
      if (match) {
        const numeric = match[0];
        this.advance(numeric.length);
        return new Token(TokenKind.NUMBER, numeric, this.line);
      }
    } else if (source.match(/^\.[0-9]+/)) {
      const match = source.match(/^\.[0-9]+/);
      if (match) {
        const numeric = match[0];
        this.advance(numeric.length);
        return new Token(TokenKind.NUMBER, numeric, this.line);
      }
    } else if (source.match(/^[a-zA-Z_][a-zA-Z0-9_]*/)) {
      const match = source.match(/^[a-zA-Z_][a-zA-Z0-9_]*/);
      if (match) {
        const identifier = match[0];
        this.advance(identifier.length);
        return new Token(TokenKind.IDENTIFIER, identifier, this.line);
      }
    } else if (source.match(/^\n/)) {
      this.advance();
      this.nextLine();
      return null;
    } else if (source.match(/^\s/)) {
      this.advance();
      return null;
    }

    return new Error(`Unrecognized character ${source[0]}`, this.line);
  }

  private advance(step: number = 1) {
    this.position += step;
  }

  private nextLine() {
    this.line += 1;
  }

  private atEnd(): boolean {
    return this.position === this.source.length;
  }
}