const { createToken: createTokenOrg, Lexer } = require("chevrotain");

// A little mini DSL for easier lexer definition.
const fragments = {};
const f = fragments;

function FRAGMENT(name, def) {
  fragments[name] = typeof def === "string" ? def : def.source;
}

function makePattern(strings, ...args) {
  let combined = "";
  for (let i = 0; i < strings.length; i++) {
    combined += strings[i];
    if (i < args.length) {
      let pattern = args[i];
      // By wrapping in a RegExp (none) capturing group
      // We enabled the safe usage of qualifiers and assertions.
      combined += `(?:${pattern})`;
    }
  }
  return new RegExp(combined);
}

const tokensArray = [];
const tokensDictionary = {};

function createToken(options) {
  const newTokenType = createTokenOrg(options);
  tokensArray.push(newTokenType);
  tokensDictionary[options.name] = newTokenType;
  return newTokenType;
}

FRAGMENT(
  "NameStartChar",
  "(:|[a-zA-Z]|_|\\u2070-\\u218F|\\u2C00-\\u2FEF|\\u3001-\\uD7FF|\\uF900-\\uFDCF|\\uFDF0-\\uFFFD)"
);

FRAGMENT(
  "NameChar",
  makePattern`${f.NameStartChar}|-|\\.|\\d|\\u00B7||[\\u0300-\\u036F]|[\\u203F-\\u2040]`
);
FRAGMENT("Name", makePattern`${f.NameStartChar}(${f.NameChar})*`);

const Comment = createToken({
  name: "Comment",
  pattern: /<!--(.|\r?\n)*?-->/,
  // A Comment may span multiple lines.
  line_breaks: true,
});

const CData = createToken({
  name: "CData",
  pattern: /<!\[CDATA\[(.|\r?\n)*?]]>/,
  line_breaks: true,
});

const DocType = createToken({
  name: "DocType",
  pattern: /<!DOCTYPE/,
  push_mode: "INSIDE",
});

const IgnoredDTD = createToken({
  name: "DTD",
  pattern: /<!.*?>/,
  group: Lexer.SKIPPED,
});

const EntityRef = createToken({
  name: "EntityRef",
  pattern: makePattern`&${f.Name};`,
});

const CharRef = createToken({
  name: "CharRef",
  pattern: /&#\d+;|&#x[a-fA-F0-9]/,
});

const SEA_WS = createToken({
  name: "SEA_WS",
  pattern: /( |\t|\n|\r\n)+/,
});

const XMLDeclOpen = createToken({
  name: "XMLDeclOpen",
  pattern: /<\?xml[ \t\r\n]/,
  push_mode: "INSIDE",
});

const SLASH_OPEN = createToken({
  name: "SLASH_OPEN",
  pattern: /<\//,
  push_mode: "INSIDE",
});

const INVALID_SLASH_OPEN = createToken({
  name: "INVALID_SLASH_OPEN",
  pattern: /<\//,
  categories: [SLASH_OPEN],
});

const PROCESSING_INSTRUCTION = createToken({
  name: "PROCESSING_INSTRUCTION",
  pattern: makePattern`<\\?${f.Name}.*\\?>`,
});

const OPEN = createToken({ name: "OPEN", pattern: /</, push_mode: "INSIDE" });
// Meant to avoid skipping '<' token in a partial sequence of elements.
// Example of the problem this solves:
// <
// <from>john</from>
//  - The second '<' will be skipped because in the mode "INSIDE" '<' is not recognized.
//  - This means the AST will include only a single element instead of two
const INVALID_OPEN_INSIDE = createToken({
  name: "INVALID_OPEN_INSIDE",
  pattern: /</,
  categories: [OPEN],
});

const TEXT = createToken({ name: "TEXT", pattern: /[^<&]+/ });

const CLOSE = createToken({ name: "CLOSE", pattern: />/, pop_mode: true });

const SPECIAL_CLOSE = createToken({
  name: "SPECIAL_CLOSE",
  pattern: /\?>/,
  pop_mode: true,
});

const SLASH_CLOSE = createToken({
  name: "SLASH_CLOSE",
  pattern: /\/>/,
  pop_mode: true,
});

const SLASH = createToken({ name: "SLASH", pattern: /\// });

const STRING = createToken({
  name: "STRING",
  pattern: /"[^<"]*"|'[^<']*'/,
});

const EQUALS = createToken({ name: "EQUALS", pattern: /=/ });

const Name = createToken({ name: "Name", pattern: makePattern`${f.Name}` });

const S = createToken({
  name: "S",
  pattern: /[ \t\r\n]/,
  group: Lexer.SKIPPED,
});

const xmlLexerDefinition = {
  defaultMode: "OUTSIDE",

  modes: {
    OUTSIDE: [
      Comment,
      CData,
      DocType,
      IgnoredDTD,
      EntityRef,
      CharRef,
      SEA_WS,
      XMLDeclOpen,
      SLASH_OPEN,
      PROCESSING_INSTRUCTION,
      OPEN,
      TEXT,
    ],
    INSIDE: [
      // Tokens from `OUTSIDE` to improve error recovery behavior
      Comment,
      INVALID_SLASH_OPEN,
      INVALID_OPEN_INSIDE,
      // "Real" `INSIDE` tokens
      CLOSE,
      SPECIAL_CLOSE,
      SLASH_CLOSE,
      SLASH,
      EQUALS,
      STRING,
      Name,
      S,
    ],
  },
};

const xmlLexer = new Lexer(xmlLexerDefinition, {
  // Reducing the amount of position tracking can provide a small performance boost (<10%)
  // Likely best to keep the full info for better error position reporting and
  // to expose "fuller" ITokens from the Lexer.
  positionTracking: "full",
  ensureOptimizations: false,

  // TODO: inspect definitions for XML line terminators
  lineTerminatorCharacters: ["\n"],
  lineTerminatorsPattern: /\n|\r\n/g,
});

module.exports = {
  xmlLexer,
  tokensDictionary,
};