retag_misnested_dsl.js

/******************************************************************************/
'use strict';
/******************************************************************************/
const [, , fpath] = process.argv;
if (!fpath) {
  console.error('Add the file path, please.');
  process.exit(1);
}
/******************************************************************************/
const fs = require('fs');
const rl = require('readline');

const inCoding = guessEncoding(fpath);
if (!inCoding) process.exit(1);

const outCoding = inCoding;

const toChangeFileNameRE = /(\.[^.]+)?$/;
const bomRE = /^\uFEFF/;

const rli = rl.createInterface({ input: fs.createReadStream(fpath, inCoding) });

const outFileName = fpath.replace(toChangeFileNameRE, '.retagged$1');
const outFile =  fs.openSync(outFileName, 'w');
fs.writeSync(outFile, '\uFEFF', null, outCoding);
/******************************************************************************/
const r = String.raw;

const isDirective = /^#/;
const isHeadword  = /^[^#\s]/;
const isBody      = /^\s/;

let   cardBuf = '';

const startOfString = '^';
const notEscapeSymbol = r`[^\x5c]`;
const escapedEscapeSymbols = r`(?:${startOfString}|${notEscapeSymbol})(?:\x5c{2})+`;
const DSLTag = r`\x5b([^\x5d]+?)(?: ([^\x5d]+))*\x5d`;

const notEscapedDSLTagRE = new RegExp(
  `(?<=${startOfString}|${notEscapeSymbol}|${escapedEscapeSymbols})${DSLTag}`, 'g'
);

const HTMLTagRE = /<([^>]+?)(?: title="([^\x22]+)")*>/g;

/* eslint-disable array-bracket-spacing */
const DSLtoHTMLmap = new Map([
  ['b',    'b'     ],
  ['i',    'i'     ],
  ['u',    'u'     ],
  ['sub',  'small' ],
  ['sup',  'big'   ],
  ['p',    'a'     ],
  ['!trs', 'strike'],
  ['ex',   'tt'    ],
  ['com',  'code'  ],
  ['trn',  'strong'],
  ["'",    'nobr'  ],
  ['*',    's'     ],
  ['c',    'em'    ],
  ['lang', 'font'  ],
]);
/* eslint-enable array-bracket-spacing */

[...DSLtoHTMLmap.entries()].forEach(
  (pair) => { DSLtoHTMLmap.set(`/${pair[0]}`, `/${pair[1]}`); }
);

const HTMLtoDSLmap = new Map(
  [...DSLtoHTMLmap.entries()].map(pair => pair.reverse())
);

let retagged = 0;
/******************************************************************************/
const { JSDOM } = require('jsdom');

const doc = (new JSDOM()).window.document;
doc.body.innerHTML = '';
/******************************************************************************/
console.log('Processing dictionary...\n');
const pbRead = pb(fs.statSync(fpath).size);
pbRead.start();

let lineNumber = 0;
rli.on('line', (line) => {
  pbRead.stat +=  Buffer.byteLength(line, inCoding) + 1;

  if (++lineNumber === 1) {
    line = line.replace(bomRE, '');  // eslint-disable-line no-param-reassign
  }

  if (isDirective.test(line)) {
    fs.writeSync(outFile, `${line}\n`, null, outCoding);
  } else if (isHeadword.test(line)) {
    if (cardBuf) {
      fs.writeSync(outFile, `${dsl2html2dsl(cardBuf)}`, null, outCoding);
      cardBuf = '';
    }
    fs.writeSync(outFile, `${dsl2html2dsl(line)}\n`, null, outCoding);
  } else if (isBody.test(line)) {
    cardBuf += `${line}\n`;
  } else if (!line) {
    if (cardBuf) cardBuf += '\n';
    else fs.writeSync(outFile, '\n', null, outCoding);
  }
}).on('close', () => {
  if (cardBuf) {
    fs.writeSync(outFile, `${dsl2html2dsl(cardBuf)}`, null, outCoding);
  }

  pbRead.end();
  fs.closeSync(outFile);

  fs.renameSync(
    outFileName, outFileName.replace(toChangeFileNameRE, `_${retagged}$1`)
  );
  console.log(`Retagged headwords or cards: ${retagged}.`);
});
/******************************************************************************/
function dsl2html2dsl(str) {
  const [, startSpaces, endSpaces] = str.match(/^(\s*)[^]*?(\s*)$/);

  doc.body.innerHTML = '';
  doc.body.appendChild(doc.createTextNode(str.trim()));
  const escapedStr = doc.body.innerHTML;

  doc.body.innerHTML = escapedStr.replace(notEscapedDSLTagRE, (match, tag, attr) => {
    if (DSLtoHTMLmap.has(tag)) {
      return `<${DSLtoHTMLmap.get(tag)}${attr ? ` title='${attr}'` : ''}>`;
    }
    return match;
  });

  doc.body.innerHTML = doc.body.innerHTML.replace(HTMLTagRE, (match, tag, attr) => {
    if (HTMLtoDSLmap.has(tag)) {
      return `[${HTMLtoDSLmap.get(tag)}${attr ? ` ${attr}`  : ''}]`;
    }
    return match;
  });

  if (str.trim() !== doc.body.textContent) retagged++;

  return `${startSpaces}${doc.body.textContent}${endSpaces}`;
}
/******************************************************************************/
function guessEncoding(path) {
  const BOM_0 = 0xFF;
  const BOM_1 = 0xFE;

  try  {
    const fd = fs.openSync(path, 'r');
    const bf = Buffer.alloc(2);
    fs.readSync(fd, bf, 0, 2, 0);
    fs.closeSync(fd);
    return bf[0] === BOM_0 && bf[1] === BOM_1 ? 'utf16le' : 'utf8';
  } catch (e) {
    console.error(`Error: ${e.message}.`);
    return null;
  }
}
/******************************************************************************/
function pb(edge = 0) {
  const DEFAULT_FREQ = 500;
  const HUNDRED_PERCENT = 100;
  const PB_LENGTH = 50;
  const PB_SCALE = HUNDRED_PERCENT / PB_LENGTH;

  function clearLine() {
    rl.cursorTo(process.stdout, 0);
    rl.clearLine(process.stdout, 0);
  }

  return {
    edge,
    stat: 0,

    start(freq = DEFAULT_FREQ) {
      this.updater = setInterval(() => { this.update(); }, freq);
    },

    update(stat = this.stat) {
      let statPercent = Math.ceil(stat / this.edge * HUNDRED_PERCENT);
      if (statPercent > HUNDRED_PERCENT) statPercent = HUNDRED_PERCENT;

      const barsNumber = Math.floor(statPercent / PB_SCALE);
      const padsNumber = PB_LENGTH - barsNumber;

      clearLine();
      process.stdout.write(
        `${'█'.repeat(barsNumber)}${' '.repeat(padsNumber)} ${statPercent}%`
      );
    },

    end() {
      clearInterval(this.updater);
      this.stat = this.edge;
      this.update();
      console.log('\n');
    },

    clear() {
      clearInterval(this.updater);
      clearLine();
    },
  };
}
/******************************************************************************/

1 . Требуется модуль jsdom версии 10+ и Node.js версии 6+. Установка модуля из командной строки (в папке скрипта или одной из его родительских папок):

npm install jsdom

Модуль основательный, грузится при установке со всеми зависимыми подмодулями ощутимое время. В начале работы скрипта тоже запускается с небольшой задержкой.

2 . Пока lookbehind assertions остаются под флагом, запускать скрипт нужно так (изменив по необходимости пути к скрипту и словарю):

node --harmony_regexp_lookbehind retag_misnested_dsl.js test.dsl

3. Словарь может быть в кодировке UTF-8 или UTF-16 (распознаётся автоматически).

Некоторые особенности реализации скрипта см. здесь.