shelljs/src/uniq.js

var common = require('./common');
var fs = require('fs');

// add c spaces to the left of str
function lpad(c, str) {
  var res = '' + str;
  if (res.length < c) {
    res = Array((c - res.length) + 1).join(' ') + res;
  }
  return res;
}

common.register('uniq', _uniq, {
  canReceivePipe: true,
  cmdOptions: {
    'i': 'ignoreCase',
    'c': 'count',
    'd': 'duplicates',
  },
});

//@
//@ ### uniq([options,] [input, [output]])
//@ Available options:
//@
//@ + `-i`: Ignore case while comparing
//@ + `-c`: Prefix lines by the number of occurrences
//@ + `-d`: Only print duplicate lines, one for each group of identical lines
//@
//@ Examples:
//@
//@ ```javascript
//@ uniq('foo.txt');
//@ uniq('-i', 'foo.txt');
//@ uniq('-cd', 'foo.txt', 'bar.txt');
//@ ```
//@
//@ Filter adjacent matching lines from input
function _uniq(options, input, output) {
  // Check if this is coming from a pipe
  var pipe = common.readFromPipe();

  if (!pipe) {
    if (!input) common.error('no input given');

    if (!fs.existsSync(input)) {
      common.error(input + ': No such file or directory');
    } else if (common.statFollowLinks(input).isDirectory()) {
      common.error("error reading '" + input + "'");
    }
  }
  if (output && fs.existsSync(output) && common.statFollowLinks(output).isDirectory()) {
    common.error(output + ': Is a directory');
  }

  var lines = (input ? fs.readFileSync(input, 'utf8') : pipe).
              trimRight().
              split('\n');

  var compare = function (a, b) {
    return options.ignoreCase ?
           a.toLocaleLowerCase().localeCompare(b.toLocaleLowerCase()) :
           a.localeCompare(b);
  };
  var uniqed = lines.reduceRight(function (res, e) {
    // Perform uniq -c on the input
    if (res.length === 0) {
      return [{ count: 1, ln: e }];
    } else if (compare(res[0].ln, e) === 0) {
      return [{ count: res[0].count + 1, ln: e }].concat(res.slice(1));
    } else {
      return [{ count: 1, ln: e }].concat(res);
    }
  }, []).filter(function (obj) {
                 // Do we want only duplicated objects?
    return options.duplicates ? obj.count > 1 : true;
  }).map(function (obj) {
                 // Are we tracking the counts of each line?
    return (options.count ? (lpad(7, obj.count) + ' ') : '') + obj.ln;
  }).join('\n') + '\n';

  if (output) {
    (new common.ShellString(uniqed)).to(output);
    // if uniq writes to output, nothing is passed to the next command in the pipeline (if any)
    return '';
  } else {
    return uniqed;
  }
}

module.exports = _uniq;