@a-s8h/liblevenshtein/lib/liblevenshtein/levenshtein/transducer.js

Various utilities regarding Levenshtein transducers.

// Generated by CoffeeScript 1.7.1

/**
 * The algorithm for imitating Levenshtein automata was taken from the
 * following journal article:
 *
 * @ARTICLE{Schulz02faststring,
 *   author = {Klaus Schulz and Stoyan Mihov},
 *   title = {Fast String Correction with Levenshtein-Automata},
 *   journal = {INTERNATIONAL JOURNAL OF DOCUMENT ANALYSIS AND RECOGNITION},
 *   year = {2002},
 *   volume = {5},
 *   pages = {67--85}
 * }
 *
 * As well, this Master Thesis helped me understand its concepts:
 *
 *   www.fmi.uni-sofia.bg/fmi/logic/theses/mitankin-en.pdf
 *
 * The supervisor of the student who submitted the thesis was one of the
 * authors of the journal article, above.
 *
 * The algorithm for constructing a DAWG (Direct Acyclic Word Graph) from the
 * input dictionary of words (DAWGs are otherwise known as an MA-FSA, or
 * Minimal Acyclic Finite-State Automata), was taken and modified from the
 * following blog from Steve Hanov:
 *
 *   http://stevehanov.ca/blog/index.php?id=115
 *
 * The algorithm therein was taken from the following paper:
 *
 * @MISC{Daciuk00incrementalconstruction,
 *   author = {Jan Daciuk and Bruce W. Watson and Richard E. Watson and Stoyan Mihov},
 *   title = {Incremental Construction of Minimal Acyclic Finite-State Automata},
 *   year = {2000}
 * }
 */

(function() {
  var Transducer, global,
    __hasProp = {}.hasOwnProperty;

  Transducer = (function() {
    Transducer.prototype['minimum_distance'] = function(state, w) {
      throw new Error('minimum_distance not specified on construction');
    };

    Transducer.prototype['build_matches'] = function() {
      throw new Error('build_matches not specified on construction');
    };

    Transducer.prototype['initial_state'] = function() {
      throw new Error('initial_state not specified on construction');
    };

    Transducer.prototype['root'] = function() {
      throw new Error('root not specified on construction');
    };

    Transducer.prototype['edges'] = function(q_D) {
      throw new Error('edges not specified on construction');
    };

    Transducer.prototype['is_final'] = function(q_D) {
      throw new Error('is_final not specified on construction');
    };

    Transducer.prototype['transform'] = function(matches) {
      throw new Error('transform not specified on construction');
    };

    Transducer.prototype['transition_for_state'] = function(n) {
      throw new Error('transition_for_state not specified on construction');
    };

    Transducer.prototype['characteristic_vector'] = function(x, term, k, i) {
      throw new Error('characteristic_vector not specified on construction');
    };

    Transducer.prototype['push'] = function(matches, candidate) {
      throw new Error('push not specified on construction');
    };

    Transducer.prototype['default_edit_distance'] = function() {
      throw new Error('default_edit_distance not specified on construction');
    };

    function Transducer(attributes) {
      var attribute;
      for (attribute in attributes) {
        if (!__hasProp.call(attributes, attribute)) continue;
        this[attribute] = attributes[attribute];
      }
    }

    Transducer.prototype['transduce'] = function(term, n) {
      var M, V, a, b, distance, i, k, matches, next_M, next_V, next_q_D, q_D, stack, transition, vector, w, x, _ref, _ref1;
      if (typeof n !== 'number') {
        n = this['default_edit_distance']();
      }
      w = term.length;
      transition = this['transition_for_state'](n);
      matches = this['build_matches']();
      stack = [['', this['root'](), this['initial_state']()]];
      while (stack.length > 0) {
        _ref = stack['pop'](), V = _ref[0], q_D = _ref[1], M = _ref[2];
        i = M[0][0];
        a = 2 * n + 1;
        b = w - i;
        k = a < b ? a : b;
        _ref1 = this['edges'](q_D);
        for (x in _ref1) {
          next_q_D = _ref1[x];
          vector = this['characteristic_vector'](x, term, k, i);
          next_M = transition(M, vector);
          if (next_M) {
            next_V = V + x;
            stack.push([next_V, next_q_D, next_M]);
            if (this['is_final'](next_q_D)) {
              distance = this['minimum_distance'](next_M, w);
              if (distance <= n) {
                this['push'](matches, [next_V, distance]);
              }
            }
          }
        }
      }
      return this['transform'](matches);
    };

    return Transducer;

  })();

  global = typeof exports === 'object' ? exports : typeof window === 'object' ? window : this;

  global['levenshtein'] || (global['levenshtein'] = {});

  global['levenshtein']['Transducer'] = Transducer;

}).call(this);