nedb/lib/model.js

/**
 * Handle models (i.e. docs)
 * Serialization/deserialization
 * Copying
 * Querying, update
 */

var util = require('util')
  , _ = require('underscore')
  , modifierFunctions = {}
  , lastStepModifierFunctions = {}
  , comparisonFunctions = {}
  , logicalOperators = {}
  , arrayComparisonFunctions = {}
  ;


/**
 * Check a key, throw an error if the key is non valid
 * @param {String} k key
 * @param {Model} v value, needed to treat the Date edge case
 * Non-treatable edge cases here: if part of the object if of the form { $$date: number } or { $$deleted: true }
 * Its serialized-then-deserialized version it will transformed into a Date object
 * But you really need to want it to trigger such behaviour, even when warned not to use '$' at the beginning of the field names...
 */
function checkKey (k, v) {
  if (k[0] === '$' && !(k === '$$date' && typeof v === 'number') && !(k === '$$deleted' && v === true) && !(k === '$$indexCreated') && !(k === '$$indexRemoved')) {
    throw 'Field names cannot begin with the $ character';
  }

  if (k.indexOf('.') !== -1) {
    throw 'Field names cannot contain a .';
  }
}


/**
 * Check a DB object and throw an error if it's not valid
 * Works by applying the above checkKey function to all fields recursively
 */
function checkObject (obj) {
  if (util.isArray(obj)) {
    obj.forEach(function (o) {
      checkObject(o);
    });
  }

  if (typeof obj === 'object' && obj !== null) {
    Object.keys(obj).forEach(function (k) {
      checkKey(k, obj[k]);
      checkObject(obj[k]);
    });
  }
}


/**
 * Serialize an object to be persisted to a one-line string
 * For serialization/deserialization, we use the native JSON parser and not eval or Function
 * That gives us less freedom but data entered in the database may come from users
 * so eval and the like are not safe
 * Accepted primitive types: Number, String, Boolean, Date, null
 * Accepted secondary types: Objects, Arrays
 */
function serialize (obj) {
  var res;
  
  res = JSON.stringify(obj, function (k, v) {
    checkKey(k, v);
    
    if (v === undefined) { return undefined; }
    if (v === null) { return null; }

    // Hackish way of checking if object is Date (this way it works between execution contexts in node-webkit).
    // We can't use value directly because for dates it is already string in this function (date.toJSON was already called), so we use this
    if (typeof this[k].getTime === 'function') { return { $$date: this[k].getTime() }; }

    return v;
  });

  return res;
}


/**
 * From a one-line representation of an object generate by the serialize function
 * Return the object itself
 */
function deserialize (rawData) {
  return JSON.parse(rawData, function (k, v) {
    if (k === '$$date') { return new Date(v); }
    if (typeof v === 'string' || typeof v === 'number' || typeof v === 'boolean' || v === null) { return v; }
    if (v && v.$$date) { return v.$$date; }

    return v;
  });
}


/**
 * Deep copy a DB object
 */
function deepCopy (obj) {
  var res;

  if ( typeof obj === 'boolean' ||
       typeof obj === 'number' ||
       typeof obj === 'string' ||
       obj === null ||
       (util.isDate(obj)) ) {
    return obj;
  }

  if (util.isArray(obj)) {
    res = [];
    obj.forEach(function (o) { res.push(o); });
    return res;
  }

  if (typeof obj === 'object') {
    res = {};
    Object.keys(obj).forEach(function (k) {
      res[k] = deepCopy(obj[k]);
    });
    return res;
  }

  return undefined;   // For now everything else is undefined. We should probably throw an error instead
}


/**
 * Tells if an object is a primitive type or a "real" object
 * Arrays are considered primitive
 */
function isPrimitiveType (obj) {
  return ( typeof obj === 'boolean' ||
       typeof obj === 'number' ||
       typeof obj === 'string' ||
       obj === null ||
       util.isDate(obj) ||
       util.isArray(obj));
}


/**
 * Utility functions for comparing things
 * Assumes type checking was already done (a and b already have the same type)
 * compareNSB works for numbers, strings and booleans
 */
function compareNSB (a, b) {
  if (a < b) { return -1; }
  if (a > b) { return 1; }
  return 0;
}

function compareArrays (a, b) {
  var i, comp;

  for (i = 0; i < Math.min(a.length, b.length); i += 1) {
    comp = compareThings(a[i], b[i]);

    if (comp !== 0) { return comp; }
  }

  // Common section was identical, longest one wins
  return compareNSB(a.length, b.length);
}


/**
 * Compare { things U undefined }
 * Things are defined as any native types (string, number, boolean, null, date) and objects
 * We need to compare with undefined as it will be used in indexes
 * In the case of objects and arrays, we deep-compare
 * If two objects dont have the same type, the (arbitrary) type hierarchy is: undefined, null, number, strings, boolean, dates, arrays, objects
 * Return -1 if a < b, 1 if a > b and 0 if a = b (note that equality here is NOT the same as defined in areThingsEqual!)
 */
function compareThings (a, b) {
  var aKeys, bKeys, comp, i;

  // undefined
  if (a === undefined) { return b === undefined ? 0 : -1; }
  if (b === undefined) { return a === undefined ? 0 : 1; }

  // null
  if (a === null) { return b === null ? 0 : -1; }
  if (b === null) { return a === null ? 0 : 1; }

  // Numbers
  if (typeof a === 'number') { return typeof b === 'number' ? compareNSB(a, b) : -1; }
  if (typeof b === 'number') { return typeof a === 'number' ? compareNSB(a, b) : 1; }

  // Strings
  if (typeof a === 'string') { return typeof b === 'string' ? compareNSB(a, b) : -1; }
  if (typeof b === 'string') { return typeof a === 'string' ? compareNSB(a, b) : 1; }

  // Booleans
  if (typeof a === 'boolean') { return typeof b === 'boolean' ? compareNSB(a, b) : -1; }
  if (typeof b === 'boolean') { return typeof a === 'boolean' ? compareNSB(a, b) : 1; }

  // Dates
  if (util.isDate(a)) { return util.isDate(b) ? compareNSB(a.getTime(), b.getTime()) : -1; }
  if (util.isDate(b)) { return util.isDate(a) ? compareNSB(a.getTime(), b.getTime()) : 1; }

  // Arrays (first element is most significant and so on)
  if (util.isArray(a)) { return util.isArray(b) ? compareArrays(a, b) : -1; }
  if (util.isArray(b)) { return util.isArray(a) ? compareArrays(a, b) : 1; }

  // Objects
  aKeys = Object.keys(a).sort();
  bKeys = Object.keys(b).sort();

  for (i = 0; i < Math.min(aKeys.length, bKeys.length); i += 1) {
    comp = compareThings(a[aKeys[i]], b[bKeys[i]]);

    if (comp !== 0) { return comp; }
  }

  return compareNSB(aKeys.length, bKeys.length);
}



// ==============================================================
// Updating documents
// ==============================================================

/**
 * The signature of modifier functions is as follows
 * Their structure is always the same: recursively follow the dot notation while creating
 * the nested documents if needed, then apply the "last step modifier"
 * @param {Object} obj The model to modify
 * @param {String} field Can contain dots, in that case that means we will set a subfield recursively
 * @param {Model} value
 */

/**
 * Set a field to a new value
 */
lastStepModifierFunctions.$set = function (obj, field, value) {
  obj[field] = value;
};


/**
 * Unset a field
 */
lastStepModifierFunctions.$unset = function (obj, field, value) {
  delete obj[field];
};


/**
 * Push an element to the end of an array field
 */
lastStepModifierFunctions.$push = function (obj, field, value) {
  // Create the array if it doesn't exist
  if (!obj.hasOwnProperty(field)) { obj[field] = []; }

  if (!util.isArray(obj[field])) { throw "Can't $push an element on non-array values"; }

  if (value !== null && typeof value === 'object' && value.$each) {
    if (Object.keys(value).length > 1) { throw "Can't use another field in conjunction with $each"; }
    if (!util.isArray(value.$each)) { throw "$each requires an array value"; }

    value.$each.forEach(function (v) {
      obj[field].push(v);
    });
  } else {
    obj[field].push(value);
  }
};


/**
 * Add an element to an array field only if it is not already in it
 * No modification if the element is already in the array
 * Note that it doesn't check whether the original array contains duplicates
 */
lastStepModifierFunctions.$addToSet = function (obj, field, value) {
  var addToSet = true;

  // Create the array if it doesn't exist
  if (!obj.hasOwnProperty(field)) { obj[field] = []; }

  if (!util.isArray(obj[field])) { throw "Can't $addToSet an element on non-array values"; }

  if (value !== null && typeof value === 'object' && value.$each) {
    if (Object.keys(value).length > 1) { throw "Can't use another field in conjunction with $each"; }
    if (!util.isArray(value.$each)) { throw "$each requires an array value"; }

    value.$each.forEach(function (v) {
      lastStepModifierFunctions.$addToSet(obj, field, v);
    });
  } else {
    obj[field].forEach(function (v) {
      if (compareThings(v, value) === 0) { addToSet = false; }
    });
    if (addToSet) { obj[field].push(value); }
  }
};


/**
 * Remove the first or last element of an array
 */
lastStepModifierFunctions.$pop = function (obj, field, value) {
  if (!util.isArray(obj[field])) { throw "Can't $pop an element from non-array values"; }
  if (typeof value !== 'number') { throw value + " isn't an integer, can't use it with $pop"; }
  if (value === 0) { return; }

  if (value > 0) {
    obj[field] = obj[field].slice(0, obj[field].length - 1);
  } else {
    obj[field] = obj[field].slice(1);
  }
};


/**
 * Removes all instances of a value from an existing array
 */
lastStepModifierFunctions.$pull = function (obj, field, value) {
  var arr, i;
  
  if (!util.isArray(obj[field])) { throw "Can't $pull an element from non-array values"; }

  arr = obj[field];
  for (i = arr.length - 1; i >= 0; i -= 1) {
    if (match(arr[i], value)) {
      arr.splice(i, 1);
    }
  }
};


/**
 * Increment a numeric field's value
 */
lastStepModifierFunctions.$inc = function (obj, field, value) {
  if (typeof value !== 'number') { throw value + " must be a number"; }

  if (typeof obj[field] !== 'number') {
    if (!_.has(obj, field)) {
      obj[field] = value;
    } else {
      throw "Don't use the $inc modifier on non-number fields";
    }
  } else {
    obj[field] += value;
  }
};

// Given its name, create the complete modifier function
function createModifierFunction (modifier) {
  return function (obj, field, value) {
    var fieldParts = typeof field === 'string' ? field.split('.') : field;

    if (fieldParts.length === 1) {
      lastStepModifierFunctions[modifier](obj, field, value);
    } else {
      obj[fieldParts[0]] = obj[fieldParts[0]] || {};
      modifierFunctions[modifier](obj[fieldParts[0]], fieldParts.slice(1), value);
    }
  };
}

// Actually create all modifier functions
Object.keys(lastStepModifierFunctions).forEach(function (modifier) {
  modifierFunctions[modifier] = createModifierFunction(modifier);
});


/**
 * Modify a DB object according to an update query
 * For now the updateQuery only replaces the object
 */
function modify (obj, updateQuery) {
  var keys = Object.keys(updateQuery)
    , firstChars = _.map(keys, function (item) { return item[0]; })
    , dollarFirstChars = _.filter(firstChars, function (c) { return c === '$'; })
    , newDoc, modifiers
    ;

  if (keys.indexOf('_id') !== -1 && updateQuery._id !== obj._id) { throw "You cannot change a document's _id"; }

  if (dollarFirstChars.length !== 0 && dollarFirstChars.length !== firstChars.length) {
    throw "You cannot mix modifiers and normal fields";
  }

  if (dollarFirstChars.length === 0) {
    // Simply replace the object with the update query contents
    newDoc = deepCopy(updateQuery);
    newDoc._id = obj._id;
  } else {
    // Apply modifiers
    modifiers = _.uniq(keys);
    newDoc = deepCopy(obj);
    modifiers.forEach(function (m) {
      var keys;

      if (!modifierFunctions[m]) { throw "Unknown modifier " + m; }

      try {
        keys = Object.keys(updateQuery[m]);
      } catch (e) {
        throw "Modifier " + m + "'s argument must be an object";
      }

      keys.forEach(function (k) {
        modifierFunctions[m](newDoc, k, updateQuery[m][k]);
      });
    });
  }

  // Check result is valid and return it
  checkObject(newDoc);
  if (obj._id !== newDoc._id) { throw "You can't change a document's _id"; }
  return newDoc;
};


// ==============================================================
// Finding documents
// ==============================================================

/**
 * Get a value from object with dot notation
 * @param {Object} obj
 * @param {String} field
 */
function getDotValue (obj, field) {
  var fieldParts = typeof field === 'string' ? field.split('.') : field
    , i, objs;

  if (!obj) { return undefined; }   // field cannot be empty so that means we should return undefined so that nothing can match

  if (fieldParts.length === 0) { return obj; }

  if (fieldParts.length === 1) { return obj[fieldParts[0]]; }
  
  if (util.isArray(obj[fieldParts[0]])) {
    // If the next field is an integer, return only this item of the array
    i = parseInt(fieldParts[1], 10);
    if (typeof i === 'number' && !isNaN(i)) {
      return getDotValue(obj[fieldParts[0]][i], fieldParts.slice(2))
    }

    // Return the array of values
    objs = new Array();
    for (i = 0; i < obj[fieldParts[0]].length; i += 1) {
       objs.push(getDotValue(obj[fieldParts[0]][i], fieldParts.slice(1)));
    }
    return objs;
  } else {
    return getDotValue(obj[fieldParts[0]], fieldParts.slice(1));
  }
}


/**
 * Check whether 'things' are equal
 * Things are defined as any native types (string, number, boolean, null, date) and objects
 * In the case of object, we check deep equality
 * Returns true if they are, false otherwise
 */
function areThingsEqual (a, b) {
  var aKeys , bKeys , i;

  // Strings, booleans, numbers, null
  if (a === null || typeof a === 'string' || typeof a === 'boolean' || typeof a === 'number' ||
      b === null || typeof b === 'string' || typeof b === 'boolean' || typeof b === 'number') { return a === b; }

  // Dates
  if (util.isDate(a) || util.isDate(b)) { return util.isDate(a) && util.isDate(b) && a.getTime() === b.getTime(); }

  // Arrays (no match since arrays are used as a $in)
  // undefined (no match since they mean field doesn't exist and can't be serialized)
  if (util.isArray(a) || util.isArray(b) || a === undefined || b === undefined) { return false; }

  // General objects (check for deep equality)
  // a and b should be objects at this point
  try {
    aKeys = Object.keys(a);
    bKeys = Object.keys(b);
  } catch (e) {
    return false;
  }

  if (aKeys.length !== bKeys.length) { return false; }
  for (i = 0; i < aKeys.length; i += 1) {
    if (bKeys.indexOf(aKeys[i]) === -1) { return false; }
    if (!areThingsEqual(a[aKeys[i]], b[aKeys[i]])) { return false; }
  }
  return true;
}


/**
 * Check that two values are comparable
 */
function areComparable (a, b) {
  if (typeof a !== 'string' && typeof a !== 'number' && !util.isDate(a) &&
      typeof b !== 'string' && typeof b !== 'number' && !util.isDate(b)) {
    return false;
  }

  if (typeof a !== typeof b) { return false; }

  return true;
}


/**
 * Arithmetic and comparison operators
 * @param {Native value} a Value in the object
 * @param {Native value} b Value in the query
 */
comparisonFunctions.$lt = function (a, b) {
  return areComparable(a, b) && a < b;
};

comparisonFunctions.$lte = function (a, b) {
  return areComparable(a, b) && a <= b;
};

comparisonFunctions.$gt = function (a, b) {
  return areComparable(a, b) && a > b;
};

comparisonFunctions.$gte = function (a, b) {
  return areComparable(a, b) && a >= b;
};

comparisonFunctions.$ne = function (a, b) {
  if (a === undefined) { return true; }
  return !areThingsEqual(a, b);
};

comparisonFunctions.$in = function (a, b) {
  var i;

  if (!util.isArray(b)) { throw "$in operator called with a non-array"; }

  for (i = 0; i < b.length; i += 1) {
    if (areThingsEqual(a, b[i])) { return true; }
  }

  return false;
};

comparisonFunctions.$nin = function (a, b) {
  if (!util.isArray(b)) { throw "$nin operator called with a non-array"; }

  return !comparisonFunctions.$in(a, b);
};

comparisonFunctions.$regex = function (a, b) {
  if (!util.isRegExp(b)) { throw "$regex operator called with non regular expression"; }

  if (typeof a !== 'string') {
    return false
  } else {
    return b.test(a);
  }
};

comparisonFunctions.$exists = function (value, exists) {
  if (exists || exists === '') {   // This will be true for all values of exists except false, null, undefined and 0
    exists = true;                 // That's strange behaviour (we should only use true/false) but that's the way Mongo does it...
  } else {
    exists = false;
  }

  if (value === undefined) {
    return !exists
  } else {
    return exists;
  }
};

// Specific to arrays
comparisonFunctions.$size = function (obj, value) {
    if (!util.isArray(obj)) { return false; }
    if (value % 1 !== 0) { throw "$size operator called without an integer"; }

    return (obj.length == value);
};
arrayComparisonFunctions.$size = true;


/**
 * Match any of the subqueries
 * @param {Model} obj
 * @param {Array of Queries} query
 */
logicalOperators.$or = function (obj, query) {
  var i;

  if (!util.isArray(query)) { throw "$or operator used without an array"; }

  for (i = 0; i < query.length; i += 1) {
    if (match(obj, query[i])) { return true; }
  }

  return false;
};


/**
 * Match all of the subqueries
 * @param {Model} obj
 * @param {Array of Queries} query
 */
logicalOperators.$and = function (obj, query) {
  var i;

  if (!util.isArray(query)) { throw "$and operator used without an array"; }

  for (i = 0; i < query.length; i += 1) {
    if (!match(obj, query[i])) { return false; }
  }

  return true;
};


/**
 * Inverted match of the query
 * @param {Model} obj
 * @param {Query} query
 */
logicalOperators.$not = function (obj, query) {
  return !match(obj, query);
};


/**
 * Use a function to match
 * @param {Model} obj
 * @param {Query} query
 */
logicalOperators.$where = function (obj, fn) {
  var result;

  if (!_.isFunction(fn)) { throw "$where operator used without a function"; }

  result = fn.call(obj);
  if (!_.isBoolean(result)) { throw "$where function must return boolean"; }

  return result;
};


/**
 * Tell if a given document matches a query
 * @param {Object} obj Document to check
 * @param {Object} query
 */
function match (obj, query) {
  var queryKeys, queryKey, queryValue, i;

  // Primitive query against a primitive type
  // This is a bit of a hack since we construct an object with an arbitrary key only to dereference it later
  // But I don't have time for a cleaner implementation now
  if (isPrimitiveType(obj) || isPrimitiveType(query)) {
    return matchQueryPart({ needAKey: obj }, 'needAKey', query);
  }
    
  // Normal query
  queryKeys = Object.keys(query);
  for (i = 0; i < queryKeys.length; i += 1) {
    queryKey = queryKeys[i];
    queryValue = query[queryKey];
  
    if (queryKey[0] === '$') {
      if (!logicalOperators[queryKey]) { throw "Unknown logical operator " + queryKey; }
      if (!logicalOperators[queryKey](obj, queryValue)) { return false; }
    } else {
      if (!matchQueryPart(obj, queryKey, queryValue)) { return false; }
    }
  }

  return true;
};


/**
 * Match an object against a specific { key: value } part of a query
 * if the treatObjAsValue flag is set, don't try to match every part separately, but the array as a whole
 */
function matchQueryPart (obj, queryKey, queryValue, treatObjAsValue) {
  var objValue = getDotValue(obj, queryKey)
    , i, keys, firstChars, dollarFirstChars;

  // Check if the value is an array if we don't force a treatment as value
  if (util.isArray(objValue) && !treatObjAsValue) {
    // Check if we are using an array-specific comparison function
    if (queryValue !== null && typeof queryValue === 'object' && !util.isRegExp(queryValue)) {
      keys = Object.keys(queryValue);      
      for (i = 0; i < keys.length; i += 1) {
        if (arrayComparisonFunctions[keys[i]]) { return matchQueryPart(obj, queryKey, queryValue, true); }
      }
    }

    // If not, treat it as an array of { obj, query } where there needs to be at least one match
    for (i = 0; i < objValue.length; i += 1) {
      if (matchQueryPart({ k: objValue[i] }, 'k', queryValue)) { return true; }   // k here could be any string
    }
    return false;
  }

  // queryValue is an actual object. Determine whether it contains comparison operators
  // or only normal fields. Mixed objects are not allowed
  if (queryValue !== null && typeof queryValue === 'object' && !util.isRegExp(queryValue)) {
    keys = Object.keys(queryValue);
    firstChars = _.map(keys, function (item) { return item[0]; });
    dollarFirstChars = _.filter(firstChars, function (c) { return c === '$'; });

    if (dollarFirstChars.length !== 0 && dollarFirstChars.length !== firstChars.length) {
      throw "You cannot mix operators and normal fields";
    }

    // queryValue is an object of this form: { $comparisonOperator1: value1, ... }
    if (dollarFirstChars.length > 0) {
      for (i = 0; i < keys.length; i += 1) {
        if (!comparisonFunctions[keys[i]]) { throw "Unknown comparison function " + keys[i]; }

        if (!comparisonFunctions[keys[i]](objValue, queryValue[keys[i]])) { return false; }
      }
      return true;
    }
  }

  // Using regular expressions with basic querying
  if (util.isRegExp(queryValue)) { return comparisonFunctions.$regex(objValue, queryValue); }

  // queryValue is either a native value or a normal object
  // Basic matching is possible
  if (!areThingsEqual(objValue, queryValue)) { return false; }

  return true;
}


// Interface
module.exports.serialize = serialize;
module.exports.deserialize = deserialize;
module.exports.deepCopy = deepCopy;
module.exports.checkObject = checkObject;
module.exports.isPrimitiveType = isPrimitiveType;
module.exports.modify = modify;
module.exports.getDotValue = getDotValue;
module.exports.match = match;
module.exports.areThingsEqual = areThingsEqual;
module.exports.compareThings = compareThings;