mungedb-aggregate/lib/pipeline/matcher/MatchExpressionParser.js

"use strict";

// Autogenerated by cport.py on 2013-09-17 14:37
var MatchExpressionParser = module.exports = function (){

}, klass = MatchExpressionParser, base =  Object  , proto = klass.prototype = Object.create(base.prototype, {constructor:{value:klass}});

// DEPENDENCIES
var errors = require("../../Errors.js"),
	ErrorCodes = errors.ErrorCodes,
	AndMatchExpression = require("./AndMatchExpression.js"),
	MatchExpression = require("./MatchExpression.js"),
	OrMatchExpression = require("./OrMatchExpression.js"),
	ModMatchExpression = require("./ModMatchExpression.js"),
	NorMatchExpression = require("./NorMatchExpression.js"),
	NotMatchExpression = require("./NotMatchExpression.js"),
	LTMatchExpression = require("./LTMatchExpression.js"),
	LTEMatchExpression = require("./LTEMatchExpression.js"),
	GTMatchExpression = require("./GTMatchExpression.js"),
	GTEMatchExpression = require("./GTEMatchExpression.js"),
	InMatchExpression = require("./InMatchExpression.js"),
	SizeMatchExpression = require("./SizeMatchExpression.js"),
	TypeMatchExpression = require("./TypeMatchExpression.js"),
	ExistsMatchExpression = require("./ExistsMatchExpression.js"),
	EqualityMatchExpression = require("./EqualityMatchExpression.js"),
	ArrayMatchingMatchExpression = require("./ArrayMatchingMatchExpression.js"),
	RegexMatchExpression = require("./RegexMatchExpression.js"),
	FalseMatchExpression = require("./FalseMatchExpression.js"),
	ComparisonMatchExpression = require("./ComparisonMatchExpression.js"),
	ElemMatchValueMatchExpression = require("./ElemMatchValueMatchExpression.js"),
	ElemMatchObjectMatchExpression = require("./ElemMatchObjectMatchExpression.js"),
	AllElemMatchOp = require("./AllElemMatchOp.js"),
	AtomicMatchExpression = require("./AtomicMatchExpression.js");

/**
 *
 * Check if the input element is an expression
 * @method _isExpressionDocument
 * @param element
 *
 */
proto._isExpressionDocument = function _isExpressionDocument(element){
	// File: expression_parser.cpp lines: 340-355
	if (!(element instanceof Object))
		return false;

	if (Object.keys(element).length === 0)
		return false;

	var name = Object.keys(element)[0];
	if (name[0] != '$')
		return false;

	if ("$ref" == name)
		return false;

	return true;
};

/**
 *
 * Parse the input object into individual elements
 * @method _parse
 * @param obj
 * @param topLevel
 *
 */
proto._parse = function _parse(obj, topLevel){
	// File: expression_parser.cpp lines: 217-319
	var rest, temp, status, element, eq, real;
	var root = new AndMatchExpression();
	var objkeys = Object.keys(obj);
	var currname, currval;
	for (var i = 0; i < objkeys.length; i++) {
		currname = objkeys[i];
		currval = obj[currname];
		if (currname[0] == '$' ) {
			rest = currname.substr(1);

			// TODO: optimize if block?
			if ("or" == rest) {
				if (!(currval instanceof Array))
					return {code:ErrorCodes.BAD_VALUE, description:"$or needs an array"};
				temp = new OrMatchExpression();
				status = this._parseTreeList(currval, temp);
				if (status.code != ErrorCodes.OK)
					return status;
				root.add(temp);
			}
			else if ("and" == rest) {
				if (!(currval instanceof Array))
					return {code:ErrorCodes.BAD_VALUE, description:"and needs an array"};
				temp = new AndMatchExpression();
				status = this._parseTreeList(currval, temp);
				if (status.code != ErrorCodes.OK)
					return status;
				root.add(temp);
			}
			else if ("nor" == rest) {
				if (!(currval instanceof Array))
					return {code:ErrorCodes.BAD_VALUE, description:"and needs an array"};
				temp = new NorMatchExpression();
				status = this._parseTreeList(currval, temp);
				if (status.code != ErrorCodes.OK)
					return status;
				root.add(temp);
			}
			else if (("atomic" == rest) || ("isolated" == rest)) {
				if (!topLevel)
					return {code:ErrorCodes.BAD_VALUE, description:"$atomic/$isolated has to be at the top level"};
				if (element)
					root.add(new AtomicMatchExpression());
			}
			else if ("where" == rest) {
				/*
				if ( !topLevel )
					return StatusWithMatchExpression( ErrorCodes::BAD_VALUE, "$where has to be at the top level" );
				*/

				return {'code':'FAILED_TO_PARSE', 'desc':'Where unimplimented.'};
				/*
				status = this.expressionParserWhereCallback(element);
				if (status.code != ErrorCodes.OK)
					return status;
				root.add(status.result);*/
			}
			else if ("comment" == rest) {
				1+1;
			}
			else {
				return {code:ErrorCodes.BAD_VALUE, description:"unknown top level operator: " + currname};
			}

			continue;
		}

		if (this._isExpressionDocument(currval)) {
			status = this._parseSub(currname, currval, root);
			if (status.code != ErrorCodes.OK)
				return status;
			continue;
		}

		if (currval instanceof RegExp) {
			status = this._parseRegexElement(currname, currval);
			if (status.code != ErrorCodes.OK)
				return status;
			root.add(status.result);
			continue;
		}

		eq = new EqualityMatchExpression();
		status = eq.init(currname, currval);
		if (status.code != ErrorCodes.OK)
			return status;

		root.add(eq);
	}

	if (root.numChildren() == 1) {
		return {code:ErrorCodes.OK, result:root.getChild(0)};
	}

	return {code:ErrorCodes.OK, result:root};
};

/**
 *
 * Parse the $all element
 * @method _parseAll
 * @param name
 * @param element
 *
 */
proto._parseAll = function _parseAll(name, element){
	// File: expression_parser.cpp lines: 512-583
	var status, i;

	if (!(element instanceof Array))
		return {code:ErrorCodes.BAD_VALUE, description:"$all needs an array"};

	var arr = element;
	if ((arr[0] instanceof Object) && ("$elemMatch" == Object.keys(arr[0])[0])) {
		// $all : [ { $elemMatch : {} } ... ]

		var temp = new AllElemMatchOp();
		status = temp.init(name);
		if (status.code != ErrorCodes.OK)
			return status;

		for (i = 0; i < arr.length; i++) {
			var hopefullyElemMatchElement = arr[i];

			if (!(hopefullyElemMatchElement instanceof Object)) {
				// $all : [ { $elemMatch : ... }, 5 ]
				return {code:ErrorCodes.BAD_VALUE, description:"$all/$elemMatch has to be consistent"};
			}

			if ("$elemMatch" != Object.keys(hopefullyElemMatchElement)[0]) {
				// $all : [ { $elemMatch : ... }, { x : 5 } ]
				return {code:ErrorCodes.BAD_VALUE, description:"$all/$elemMatch has to be consistent"};
			}

			status = this._parseElemMatch("",  hopefullyElemMatchElement.$elemMatch ); // TODO: wrong way to do this?
			if (status.code != ErrorCodes.OK)
				return status;
			temp.add(status.result);
		}

		return {code:ErrorCodes.OK, result:temp};
	}

	var myAnd = new AndMatchExpression();
	for (i = 0; i < arr.length; i++) {
		var e = arr[i];

		if (e instanceof RegExp) {
			var r = new RegexMatchExpression();
			status = r.init(name, e);
			if (status.code != ErrorCodes.OK)
				return status;
			myAnd.add(r);
		}
		else if ((e instanceof Object) && (typeof(Object.keys(e)[0] == 'string' && Object.keys(e)[0][0] == '$' ))) {
			return {code:ErrorCodes.BAD_VALUE, description:"no $ expressions in $all"};
		}
		else {
			var x = new EqualityMatchExpression();
			status = x.init(name, e);
			if (status.code != ErrorCodes.OK)
				return status;
			myAnd.add(x);
		}
	}

	if (myAnd.numChildren() === 0) {
		return {code:ErrorCodes.OK, result:new FalseMatchExpression()};
	}

	return {code:ErrorCodes.OK, result:myAnd};
};

/**
 *
 * Parse the input array and add new RegexMatchExpressions to entries
 * @method _parseArrayFilterEntries
 * @param entries
 * @param theArray
 *
 */
proto._parseArrayFilterEntries = function _parseArrayFilterEntries(entries, theArray){
	// File: expression_parser.cpp lines: 445-468
	var status, e, r;
	for (var i = 0; i < theArray.length; i++) {
		e = theArray[i];

		if (e instanceof RegExp ) {
			r = new RegexMatchExpression();
			status = r.init("", e);
			if (status.code != ErrorCodes.OK)
				return status;
			status = entries.addRegex(r);
			if (status.code != ErrorCodes.OK)
				return status;
		}
		else {
			status = entries.addEquality(e);
			if (status.code != ErrorCodes.OK)
				return status;
		}
	}
	return {code:ErrorCodes.OK};
};

/**
 *
 * Parse the input ComparisonMatchExpression
 * @method _parseComparison
 * @param name
 * @param cmp
 * @param element
 *
 */
proto._parseComparison = function _parseComparison(name, cmp, element){
	// File: expression_parser.cpp lines: 34-43
	var temp = new ComparisonMatchExpression(cmp);

	var status = temp.init(name, element);
	if (status.code != ErrorCodes.OK)
		return status;

	return {code:ErrorCodes.OK, result:temp};
};

/**
 *
 * Parse an element match into the appropriate expression
 * @method _parseElemMatch
 * @param name
 * @param element
 *
 */
proto._parseElemMatch = function _parseElemMatch(name, element){
	// File: expression_parser.cpp lines: 471-509
	var temp, status;
	if (!(element instanceof Object))
		return {code:ErrorCodes.BAD_VALUE, description:"$elemMatch needs an Object"};

	if (this._isExpressionDocument(element)) {
		// value case

		var theAnd = new AndMatchExpression();
		status = this._parseSub("", element, theAnd);
		if (status.code != ErrorCodes.OK)
			return status;

		temp = new ElemMatchValueMatchExpression();
		status = temp.init(name);
		if (status.code != ErrorCodes.OK)
			return status;

		for (var i = 0; i < theAnd.numChildren(); i++ ) {
			temp.add(theAnd.getChild(i));
		}
		theAnd.clearAndRelease();

		return {code:ErrorCodes.OK, result:temp};
	}

	// object case

	status = this._parse(element, false);
	if (status.code != ErrorCodes.OK)
		return status;

	temp = new ElemMatchObjectMatchExpression();
	status = temp.init(name, status.result);
	if (status.code != ErrorCodes.OK)
		return status;

	return {code:ErrorCodes.OK, result:temp};
};

/**
 *
 * Parse a ModMatchExpression
 * @method _parseMOD
 * @param name
 * @param element
 *
 */
proto._parseMOD = function _parseMOD(name, element){
	// File: expression_parser.cpp lines: 360-387
	var d,r;
	if (!(element instanceof Array))
		return {code:ErrorCodes.BAD_VALUE, result:"malformed mod, needs to be an array"};
	if (element.length < 2)
		return {code:ErrorCodes.BAD_VALUE, result:"malformed mod, not enough elements"};
	if (element.length > 2)
		return {code:ErrorCodes.BAD_VALUE, result:"malformed mod, too many elements"};
	if ((typeof(element[0]) != 'number')) {
		return {code:ErrorCodes.BAD_VALUE, result:"malformed mod, divisor not a number"};
	} else {
		d = element[0];
	}
	if (( typeof(element[1]) != 'number') ) {
		r = 0;
	} else {
		r = element[1];
	}

	var temp = new ModMatchExpression();
	var status = temp.init( name, d, r);
	if (status.code != ErrorCodes.OK)
		return status;

	return {code:ErrorCodes.OK, result:temp};
};

/**
 *
 * Parse a NotMatchExpression
 * @method _parseNot
 * @param name
 * @param element
 *
 */
proto._parseNot = function _parseNot(name, element){
	// File: expression_parser_tree.cpp lines: 55-91
	var status;
	if (element instanceof RegExp) {
		status = this._parseRegexElement(name, element);
		if (status.code != ErrorCodes.OK)
			return status;
		var n = new NotMatchExpression();
		status = n.init(status.result);
		if (status.code != ErrorCodes.OK)
			return status;
		return {code:ErrorCodes.OK, result:n};
	}

	if (!(element instanceof Object))
		return {code:ErrorCodes.BAD_VALUE, result:"$not needs a regex or a document"};

	if (element == {})
		return {code:ErrorCodes.BAD_VALUE, result:"$not cannot be empty"};

	var theAnd = new AndMatchExpression();
	status = this._parseSub(name, element, theAnd);
	if (status.code != ErrorCodes.OK)
		return status;

	// TODO: this seems arbitrary?
	// tested in jstests/not2.js
	for (var i = 0; i < theAnd.numChildren(); i++) {
		if (theAnd.getChild(i).matchType == MatchExpression.REGEX) {
			return {code:ErrorCodes.BAD_VALUE, result:"$not cannot have a regex"};
		}
	}
	var theNot = new NotMatchExpression();
	status = theNot.init(theAnd);
	if (status.code != ErrorCodes.OK)
		return status;

	return {code:ErrorCodes.OK, result:theNot};
};

/**
 *
 * Parse a RegexMatchExpression
 * @method _parseRegexDocument
 * @param name
 * @param doc
 *
 */
proto._parseRegexDocument = function _parseRegexDocument(name, doc){
	// File: expression_parser.cpp lines: 402-442
	var regex = '', regexOptions = '', e;

	if(doc.$regex) {
		e = doc.$regex;
		if(e instanceof RegExp) {
			var str = e.toString(),
				flagIndex = 0;
			for (var c = str.length; c > 0; c--){
				if (str[c] == '/') {
					flagIndex = c;
					break;
				}
			}
			regex = (flagIndex? str : str.substr(1, flagIndex-1));
			regexOptions = str.substr(flagIndex, str.length);
		} else if (typeof(e) == 'string') {
			regex = e;
		} else {
			return {code:ErrorCodes.BAD_VALUE, description:"$regex has to be a string"};
		}
	}

	if(doc.$options) {
		e = doc.$options;
		if(typeof(e) == 'string') {
			regexOptions = e;
		} else {
			return {code:ErrorCodes.BAD_VALUE, description:"$options has to be a string"};
		}

	}

	var temp = new RegexMatchExpression();
	var status = temp.init(name, regex, regexOptions);
	if (status.code != ErrorCodes.OK)
		return status;

	return {code:ErrorCodes.OK, result:temp};
};

/**
 *
 * Parse an element into a RegexMatchExpression
 * @method _parseRegexElement
 * @param name
 * @param element
 *
 */
proto._parseRegexElement = function _parseRegexElement(name, element){
	// File: expression_parser.cpp lines: 390-399
	if (!(element instanceof RegExp))
		return {code:ErrorCodes.BAD_VALUE, description:"not a regex"};

	var str = element.toString(),
		flagIndex = 0;
	for (var c = str.length; c > 0; c--){
		if (str[c] == '/') {
			flagIndex = c;
			break;
		}
	}
	var regex = str.substr(1, flagIndex-1),
		regexOptions = str.substr(flagIndex+1, str.length),
		temp = new RegexMatchExpression(),
		status = temp.init(name, regex, regexOptions);

	if (status.code != ErrorCodes.OK)
		return status;

	return {code:ErrorCodes.OK, result:temp};
};

/**
 *
 * Parse a sub expression
 * @method _parseSub
 * @param name
 * @param sub
 * @param root
 *
 */
proto._parseSub = function _parseSub(name, sub, root){
	// File: expression_parser.cpp lines: 322-337
	var subkeys = Object.keys(sub),
		currname, currval;

	for (var i = 0; i < subkeys.length; i++) {
		currname = subkeys[i];
		currval = sub[currname];
		var deep = {};
		deep[currname] = currval;

		var status = this._parseSubField(sub, root, name, deep);
		if (status.code != ErrorCodes.OK)
			return status;

		if (status.result)
			root.add(status.result);
	}

	return {code:ErrorCodes.OK, result:root};
};

/**
 *
 * Parse a sub expression field
 * @method _parseSubField
 * @param context
 * @param andSoFar
 * @param name
 * @param element
 *
 */
proto._parseSubField = function _parseSubField(context, andSoFar, name, element){
	// File: expression_parser.cpp lines: 46-214
	// TODO: these should move to getGtLtOp, or its replacement
	var currname = Object.keys(element)[0];
	var currval = element[currname];
	if ("$eq" == currname)
		return this._parseComparison(name, 'EQ', currval);

	if ("$not" == currname) {
		return this._parseNot(name, currval);
	}

	var status, temp, temp2;
	switch (currname) {
	case '$lt':
		return this._parseComparison(name, 'LT', currval);
	case '$lte':
		return this._parseComparison(name, 'LTE', currval);
	case '$gt':
		return this._parseComparison(name, 'GT', currval);
	case '$gte':
		return this._parseComparison(name, 'GTE', currval);
	case '$ne':
		status = this._parseComparison(name, 'EQ', currval);
		if (status.code != ErrorCodes.OK)
			return status;
		var n = new NotMatchExpression();
		status = n.init(status.result);
		if (status.code != ErrorCodes.OK)
			return status;
		return {code:ErrorCodes.OK, result:n};
	case '$eq':
		return this._parseComparison(name, 'EQ', currval);

	case '$in':
		if (!(currval instanceof Array))
			return {code:ErrorCodes.BAD_VALUE, description:"$in needs an array"};
		temp = new InMatchExpression();
		status = temp.init(name);
		if (status.code != ErrorCodes.OK)
			return status;
		status = this._parseArrayFilterEntries(temp.getArrayFilterEntries(), currval);
		if (status.code != ErrorCodes.OK)
			return status;
		return {code:ErrorCodes.OK, result:temp};

	case '$nin':
		if (!(currval instanceof Array))
			return {code:ErrorCodes.BAD_VALUE, description:"$nin needs an array"};
		temp = new InMatchExpression();
		status = temp.init(name);
		if (status.code != ErrorCodes.OK)
			return status;
		status = this._parseArrayFilterEntries(temp.getArrayFilterEntries(), currval);
		if (status.code != ErrorCodes.OK)
			return status;

		temp2 = new NotMatchExpression();
		status = temp2.init(temp);
		if (status.code != ErrorCodes.OK)
			return status;

		return {code:ErrorCodes.OK, result:temp2};

	case '$size':
		var size = 0;
		if ( typeof(currval) === 'string')
			// matching old odd semantics
			size = 0;
		else if (typeof(currval) === 'number')
			size = currval;
		else {
			return {code:ErrorCodes.BAD_VALUE, description:"$size needs a number"};
		}

		temp = new SizeMatchExpression();
		status = temp.init(name, size);
		if (status.code != ErrorCodes.OK)
			return status;

		return {code:ErrorCodes.OK, result:temp};

	case '$exists':
		if (currval == {})
			return {code:ErrorCodes.BAD_VALUE, description:"$exists can't be eoo"};
		temp = new ExistsMatchExpression();
		status = temp.init(name);
		if (status.code != ErrorCodes.OK)
			return status;
		if (currval)
			return {code:ErrorCodes.OK, result:temp};
		temp2 = new NotMatchExpression();
		status = temp2.init(temp);
		if (status.code != ErrorCodes.OK)
			return status;
		return {code:ErrorCodes.OK, result:temp2};

	case '$type':
		if (typeof(currval) != 'number')
			return {code:ErrorCodes.BAD_VALUE, description:"$type has to be a number"};
		var type = currval;
		temp = new TypeMatchExpression();
		status = temp.init(name, type);
		if (status.code != ErrorCodes.OK)
			return status;
		return {code:ErrorCodes.OK, result:temp};

	case '$mod':
		return this._parseMOD(name, currval);

	case '$options':
		// TODO: try to optimize this
		// we have to do this since $options can be before or after a $regex
		// but we validate here
		for(var i = 0; i < Object.keys(context).length; i++) {
			temp = Object.keys(context)[i];
			if (temp == '$regex')
				return {code:ErrorCodes.OK, result:null};
		}

		return {code:ErrorCodes.BAD_VALUE, description:"$options needs a $regex"};

	case '$regex':
		return this._parseRegexDocument(name, context);

	case '$elemMatch':
		return this._parseElemMatch(name, currval);

	case '$all':
		return this._parseAll(name, currval);

	case '$geoWithin':
	case '$geoIntersects':
	case '$near':
	case '$nearSphere':
		var x = 'Temporary value until Geo fns implimented.';
		return this.expressionParserGeoCallback(name, x, context);
	default:
		return {code:ErrorCodes.BAD_VALUE, description:"not handled: " + element};
	} // end switch


	return {code:ErrorCodes.BAD_VALUE, description:"not handled: " + element};
};

/**
 *
 * Parse a list of parsable elements
 * @method _parseTreeList
 * @param arr
 * @param out
 *
 */
proto._parseTreeList = function _parseTreeList(arr, out){
	// File: expression_parser_tree.cpp lines: 33-52
	if (arr.length === 0)
		return {code:ErrorCodes.BAD_VALUE, description:"$and/$or/$nor must be a nonempty array"};

	var status, element;
	for (var i = 0; i < arr.length; i++) {
		element = arr[i];

		if (!(element instanceof Object))
			return {code:ErrorCodes.BAD_VALUE, description:"$or/$and/$nor entries need to be full objects"};

		status = this._parse(element, false);
		if (status.code != ErrorCodes.OK)
			return status;

		out.add(status.result);
	}
	return {code:ErrorCodes.OK};
};

/**
 *
 * Wrapper for _parse
 * @method parse
 * @param obj
 *
 */
proto.parse = function parse(obj){
	// File: expression_parser.h lines: 40-41
    return this._parse(obj, true);
};