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");

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _isExpressionDocument
 * @param
 *
 */
proto._isExpressionDocument = function _isExpressionDocument(element){
	// File: expression_parser.cpp lines: 340-355
	if (!(element instanceof Object))
		return false;

	if (element.isEmpty())
		return false;

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

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

	return true;
};

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parse
 * @param
 *
 */
proto._parse = function _parse(obj, topLevel){
	// File: expression_parser.cpp lines: 217-319
	var rest, temp, status, element, eq, real;
	var root = new AndMatchExpression();

	for (var i = 0; i < obj.length; i++) {
		element = obj[i];
		if (element[0] == '$' ) {
			rest = element.substr(1, element.length);

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

			continue;
		}

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

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

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

		root.add(eq.release());
	}

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

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

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseAll
 * @param
 *
 */
proto._parseAll = function _parseAll(name, element){
	// File: expression_parser.cpp lines: 512-583
	var status, i;

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

	var arr = element;
	if ((arr[0] instanceof Object) && ("$elemMatch" == arr[0].keys()[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 : [ { $elemMatch : ... }, 5 ]
				return {code:ErrorCodes.BadValue, description:"$all/$elemMatch has to be consistent"};
			}

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

			status = this._parseElemMatch("", hopefullyElemMatchElement[hopefullyElemMatchElement.keys()[0]]); // TODO: wrong way to do this?
			if (status.code != ErrorCodes.OK)
				return status;
			temp.add(new ArrayMatchingMatchExpression(status.result));
		}

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

	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.release());
		}
		else if ((e instanceof Object) && (e.keys()[0].getGtLtOp(-1) != -1)) {
			return {code:ErrorCodes.BadValue, 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.release());
		}
	}

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

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

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseArrayFilterEntries
 * @param
 *
 */
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.release());
			if (status.code != ErrorCodes.OK)
				return status;
		}
		else {
			status = entries.addEquality(e);
			if (status.code != ErrorCodes.OK)
				return status;
		}
	}
	return {code:ErrorCodes.OK};
};

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseComparison
 * @param
 *
 */
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.release()};
};

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseElemMatch
 * @param
 *
 */
proto._parseElemMatch = function _parseElemMatch(name, element){
	// File: expression_parser.cpp lines: 471-509
	var temp, status;
	if (!(element instanceof Object))
		return {code:ErrorCodes.BadValue, 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.release()};
	}

	// 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.release()};
};

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseMOD
 * @param
 *
 */
proto._parseMOD = function _parseMOD(name, element){
	// File: expression_parser.cpp lines: 360-387

	if (!(element instanceof Array))
		return {code:ErrorCodes.BadValue, result:"malformed mod, needs to be an array"};
	if (element.length < 2)
		return {code:ErrorCodes.BadValue, result:"malformed mod, not enough elements"};
	if (element.length > 2)
		return {code:ErrorCodes.BadValue, result:"malformed mod, too many elements"};
	if (!(element[0] instanceof Number))
		return {code:ErrorCodes.BadValue, result:"malformed mod, divisor not a number"};
	if (!(element[1] instanceof Number))
		return {code:ErrorCodes.BadValue, result:"malformed mod, remainder not a number"};

	var temp = new ModMatchExpression();
	var status = temp.init( name, element[0], element[1]);
	if (status.code != ErrorCodes.OK)
		return status;

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

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseNot
 * @param
 *
 */
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.release()};
	}

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

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

	var theAnd = new AndMatchExpression();
	status = this._parseSub(name, element, theAnd.get());
	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.BadValue, result:"$not cannot have a regex"};

	var theNot = new NotMatchExpression();
	status = theNot.init(theAnd.release());
	if (status.code != ErrorCodes.OK)
		return status;

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

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseRegexDocument
 * @param
 *
 */
proto._parseRegexDocument = function _parseRegexDocument(name, doc){
	// File: expression_parser.cpp lines: 402-442
	var regex, regexOptions, e;

	for (var i = 0; i < doc.length; i++) {
		e = doc[i];
		switch (e.getGtLtOp()) {
		case 'opREGEX':
			if (e instanceof String) {
				regex = e;
			}
			else 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 = str.substr(1, flagIndex-1);
				regexOptions = str.substr(flagIndex, str.length);
			}
			else {
				return {code:ErrorCodes.BadValue, description:"$regex has to be a string"};
			}

			break;
		case 'opOPTIONS':
			if (!(e instanceof String))
				return {code:ErrorCodes.BadValue, description:"$options has to be a string"};
			regexOptions = e;
			break;
		default:
			break;
		}
	}

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

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

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseRegexElement
 * @param
 *
 */
proto._parseRegexElement = function _parseRegexElement(name, element){
	// File: expression_parser.cpp lines: 390-399
	if (!(element instanceof RegExp))
		return {code:ErrorCodes.BadValue, 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, str.length),
		temp = new RegexMatchExpression(),
		status = temp.init(name, regex, regexOptions);

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

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

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseSub
 * @param
 *
 */
proto._parseSub = function _parseSub(name, sub, root){
	// File: expression_parser.cpp lines: 322-337
	for (var i = 0; i < sub.length; i++) {
		var deep = sub[i];

		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};
};

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseSubField
 * @param
 *
 */
proto._parseSubField = function _parseSubField(context, andSoFar, name, element){
	// File: expression_parser.cpp lines: 46-214
	// TODO: these should move to getGtLtOp, or its replacement

	if ("$eq" == element)
		return this._parseComparison(name, new EqualityMatchExpression(), element);

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

	var x = element.getGtLtOp(-1),
		status, temp, temp2;
	switch (x) {
	case -1:
		return {code:ErrorCodes.BadValue, description:"unknown operator: " + element};
	case 'LT':
		return this._parseComparison(name, new LTMatchExpression(), element);
	case 'LTE':
		return this._parseComparison(name, new LTEMatchExpression(), element);
	case 'GT':
		return this._parseComparison(name, new GTMatchExpression(), element);
	case 'GTE':
		return this._parseComparison(name, new GTEMatchExpression(), element);
	case 'NE':
		status = this._parseComparison(name, new EqualityMatchExpression(), 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.release()};
	case 'Equality':
		return this._parseComparison(name, new EqualityMatchExpression(), element);

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

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

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

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

	case 'opSIZE':
		var size = 0;
		if (element instanceof String)
			// matching old odd semantics
			size = 0;
		else if (element instanceof Number)
			size = element;
		else {
			return {code:ErrorCodes.BadValue, 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.release()};

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

	case 'opTYPE':
		if (!(element instanceof Number))
			return {code:ErrorCodes.BadValue, description:"$type has to be a number"};
		var type = element;
		temp = new TypeMatchExpression();
		status = temp.init(name, type);
		if (status.code != ErrorCodes.OK)
			return status;
		return {code:ErrorCodes.OK, result:temp.release()};

	case 'opMOD':
		return this._parseMOD(name, element);

	case 'opOPTIONS':
		// 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 < context.length; i++) {
			temp = context[i];
			if (temp.getGtLtOp(-1) == 'opREGEX')
				return {code:ErrorCodes.OK, result:null};
		}

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

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

	case 'opELEM_MATCH':
		return this._parseElemMatch(name, element);

	case 'opALL':
		return this._parseAll(name, element);

	case 'opWITHIN':
	case 'opGEO_INTERSECTS':
	case 'opNEAR':
		return this.expressionParserGeoCallback(name, x, context);

	} // end switch

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

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method _parseTreeList
 * @param
 *
 */
proto._parseTreeList = function _parseTreeList(arr, out){
	// File: expression_parser_tree.cpp lines: 33-52
	if (arr.length === 0)
		return {code:ErrorCodes.BadValue, 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.BadValue, 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};
};

/**
 *
 * This documentation was automatically generated. Please update when you touch this function.
 * @method parse
 * @param
 *
 */
proto.parse = function parse(obj){
	// File: expression_parser.h lines: 40-41
    return this._parse(obj, true);
};