json-easy-filter

Javascript node module for programmatic filtering and validation of Json objects.

Installation

$ npm install json-easy-filter

Usage

plunkr

var JefNode = require('json-easy-filter').JefNode;
 
var obj = {
        v1: 100,
        v2: 'v2',
        v3: {
                v4: 'v4',
                v5: 400
        }
};
var numbers = new JefNode(obj).filter(function(node) {
        if (node.type()==='number') {
            return node.key + ' ' + node.value;
        }
    });
 
console.log(numbers);
>> [ 'v1 100', 'v5 400' ]

How it works

Any newly instantiated JefNode object is actually a structure wrapping the real Json object so that for each Json node there will be a corresponding JefNode. The purpose of this structure is to allow easy tree navigation. Each JefNode maintains properties such as 'parent' which returns the ancestor or get(path) which returns a child based on its relative path. In fact 'new JefNode(obj)' returns the root JefNode which is further used to filter(), validate() or remove().

A word on performance

It is obvious already that json-easy-filter is designed more towards convenience rather than being performance wise. Particularly using it on server side or feeding large files may pose a problem for high request rate apps. If this is the case, Jef exposes its own internal traversal mechanism or you may try one of the similar projects presented in links section.

Filter, validate, remove

Tree traversal is provided by JefNode.filter(callback) . It will recursively iterate each node and trigger the callback method which receives the currently traveled JefNode. Use node.value and node.key to get access to the real json object. Use parent, path and get() to navigate the tree. Use isRoot, isLeaf, isCircular for information about current node. level provides the traversal depth.

IMPORTANT - Do not change Json object during filter() call. Keep a separate list of changes and apply it after filter has finished. For convenience, remove() will iterate the tree and delete nodes passed back by the callback. Following example will structure of 'text' node.

var obj = {text : 't'};
var modif = [];
var res = new JefNode(obj).filter(function(node) {
    if (node.has('text')){
        modif.push({
            parent: node.value, 
            newVal: {'new':'val'}})
    }
});
for (var i = 0; i < modif.length; i++) {
    var elem = modif[i];
    elem.parent.text = elem.newVal; 
}
console.log(JSON.stringify(obj, null, 2));
>>
{
  "text": {
    "new": "val"
  }
}
 

Aside from filter and remove, there is also a validate() method. Returning false from callback will cause the whole validation to fail.

Check out the examples and API for more info.

Examples

Use the sample data to follow this section.

Filter

#1. node.has() plunkr

var res = new JefNode(sample1).filter(function(node) {
    if (node.has('username')) {
        return node.value.username;
    }
});
console.log(res);
 
>> [ 'john', 'adams', 'lee', 'scott', null ] 

#2. node.value plunkr

var res = new JefNode(sample1).filter(function(node) {
    if (node.has('salary') && node.value.salary > 200) {
        return node.value.username + ' ' + node.value.salary;
    }
});
console.log(res);
>> [ 'lee 300', 'scott 400' ] 

#3. Paths, node.has(RegExp), level plunkr

var res = new JefNode(sample1).filter(function(node){
    if(node.has(/^(phone|email|city)$/)){
        return 'contact: '+node.path;
    }
    if(node.pathArray[0]==='departments' && node.pathArray[1]==='admin' && node.level===3){
        return 'department '+node.key+': '+node.value;
    }
});
console.log(res);
>> 
[ 'department name: Administrative',
  'department manager: john',
  'department employees: john,lee',
  'contact: employees.0.contact.0',
  'contact: employees.0.contact.1',
  'contact: employees.0.contact.2.address' ]

When has(propertyName) receives a string it calls node.value[propertyName]. If RegExp is passed, all properties of node.value are iterated and tested against it.

#4. node.key, node.parent and node.get() plunkr

var res = new JefNode(sample1).filter(function(node){
    if(node.key==='email' && node.value==='a@b.c'){
        var res = [];
        res.push('Email: key - '+node.key+', value: '+node.value+', path: '+node.path);
 
        if(node.parent){ // Test parent exists
            var emailContainer = node.parent;
            res.push('Email parent: key - '+emailContainer.key+', type: '+emailContainer.type()+', path: '+emailContainer.path);
        }
 
        if(node.parent && node.parent.parent){
            var contact = node.parent.parent;
            res.push('Contact: key - '+contact.key+', type: '+contact.type()+', path: '+contact.path);
 
            var city = contact.get('2.address.city');
            if(city){ // Test relative path exists. node.get() returns 'undefined' otherwise.
                res.push('City: key - '+city.key+', type: '+city.value+', path: '+city.path);
            }
        }
 
        return res;
    }
});
console.log(res);
>>
[ [ 'Email: key - email, value: a@b.c, path: employees.0.contact.1.email',
    'Email parent: key - 1, type: object, path: employees.0.contact.1',
    'Contact: key - contact, type: array, path: employees.0.contact',
    'City: key - city, type: NY, path: employees.0.contact.2.address.city' ] ]

#5. Array handling plunkr

var res = new JefNode(sample1).filter(function(node){
    if(node.parent && node.parent.key==='employees'){
        if(node.type()==='object'){
            return 'key: '+node.key+', username: '+node.value.username+', path: '+node.path;
        } else{
            return 'key: '+node.key+', username: '+node.value+', path: '+node.path;
        }
    }
});
console.log(res);
>>
[ 'key: 0, username: john, path: departments.admin.employees.0',
  'key: 1, username: lee, path: departments.admin.employees.1',
  'key: 0, username: scott, path: departments.it.employees.0',
  'key: 1, username: john, path: departments.it.employees.1',
  'key: 2, username: lewis, path: departments.it.employees.2',
  'key: 0, username: adams, path: departments.finance.employees.0',
  'key: 1, username: scott, path: departments.finance.employees.1',
  'key: 2, username: lee, path: departments.finance.employees.2',
  'key: 0, username: john, path: employees.0',
  'key: 1, username: adams, path: employees.1',
  'key: 2, username: lee, path: employees.2',
  'key: 3, username: scott, path: employees.3',
  'key: 4, username: null, path: employees.4',
  'key: 5, username: undefined, path: employees.5' ]

#6. Circular references plunkr

var data = {
    x: {
        y: null  
    },
    z: null,
    t: null
};
data.z = data.x;
data.x.y = data.z;
data.t = data.z;
var res = new JefNode(data).filter(function(node) {
    if(node.isRoot){
        return 'root';
    } else if (node.isCircular) {
        return 'circular key: '+node.key + ', path: '+node.path;
    } else{
        return 'key: '+node.key + ', path: '+node.path;
    }
});
console.log(res);
>>
[   "root",
    "key: x, path: x",
    "circular key: y, path: x.y",
    "circular key: z, path: z",
    "circular key: t, path: t" ]

Validate

#1. node.validate() plunkr

var res = new JefNode(sample1).validate(function(node) {
    if (node.parent && node.parent.key==='departments' && !node.has('manager')) {
        // current department is missing the mandatory 'manager' property
        return false;
    }
});
console.log(res);
>> false

#2. Validation info plunkr

var info = [];
var res = new JefNode(sample1).validate(function(node) {
var valid = true;
if (node.parent && node.parent.key==='departments' ) {
    // Inside department
    if(!node.has('manager')){
        valid = false;
        info.push('Error: '+node.key+' department is missing mandatory manager property');
    }
    if(!node.has('employees')){
        valid = false;
        info.push('Error: '+node.key+' department is missing mandatory employee list');
    } else if(node.get('employees').type()!=='array'){
        valid = false;
        info.push('Error: '+node.key+' department has wrong employee list type "'+node.get('employees').type()+'"');
    } else if(node.value.employees.length===0){
        info.push('Warning: '+node.key+' department has no employees');
    }
}
if (node.parent && node.parent.key==='employees' && node.type()==='object') {
    // Inside employee
    if(!node.has('username') || node.get('username').type()!=='string'){
        valid = false;
        info.push('Error: Employee '+node.path+' does not have username');
    } else if(!node.has('gender')){
        info.push('Warning: Employee '+node.value.username+' does not have gender');
    }
}
 
return valid;
});
console.log(res.toString());
console.log(info);
>>
false
[ 'Error: marketing department is missing mandatory manager property',
  'Warning: marketing department has no employees',
  'Error: hr department is missing mandatory manager property',
  'Error: hr department is missing mandatory employee list',
  'Error: supply department is missing mandatory manager property',
  'Error: supply department has wrong employee list type "string"',
  'Warning: Employee scott does not have gender',
  'Error: Employee employees.4 does not have username',
  'Error: Employee employees.5 does not have username' ]

#3. Sub validator plunkr

var info = [];
var res = new JefNode(sample1).get('departments').validate(function (node, local) {
    var valid = true;
    if (local.level === 1) {
        // Inside department
        if (!node.has('manager')) {
            valid = false;
            info.push('Error: ' + local.path + '(' + node.path + ')' + ' department is missing mandatory manager property');
        }
    }
    return valid;
});
console.log(res);
console.log(info);
>>
false
[ 'Error: marketing(departments.marketing) department is missing mandatory manager property',
  'Error: hr(departments.hr) department is missing mandatory manager property',
  'Error: supply(departments.supply) department is missing mandatory manager property' ]

Remove

Instead of using filter() for deleting certain nodes, remove() makes it easy by just requiring to return the nodes to be deleted from the callback.

plunkr

var sample = JSON.parse(JSON.stringify(sample1));
var success = new JefNode(sample).remove(function(node) {
    if(node.parent && node.parent.key==='departments'){
        var isITDepartment = node.has('name') && node.value.name==='IT'; 
        if(isITDepartment){
            // remove manager and first employee from IT department.
            return [node.get('manager'), node.get('employees.0')] ;
        } else{
            // remove all but IT department
            return node;
        }
    }
    if(node.parent && node.parent.key==='employees' && node.type()==='object'){
        if(node.has('salary') && node.get('salary').type()==='number' && node.value.salary<400){
            return node;
        }
    }
});
console.log(JSON.stringify(sample, null, 4));
console.log(success);
>> 
{
    "departments": {
        "it": {
            "name": "IT",
            "employees": [
                "john",
                "lewis"
            ]
        }
    },
    "employees": [
        {
            "username": "scott",
            "firstName": "Scott",
            "lastName": "SCOTT",
            "salary": 400,
            "birthDate": "1993/11/20"
        },
        {
            "firstName": "Unknown2",
            "lastName": "Unknown2"
        }
    ]
}
true

Traverse

Internal Json traversal mechanism is exposed for cases where performance is an issue. plunkr

var traverse = require('json-easy-filter').traverse;
var res = [];
traverse(sample1, function (key, val, path, parentKey, parentVal, level, isRoot, isLeaf, isCircular) {
    debugger;
    if (parentKey && parentKey === 'departments') {
        // inside department
        res.push('key: ' + key + ', val: ' + val.name + ', path: ' + path);
    }
})
console.log(res);
 
>> [  'key: admin, val: Administrative, path: departments,admin',
      'key: it, val: IT, path: departments,it',
      'key: finance, val: Financiar, path: departments,finance',
      'key: marketing, val: Commercial, path: departments,marketing',
      'key: hr, val: Human resources, path: departments,hr',
      'key: supply, val: undefined, path: departments,supply' ]

Refresh

refresh() is used to update Jef internal structure when structure of wrapped json changes.

var root = new JefNode(obj);
var res = root.filter(function(node) {
    if (node.key==='text1'){
        return node.value;
    }
});
console.log(res);
 
obj.text1 = {'new': 'val'};
root.refresh();
res = root.filter(function(node) {
    if (node.key==='text1'){
        return node.value;
    }
});
console.log(res);
 
>>
[ 't1' ]
[ { new: 'val' } ]

Tests

Make sure it's all working with 'npm test'. The awesome istanbul tool provides code coverage.

API

JefNode class

• node.key - node's key. For root object it is undefined.
• node.value - the real Json value behind node.
• node.parent - node's parent. Root's parent points to itself so that node.parent is never undefined.
• node.isRoot - true if current node is the root of the object tree.
• node.pathArray - string array containing the path to current node.
• node.path - string representation of node.pathArray.
• node.root - root JefNode.
• node.level - level of the current node. Root node has level 0.
• node.isLeaf - true if it is a leaf node. Primitives are considered leafs, empty objects (ie. a: { }) are not.
• node.isCircular - indicates a circular reference
• node.count - number of first level child nodes. For array indicates nuber of elements.
• node.has(propertyName) - returns true if node.value has that property. If a regular expression is passed, all node.value property names are iterated and matched against pattern.
• node.get(relativePath) - returns the JefNode relative to current node or 'undefined' if path cannot be found.
• node.type() - returns the type of node.value as one of 'string', 'array', 'object', 'function', 'number', 'boolean', 'undefined', 'null'.
• node.hasType(types) - compares against multiple types - node.hasType('number', 'object') returns true if node is either of the two types.
• node.isEmpty() - returns true if this object/array has no children/elements.
• node.filter(callback) - traverses node's children and triggers callback(childNode, localContext). The result of callback call is added to an array which is later returned by filter method. When filter method is called for a node other than root, localContext holds info relative to that node. If it is called for root, there is no reason to use localContext. See JefLocalContext class below.
• node.filterFirst(callback) - use this to traverse the first level (direct children) of node.
• node.filterLevel(level, callback) - iterates only nodes at specified level.
• node.validate(callback) - traverses node's children and triggers callback(childNode, localContext). If any of the calls to callback method returns false, validate method will also return false. localContext is treated the same as for filter method.
• node.remove(callback) - traverses node's children and triggers callback(childNode, localContext). Callback method is expected to return the nodes to be deleted. Either a JefNode or an array of JefNode objects may be returned. After traversal is complete the nodes are removed from Js tree. The root object is never deleted.
• node.refresh() - call this to update Jef object after any of node's content have been created/updated/deleted. Shall not be used inside node.filter(), node.validate(), node.remove().

JefLocalContext class

• localContext.isRoot - true if current node is the one that started filter/validate/remove operation.
• localContext.pathArray - string array containing the path to this node relative to current filter/validate/remove operation.
• localContext.path - string representation of localContext.pathArray.
• localContext.level - level of this node relative to current filter/validate operation.
• localContext.root - node that started filter/validate/remove operation.

Changelog

v0.3.0

• exposed internal traverse() mechanism. Instead of require('json-easy-filter') use either require('json-easy-filter').JefNode or require('json-easy-filter').traverse.
• node.getType() is deprecated in favour of node.type()
• addedd node.remove()
• node.isLeaf behavour no longer works as in 0.3.0. See API.
• removed dependecy on traverse
• added node.count, node.isEmpty(), node.root, filterFirst(), filterLevel()
• added node.refresh() to support json content modification
• bug fixes

Links

• XPath like query for json - JsonPath, SpahQL
• Filter, map, reduce - traverse
• Json validator - json-filter, json-validator
• Linq - jLinq, jslinq