Skip to content

Latest commit

 

History

History
665 lines (533 loc) · 18.6 KB

README.md

File metadata and controls

665 lines (533 loc) · 18.6 KB

Universe

Join the chat at https://gitter.im/crossfilter/universe

Build Status Code Climate

The easiest and fastest way to explore your data

Before Universe, exploring and filtering large datasets in javascript meant constant data looping, complicated indexing, and countless lines of code to dissect your data.

With Universe, you can be there in just a few lines of code. You've got better things to do than write intense map-reduce functions or learn the intricate inner-workings of Crossfilter ;)

Features

  • Simple, yet powerful query syntax
  • Built on, and tightly integrated with Crossfilter, and Reductio - the fastest multi-dimensional JS data frameworks available
  • Real-time updates to query results as you filter
  • Flexible filtering system
  • Automatic and invisible management of data indexing and memory
  • Post Aggregation

Features in the Pipeline

  • Query Joins
  • Query Macros
  • Sub Queries
  • To help contribute, join us at Join the chat at https://gitter.im/crossfilter/universe

Demos

Getting Started

Installation

NPM

npm install universe --save

CDN or Download from the npmcdn load or download universe.js or universe.min.js file as part of your application.

Create a new Universe

Pass universe an array of objects or a Crossfilter instance:

var universe = require('universe')

var myUniverse = universe([
    {date: "2011-11-14T16:17:54Z", quantity: 2, total: 190, tip: 100, type: "tab", productIDs: ["001"]},
    {date: "2011-11-14T16:20:19Z", quantity: 2, total: 190, tip: 100, type: "tab", productIDs: ["001",  "005"]},
    {date: "2011-11-14T16:28:54Z", quantity: 1, total: 300, tip: 200, type: "visa", productIDs: ["004", "005"]},
    ...
  ])
  .then(function(myUniverse){
    // And now you're ready to query! :)
    return myUniverse
  })

Query your data

.then(function(myUniverse){
  myUniverse.query({
    groupBy: 'type' // GroupBy the type key
    columns: {
      $count: true, // Count the number of records
      quantity: { // Create a custom 'quantity' column
        $sum: 'quantity' // Sum the quantity column
      },
    },
    // Limit selection to rows where quantity is greater than 50
    filter: {
      quantity: {
        $gt: 50
      }
    },
  })

  // Optionally post-aggregate your data
  // Reduce all results after 5 to a single result using sums
  myUniverse.squash(5, null, {
    count: '$sum',
    quantity: {
      sum: '$sum'
    }
  })

  // See Post-Aggregations for more information
})

Use your data

.then(function(res) {
  // Use your data for tables, charts, data visualiztion, etc.
  res.data === [
    {"key": "cash","value": {"count": 2,"quantity": {"sum": 3}}},
    {"key": "tab","value": {"count": 8,"quantity": {"sum": 16}}},
    {"key": "visa","value": {"count": 2,"quantity": {"sum": 2}}}
  ]

  // Or plost the data in DC.js using the underlying crossfilter dimension and group
  dc.pieChart('#chart')
    .dimension(res.dimension)
    .group(res.group)

  // Pass the query's universe instance to keep chaining
  return res.universe
})

Explore your data

As you filter your data on the universe level, every query's result is updated in real-time to reflect changes in aggregation

// Filter records where 'type' === 'visa'
.then(function(myUniverse) {
  return myUniverse.filter('type', 'visa')
})

// Filter records where 'type' === 'visa' or 'tab'
.then(function(myUniverse) {
  return myUniverse.filter('type', ['visa', 'tab'])
})

// Filter records where 'total' is between 50 and 100
.then(function(myUniverse) {
  return myUniverse.filter('total', [50, 10], true)
})

// Filter records using an expressive and JSON friendly query syntax
.then(function(myUniverse) {
  return myUniverse.filter('total', {
    $lt: { // Filter to results where total is less than
      '$get(total)': { // the "total" property from
        '$nthLast(3)': { // the 3rd to the last row from
          $column: 'date' // the dataset sorted by the date column
        }
      }
    }
  })
})

// Or if you're feeling powerful, just write your own custom filter function
.then(function(myUniverse){
  return myUniverse.filter({
    total: function(row){
      return (row.quantity * row.sum) > 50
    }
  })
})

// Clear the filters for the 'type' column
.then(function(myUniverse){
  return myUniverse.filter('type')
})

// Apply many filters in one go
.then(function(myUniverse){
  return myUniverse.filterAll([{
    column: 'type',
    value: 'visa',
  }, {
    column: 'quantity',
    value: [200, 500],
    isRange: true,
  }])
})

// Clear all of the filters
.then(function(myUniverse){
  return myUniverse.filterAll()
})

Clean Up

// Remove a column index
.then(function(myUniverse){
  return myUniverse.clear('total')
})

// Remove all columns
.then(function(myUniverse){
  return myUniverse.clear()
})

API #

universe( [data] , {config} ) #

  • Description

    • Creates a new universe instance
  • Parameters

    • [data] - An array of objects
    • {config} - Optional configurations for this Universe instance
      • {generatedColumns} - An object of keys and their respective accessor functions used to dynamically generate columns.
  • Returns a promise that is resolved with the universe instance

  • [Example](Create a new Universe)

    • Generated Columns Example
      universe([
        {day: '1', price: 30, quantity: 3},
        {day: '2', price: 40, quantity: 5}
      ], {
        generatedColumns: {
          total: function(row){return row.price * row.quantity}
        }
      })
        .then(function(myUniverse){
          // data[0].total === 90
          // data[1].total === 200
        })

.query( {queryObject} ) #

  • Description

    • Creates a new query from a universe instance
  • Parameters

    • queryObject:
      • groupBy - Property name, property string representation, or even a function! (see .column() method),
      • select - An object of column aggregations and/or column names
        • $aggregation - Aggregations are prefixed with a $
        • columnName - Creates a nested column with the name provided
      • filter - A filter object that is applied to the query (similar to a where clause in mySQL)
  • Returns

    • promise, resolved with a query results object
      • data - The result of the query
      • group - The crossfilter/reductio group used to build the query
      • dimension - The crossfilter dimension used to build the query
      • crossfilter - The crossfilter that runs this universe
      • universe - The current instance of the universe. Return this to keep chaining via promises
  • [Example](#Explore your data)

.filter( columnKey, filterObject, isArray, replace ) #

  • Description

    • Filters everything in the universe to only include rows that match certain conditions. Queries automatically and instantaneously update their values and aggregations.
  • Parameters

    • columnKey - The object property to filter on,
  • Returns

    • promise resolved with
      • universe instance
  • [Example](#Query your data)

.filterAll() #

  • Description
    • Clears all filters accross all dimensiona.
  • Returns
    • promise resolved with
      • universe instance

.column( columnKey/columnObject ) #

  • Description

    • Use to optionally pre-index a column. Accepts either:
      • String or number corresponding to the key or index of the column. eg. propertyName or 2
      • A nested string representation of the property. eg. a.nested.property, a.nested[number]
      • Multiple singular key shorthand eg. ['prop1', 'prop2', 'prop3']
      • A callback function that returns the key (very powerful) eg. function(d){return d.myProperty}
  • Parameters

    • columnKey - the column property or array index you would like to pre-compile eg.
      .then(function(universe){
        return universe.column('total')
      })
    • columnObject allows you to override the column type, otherwise it is calculated automatically:
      .then(function(universe){
        return universe.column({
          key: columnKey,
          type: 'number'
        })
      })
  • Returns

    • promise resolved with
      • universe instance
  • [Example](#Pre-compile Columns)

.clear( columnKey/columnObject/[columnKeys/columnObjects] ) #

  • Description

    • Clears individual or all column definitions and indexes
  • Parameters

    • columnKey - the column property or array of columns you would like to clear eg.
      .then(function(universe){
        // Single Key
        return universe.clear('total')
        // Complex Key
        return universe.clear({key: ['complex', 'key']})
        // Multiple Single Keys
        return universe.clear(['total', 'quantity'])
        // Multiple Complex Keys
        return universe.clear([{key: ['complex', 'key']}, {key: ['another', 'one']}])
      })
  • Returns

    • promise resolved with
      • universe instance
  • [Example](#Clean Up)

.add( [data] ) #

  • Description
    • Adds additional data to a universe instance. This data will be indexed, aggregated and queries/filters immediately updated when added.
  • Parameters
    • [data] - An new array of objects similar to the original dataset
  • Returns
    • promise resolved with
      • universe instance

Post Aggregation #

Post aggregation methods can be run on query results to further modify your data. Just like queries, the results magically and instantly respond to filtering.

  • Each post aggregation is very powerful, but not all post aggregations can be chained together.

Locking a query

A majority of the time, you're probably only interested in the end result of a query chain. For this reason, Post Aggregations default to mutating the data of their direct parent (unless the parent is the original query), thereby avoiding unnecessary copying of data. On the other hand, if you plan on accessing data at any point in the middle of a query chain, you will need to lock() that query's results. This ensure's it won't be overwritten or mutated by any further post aggregation.

Note: Running more than 1 post aggregation on a query will automatically lock the parent query.

.then(function(universe){
  return universe.query({
    groupBy: 'tag'
  })
})
.then(function(query){
  query.lock()
  var all = query.data
  return query.limit(5)
})
.then(function(query){
  var only5 = query.data

  all.length === 10
  only5.length === 5
})

Without locking the above query before using .limit(5), the all data array would have been mutated by .limit(5)

.sortByKey(descending) #

  • Description
    • Sort results by key (ascending or descending)
  • Parameters
    • descending - Pass true to sortKeys in descending order
      .then(function(query){  
        return query.sortByKey(true)
      })
  • Returns
    • promise resolved with
      • query instance

.limit(n, n2) #

  • Description
    • Limit results to those betweenn and n2. If n2 is not passed, will limit to the first n records
  • Parameters
    • n - Start index. Defaults to 0 if null or undefined,
    • n2 - End index. Defaults to query.data.length if null. If undefined, will limit to the first n records instead.
      .then(function(query){
        // limits results to the first 5 records
        return query.limit(5)
        // limits results to records 5 through 10
        return query.limit(4, 10)
      })
  • Returns
    • promise resolved with
      • query instance

.squash(n, n2, aggregationMap, keyName) #

  • Description

    • Takes records from n to n2 and reduces them to a single record using the aggregationMap
  • Parameters

    • n - Start index. Defaults to 0 if false-y
    • n2 - End index. Defaults to query.data.length if false-y
    • aggregationMap - A 1:1 map of property to the aggregation to be used when combining the records
    • keyName (optional) - The key to be used for the new record. Defaults to Other
    .then(function(universe){
      universe.query({
      groupBy: 'type',
      select: {
        $sum: 'total',
        otherColumn: {
          $avg: 'tip'
        }
      })
    })
    .then(function(query){
      // Will squash all records after the 5 record
      query.squash(5, null, {
        // Sum the sum column
        sum: '$sum',
        othercolumn: {
          // Average the avg column
          avg: '$avg'
        }
      }, 'Everything after 5')
      // Give the squashed record a new key
    })
  • Returns

    • promise resolved with
      • query instance

.change(n, n2, changeFields) #

  • Description

    • Determines the change from the n to n2 using the keys in changeFields
  • Parameters

    • n - Start index. Defaults to 0 if false-y
    • n2 - End index. Defaults to query.data.length if false-y
    • changeFields - An object or array, referencing the fields to measure for change
    .then(function(universe){
      universe.query({
        groupBy: 'type',
        select: {
          $sum: 'total',
          otherColumn: {
            $avg: 'tip'
          }
        }
      })
    })
    .then(function(query){
      // Measure the change for sum and avg from result 0 to 10
      query.change(0, 10, {
        sum: true
        otherColumn: {
          avg: true
        }
      })
    })
  • Returns

    • promise resolved with
      • query instance
        • query.data is now an object:
        {
          key: ['nKey', 'n2Key'],
          value: {
            sumChange: 7,
            otherColumn: {
              avgChange: 4
            }
          }
        }

.changeMap(changeMapObj) #

  • Description

    • Determines incremental change for each record across the fields defined in changeMapObj
  • Parameters

    • changeMapObj - An object or array, referencing the fields to measure for change
    .then(function(universe){
      universe.query({
        groupBy: 'type',
        select: {
          $sum: 'total',
          otherColumn: {
            $avg: 'tip'
          }
        }
      })
    })
    .then(function(query){
      // Measure the change for sum and avg from result 0 to 10
      query.change({
        sum: true
        otherColumn: {
          avg: true
        }
      })
    })
  • Returns

    • promise resolved with
      • query instance
        • query.data records are now decorated with incremental change data:
        [...{
          key: 'tag5'
          value: {
            sum: 5
            sumChange: 7,
            sumChangeFromStart: 0,
            sumChangeFromEnd: 30,
            otherColumn: {
              avgChange: 4
              avgChangeFromStart: -4
              avgChangeFromEnd: -20
            }
          }
        }...]

.post(callback) #

  • Description
    • Use a custom callback function to perform your own post aggregations.
  • Parameters
    • callback - the callback function to execute. It accepts the following parameters:
      • query - the new query object. A fresh reference (or copy, if the parent is locked) is located at query.data. It is highly discouraged to change any other property on this object
      • parentQuery - the parent query.
    • You may optionally return a promise-like value for asynchronous processing
      .post(function(query, parentQuery){
        query.data[0].key = 'newKeyName'
        return Promise.resolve(doSomethingSpecial(query.data))
      })
  • Returns
    • promise resolved with
      • query instance

Pro Tips #

No Arrays Necessary

Don’t want to use arrays in your aggregations? No problem, because this:

.then(function(universe){
  universe.query({
    select: {
      $sum: {
        $sum: [
          {$max: ['tip', 'total']},
          {$min: ['quantity', 'total']}
        ]
      },
    }
  })
})

… is now easier written like this:

.then(function(universe){
  universe.query({
    select: {
      $sum: {
        $sum: {
          $max: ['tip', 'total'],
          $min: ['quantity', 'total']
        }
      },
    }
  })
})

No Objects Necessary, either!

What’s that? Don’t like the verbosity of objects or arrays? Use the new string syntax!

.then(function(universe){
  universe.query({
    select: {
      $sum: '$sum($max(tip,total), $min(quantity,total))'
    }
  })
})

Pre-compile Columns

Pro-Tip: You can also pre-compile column indices before querying. Otherwise, ad-hoc indices are created and managed automagically for you anyway.

.then(function(myUniverse){
  return myUniverse.column('a')
  return myUniverse.column(['a', 'b', 'c'])
  return myUniverse.column({
    key: 'd',
    type: 'string' // override automatic type detection
  })
})