A tiny in-memory database with an tuple space inspired API aimed to be used at small to medium sized data sets.
yocto.js is a tiny in-memory database with an tuple space inspired API. In a way its a set of convinient methods to work on a list of objects.
In many cases we do operate on small to medium sized data sets and in the browser this is mostly the case. yocto.js is intended to be used on such data sets in the browser and provides an easy way to query data. yocto.js does also have methods for persisting and retreiving data to and from local storage.
yocto.js can also be used in node.js.
yocto.js operate on lists of objects. All objects put into the base is looked up by matching with a template object that contains the keys and values one want to match on.
Lets say we want to have a database of LP Records. This can be represented as a list of objects like this:
id: 0 album: 'High Visibility' artist: 'The Hellacopters' year: 2002 style: 'Punk rock'id: 1 album: 'Tender Is the Savage' artist: 'Gluecifer' year: 2000 style: 'Rock'id: 2 album: 'Deathcrush' artist: 'Mayhem' year: 1987 style: 'Black metal'id: 3 album: 'Supershitty to the Max!' artist: 'The Hellacopters' year: 1996 style: 'Punk rock'id: 4 album: 'Money for Soul' artist: 'Baby Woodrose' year: 2003 style 'Psychedelic rock'
A template object used to look up objects in the database can then look like this:
artist: 'The Hellacopters'year: 1996
If this template is passed on to one of the methods for finding objects in yocto.js, it will match with one object in the above example database. The method executing the match will then fire a callback method where the following list is provided to the callback:
id: 3 album: 'Supershitty to the Max!' artist: 'The Hellacopters' year: 1996 style: 'Punk rock'
The interface of yocto.js aims to be fluent so almost all methods can be chained. When using chaining each method in the chain work on the returned result from the previous method in the chain. The chain is terminated by executing a callback where one want to do something on the final result.
In this example, again working on the above dataset, we get all objects in the database where the artist is "The Hellacopters" and then we sort the result on the property "year" before we loop over each item:
dbgetartist: 'The Hellacopters'sort'year'eachconsole.logobjartist + ' did release ' + objalbum + ' in ' + objyeartoString;
Internally yocto.js tries to compose methods to reduce the number of loops it has to do in a chain. In other words; chaining, as an example, get() and each() will result in one internal loop over the data set where both get() and the callback to each() are executed in the loop. This is faster than going a get() and then passing the returned result to a separate loop.
Since a chain is terminated by a callback one can write the chain without dotting methods together. The above chain example can also be written like this:
dbgetartist: 'The Hellacopters';dbsort'year';dbeachconsole.logobjartist + ' did release ' + objalbum + ' in ' + objyeartoString;
The result is exactly the same as in the example above.
yocto.js has the following methods:
You create a database by invoking the db method:
var db = yoctodb;
All further methods does then operate on the assigned variable.
The db method can take a, optional, configuration object to configure certain behaviour in yocto.js.
var db = yoctodbfeature : 'value';
The config object has the following parameters:
- uuid - Defines one key in each object to be treated as a unique id for each object. This allows yocto.js have an ultra fast look-up of single objects. Please see the paragraph enitled "uuid" for further information.
- timestamp - Enables timestamping of each object put into the database. The value provided to this key will be the name of the key holding the timestamp in each object. Please see the paragraph enitled "Timestamps" for further information.
One can put a single object into the database by passing it as the first parameter to the put() method. The second parameter to the put() method is a callback where the value for the first argument to the callback is an array of the data put into the database.
Instead of putting just a single object into the database one can also put a whole array of objects:
Multiple calls to the put method will append multiple sets of data to the database.
It is only possible to put objects into the database. Any non object values will be filtered out and not be added to the database.
One can get data from the database by passing a template object as the first parameter to the get() method. The key/values given in the template object will be used to match with keys/values objects in the database. The second parameter to the get method is a callback where the value for the first argument to the callback is an array of the data the get method found in the database.
Takes data from the database by passing a template object as the first parameter to the take() method. The key/values given in the template object will be used to match with keys/values objects in the database. The second parameter to the take method is a callback where the value for the first argument to the callback is an array of the data the take method found in the database.
The take() method takes data out of the database. All matching objects of the take method is removed from the database. take() can also be looked upon as deleting data in the database.
Sorts a query result by passing which object parameter to sort on to the first parameter to the sort() method. The second parameter to the sort() method is a callback where the value for the first argument to the callback is an array of the sorted list of objects.
Loops over each object in a query result. The first argument to the each() method is a callback where the value for the first argument is the object in the iteration.
Saves the whole database or an array of objects to localstorage if localstorage is available in the run time. The first argument to the save() method is an config object for interacting with the storage. The second parameter to the save() method is a callback where the value for the first argument to the callback is an array of the data the save() method stored.
The config object has the following parameters:
- name - A string with the name the data should be stored under. Required.
- type - 'local' or 'session'. If 'local' is provided the data will be stored on the 'localStorage' object. If 'session' is provided, the data will be stored on the 'sessionStorage' object.
The save() method will save data as an stringified object in localstorage. The object looks like this:
creator : 'yocto'timestamp : numobjects :
'creator' is a reference to what wrote the object. 'timestamp' is number of milliseconds since the epoch indicating when the data was save and 'objects' is the list of objects stored.
Loads a dataset previously saved by the save() method from localstorage into memory if localstorage is available in the run time. The first argument to the load() method is an config object for interacting with the storage. The second parameter to the load() method is a callback where the value for the first argument to the callback is an array of the data the load() method loaded.
The config object has the following parameters:
- name - A string with the name of what name the data is stored under. Required.
- type - 'local' or 'session'. If 'local' is provided the data will be loaded from the 'localStorage' object. If 'session' is provided, the data will be loaded from the 'sessionStorage' object.
Loaded dataset will be appended to any existing data in memory.
Deletes all database records in memory. Persisted data are not deleted. The callback function is not passed any variables.
The drop() method is intended to be a fast way to delete all database records in memory. Therefor it does not pass any values to the callback.
In some cases one would like to do something upon all elements when the database is deleted. This can be achieved by passing an empty template to the take() method. The take() method will then return all records in the database and perform a delete on each record and return the record.
This will be a tad slower than drop().
Deletes all database records in memory and all persisted data in a named storage. The first argument to the destroy() method is an config object for interacting with the storage. The second parameter to the destroy() method is a callback. The callback function is not passed any variables.
Returns an object with a status about the database.
The status object has the following values:
- count - A numeric value of how many objects the database holds in memory.
- updated - A timestamp of when something was last put into the database.
To be implemented
To be implemented
All queries in yocto.js is done by passing objects that will match with what you want to look up. A template object can be as simple as this:
The above template will match all object in the database where the key "year" has the value "1996". Objects that does not has the key "year" will be ignored when traversing the database.
It is worth noticing that this is done by exact matching. That implies that the value given in the template must match exactly with the vaules in the objects. This is also case sensitive.
It is possible to do finer queries by providing a query function as the key value in templates. The query function must return "true" if the key evaluates as a match for what your quering and "false" if it does not.
The query function is passed two method arguments. The first argument is the value of the key which the query function operate on. The second parameter is an object with convenient "is" functions which can be used to evaluate values in the query function.
These "is" functions are:
- is.arr(val) - Check if a value is an Array.
- is.obj(val) - Check if a value is an Object.
- is.fn(val) - Check if a value is a Function.
- is.str(val) - Check if a value is a String.
- is.num(val) - Check if a value is an Number.
- is.empty(obj) - Check if a Array, Object or Sting is empty.
All checks return a Boolean value.
A template with a query function can look like this:
return value >= 1985 && value <= 1990;
The above template will match all objects where they value for the key "year" is between "1985" and "1990".
In many cases we only need to look up an object based on a single unique key value (an uuid). The performance of looking up a single object based on an uuid can be increased dramaticly by telling yocto.js what object key in our objects that holds an unique value.
We can tell yocto.js this by providing the key name for what we want to use as a uuid in our objects to the "uuid" key in the configuration that can be passed on to the yocto.db() method.
In other words, lets say that we are working with the objects shown under the paragraph "Working with yocto.js". They hold an "id" key which has an unique value for each object. By telling yocto.js which key that holds an unique value like this:
var db = yoctodbuuid : 'id';
a query like this will be dramaticly faster:
dbgetid : 3// do something;
Do note the following when using "uuids":
- If multiple objects has the same "uuid" the get() method will only return one object which is the last object put into the database. The take() method will in this case return all multiple objects in an array to the callback function.
- A faster lookup due to the precense of "uuid" only applies to the get() method.
By enabling timestamping in the configuration object passed on to the yocto.db(), method yocto.js can append a key holding a timestamp value for when a object are put into the database.
In other words, if we set up a database and put a object into the database like this:
var db = yoctodbtimestamp : 'tstamp';dbputfoo:'bar'// do something;
Then the inserted object will contain a 'tstamp' key when retrieved:
foo : 'bar' tstamp : 1356799270734
If an object already contains a key matching the configured timestamp key name, the existing key in the object will not be modified.
All timestamps created by yocto.js is number of milliseconds since the epoch.
To be made!
yocto.js utalize some new ES5 methods when working with lists and objects. These are not nesseserly pressent in older browsers. The good news is that these ES5 methods can be added by shims to older browsers.
yocto.js use the following ES5 methods:
yocto.js provide a separate es5.js file containing just these methods as a shim for older browsers. This shim extends the native types in the run time. If you need to support older browsers and don't already have a shim for the above methods, please include the es5.js shim file in your project also.
To see if you need to include the ES5 shims in your build, please see Kangax's ECMAScript 5 compatibility table.
yocto.js does also use native JSON.parse() and JSON.stringify() to persist data in the client. This is also absent in some older browsers but can also be shimmed in in the same way as the ES5 features. If you need to support older browsers and don't already have a shim for JSON.parse() and JSON.stringify() in your project, I recommend that you include JSON-js in your project.
This small library did spinn out of the fact that I saw myself doing a lot of simmilar tasks on lists in different applications. I've found myself fetching a lot of generic lists of objects from a server and then wanting to select subsets in them without doing a round trip to the server. I've also found myself using localstorage to store these lists and object in the client to reduce round trips to a server. So; this small library came out of small needs I had and I found the API used in a tuple space to kinda fit the operations I do.
There is probably plenty of room to improve both the API, the code, performance, tests, doc etc, etc so if you have any ideas and feedback on how to make this small library better, please feel free to create an issue, submit a pull request or ping me in any way :)
MIT License Copyright (c) Trygve Lie