**Embedded persistent or in memory database for Node.js, nw.js, Electron and browsers, 100% Javascript, no binary dependency**. API is a subset of MongoDB's and it's <ahref="#speed">plenty fast</a>.
**IMPORTANT NOTE**: Please don't submit issues for questions regarding your code. Only actual bugs or feature requests will be answered, all others will be closed without comment. Also, please follow the <ahref="#bug-reporting-guidelines">bug reporting guidelines</a> and check the <ahref="https://github.com/louischatriot/nedb/wiki/Change-log"target="_blank">change log</a> before submitting an already fixed bug :)
Money: [](https://www.paypal.com/cgi-bin/webscr?cmd=_donations&business=louis%2echatriot%40gmail%2ecom&lc=US¤cy_code=EUR&bn=PP%2dDonationsBF%3abtn_donate_LG%2egif%3aNonHostedGuest)
You can use NeDB as an in-memory only datastore or as a persistent datastore. One datastore is the equivalent of a MongoDB collection. The constructor is used as follows `new Datastore(options)` where `options` is an object with the following fields:
*`filename` (optional): path to the file where the data is persisted. If left blank, the datastore is automatically considered in-memory only. It cannot end with a `~` which is used in the temporary files NeDB uses to perform crash-safe writes
*`timestampData` (optional, defaults to false): timestamp the insertion and last update of all documents, with the fields `createdAt` and `updatedAt`. User-specified values override automatic generation, usually useful for testing.
*`onload` (optional): if you use autoloading, this is the handler called after the `loadDatabase`. It takes one `error` argument. If you use autoloading without specifying this handler, and an error happens during load, an error will be thrown.
*`afterSerialization` (optional): hook you can use to transform data after it was serialized and before it is written to disk. Can be used for example to encrypt data before writing database to disk. This function takes a string as parameter (one line of an NeDB data file) and outputs the transformed string, **which must absolutely not contain a `\n` character** (or data will be lost)
*`beforeDeserialization` (optional): reverse of `afterSerialization`. Make sure to include both and not just one or you risk data loss. For the same reason, make sure both functions are inverses of one another. Some failsafe mechanisms are in place to prevent data loss if you misuse the serialization hooks: NeDB checks that never one is declared without the other, and checks that they are reverse of one another by testing on random strings of various lengths. In addition, if too much data is detected as corrupt, NeDB will refuse to start as it could mean you're not using the deserialization hook corresponding to the serialization hook used before (see below)
*`corruptAlertThreshold` (optional): between 0 and 1, defaults to 10%. NeDB will refuse to start if more than this percentage of the datafile is corrupt. 0 means you don't tolerate any corruption, 1 means you don't care
*`nodeWebkitAppName` (optional, **DEPRECATED**): if you are using NeDB from whithin a Node Webkit app, specify its name (the same one you use in the `package.json`) in this field and the `filename` will be relative to the directory Node Webkit uses to store the rest of the application's data (local storage etc.). It works on Linux, OS X and Windows. Now that you can use `require('nw.gui').App.dataPath` in Node Webkit to get the path to the data directory for your application, you should not use this option anymore and it will be removed.
Under the hood, NeDB's persistence uses an append-only format, meaning that all updates and deletes actually result in lines added at the end of the datafile, for performance reasons. The database is automatically compacted (i.e. put back in the one-line-per-document format) everytime your application restarts.
You can manually call the compaction function with `yourDatabase.persistence.compactDatafile` which takes no argument. It queues a compaction of the datafile in the executor, to be executed sequentially after all pending operations.
You can also set automatic compaction at regular intervals with `yourDatabase.persistence.setAutocompactionInterval(interval)`, `interval` in milliseconds (a minimum of 5s is enforced), and stop automatic compaction with `yourDatabase.persistence.stopAutocompaction()`.
Keep in mind that compaction takes a bit of time (not too much: 130ms for 50k records on a typical development machine) and no other operation can happen when it does, so most projects actually don't need to use it.
Durability works similarly to major databases: compaction forces the OS to physically flush data to disk, while appends to the data file do not (the OS is responsible for flushing the data). That guarantees that a server crash can never cause complete data loss, while preserving performance. The worst that can happen is a crash between two syncs, causing a loss of all data between the two syncs. Usually syncs are 30 seconds appart so that's at most 30 seconds of data. <ahref="http://oldblog.antirez.com/post/redis-persistence-demystified.html"target="_blank">This post by Antirez on Redis persistence</a> explains this in more details, NeDB being very close to Redis AOF persistence with `appendfsync` option set to `no`.
If the document does not contain an `_id` field, NeDB will automatically generated one for you (a 16-characters alphanumerical string). The `_id` of a document, once set, cannot be modified.
You can also bulk-insert an array of documents. This operation is atomic, meaning that if one insert fails due to a unique constraint being violated, all changes are rolled back.
Use `find` to look for multiple documents matching you query, or `findOne` to look for one specific document. You can select documents based on field equality or use comparison operators (`$lt`, `$lte`, `$gt`, `$gte`, `$in`, `$nin`, `$ne`). You can also use logical operators `$or`, `$and`, `$not` and `$where`. See below for the syntax.
*`$regex`: checks whether a string is matched by the regular expression. Contrary to MongoDB, the use of `$options` with `$regex` is not supported, because it doesn't give you more power than regex flags. Basic queries are more readable so only use the `$regex` operator when you need to use another operator with it (see example below)
When a field in a document is an array, NeDB first tries to see if there is an array-specific comparison function (for now there is only `$size`) being used
and tries it first. If there isn't, the query is treated as a query on every element and there is a match if at least one element matches.
If you don't specify a callback to `find`, `findOne` or `count`, a `Cursor` object is returned. You can modify the cursor with `sort`, `skip` and `limit` and then execute it with `exec(callback)`.
You can give `find` and `findOne` an optional second argument, `projections`. The syntax is the same as MongoDB: `{ a: 1, b: 1 }` to return only the `a` and `b` fields, `{ a: 0, b: 0 }` to omit these two fields. You cannot use both modes at the time, except for `_id` which is by default always returned and which you can choose to omit.
*`update` specifies how the documents should be modified. It is either a new document or a set of modifiers (you cannot use both together, it doesn't make sense!)
* The modifiers create the fields they need to modify if they don't exist, and you can apply them to subdocs. Available field modifiers are `$set` to change a field's value, `$unset` to delete a field and `$inc` to increment a field's value. To work on arrays, you have `$push`, `$pop`, `$addToSet`, `$pull`, and the special `$each`. See examples below for the syntax.
*`upsert` (defaults to `false`) if you want to insert a new document corresponding to the `update` rules if your `query` doesn't match anything. If your `update` is a simple object with no modifiers, it is the inserted document. In the other case, the `query` is stripped from all operator recursively, and the `update` is applied to it.
NeDB supports indexing. It gives a very nice speed boost and can be used to enforce a unique constraint on a field. You can index any field, including fields in nested documents using the dot notation. For now, indexes are only used to speed up basic queries and queries using `$in`, `$lt`, `$lte`, `$gt` and `$gte`. The indexed values cannot be of type array of object.
To create an index, use `datastore.ensureIndex(options, cb)`, where callback is optional and get passed an error if any (usually a unique constraint that was violated). `ensureIndex` can be called when you want, even after some data was inserted, though it's best to call it at application startup. The options are:
* **fieldName** (required): name of the field to index. Use the dot notation to index a field in a nested document.
* **unique** (optional, defaults to `false`): enforce field uniqueness. Note that a unique index will raise an error if you try to index two documents for which the field is not defined.
* **sparse** (optional, defaults to `false`): don't index documents for which the field is not defined. Use this option along with "unique" if you want to accept multiple documents for which it is not defined.
You can remove a previously created index with `datastore.removeIndex(fieldName, cb)`.
If your datastore is persistent, the indexes you created are persisted in the datafile, when you load the database a second time they are automatically created for you. No need to remove any `ensureIndex` though, if it is called on a database that already has the index, nothing happens.
**Note:** the `ensureIndex` function creates the index synchronously, so it's best to use it at application startup. It's quite fast so it doesn't increase startup time much (35 ms for a collection containing 10,000 documents).
The browser version and its minified counterpart are in the `browser-version/out` directory. You only need to require `nedb.js` or `nedb.min.js` in your HTML file and the global object `Nedb` can be used right away, with the same API as the server version:
If you specify a `filename`, the database will be persistent, and automatically select the best storage method available (IndexedDB, WebSQL or localStorage) depending on the browser. In most cases that means a lot of data can be stored, typically in hundreds of MB. **WARNING**: the storage system changed between v1.3 and v1.4 and is NOT back-compatible! Your application needs to resync client-side when you upgrade NeDB.
NeDB is compatible with all major browsers: Chrome, Safari, Firefox, IE9+. Tests are in the `browser-version/test` directory (files `index.html` and `testPersistence.html`).
NeDB is not intended to be a replacement of large-scale databases such as MongoDB, and as such was not designed for speed. That said, it is still pretty fast on the expected datasets, especially if you use indexing. On a typical, not-so-fast dev machine, for a collection containing 10,000 documents, with indexing:
* If you mostly use NeDB for logging purposes and don't want the memory footprint of your application to grow too large, you can use <ahref="https://github.com/louischatriot/nedb-logger"target="_blank">NeDB Logger</a> to insert documents in a NeDB-readable database
* If you've outgrown NeDB, switching to MongoDB won't be too hard as it is the same API. Use <ahref="https://github.com/louischatriot/nedb-to-mongodb"target="_blank">this utility</a> to transfer the data from a NeDB database to a MongoDB collection
If you report a bug, thank you! That said for the process to be manageable please strictly adhere to the following guidelines. I'll not be able to handle bug reports that don't:
* Your bug report should be a self-containing gist complete with a package.json for any dependencies you need. I need to run through a simple `git clone gist; npm install; node bugreport.js`, nothing more.
* It should use assertions to showcase the expected vs actual behavior and be hysteresis-proof. It's quite simple in fact, see this example: https://gist.github.com/louischatriot/220cf6bd29c7de06a486
* Simplify as much as you can. Strip all your application-specific code. Most of the time you will see that there is no bug but an error in your code :)
* 50 lines max. If you need more, read the above point and rework your bug report. If you're **really** convinced you need more, please explain precisely in the issue.
