Skip to main content


Description — Couchbase Lite Concepts — Data Model — Documents
Related Content — Databases | Blobs | Indexing


Document Structure

In Couchbase Lite the term document refers to an entry in the database; a record, or row in a table if you like.

Each document has an ID (primary key in other databases) by which it can be located. This ID can be automatically generated (as a UUID) or specified programmatically; the only constraints are that it must be unique within the database, and it can't be changed. The document also has a value which contains the actual application data. This value is stored as a dictionary collection of key-value (k-v) pairs where the values themselves may comprise different types of data such as numbers, strings, arrays or even nested objects.

Data Encoding

The document body is stored in an internal, efficient, binary form (Fleece). This internal form is easily converted into a manageable native dictionary format for manipulation in applications.

Fleece data is stored in the smallest format that will hold the value, whilst maintaining the integrity of the value.

Fleece Data Encoding

When working with Dart, the Fleece encoding cycle can result in the Dart type information being lost. Specifically, DateTimes are stored as strings. So, care should be taken when storing and recovering data in a document - or converting that document to JSON and back - using non-explicit functions such as DictionaryInterface.toPlainMap.

Always use explicit creation of the expected type, whenever the type of result is not itself explicit. For example:

final doc = (await collection.document(documentId))!;
final map = doc.toPlainMap();

final createdAtFromTypedGetter ='createdAt');
final createdAtFromMap = DateTime.parse(map['createdAt']! as String);

Data Types

The Document class offers a set of property accessors for various scalar types, including boolean, integers, floating-point and strings. These accessors take care of converting to/from JSON encoding, and make sure you get the type you're expecting.

So your document content may well comprise one or more supporting data types such as:

  • Boolean
  • Date
  • Double
  • Float
  • Integer
  • String

In addition to these basic data types Couchbase Lite provides for the following:

  • Dictionary — Represents a read-only key-value pair collection.
  • MutableDictionary — Represents a writeable key-value pair collection.
  • Array — Represents a readonly ordered collection of objects.
  • MutableArray — Represents a writeable collection of objects.
  • Blob — Represents an arbitrary piece of binary data.


Couchbase Lite also provides for the direct handling of JSON data implemented in most cases by the provision of a toJson() method on appropriate API classes (for example, on MutableDocument, Dictionary, Array and Blob) — see Working with JSON Data.

Constructing a Document

An individual document often represents a single instance of an object in application code. A document might be considered equivalent to a row in a relational table; with each of the document's attributes being equivalent to a column.

Documents can contain nested structures. This allows developers to express many-to-many relationships without requiring a reference or junction table; and is naturally expressive of hierarchical data.

Most apps will work with one or more documents, persisting them to a local database and optionally syncing them, either centrally or to the cloud.

In this section we provide an example of how you might create a hotel document, which provides basic contact details and price data.

Data Model

hotel: {
type: string (value = `hotel`)
name: string
address: dictionary {
street: string
city: string
state: string
country: string
code: string
phones: array
rate: float

Open a Database

First we open your database. If the database does not already exist, Couchbase Lite will create it for us.

Couchbase documents are assigned to a Collection. All the CRUD examples in this document operate on a collection object (here, the Default Collection).

// Get the database (and create it if it doesn't exist).
final database = await Database.openAsync('hoteldb', dbConfig);
final collection = await database.defaultCollection;

See: Databases for more information.

Create a Document

Now we create a new document to hold our application's data.

Because we will be adding data to the document we must use its mutable form.

// Create your new document.
final mutableDoc = MutableDocument.withId('hotel::1');

For more on using Documents, see: Document Initializers and Mutability.

Create a Dictionary

Here we create a dictionary (address). Because we want to add values into the dictionary, we must create it in mutable form.

When the dictionary is retrieved, each element's value is directly accessible via its own key.

// Create and populate mutable dictionary.
// Create a new mutable dictionary and populate some keys/values.
final address = MutableDictionary()
..setString('1 Main Street', key:'street')
..setString('San Francisco', key:'city')
..setString('CA.', key:'state')
..setString('USA', key:'country')
..setString('90210', key:'code');

For more on using Dictionaries see: Using Dictionaries.

Create an Array

Since our hotel may have multiple lines we provide an array (phones) to hold contact numbers. Again, because we want to add values into the array, we create it in mutable form.

// Create and populate mutable array.
final phones = MutableArray()

For more on using Arrays see: Using Arrays.

Populate a Document

Here we add our data to the mutable document we created earlier. Each data item is stored as a key-value pair.

// Initialize and populate the document

// <1> Add document type to document properties.
..setString('hotel', key: 'type')

// <2> Add hotel name string to document properties.
..setString('Hotel Dart Mo', key: 'name')

// <3> Add float to document properties.
..setFloat(121.75, key: 'room_rate')

// <4> Add dictionary to document's properties.
..setDictionary(address, key: 'address')

// <5> Add array to document's properties.
..setString(phones, key: 'phones');
  1. Add hotel name (string).
  2. Add average room rate (float).
  3. Add document type (string).
    Couchbase recommend using a type attribute to define each logical document type.
  4. Add address (dictionary).
    The address dictionary is added to the document and stored under the key address. We will use this to retrieve it when needed.
  5. Add phone numbers (array).
    The phones arrary is added to the document and stored under the key phones. We will use this to retrieve it when needed.

Save a Document

With the document now populated, we can persist to our Couchbase Lite database.

// Save the document to the database.
await collection.saveDocument(mutableDoc);

Close the Database

With our document saved, we can now close our Couchbase Lite database.

await database.close();

Working with Data

Checking a Document's Properties

To check whether a given property exists in the document, you should use the DictionaryInterface.contains method (Document implements DictionaryInterface).

If the property doesn't exist, the call will return the default for that that method (0 for DictionaryInterface.integer, 0.0 for DictionaryInterface.float, etc.).


Care should be taken when storing and recovering data in a document or converting that document to JSON and back.

Data encoding (Fleece) can result in Long values being converted to Float instead of Double. Interpreting data as boolean can also give inconsistent results.

Date accessors

As a convenience Couchbase Lite offers Date accessors. Dates are a common data type, but JSON doesn't natively support them, so the convention is to store them as strings in ISO-8601 format.

Example 1. Date Getter

This example sets the date on the createdAt property and reads it back using the accessor method.

mutableDoc.setValue(, key: 'createdAt');
final date ='createdAt');

Using Dictionaries

API References

Example 2. Read Only
final document = await collection.document('hotel::1');

// Get a dictionary from the document's properties.
final dictionary = document?.dictionary('address')

// Access a value with a key from the dictionary.
final street = dictionary?.string("street")

// Iterate the dictionary.
for (final key in dictionary!) {
print("Key $key = ${dictionary.value(key)}");

// Create a mutable copy of the dictionary.
final mutableDictionary = dictionary.toMutable();
Example 3. Mutable
// Create a new mutable dictionary and populate some keys/values.
final mutableDictionary = MutableDictionary()
..setString('1 Main Street', key: 'street')
..setString('San Francisco', key: 'city');

// Add the dictionary to a document's properties and save the document.
final mutableDocumemt = MutableDocument.withId('hotel::1')
..setDictionary(mutableDictionary, key: 'address');
await collection.saveDocument(mutableDocumemt);

Using Arrays

API References

Example 4. Read Only
final document = await collection.document('hotel::1');

// Get an array from the document's properties.
final array = document?.array('phones');

// Get the element count.
final count = array?.length;

// Access an array element by index.
final phone = array?.string(1);

// Iterate the array.
for (final element in array!) {
print('Element $element');

// Create a mutable copy of the array.
final mutableArray = array.toMutable();
Example 5. Mutable
// Create a new mutable array and populate it with data.
final mutableArray = MutableArray()

// Set the array to document's properties and save the document.
final mutableDocumemt = MutableDocument.withId('hotel::1')
..setArray(mutableArray, key: 'phones');
await collection.saveDocument(mutableDocumemt);

Using Blobs

For more on working with blobs — see Blobs.

Document Initializers

The following methods/initializers can be used:

The MutableDocument constructor can be used to create a new document where the document ID is randomly generated by the database.

The MutableDocument.withId constructor can be used to create a new document with a specific document ID.

The Database.document method can be used to get a document. If it doesn't exist in the database, it will return null. This method can be used to check if a document with a given ID already exists in the database.

Example 6. Persist a Document

The following code example creates a document and persists it to the database.

final document = MutableDocument()
..setString('task', key: 'type')
..setString('todo', key: 'owner')
..setDate(, key: 'createdAt');
await collection.saveDocument(document);


By default, when a document is read from the database it is immutable. The Document.toMutable method should be used to create an instance of the document which can be updated.

Example 7. Make a Mutable Document

Changes to the document are persisted to the database when the save method is called.

final document = await collection.document('xyz');
final mutableDocument = document!.toMutable()
..setString('new value', key: 'key');
await collection.saveDocument(mutableDocument);

Any user change to the value of reserved keys (_id, _rev or _deleted) will be detected when a document is saved and will result in an exception (DatabaseErrorCode.corruptRevisionData) — see also Document Constraints.

Document Conversion

A Document can be converted to a plain dictionary type and-or to a JSON string. This can often be useful to pass the document contents as a plain object to another method.

Example 8. Convert Document
/// Convert the document to a plain dictionary of type Map<String, Object?>.

/// Convert the document to a JSON string.

Batch operations

If you're making multiple changes to a database at once, it's faster to group them together. The following example persists a few documents in batch.

Example 8. Batch operations
Example 9. Batch Operations
await database.inBatch(() async {
for (final i in Iterable.generate(10)) {
final document = MutableDocument()
..setValue('user', key: 'type')
..setValue('user $i' key: 'name')
..setBoolean(false, key: 'admin');

await collection.saveDocument(document);

At the local level this operation is still transactional: no other Database instances, including ones managed by the replicator can make changes during the execution of the block, and other instances will not see partial changes. Couchbase Mobile is a distributed system, and due to the way replication works, there's no guarantee that Capella App Services or Sync Gateway will receive your changes all at once.

Document change events

It is possible to register for document changes. The following example registers for changes to the document with ID user.john and prints the verified_account property when a change is detected.

Example 10. Document Change Events
final token = collection.addDocumentChangeListener('user.john', (change) async {
final document = await collection.document(change.documentId);
if (document != null){
print('Status: ${document.string('verified_account')}');

Change Streams

Streams are a convenient alternative to listen for changes.


When multiple databases are involed, making sure that a stream is able to observe all changes requires waiting for the stream to be ready. See General Concepts - Change Streams for more information.

Example 11. Document Change Streams
final stream = collection.changes();

stream.listen((change) {
print('Ids of changed documents: ${change.documentIds}'):

To stop listening to changes just cancel the subscription, like with any other stream.

Document Expiration

Document expiration allows users to set the expiration date for a document. When the document expires, it is purged from the database. The purge is not replicated to Capella App Services or Sync Gateway.

Example 12. Set Document Expiration
// Purge the document one day from now.
final ttl = Duration(days: 1));
await collection.setDocumentExpiration('hotel::1', ttl);

// Reset the expiration.
await collection.setDocumentExpiration('hotel::1', null);

// Query documents that will be expired in less than five minutes.
final fiveMinutesFromNow = Duration(minutes: 5));
final query = const QueryBuilder()

Document Constraints

Couchbase Lite APIs do not explicitly disallow the use of attributes with the underscore prefix at the top level of document. This is to facilitate the creation of documents for use either in local only mode where documents are not synced.


_id, _rev and _sequence are reserved keywords and must not be used as top-level attributes — see Figure 13.

Users are cautioned that any attempt to sync such documents to Sync Gateway will result in an error. To be future proof, you are advised to avoid creating such documents. Use of these attributes for user-level data may result in undefined system behavior

For more guidance — see: Sync Gateway - Data Modeling Guidelines

Figure 13. Reserved Keys List
  • _attachments
  • _deleted
  • _id
  • _removed
  • _rev
  • _sequence

Working with JSON Data

The toJson typed-accessor means you can easily work with JSON data, native and Couchbase Lite objects.


Convert a Document to a JSON string using the toJson method — see Example 14.

Example 14. Documents as JSON Strings
final document = await collection.document('hotel::1');
if (document != null) {
final json = document.toJson();


Convert a Dictionary to a JSON string using the toJson method — see Example 15.

Example 15. Dictionaries as JSON Strings
final document = await collection.document('hotel::1');
if (document != null) {
final dictionary = document.dictionary('address');
if (dictionary != null) {
final json = dictionary.toJson();


Convert an Array to a JSON string using the toJson method — see Example 16.

Example 16. Arrays as JSON Strings
final document = await collection.document('hotel::1');
if (document != null) {
final array = document.array('phones');
if (array != null) {
final json = array.toJson();


Convert a Blob to JSON using the toJson method — see Example 17.

You can also check whether a given plain Map is a blob, or not, using Blob.isBlob — again, see Example 17.


The blob object must first be saved to the database (generating required metadata) before you can use the toJson method.

Example 17. Blobs as JSON Strings
final document = await collection.document('user.john');
if (document != null) {
final blob = document.blob('avatar');
if (blob != null) {
final json = blob.toJson();

Query Results

Convert a Result to JSON using its toJson method — see Example 18.

Example 18. Results as JSON Strings
final result = await query.execute();
final results = await result.allResults();
for (final result in results){
final json = result.toJson();

JSON String format

If your query selects ALL then the JSON format will be:

<collection-name>: {
"key1": "value1",
"keyx": "valuex"

If your query selects a sub-set of available properties then the JSON format will be:

"key1": "value1",
"keyx": "valuex"