Returns a sorted list of changes made to documents in the database, in time order of application, can be obtained from the database’s _changes resource. Only the most recent change for a given document is guaranteed to be provided, for example if a document has had fields added, and then deleted, an API client checking for changes will not necessarily receive the intermediate state of added documents.
This can be used to listen for update and modifications to the database for post processing or synchronization, and for practical purposes, a continuously connected _changes feed is a reasonable approach for generating a real-time log for most applications.
Parameters: |
|
---|---|
Request Headers: | |
|
|
Query Parameters: | |
|
|
Response Headers: | |
|
|
Response JSON Object: | |
|
|
Status Codes: |
|
The result field of database changes
JSON Object: |
|
---|
Request:
GET /db/_changes?style=all_docs HTTP/1.1
Accept: application/json
Host: localhost:5984
Response:
HTTP/1.1 200 OK
Cache-Control: must-revalidate
Content-Type: application/json
Date: Mon, 12 Aug 2013 00:54:58 GMT
ETag: "6ASLEKEMSRABT0O5XY9UPO9Z"
Server: CouchDB (Erlang/OTP)
Transfer-Encoding: chunked
{
"last_seq": 11,
"results": [
{
"changes": [
{
"rev": "2-7051cbe5c8faecd085a3fa619e6e6337"
}
],
"id": "6478c2ae800dfc387396d14e1fc39626",
"seq": 6
},
{
"changes": [
{
"rev": "3-7379b9e515b161226c6559d90c4dc49f"
}
],
"deleted": true,
"id": "5bbc9ca465f1b0fcd62362168a7c8831",
"seq": 9
},
{
"changes": [
{
"rev": "6-460637e73a6288cb24d532bf91f32969"
},
{
"rev": "5-eeaa298781f60b7bcae0c91bdedd1b87"
}
],
"id": "729eb57437745e506b333068fff665ae",
"seq": 11
}
]
}
Changed in version 0.11.0: added include_docs parameter
Changed in version 1.2.0: added view parameter and special value _view for filter one
Changed in version 1.3.0: since parameter could take now value to start listen changes since current seq number.
Changed in version 1.3.0: eventsource feed type added.
Changed in version 1.4.0: Support Last-Event-ID header.
Changed in version 1.6.0: added attachments and att_encoding_info parameters
Warning
Using the attachments parameter to include attachments in the changes feed is not recommended for large attachment sizes. Also note that the Base64-encoding that is used leads to a 33% overhead (i.e. one third) in transfer size for attachments.
Requests the database changes feed in the same way as GET /{db}/_changes does, but is widely used with ?filter=_doc_ids query parameter and allows one to pass a larger list of document IDs to filter.
Request:
POST /recipes/_changes?filter=_doc_ids HTTP/1.1
Accept: application/json
Content-Length: 40
Content-Type: application/json
Host: localhost:5984
{
"doc_ids": [
"SpaghettiWithMeatballs"
]
}
Response:
HTTP/1.1 200 OK
Cache-Control: must-revalidate
Content-Type: application/json
Date: Sat, 28 Sep 2013 07:23:09 GMT
ETag: "ARIHFWL3I7PIS0SPVTFU6TLR2"
Server: CouchDB (Erlang OTP)
Transfer-Encoding: chunked
{
"last_seq": 38,
"results": [
{
"changes": [
{
"rev": "13-bcb9d6388b60fd1e960d9ec4e8e3f29e"
}
],
"id": "SpaghettiWithMeatballs",
"seq": 38
}
]
}
By default all changes are immediately returned within the JSON body:
GET /somedatabase/_changes HTTP/1.1
{"results":[
{"seq":1,"id":"fresh","changes":[{"rev":"1-967a00dff5e02add41819138abb3284d"}]},
{"seq":3,"id":"updated","changes":[{"rev":"2-7051cbe5c8faecd085a3fa619e6e6337"}]},
{"seq":5,"id":"deleted","changes":[{"rev":"2-eec205a9d413992850a6e32678485900"}],"deleted":true}
],
"last_seq":5}
results is the list of changes in sequential order. New and changed documents only differ in the value of the rev; deleted documents include the "deleted": true attribute. (In the style=all_docs mode, deleted applies only to the current/winning revision. The other revisions listed might be deleted even if there is no deleted property; you have to GET them individually to make sure.)
last_seq is the sequence number of the last update returned. (Currently it will always be the same as the seq of the last item in results.)
Sending a since param in the query string skips all changes up to and including the given sequence number:
GET /somedatabase/_changes?since=3 HTTP/1.1
The return structure for normal and longpoll modes is a JSON array of changes objects, and the last update sequence number.
In the return format for continuous mode, the server sends a CRLF (carriage-return, linefeed) delimited line for each change. Each line contains the JSON object described above.
You can also request the full contents of each document change (instead of just the change notification) by using the include_docs parameter.
{
"last_seq": 5
"results": [
{
"changes": [
{
"rev": "2-eec205a9d413992850a6e32678485900"
}
],
"deleted": true,
"id": "deleted",
"seq": 5,
}
]
}
The longpoll feed, probably most applicable for a browser, is a more efficient form of polling that waits for a change to occur before the response is sent. longpoll avoids the need to frequently poll CouchDB to discover nothing has changed!
The request to the server will remain open until a change is made on the database and is subsequently transferred, and then the connection will close. This is low load for both server and client.
The response is basically the same JSON as is sent for the normal feed.
Because the wait for a change can be significant you can set a timeout before the connection is automatically closed (the timeout argument). You can also set a heartbeat interval (using the heartbeat query argument), which sends a newline to keep the connection active.
Continually polling the CouchDB server is not ideal - setting up new HTTP connections just to tell the client that nothing happened puts unnecessary strain on CouchDB.
A continuous feed stays open and connected to the database until explicitly closed and changes are sent to the client as they happen, i.e. in near real-time.
As with the longpoll feed type you can set both the timeout and heartbeat intervals to ensure that the connection is kept open for new changes and updates.
The continuous feed’s response is a little different than the other feed types to simplify the job of the client - each line of the response is either empty or a JSON object representing a single change, as found in the normal feed’s results.
GET /somedatabase/_changes?feed=continuous HTTP/1.1
{"seq":1,"id":"fresh","changes":[{"rev":"1-967a00dff5e02add41819138abb3284d"}]}
{"seq":3,"id":"updated","changes":[{"rev":"2-7051cbe5c8faecd085a3fa619e6e6337"}]}
{"seq":5,"id":"deleted","changes":[{"rev":"2-eec205a9d413992850a6e32678485900"}],"deleted":true}
... tum tee tum ...
{"seq":6,"id":"updated","changes":[{"rev":"3-825cb35de44c433bfb2df415563a19de"}]}
Obviously, ... tum tee tum ... does not appear in the actual response, but represents a long pause before the change with seq 6 occurred.
The eventsource feed provides push notifications that can be consumed in the form of DOM events in the browser. Refer to the W3C eventsource specification for further details. CouchDB also honours the Last-Event-ID parameter.
GET /somedatabase/_changes?feed=eventsource HTTP/1.1
// define the event handling function
if (window.EventSource) {
var source = new EventSource("/somedatabase/_changes?feed=eventsource");
source.onerror = function(e) {
alert('EventSource failed.');
};
var results = [];
var sourceListener = function(e) {
var data = JSON.parse(e.data);
results.push(data);
};
// start listening for events
source.addEventListener('message', sourceListener, false);
// stop listening for events
source.removeEventListener('message', sourceListener, false);
}
Note
EventSource connections are subject to cross-origin resource sharing restrictions. You might need to configure CORS support to get the EventSource to work in your application.
You can filter the contents of the changes feed in a number of ways. The most basic way is to specify one or more document IDs to the query. This causes the returned structure value to only contain changes for the specified IDs. Note that the value of this query argument should be a JSON formatted array.
You can also filter the _changes feed by defining a filter function within a design document. The specification for the filter is the same as for replication filters. You specify the name of the filter function to the filter parameter, specifying the design document name and filter name. For example:
GET /db/_changes?filter=design_doc/filtername
Additionally, there are couple of builtin filters are available and described below.
This filter accepts only changes for documents which ID in specified in doc_ids query parameter or payload’s object array. See POST /{db}/_changes for an example.
The _design filter accepts only changes for any design document within the requested database.
Request:
GET /recipes/_changes?filter=_design HTTP/1.1
Accept: application/json
Host: localhost:5984
Response:
HTTP/1.1 200 OK
Cache-Control: must-revalidate
Content-Type: application/json
Date: Sat, 28 Sep 2013 07:28:28 GMT
ETag: "ARIHFWL3I7PIS0SPVTFU6TLR2"
Server: CouchDB (Erlang OTP)
Transfer-Encoding: chunked
{
"last_seq": 38,
"results": [
{
"changes": [
{
"rev": "10-304cae84fd862832ea9814f02920d4b2"
}
],
"id": "_design/ingredients",
"seq": 29
},
{
"changes": [
{
"rev": "123-6f7c1b7c97a9e4f0d22bdf130e8fd817"
}
],
"deleted": true,
"id": "_design/cookbook",
"seq": 35
},
{
"changes": [
{
"rev": "6-5b8a52c22580e922e792047cff3618f3"
}
],
"deleted": true,
"id": "_design/meta",
"seq": 36
}
]
}
New in version 1.2.
The special filter _view allows to use existing map function as the filter. If the map function emits anything for the processed document it counts as accepted and the changes event emits to the feed. For most use-practice cases filter functions are very similar to map ones, so this feature helps to reduce amount of duplicated code.
Warning
While map functions doesn’t process the design documents, using _view filter forces them to do this. You need to be sure, that they are ready to handle documents with alien structure without panic crush.
Note
Using _view filter doesn’t queries the view index files, so you cannot use common view query parameters to additionally filter the changes feed by index key. Also, CouchDB doesn’t returns the result instantly as it does for views - it really uses the specified map function as filter.
Moreover, you cannot make such filters dynamic e.g. process the request query parameters or handle the User Context Object - the map function is only operates with the document.
Request:
GET /recipes/_changes?filter=_view&view=ingredients/by_recipe HTTP/1.1
Accept: application/json
Host: localhost:5984
Response:
HTTP/1.1 200 OK
Cache-Control: must-revalidate
Content-Type: application/json
Date: Sat, 28 Sep 2013 07:36:40 GMT
ETag: "ARIHFWL3I7PIS0SPVTFU6TLR2"
Server: CouchDB (Erlang OTP)
Transfer-Encoding: chunked
{
"last_seq": 38,
"results": [
{
"changes": [
{
"rev": "13-bcb9d6388b60fd1e960d9ec4e8e3f29e"
}
],
"id": "SpaghettiWithMeatballs",
"seq": 38
}
]
}