Table of Contents
LightCouch provides a Java interface for communicating with CouchDB RESTful JSON APIs.
APIs in LightCouch is accessed by establishing a 'client' towards a CouchDB database server. A client is the main object which performs every request to the database, under it's context, such as document persistence or querying Views.
The APIs focus on Documents CRUD, Views, Attachments, Design Documents, Bulk operations, Changes notification, and database specific such as Compaction and Replication.
API usage and use cases examples is covered by integration unit tests, that runs against a running database. Download.
Internally LightCouch depends on Apache HttpClient for handling the HTTP communications & Gson for JSON / Java mappings.
Maven
<dependency> <groupId>org.lightcouch</groupId> <artifactId>lightcouch</artifactId> <version>0.2.0</version> </dependency>
LightCouch client, i.e. org.lightcouch.CouchDbClient can be configured and constructed in several ways as preferred.
LightCouch could run on Android
platform via org.lightcouch.CouchDbClientAndroid
Configuration parameters can be supplied to a client via .properties
files, or through convenient constructors.
Typical usage is creating a file named couchdb.properties
in the default classpath of your application,
containing the following parameters:
### Required couchdb.name=db-test couchdb.createdb.if-not-exist=true # The protocol: http | https couchdb.protocol=http couchdb.host=127.0.0.1 # The port e.g: 5984 | 6984 couchdb.port=5984 # Blank username/password for no login couchdb.username= couchdb.password= ### Optional/Advanced # Timeout to wait for a response in ms. Defaults to 0 (no timeout). couchdb.http.socket.timeout= # Timeout to establish a connection in ms. Defaults to 0 (no timeout). couchdb.http.connection.timeout= # Max connections. couchdb.max.connections=100 # Connect through proxy couchdb.proxy.host= couchdb.proxy.port= # path to append to DB URI couchdb.path=
Next, create a client instance, using the default constructor:
CouchDbClient dbClient = new CouchDbClient();
CouchDbClient
provides additional constructors for instantiating a client.
One which accepts a .properties
file name e.g, mycouchdb.properties
,
another that accepts individual parameters, the final uses an object with properties CouchDbProperties.
// couchdb-2.properties is on the classpath CouchDbClient dbClient1 = new CouchDbClient("couchdb-2.properties"); CouchDbClient dbClient2 = new CouchDbClient("db-name", true, "http", "127.0.0.1", 5984, "username", "secret"); CouchDbProperties properties = new CouchDbProperties() .setDbName("db-name") .setCreateDbIfNotExist(true) .setProtocol("https") .setHost("example.com") .setPort(443) .setUsername("username") .setPassword("secret") .setMaxConnections(100) .setConnectionTimeout(0); CouchDbClient dbClient3 = new CouchDbClient(properties); // Client is ready to use
Typically a client instance is created once per an application, and reused.
Multiple client instances within an application; can be created to handle multiple databases, each instance runs in isolation.
After usage, it might be useful to shutdown a client's underlying connection manager, for proper release of resources:
dbClient.shutdown();
Example use in Spring application.
<bean id="dbClient" class="org.lightcouch.CouchDbClient" init-method="syncDesignDocsWithDb" destroy-method="shutdown"/>
@Autowired private CouchDbClient dbClient;
Documents API is accessed via a client instance which acts like JPA EntityManager
.
The API provides CRUD
kind of operations for documents.
Model classes that represent documents in CouchDB, can choose to extend
org.lightcouch.Document
for conventience.
Plain objects need to define id and revision fields, i.e. _id
& _rev
.
Gson Mapping annotations eg. @SerializedName("_id") String id;
could be used to indicate the JSON
name of a field.
Fields with null
value are not serialized by defualt.
An instance of Gson
may be obtained by a client, for use in an application when desired.
Saving / Updating Documents is handled by save(Object)
and update(Object)
.
An object can be:
1. Plain Java Object.
2. java.util.Map
.
3. com.google.gson.JsonObject
.
Saving a new object internally is handled by HTTP PUT
,
a case where a document must be assigned an id
prior to sending to the database.
If the new object does not define an id
, the API will generate an
UUID
similar to database generated.
UUID
s can also by obtained from the database,
an example below on Databse API section.
Documents can also be saved via batch(Object)
.
The Result of save or update operations returns a org.lightcouch.Response object that represents the database response.
CouchDbClient dbClient = new CouchDbClient(); Foo foo = new Foo(); Response response = dbClient.save(foo); // dbClient.save(foo); // save, ignore response // dbClient.batch(foo); // saves batch dbClient.update(foo); Map<String, Object> map = new HashMap<>(); map.put("_id", "doc-id"); map.put("title", "value"); dbClient.save(map); JsonObject json = new JsonObject(); json.addProperty("_id", "doc-id"); json.add("array", new JsonArray()); dbClient.save(json); String jsonstr = "{\"title\":\"val\"}"; JsonObject jsonobj = dbClient.getGson().fromJson(jsonstr, JsonObject.class); dbClient.save(jsonobj);
Finding documents API is available through find()
s,
that expect the document id
, or w/ rev
,
and optional query parameters through org.lightcouch.Params
An additional finder findAny()
gives the option to get any document from the database,
given a URI
, as string, encoding may be required.
Documents returned as:
1. Plain Java Objects.
2. com.google.gson.JsonObject
.
3. java.io.InputStream
.
java.util.Map
, however does not qualify as a return type.
Note: Input streams need to be closed properly; to release the connection.
Foo foo = dbClient.find(Foo.class, "doc-id"); foo = dbClient.find(Foo.class, "doc-id", "doc-rev"); foo = dbClient.find(Foo.class, "doc-id", new Params().revsInfo().localSeq()); boolean found = dbClient.contains("doc-id"); InputStream in = dbClient.find("doc-id"); // .. in.close(); // close stream JsonObject json = dbClient.find(JsonObject.class, "doc-id"); String baseURI = dbClient.getBaseUri().toString(); String dbURI = dbClient.getDBUri().toString(); String uri = baseURI + "_stats"; JsonObject stats = dbClient.findAny(JsonObject.class, uri);
The API provides a remove()
for deleting documents,
expected parameters:
1. An object containing _id
and _rev
.
2. id
and revision
values.
Response response = dbClient.remove(object); dbClient.remove("doc-id", "doc-rev");
Configuring a custom serializer gives the option to register classes
Gson
may not be able to handle by default,
eg. JodaTime
DateTime
class.
GsonBuilder gsonBuilder = new GsonBuilder(); gsonBuilder.registerTypeAdapter(DateTime.class, new JsonSerializer<DateTime>() { public JsonElement serialize(DateTime src, Type typeOfSrc, JsonSerializationContext context) { return new JsonPrimitive(src.toString()); } }); gsonBuilder.registerTypeAdapter(DateTime.class, new JsonDeserializer<DateTime>() { public DateTime deserialize(JsonElement json, Type typeOfT, JsonDeserializationContext context) throws JsonParseException { return new DateTime(json.getAsJsonPrimitive().getAsString()); } }); CouchDbClient dbClient = new CouchDbClient(); dbClient.setGsonBuilder(gsonBuilder); // .. client is ready to use
The API supports both inline and standalone Attachments.
org.lightcouch.Attachment represents an inline attachment enclosed in a document.
The base64
data of an attachment may be encoded utilizing
the included dependency on Apache Codec
Base64.encodeBase64String(bytes)
.
Model classes that extend org.lightcouch.Document inherit the support for inline attachments.
To retrieve the base64
data of an attachment, include a parameter to the find()
byte[] bytes = "binary string".getBytes(); String data = Base64.encodeBase64String(bytes); Attachment attachment = new Attachment(data, "text/plain"); Bar bar = new Bar(); bar.addAttachment("bar.txt", attachment); dbClient.save(bar); bar = dbClient.find(Bar.class, "doc-id", new Params().attachments()); String base64Data = bar.getAttachments().get("bar.txt").getData();
Standalone attachments could be saved to an existing, or to a new document.
The attachment data is provided as java.io.InputStream
.
InputStream in = // .. init stream // save under a new document, a generated UUID is assigned as id. Response response = dbClient.saveAttachment(in, "photo.png", "image/png"); // save to an existing document dbClient.saveAttachment(in, "file.pdf", "application/pdf", "doc-id", "doc-rev"); // save under a new document with the given id. dbClient.saveAttachment(in, "photo.jpeg", "image/jpeg", "doc-id", null); // get attachment InputStream in = dbClient.find("doc-id/photo.jpeg"); // .. in.close(); // close the stream
LightCouch logs information about API usage, such as a request URI, and the returned HTTP status code
from the database server, such information is logged at the level INFO
.
Log4j
configuration:
# LightCouch log4j.logger.org.lightcouch=INFO # Apache HttpClient log4j.logger.org.apache.http=ERROR
In the event of API call failure, an exception of a specific type is thrown to report the failure.
Defined Exceptions:
1. NoDocumentException
when a requested document could not be found.
2. DocumentConflictException
when a conflict is detected during a save or update.
3. CouchDbException
is the super class for the former two exceptions,
conventiently it is thrown in cases of unexcpected exceptions, such as database errors 4xx
or 5xx
.
Views API is accessible under the context view("design_name/my_view")
.
Views can be queried in a number of ways:
1. List<T>
A list of specified type.
2. Scalar values, int
, boolean
etc.
3. View entries, key/value pairs as stored in the Database B-Tree.
4. java.io.InputStream
.
5. Pagination.
// List<T> List<Foo> list = dbClient.view("example/foo") .includeDocs(true) .startKey("start-key") .endKey("end-key") .limit(10) .query(Foo.class); // primitive types int count = dbClient.view("example/by_tag").key("couchdb").queryForInt(); // View entries keys, values and documents View view = dbClient.view("example/by_date") .key(2011, 10, 15) // complex key as: values or array .reduce(false) .includeDocs(true); ViewResult<int[], String, Foo> entries = view.queryView(int[].class, String.class, Foo.class);
Views API extend to support special Views such as _all_docs
.
// _all_docs List<Document> allDocs = dbClient.view("_all_docs").query(Document.class); // _all_docs paging Page<JsonObject> allDocsPage = dbClient.view("_all_docs").queryPage(10, pageParam, JsonObject.class);
Paging over a View results is offered by an API that features previous & next
navigation model.
A View page is respresented by Page<T>
Usage example in web application.
// Servlet public void doGet(HttpServletRequest request, HttpServletResponse r) { String param = request.getParameter("param"); int resultsPerPage = 15; // null param gets the first page Page<Foo> page = dbClient.view("example/foo") .reduce(false) .queryPage(resultsPerPage, param, Foo.class); request.setAttribute("page", page); // .. forward to a JSP for display }
<%-- JSP --%> <%-- iterate over documents --%> <c:forEach items="${page.resultList}" var="foo"> <p>${foo.title} </c:forEach> <%-- show paging status --%> <p>Showing: ${page.resultFrom} - ${page.resultTo} of total ${page.totalResults} <%-- handle navigation --%> <p> <c:choose> <c:when test="${page.hasPrevious}"> <a href="/getpage?param=${page.previousParam}"> Previous </a> </c:when> <c:otherwise> Previous </c:otherwise> </c:choose> ${page.pageNumber} <c:choose> <c:when test="${page.hasNext}"> <a href="/getpage?param=${page.nextParam}"> Next </a> </c:when> <c:otherwise> Next </c:otherwise> </c:choose>
Bulk documents API performs two operations: Modify & Fetch for bulk documents.
// insert/update docs List<Object> newDocs = new ArrayList<>(); newDocs.add(new Foo()); newDocs.add(new JsonObject()); boolean newEdits = true; List<Response> bulkResponse = dbClient.bulk(newDocs, newEdits);
// fetch multiple documents List<String> keys = Arrays.asList(new String[]{"doc-id-1", "doc-id-2"}); List<Foo> docs = dbClient.view("_all_docs") .includeDocs(true) .keys(keys) .query(Foo.class);
The API for Design Documents is accessible under the context design()
.
A Design Document is represented by
org.lightcouch.DesignDocument.
Design Documents may be maintained on local .js
files; saved in an application classpath.
An API provides a kind of synchronization
to copy from local files to the database.
org.lightcouch.CouchDbDesign
lists all the supported operations.
An example design document can be found under the test package of the source code.
Design documents directory structure:
design-docs |- example |- filters |- filter1.js |- .. |- lists |- list1.js |- .. |- shows |- show1.js |- .. |- updates |- update1.js |- .. |- validate_doc_update |- validate_doc_update.js |- rewrites |- rewrites.js |- views |- by_date |- map.js |- by_tag |- map.js |- reduce.js |- example2 ...
// find by dir name DesignDocument designDoc = dbClient.design().getFromDesk("example"); dbClient.design().synchronizeWithDb(designDoc); // find from Db DesignDocument designDoc1 = dbClient.design().getFromDb("_design/example"); // synchronize all dbClient.design().synchronizeAllWithDb(); // dbClient.syncDesignDocsWithDb(); // or use a client
Database specific APIs is accessible under the context context()
.
CouchDbInfo dbInfo = dbClient.context().info(); String version = dbClient.context().serverVersion(); dbClient.context().compact(); dbClient.context().ensureFullCommit(); dbClient.context().createDB("new-db"); List<String> uuids = dbClient.context().uuids(100);
Change notifications API is accessible under the context changes(). The API supports Changes feed of both types normal and continuous. A Row in the Changes feed is returned as com.google.gson.JsonObject
// get latest update seq CouchDbInfo dbInfo = dbClient.context().info(); String since = dbInfo.getUpdateSeq(); // feed type normal ChangesResult changes = dbClient.changes() .includeDocs(true) .limit(20) .since(since) .getChanges(); List<ChangesResult.Row> rows = changes.getResults(); for (ChangesResult.Row row : rows) { String docId = row.getId() JsonObject doc = row.getDoc(); } // feed type continuous Changes changes = dbClient.changes() .includeDocs(true) .since(since) .heartBeat(30000) .filter("example/filter") .continuousChanges(); // live access to continuous feeds while (changes.hasNext()) { ChangesResult.Row feed = changes.next(); String docId = feed.getId(); JsonObject doc = feed.getDoc(); // changes.stop(); // stop the running feed }
Update handler is invoked with invokeUpdateHandler().
Params params = new Params() .addParam("field", "title") .addParam("value", "foo bar"); String output = dbClient.invokeUpdateHandler("example/update1", "doc-id", params);
Example Update Function:
design-docs |- example |- updates |- update1.js function(doc, req) { var field = req.query.field; var value = req.query.value; var message = 'set '+field+' to '+value; doc[field] = value; return [doc, message]; }
Replication is handled by two sets of API, one that replicate by POSTing to
_replicate
URI, and the newer replicator database
(i.e _replicator
) introduced with CouchDB 1.1.0
.
// query params for filtered replication // String queryParams = "{\"key1\":\"val\"}"; Map<String, Object> queryParams = new HashMap<>(); queryParams.put("key1", "val"); // normal replication ReplicationResult result = dbClient.replication() .source("source-db") .target("http://user:secret@127.0.0.1:5984/target-db") .createTarget(true) .sinceSeq(100) // as of CouchDB 1.2 // .cancel() // cancel a replication .filter("example/filter1") .queryParams(queryParams); .trigger(); List<ReplicationHistory> histories = result.getHistories(); // _replicator database Replicator replicator = dbClient.replicator() .source("source-db") .target("target-db") .continuous(true) .createTarget(true) .sinceSeq(100) // as of CouchDB 1.2 .replicatorDB("replicator_2") // optional, defaults to _replicator .replicatorDocId("doc-id") // optional, defaults to UUID .filter("example/filter1") .queryParams(queryParams) .userCtxName("user") // for delegated requests .userCtxRoles("admin", "manager"); replicator.save(); // triggers a replication ReplicatorDocument findDocument = dbClient.replicator() .replicatorDocId("doc-id") .find(); List<ReplicatorDocument> docs = dbClient.replicator().findAll(); dbClient.replicator() .replicatorDocId("doc-id") .replicatorDocRev("doc-rev") .remove(); // cancels an ongoing replication
FullText search is enabled by including the relevant data to design documents, then running the query using the usual API.
couchdb-lucene example:
design-docs |- example |- fulltext |- fulltext.js { "by_title": { "index": "function(doc) { var ret = new Document(); ret.add( doc.title ); return ret; }" } }
String uri = dbClient.getBaseUri() + "_fti/local/mydb/_design/example/by_title?q=hello"; JsonObject result = dbClient.findAny(JsonObject.class, uri);
The second example uses Cloudant Search. Example design doc can be found here.
design-docs |- views101 |- indexes |- indexes.js |- views |- ...
String uri = dbClient.getDBUri() + "_design/views101/_search/animals?q=kookaburra"; JsonObject result = dbClient.findAny(JsonObject.class, uri);
This API enables extending LightCouch internal API by allowing a user-defined raw HTTP request to execute against a database.
An example get document current revision is an API not provided by LightCouch.
import org.apache.http.client.methods.HttpHead; import org.apache.http.HttpResponse; import org.apache.http.client.utils.HttpClientUtils; HttpHead head = new HttpHead(dbClient.getDBUri() + "doc-id"); HttpResponse response = dbClient.executeRequest(head); String revision = response.getFirstHeader("ETAG").getValue(); HttpClientUtils.closeQuietly(response); // closes the response