HTTP Interface for bulk imports
ArangoDB provides an HTTP interface to import multiple documents at once into a collection. This is known as a bulk import.
The data uploaded must be provided in JSON format. There are two mechanisms to import the data:
- self-contained documents: in this case, each document contains all attribute names and values. Attribute names may be completely different among the documents uploaded
- attribute names plus document data: in this case, the first document must be a JSON list containing the attribute names of the documents that follow. The following documents must be lists containing only the document data. Data will be mapped to the attribute names by attribute positions.
The endpoint address is /_api/import for both input mechanisms. Data must be sent to this URL using an HTTP POST request. The data to import must be contained in the body of the POST request.
The collection URL parameter must be used to specify the target collection for the import. The optional URL parameter createCollection can be used to create a non-existing collection during the import. If not used, importing data into a non-existing collection will produce an error.
Importing self-contained documents
This import method allows uploading self-contained JSON documents. The documents must be uploaded in the body of the HTTP POST request. Each line of the body will be interpreted as one stand-alone document. Empty lines in the body are allowed and will be skipped.
To use this method, the type URL parameter should be set to documents.
Examples
curl --data-binary @- -X POST --dump - "http://localhost:8529/_api/import?type=documents&collection=test&createCollection=true"
{ "name" : "test", "gender" : "male", "age" : 39 }
{ "type" : "bird", "name" : "robin" }
HTTP/1.1 201 Created
server: triagens GmbH High-Performance HTTP Server
connection: Keep-Alive
content-type: application/json; charset=utf-8
{"error":false,"created":2,"errors":0}
The server will respond with an HTTP 201 if everything went well. The number of documents imported will be returned in the created attribute of the response. If any documents were skipped or incorrectly formatted, this will be returned in the errors attribute.
Importing headers and values
When using this type of import, the attribute names of the documents to be imported are specified separate from the actual document value data. The first line of the HTTP POST request body must be a JSON list containing the attribute names for the documents that follow. The following lines are interpreted as the document data. Each document must be a JSON list of values. No attribute names are needed or allowed in this data section.
Examples
curl --data-binary @- -X POST --dump - "http://localhost:8529/_api/import?collection=test&createCollection=true"
[ "firstName", "lastName", "age", "gender" ]
[ "Joe", "Public", 42, "male" ]
[ "Jane", "Doe", 31, "female" ]
HTTP/1.1 201 Created
server: triagens GmbH High-Performance HTTP Server
connection: Keep-Alive
content-type: application/json; charset=utf-8
{"error":false,"created":2,"errors":0}
The server will again respond with an HTTP 201 if everything went well. The number of documents imported will be returned in the created attribute of the response. If any documents were skipped or incorrectly formatted, this will be returned in the errors attribute.