HTTP interface for data import

The Import HTTP API allows you to load JSON data in bulk into ArangoDB

Import JSON data as documents

POST /_db/{database-name}/_api/import

Load JSON data and store it as documents into the specified collection.

If you import documents into edge collections, all documents require a _from and a _to attribute.

Path Parameters
  • The name of the database.

Query Parameters
  • The name of the target collection. The collection needs to exist already.

  • Possible values: "", "documents", "array", "auto"

    Determines how the body of the request is interpreted.

    • documents: JSON Lines (JSONL) format. Each line is expected to be one JSON object.

      Example:

      {"_key":"john","name":"John Smith","age":35}
      {"_key":"katie","name":"Katie Foster","age":28}
      
    • array (or list): JSON format. The request body is expected to be a JSON array of objects. This format requires ArangoDB to parse the complete array and keep it in memory for the duration of the import. This is more resource-intensive than the line-wise JSONL processing.

      Any whitespace outside of strings is ignored, which means the JSON data can be a single line or be formatted as multiple lines.

      Example:

      [
        {"_key":"john","name":"John Smith","age":35},
        {"_key":"katie","name":"Katie Foster","age":28}
      ]
      
    • auto: automatically determines the type (either documents or array).

    • Omit the type parameter entirely (or set it to an empty string) to import JSON arrays of tabular data, similar to CSV.

      The first line is an array of strings that defines the attribute keys. The subsequent lines are arrays with the attribute values. The keys and values are matched by the order of the array elements.

      Example:

      ["_key","name","age"]
      ["john","John Smith",35]
      ["katie","Katie Foster",28]
      

  • When importing JSON arrays of tabular data (type parameter is omitted), the first line of the request body defines the attribute keys and the subsequent lines the attribute values for each document. Subsequent lines with a different number of elements than the first line are not imported by default.

    ["attr1", "attr2"]
    [1, 2]     // matching number of elements
    [1]        // misses 2nd element
    [1, 2, 3]  // excess 3rd element
    

    You can enable this option to import them anyway. For the missing elements, the document attributes are omitted. Excess elements are ignored.

  • The collection name prefix to prepend to all values in the _from attribute that only specify a document key.

  • The collection name prefix to prepend to all values in the _to attribute that only specify a document key.

  • Force the fromPrefix and toPrefix, possibly replacing existing collection name prefixes.

  • If enabled, then all data in the collection is removed prior to the import. Any existing index definitions are preserved.

  • Wait until documents have been synced to disk before returning.

  • Possible values: "error", "update", "replace", "ignore"

    Controls what action is carried out in case of a unique key constraint violation.

    • error: this will not import the current document because of the unique key constraint violation. This is the default setting.
    • update: this will update an existing document in the database with the data specified in the request. Attributes of the existing document that are not present in the request will be preserved.
    • replace: this will replace an existing document in the database with the data specified in the request.
    • ignore: this will not update an existing document and simply ignore the error caused by a unique key constraint violation.

    Note that update, replace and ignore will only work when the import document in the request contains the _key attribute. update and replace may also fail because of secondary unique key constraint violations.

  • If set to true, the whole import fails if any error occurs. Otherwise, the import continues even if some documents are invalid and cannot be imported, skipping the problematic documents.

  • If set to true, the result includes a details attribute with information about documents that could not be imported.

HTTP Headers
    Request Body text/plain; charset=utf-8

      The request body can have different JSON formats depending on the type parameter:

      • One JSON object per line (JSONL)
      • A JSON array of objects
      • One JSON array per line (CSV-like)
    Responses
    • is returned if all documents could be imported successfully.

      The response is a JSON object with the following attributes:

        Response Body application/json object
      • The number of imported documents.

      • An array with the error messages caused by documents that could not be imported. Only present if details is set to true.

      • The number of empty lines found in the input. Only greater than zero for the types documents and auto.

      • The number of documents that were not imported due to errors.

      • The number of failed but ignored insert operations. Only greater than zero if onDuplicate is set to ignore.

      • The number of updated/replaced documents. Only greater than zero if onDuplicate is set to either update or replace.

    • The type contains an invalid value, no collection is specified, the documents are incorrectly encoded, or the request is malformed.

    • The collection parameter or the _from or _to attributes of an imported edge refer to an unknown collection.

    • The complete option is enabled and the import triggers a unique key violation.

    • The complete option is enabled and the input is invalid, or the server cannot auto-generate a document key (out of keys error) for a document with no user-defined key.

    Examples

    Importing documents with heterogenous attributes from an array of JSON objects:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products&type=list' <<'EOF'
    [
      {
        "_key": "abc",
        "value1": 25,
        "value2": "test",
        "allowed": true
      },
      {
        "_key": "foo",
        "name": "baz"
      },
      {
        "name": {
          "detailed": "detailed name",
          "short": "short name"
        }
      }
    ]
    EOF
    Show output

    Importing documents using JSON objects separated by new lines (JSONL):

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products&type=documents' <<'EOF'
    { "_key": "abc", "value1": 25, "value2": "test","allowed": true }
    { "_key": "foo", "name": "baz" }
    
    { "name": { "detailed": "detailed name", "short": "short name" } }
    
    EOF
    Show output

    Using the auto type detection:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products&type=auto' <<'EOF'
    [
      {
        "_key": "abc",
        "value1": 25,
        "value2": "test",
        "allowed": true
      },
      {
        "_key": "foo",
        "name": "baz"
      },
      {
        "name": {
          "detailed": "detailed name",
          "short": "short name"
        }
      }
    ]
    EOF
    Show output

    Importing JSONL into an edge collection, with _from, _to and name attributes:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=links&type=documents' <<'EOF'
    { "_from": "products/123", "_to": "products/234" }
    {"_from": "products/332", "_to": "products/abc",   "name": "other name" }
    EOF
    Show output

    Importing an array of JSON objects into an edge collection, omitting _from or _to:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=links&type=list&details=true' <<'EOF'
    [
      {
        "name": "some name"
      }
    ]
    EOF
    Show output

    Violating a unique constraint, but allowing partial imports:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products&type=documents&details=true' <<'EOF'
    { "_key": "abc", "value1": 25, "value2": "test" }
    { "_key": "abc", "value1": "bar", "value2": "baz" }
    EOF
    Show output

    Violating a unique constraint, not allowing partial imports:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products&type=documents&complete=true' <<'EOF'
    { "_key": "abc", "value1": 25, "value2": "test" }
    { "_key": "abc", "value1": "bar", "value2": "baz" }
    EOF
    Show output

    Using a non-existing collection:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products&type=documents' <<'EOF'
    { "name": "test" }
    EOF
    Show output

    Using a malformed body with an array of JSON objects being expected:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products&type=list' <<'EOF'
    { }
    EOF
    Show output

    Importing two documents using the JSON arrays format. The documents have a _key, value1, and value2 attribute each. One line in the import data is empty and skipped:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products' <<'EOF'
    [ "_key", "value1", "value2" ]
    [ "abc", 25, "test" ]
    
    [ "foo", "bar", "baz" ]
    EOF
    Show output

    Importing JSON arrays into an edge collection, with _from, _to, and name attributes:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=links' <<'EOF'
    [ "_from", "_to", "name" ]
    [ "products/123","products/234", "some name" ]
    [ "products/332", "products/abc", "other name" ]
    EOF
    Show output

    Importing JSON arrays into an edge collection, omitting _from or _to:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=links&details=true' <<'EOF'
    [ "name" ]
    [ "some name" ]
    [ "other name" ]
    EOF
    Show output

    Using a malformed body with JSON arrays being expected:

    curl -X POST --header 'accept: application/json' --data-binary @- --dump - 'http://localhost:8529/_api/import?collection=products' <<'EOF'
    { "_key": "foo", "value1": "bar" }
    EOF
    Show output