Permalink
Switch branches/tags
Nothing to show
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
664 lines (478 sloc) 20.3 KB

API

The Phantombuster API gives you control over your account. It is composed of HTTPS endpoints returning JSON data.

Here's a short list of what the API allows:

  • Launch and abort agents
  • Get console output, status, progress and messages from an agent
  • Get real-time console output from an agent
  • Get user, agent and script records
  • Write scripts
  • ...

We deliberately made the API extremely simple to use. Any developer should be able to get responses in a matter of minutes.

Versioning

All API endpoints URLs start with https://phantombuster.com/api/v1/.

Only version 1 exists for now.

Response format

All endpoints return JSON following the JSend specification. It basically means all successfull responses have HTTP code 2XX and look like this:

{
    "status": "success",
    "data": {
        ...
    }
}

Date/time fields are Unix/POSIX timestamps (in seconds).

Authentication and request format

Authentication is dead simple: put your API key in the X-Phantombuster-Key-1 HTTP header (or in the key parameter) of every request you make.

To get your API key, simply go to your settings page and click Reveal.

Parameters can be put in the query string or in the request body for POST requests. Here is how a typical request looks like:

GET /api/v1/agent/785/launch?command=casperjs&saveLaunchOptions=1 HTTP/1.1
Host: phantombuster.com
X-Phantombuster-Key-1: YOUR_API_KEY

You can also put your API key as a parameter. This is not recommended because your key might show up in log files:

GET /api/v1/agent/785/launch?command=casperjs&saveLaunchOptions=1&key=YOUR_API_KEY HTTP/1.1
Host: phantombuster.com

Please be aware that your key is precious as anyone who knows it can launch your agents (and do other mean things). Do not hesitate to generate a new one if you think it has been compromised.

Errors

If something bad happens, the HTTP code will be 4XX or 5XX and the response will look like this:

{
    "status": "error",
    "message": "Description of what happened"
}

The error response might also contain a code field and a data field describing the error in more details.

Here are some error HTTP codes you might encounter:

  • 400: missing parameter or something else wrong with the given parameters
  • 401: missing API key or wrong API key
  • 404: the requested object was not found (bad ID?)
  • 500: for some reason our servers could not handle your request

Get an agent record

GET /api/v1/agent/{id}.json

Get an agent record.

{id} (Number)
ID of the agent to retrieve.
withScript (String)
If present and not empty, and if the agent has an associated script, also return the script record.

Sample response:

{
    "status": "success",
    "data": {
        "id": 1763,
        "name": "Nice Agent",
        "scriptId": 1902,
        "proxy": "none",
        "proxyAddress": null,
        "proxyUsername": null,
        "proxyPassword": null,
        "disableWebSecurity": false,
        "ignoreSslErrors": false,
        "loadImages": true,
        "launch": "manually",
        "nbLaunches": 94,
        "showDebug": true,
        "awsFolder": "nVFRid8kvsuPeuCL80DnBg",
        "executionTimeLimit": 5,
        "fileMgmt": "folders",
        "fileMgmtMaxFolders": 10,
        "lastEndMessage": "Execution time limit reached",
        "lastEndStatus": "error",
        "maxParallelExecs": 1,
        "userAwsFolder": "QwYH17CB0Xj",
        "nonce": 123,
        "script": {
            "id": 1902,
            "name": "nice_agent.coffee",
            "source": "phantombuster",
            "url": null,
            "text": " ... script contents ... ",
            "httpHeaders": null,
        }
    }
}

Launch an agent

POST /api/v1/agent/{id}/launch

Add an agent to the launch queue.

This endpoint supports three types of outputs:

  • Standard JSON output (by setting output to json, which is the default) to get back a containerId in JSON.

    This ID can later be used to track this launch and get console output by calling /api/v1/agent/{agentId}/output.json?containerId={containerId}.

~ or ~

  • Result object output (by setting output to result-object) to get a blocking JSON response which will close when your agent finishes.

    The response will contain your agent's exit code (Number) and its result object (PlainObject) if it was set (using :ref:`buster-set-result-object`). This endpoint is very useful for getting a response from your agents "synchronously" — just make a single HTTP request and wait for your result object/exit code.

    Use first-result-object instead to have the request terminate immediately after the first call to :ref:`buster-set-result-object`. This is the fastest way to get a response from an agent using the API. However you will only get the result object and nothing else (no exit code or console output for example).

    Use result-object-with-output instead to get the console output of your agent in addition to all the other fields.

    Obviously this endpoint can be very slow to terminate (if your agent takes a long time or is queued). To prevent any risk of timeout, a space character is sent every 10 seconds to keep the HTTP socket alive (spaces do not prevent JSON parsing).

    Note: The HTTP headers are sent before your agent finishes, so you'll get a 200 OK even if your agent fails during execution (but not if it fails to queue).

~ or ~

  • Event stream output (by setting output to event-stream) to get a text/event-stream HTTP response.

    Each line of console output is sent as an event stream message starting with data:. When you receive the first message, you know the agent has started. When the agent has finished, the connection is closed. At regular intervals, event stream comments (starting with :) are sent to keep the connection alive.

    See a demo of this endpoint in action.

~ or ~

  • Raw output (by setting output to raw) to get an HTTP text/plain, chunked, streaming response of the raw console output of the agent.

    This is not recommended as almost all HTTP clients will timeout at one point or another, especially if your agent stays in queue for a few minutes (in which case the endpoint will send zero bytes for a few minutes, waiting for the agent to start — even cURL and Wget struggle to handle non-transmitting HTTP responses).

{id} (Number)
ID of the agent to launch.
output (String)
One of json, result-object, first-result-object, result-object-with-output, event-stream or raw (optional, default to json). This allows you to choose what type of response to receive.
command (String)
Command to use when launching the agent (optional). Can be either casperjs, phantomjs or node.
argument (String)
JSON argument as a String (optional). The argument can be retrieved with buster.argument in the agent's script.
saveLaunchOptions (String)
If present and not empty, command and argument will be saved as the default launch options for the agent.

Note: command and argument work together. When setting one, always set the other. When one or both are set, the saved launch options of the agent are ignored.

Note: The GET HTTP method is also allowed for this endpoint.

Sample response of json output:

{
    "status": "success",
    "data": {
        "containerId": 76426
    }
}

Sample response of result-object output:

{
    "status": "success",
    "message": "Agent finished (success)",
    "data": {
        "containerId": 76426,
        "executionTime": 17,
        "exitCode": 0,
        "resultObject": {
            "your": "data",
            "is": {
                "here": [123]
            }
        }
    }
}

Sample response of first-result-object output:

{
    "status": "success",
    "data": {
        "containerId": 76426,
        "resultObject": {
            "your": "data",
            "is": {
                "here": [123]
            }
        }
    }
}

Sample response of result-object-with-output output:

{
    "status": "success",
    "message": "Agent finished (success)",
    "data": {
        "containerId": 76426,
        "executionTime": 17,
        "exitCode": 0,
        "resultObject": {
            "your": "data",
            "is": {
                "here": [123]
            }
        }
        "output": "This is a console output line!\r\nAnd this is another one :)\r\n"
    }
}

Sample response of event-stream output:

: container 76426 in queue

: container 76426 in queue

data: This a console output line!
data:

: container 76426 still running

data: And this is

data: another one :)
data:

: container 76426 ended

Sample response of raw output:

This is a console output line!
And this is another one :)

Abort an agent

POST /api/v1/agent/{id}/abort.json

Abort all running instances of the agent.

{id} (Number)
ID of the agent to stop.

Note: The GET HTTP method is also allowed for this endpoint.

Sample response:

{
    "status": "success",
    "data": null
}

Get data from a running agent

GET /api/v1/agent/{id}/output.json

Get data from an agent: console output, status, progress and messages. This API endpoint is specifically designed so that it's easy to get incremental data from an agent.

This endpoint has two modes:

  • "Track" mode (by setting mode to track, which is the default when a containerId is specified) to get console output from a particular instance of the agent.

    In this mode, requests must have the containerId parameter set to the instance's ID from which you wish to get console output.

~ or ~

  • "Most Recent" mode (by setting mode to most-recent, which is the default when containerId is left at 0) to get console output from the most recent instance of the agent.

    In this mode, your first call should have parameter containerId left at 0. From then on, all subsequent calls must have parameter containerId set to the previously returned container ID (when a new instance of the agent is started, a different containerId will be returned).

{id} (Number)
ID of the agent from which to retrieve the output, status and messages.
mode (String)
Either track or most-recent (optional, defaults to most-recent if containerId is left at 0, otherwise defaults to track). This controls from which instance of the agent the console output is returned. In "Most Recent" mode, the most recent instance is selected each time a request is made. In "Track" mode, the console output from a particular instance is returned, as specified by the containerId parameter.
containerId (Number)
ID of the instance from which to get console output (optional, 0 by default). In "Most Recent" mode, always use the last containerId you received on a previous call or 0 for the first call. In "Track" mode, always set this parameter to the instance's ID from which you wish to get console ouput.
fromMessageId (Number)
Return the agent's messages starting from this ID (optional, -1 by default). If not present or -1, no messages are returned. Use the biggest message ID you received on a previous call to only get fresh messages.
fromOutputPos (Number)
Return the agent's console output starting from this position (optional, 0 by default). This number corresponds to the number of bytes emitted by the agent. Use the last outputPos you received on a previous call to only get new output data.

Note: The agentStatus and containerStatus fields have 3 possible values: running, queued or not running.

Note: The containerStatus field is only present in "Track" mode and represents the status of the tracked agent instance.

Note: The resultObject field is only present when a result object was set using :ref:`buster.setResultObject() <buster-set-result-object>`.

Sample response:

{
    "status": "success",
    "data": {
        "agentStatus": "running",
        "containerStatus": "running",
        "runningContainers": 1,
        "queuedContainers": 0,
        "containerId": 76427,
        "progress": {
            "progress": 0.1,
            "label": "Initializing...",
            "runtime": 3
        },
        "messages": [
            {
                "id": 65444,
                "date": 1414080820,
                "text": "Agent started",
                "type": "normal",
                "context": [
                    "Launch type: manual",
                    "Execution time limit: 60s"
                ]
            }
        ],
        "output": "* Container a255b8220379 started in directory /home/phantom/agent",
        "outputPos": 245,
        "resultObject": {
            "your": "data",
            "is": {
                "here": [123]
            }
        }
    }
}

Get container records

GET /api/v1/agent/{id}/containers.json

Get a list of ended containers for an agent, ordered by date. Useful for listing the last available output logs from an agent.

Container history is saved for up to 7 days.

{id} (Number)
ID of the agent from which to retrieve the containers.

Sample response:

{
    "status": "success",
    "data": [
        {
            "id": 195119,
            "queueDate": 1427810471,
            "launchDate": 1427810471,
            "launchType": "automatic",
            "launchNumber": 476,
            "endDate": 1427812088,
            "lastEndMessage": "Agent finished (error)",
            "lastEndStatus": "error",
            "exitCode": 1
        },
        {
            "id": 195050,
            "queueDate": 1427806874,
            "launchDate": 1427806874,
            "launchType": "automatic",
            "launchNumber": 475,
            "endDate": 1427810029,
            "lastEndMessage": "Agent finished (success)",
            "lastEndStatus": "success",
            "exitCode": 0
        }
    ]
}

Get a script by its ID

GET /api/v1/script/by-id/{mode}/{id}

Get a script record by its ID.

{id} (Number)
ID of the script to retrieve.
{mode} (String)
Either json or raw. If raw is used, the script is returned as raw text data, without any JSON.
withoutText (String)
If present and not empty, do not send the script's contents but only its metadata (only in JSON mode).

Sample response:

{
    "status": "success",
    "data": {
        "id": 1902,
        "name": "nice_agent.coffee",
        "source": "phantombuster",
        "url": null,
        "text": " ... script contents ... ",
        "httpHeaders": null,
        "lastSaveDate": 1427806874,
        "nonce": 123
    }
}

Get a script by its name

GET /api/v1/script/by-name/{mode}/{name}

Get a script record by its name.

{name} (String)
Name of the script to retrieve, with its extension (.js or .coffee).
{mode} (String)
Either json or raw. If raw is used, the script is returned as raw text data, without any JSON.
withoutText (String)
If present and not empty, do not send the script's contents but only its metadata (only in JSON mode).

Sample response:

{
    "status": "success",
    "data": {
        "id": 1902,
        "name": "nice_agent.coffee",
        "source": "phantombuster",
        "url": null,
        "text": " ... script contents ... ",
        "httpHeaders": null,
        "lastSaveDate": 1427806874,
        "nonce": 123
    }
}

List scripts

GET /api/v1/scripts.json

Get the list of all your scripts without text. To get a script contents, fetch it individually by its :ref:`ID <get-script-by-id>` or :ref:`name <get-script-by-name>`.

Sample response:

{
    "status": "success",
    "data": [
        {
            "id": 450,
            "name": "script1.coffee",
            "source": "phantombuster",
            "url": "",
            "httpHeaders": null,
            "lastSaveDate": 1446562593,
            "nonce": 12
        },
        {
            "id": 452,
            "name": "script2.js",
            "source": "sdk",
            "url": "",
            "httpHeaders": null,
            "lastSaveDate": 1446562789,
            "nonce": 4
        }
    ]
}

Delete a script

DELETE /api/v1/script/{id}.json

Delete one of your script.

{id} (Number)
ID of the script to delete.

Sample response:

{
    "status": "success",
    "data": null
}

Update or create a script

POST /api/v1/script/{name}

Update an existing script or create a new one if it does not exist (in this case, the new script ID is returned in the data field).

{name} (String)
Name of the script to update or create, with its extension (.js or .coffee).
text (String)
Full text contents of the script. This parameter must be in the request body in x-www-form-urlencoded format.
insertOnly (String)
If present and not empty, make sure that we don't update an existing script (optional). An error will be returned if a script with the same name already exists.
source (String)
Optional String describing from where the script comes from. Reserved sources keywords are phantombuster, web, sdk and bot builder. Only 20 alpha-numeric characters (and space) are allowed.

Sample response:

{
    "status": "success",
    "data": 345
}

List running agents and account info

GET /api/v1/user.json

Get information about your Phantombuster account and your agents.

Sample response:

{
    "status": "success",
    "data": {
        "email": "excellent.customer@gmail.com",
        "timeLeft": 14087,
        "emailsLeft": 100,
        "captchasLeft": 10,
        "storageLeft": 9991347906,
        "databaseLeft": 239222784,
        "agents": [
            {
                "id": 1388,
                "name": "My first agent",
                "scriptId": 0,
                "lastEndMessage": "Agent has no associated script",
                "lastEndStatus": "launch failed",
                "queuedContainers": 2,
                "runningContainers": 0
            },
            {
                "id": 1713,
                "name": "My second agent",
                "scriptId": 2003,
                "lastEndMessage": "Agent finished with exit code 0",
                "lastEndStatus": "success",
                "queuedContainers": 0,
                "runningContainers": 1,
                "progress": {
                    "progress": 0.544,
                    "label": "A progress label",
                    "runtime": 477
                }
            }
        ]
    }
}