IBrokers (version 0.10-2)

eWrapper: eWrapper Closure For Message Processing

Description

Create an eWrapper closure to allow for custom incoming message management.

Usage

eWrapper(debug = FALSE, errfile=stderr())
eWrapper.data(n)
eWrapper.MktData.CSV(n=1)
eWrapper.RealTimeBars.CSV(n=1)

Value

A list of functions [and optionally data] to be used for the eventWrapper argument to reqMktData and reqMktDepth

Arguments

debug

should debugging be enabled

errfile

where error messages are directed (stderr)

n

number of contracts being watched

Author

Jeffrey A. Ryan

Details

IBrokers implements an eWrapper scheme similar to that provided by the official Java API.

The general idea is that each real-time data capture function must manage all incoming signals correctly, while allowing for the end user to create custom handlers for each specific event.

Internal to the reqRealTimeBars, reqMktData, and reqMktDepth functions is a single call to the CALLBACK routine passed to it. By default this is twsCALLBACK (see also). A standard argument to this callback is an eventWrapper --- which is an instance of eWrapper.

eWrapper is an R closure that contains a list of functions to manage all incoming message type, as found in .twsIncomingMSG. Each message has a corresponding function in the eWrapper designed to handle the particular details of each incoming message type.

There is also an embedded environment in which data can be saved and retrieved via a handful of accessor functions mimicking the standard R tools.

The data environment is .Data, with accessor methods get.Data, assign.Data, and remove.Data.

These methods can be called from the closure object eWrapper$get.Data, eWrapper$assign.Data, etc.

The basic eWrapper call simply produces a visually informative display of the incoming stream. E.g. bidSize data would be represented with a bidSize label, instead of the internal TWS code(s) returned by the TWS.

By creating an instance of an eWrapper, accomplished by calling it as a function call, one can then modify any or all the particular methods embedded in the object.

This allows for rapid customization, as well as a built in assurance that all incoming messages will be handled appropriately without additional programmer time and resources.

An example of this ability to modify the object is given in the eWrapper.MktData.CSV code. This object produces output deisgned to be space efficient, as well as easily read back into any R session as a standard CSV file.

Setting debug=NULL will cause empty function objects to be created within the eWrapper object returned. This object can be treated as a template to implement only the methods that are needed. By default, all functions silently return the entire message they would normally parse. This includes empty functions created by setting debug to NULL.

eWrapper.data() allows for data states to be maintained from call to call, as an xts history of updates/messages is stored within the object. This is designed to minimize calling overhead by removing unneeded function calls from each message parsed.

Additional, but creating methods that update the internal environment of the eWrapper object, it is possible to maintain a snapshot of last k values for any field of interest. This is directly applicable to implementing an automated strategy from within a custom twsCALLBACK method.

See Also

twsCALLBACK, processMsg

Examples

myWrapper <- eWrapper()

str(myWrapper)

# remove tickPrice action
myWrapper$tickPrice <- function(msg, timestamp, file, ...) {}

# add new tickPrice action
myWrapper$tickPrice <- function(msg, timestamp, file, ...) { cat("tickPrice",msg) }

# add new data into the object, and retrieve
myWrapper$assign.Data("myData", 1010)
myWrapper$get.Data("myData")

if (FALSE) {
tws <- twsConnect()
reqMktData(tws, twsSTK("SBUX"))
reqMktData(tws, twsSTK("SBUX"), eventWrapper=myWrapper)
twsDisconnect(tws)
}