R/query.R

In alq666/rdog: Connect R to Datadog

#' Querying
#'
#' @param api_key Your Datadog API key
#' @param application_key Your Application Key
#' @param query You search query
#' @param from_t Start of time range 
#' @param to_t End of time range
#' @param as_df Boolean, returns xts when FALSE, data.fame when TRUE
#' @export
#' @examples
#' \dontrun{dog_query('api_key', 'app_key', 1478872763, as.integer(Sys.time()), 'avg:system.cpu.idle{*}')}

dog_query <- function(api_key, application_key, query, from_t, to_t, as_df = FALSE) {
    res <- RCurl::getForm("https://app.datadoghq.com/api/v1/query", 
        api_key = api_key, 
        application_key = application_key, 
        from = from_t, 
        to = to_t, 
        query = query
    )
    parsed <- jsonlite::fromJSON(res)
    if (parsed[["status"]] != "ok") {
        # FIXME: Probably should handle failure better
        return(xts())
    }
    timeseries <- parsed[["series"]]  # data frame
    scope <- timeseries[["scope"]]  # names of each group, if any

    # If * is the only scope, rename it to something else
    if (length(scope) == 1 && scope == "*") {
        scope <- "all"
    }
    pointlist <- timeseries$pointlist
    # collect time parameters
    start <- max(timeseries$start)
    end <- max(timeseries$end)
    interval <- min(timeseries$interval)

    # Extract timestamps from the first list They will be identical across all groups
    timestamps <- to_epoch(pointlist[[1]][, 1])

    # Collect all series values (accessible as [, 2])
    v <- mapply(values, pointlist)

    # build the time series
    if (as_df) {
        df <- data.frame(timestamps, v)
        colnames(df) <- c('time', scope)
    } else {
        df <- xts::xts(v, order.by = timestamps, frequency = freq(interval))
        colnames(df) <- scope
    }
    return(df)
}