R/aw_report.R

In benrwoodard/awoberonr: Use Oberon JSON string from the debugger to pull data into R

#' Use a prebuilt json query to pull a ranked report
#'
#' Organizes the arguments into a json string and then structures the data after the internal function makes
#' the api call. Only runs a single dimension with as many metrics as you want.
#'
#' @param req_body The json string copied from Workspace
#' @param company_id Company Id.  Taken from the global environment by default if not provided.
#' @param client_id Set in environment args, or pass directly here
#' @param client_secret Set in environment args, or pass directly here
#'
#' @export
#' @import assertthat httr dplyr tidyr
#'
aw_report <- function(req_body = '',
                      company_id = Sys.getenv('AW_COMPANY_ID'),
                      client_id = Sys.getenv("AW_CLIENT_ID"),
                      client_secret = Sys.getenv("AW_CLIENT_SECRET")) {

#assert for better error control
  assertthat::assert_that(
    file.exists(req_body),
    is.string(company_id),
    is.string(client_id),
    is.string(client_secret)
  )

# creates token to aa.oauth if not present
token <- aw_token(client_id, client_secret)

#grab the dimensions and metric names from the query
query <-jsonlite::fromJSON(txt=req_body)

#build the necessary naming items for the result
#build the metric column names
metrics <-  gsub(".*/", "", query$metricContainer$metrics$id)
#build the dimension column names
dimensions <- gsub(".*/", "", query$dimension)

#if metricFilters exist then concatenate the names of the fitler and metric for final metric name
if(!is.null(query$metricContainer$metricFilters)) {
  #if the the metric filter id is on the second or 30th metric then the filter
  #name needs to be applied on that level and the others need to have '' applied
  mf <- query$metricContainer$metricFilters  %>%
    dplyr::rename('filtername' = 3)
  #unnest the metrics to include multiple filter columns if needed
  mets <- tidyr::unnest(query$metricContainer$metrics, cols = c(columnId, filters))

  metricsinfo <- mets %>%
    dplyr::left_join(mf, by = c('filters' = 'id'))

  finalmnames <- metricsinfo %>%
    dplyr::mutate(id = gsub(".*/", "", id)) %>%
    dplyr::mutate(mfinalname = ifelse(!is.na(filtername), paste0(filtername,'_', id), id) ) %>%
    dplyr::group_by(columnId) %>%
    dplyr::summarise(mfinalname = paste0(mfinalname, collapse = "-")) %>%
    dplyr::pull(mfinalname)

}

request_url <- sprintf("https://analytics.adobe.io/api/%s/reports/ranked",
                       company_id)

req <- httr::RETRY("POST",
                   url = request_url,
                   body = upload_file(req_body),
                   encode = "json",
                   config(token = token),
                   verbose(data_out = TRUE),
                   httr::add_headers(
                     `x-api-key` = client_id,
                     `x-proxy-global-company-id` = company_id
                   ))

  httr::stop_for_status(req)

res <- httr::content(req, as = "text",encoding = "UTF-8")

#reformat from JSON
res <- jsonlite::fromJSON(res)
# Clean up and return only data rows
res_df <- res$rows
# If more than one metric the value list needs to be spread to individual columns
res_df <- res_df %>%
  tidyr::unnest(data) %>%
  dplyr::group_by(itemId,value) %>%
  dplyr::mutate(col = seq_along(data)) %>%
  tidyr::spread(key=col, value=data) %>%
  dplyr::ungroup()%>%
  dplyr::select(-itemId)

# Add column names to the dataset based on the metrics and dimensions
colnames(res_df) <- c(dimensions,finalmnames)

#return it as a datafram
df <- data.frame(res_df)

return(df)
}