R/http-handlers.R

Defines functions json_rpc_result json_rpc_input rpc_handler register_http_handlers

register_http_handlers <- function(session, metadata) {
  session$userData$learnr_state <- reactiveVal("start")

  # parent environment for completions (see discussion in setup_exercise_handler
  # for why this is chosen as the completion/execution parent)
  server_envir <- parent.env(parent.env(parent.frame()))

  # environment used for hosting state (e.g. for chunks)
  state <- new.env(parent = emptyenv())

  # initialize handler
  session$registerDataObj("initialize", NULL,  function(data, req) {

    # parameters
    location <- json_rpc_input(req)$location

    # initialize session state
    identifiers <- initialize_session_state(session, metadata, location, req)

    # data payload to return
    data <- list(
      identifiers = identifiers
    )

    # Now that we've initialized the session state, emit the start event
    event_trigger(session, "session_start")

    session$userData$learnr_state("initialized")
    # return identifers
    list(
      status = 200L,
      headers = list(
        'Content-Type' = 'application/json'
      ),
      body = json_rpc_result(data)
    )
  })

  # restore state handler
  session$registerDataObj("restore_state", NULL, rpc_handler(function(input) {

    # forward any client stored objects into our own storage
    if (!is.null(input))
      initialize_objects_from_client(session, input)

    # get state objects
    state_objects <- get_all_state_objects(session, exercise_output = FALSE)

    # create submissions from state objects
    submissions <- submissions_from_state_objects(state_objects)

    # create video progress from state objects
    video_progress <- video_progress_from_state_objects(state_objects)

    # create progress events from state objects
    progress_events <- progress_events_from_state_objects(state_objects)

    # get client state
    client_state <- get_client_state(session)

    session$userData$learnr_state("restored")

    # return data
    list(
      client_state = client_state,
      submissions = submissions,
      video_progress = video_progress,
      progress_events = progress_events
    )
  }))

  # remove state handler
  session$registerDataObj("remove_state", NULL, rpc_handler(function(input) {
    remove_all_objects(session)
  }))

  # event recording
  session$registerDataObj("record_event", NULL, rpc_handler(function(input) {
    # Augment the data with the label
    input$data$label <- input$label

    record_event(session = session,
                 event = input$event,
                 data = input$data)
  }))

  # # question submission handler
  # session$registerDataObj("question_submission", NULL, rpc_handler(function(input) {
  #   cat("--- question_submission handler called\n")
  #
  #   # extract inputs
  #   label <- input$label
  #   question <- input$question
  #   answers <- input$answers
  #   correct <- input$correct
  #
  #   # fire event
  #   question_submission_event(session = session,
  #                             label = label,
  #                             question = question,
  #                             answers = answers,
  #                             correct = correct)
  # }))

  # video progress handler
  session$registerDataObj("video_progress", NULL, rpc_handler(function(input) {

    # extract inputs
    video_url <- input$video_url
    time <- input$time
    total_time <- input$total_time

    # fire event
    event_trigger(
      session,
      "video_progress",
      data = list(
        video_url  = video_url,
        time       = time,
        total_time = total_time
      )
    )
  }))

  # exercise skipped event
  session$registerDataObj("section_skipped", NULL, rpc_handler(function(input) {

    # extract inputs
    sectionId <- input$sectionId

    # fire event
    event_trigger(
      session,
      "section_skipped",
      data = list(sectionId = sectionId)
    )

  }))

  # client state handler
  session$registerDataObj("set_client_state",  NULL, rpc_handler(function(input) {
    save_client_state(session, input)
  }))

  # help handler
  session$registerDataObj("help",  NULL, rpc_handler(function(input) {

  }))

  # setup chunk handler
  session$registerDataObj("initialize_chunk", NULL, rpc_handler(function(input) {
    params <- input

    # evaluate setup code to prime environment
    label <- as.character(params$label)
    code <- paste(params$setup_code, collapse = "\n")

    # no setup chunk / label? nothing to do
    if (!(nzchar(label) && nzchar(code)))
      return()

    # evaluate code in environment to prime
    Encoding(code) <- "UTF-8"
    state[[label]] <- new.env()
    eval(parse(text = code, encoding = "UTF-8"), envir = state[[label]])

  }))

  # completion handler
  session$registerDataObj("completion", NULL, rpc_handler(function(input) {

    # read params
    line <- as.character(input$contents)
    label <- as.character(input$label)

    Encoding(line) <- "UTF-8"

    auto_complete_r(line, label, state)
  }))

  # diagnostics handler
  session$registerDataObj("diagnotics",  NULL, rpc_handler(function(input) {

  }))

  # this is a "bat signal" to let the JS side know that the Shiny
  # server is ready to handle http requests
  session$sendCustomMessage("tutorial_isServerAvailable", "true")
}


# return a rook wrapper for a function that takes a list and returns a list
# (list contents are automatically converted to/from JSON for rook as required)
rpc_handler <- function(handler) {

  function(data, req) {

    # get the input
    input <- json_rpc_input(req)

    # call the handler
    result <- handler(input)

    # return the result as JSON
    list(
      status = 200L,
      headers = list(
        'Content-Type' = 'application/json'
      ),
      body = json_rpc_result(result)
    )
  }
}

# get the json from a request body
json_rpc_input <- function(req) {
  input_stream <- req[["rook.input"]]
  jsonlite::fromJSON(input_stream$read_lines())
}

# helper for returning JSON
json_rpc_result <- function(x) {
  jsonlite::toJSON(x, auto_unbox = TRUE, null = "null", force = TRUE)
}

Try the learnr package in your browser

Any scripts or data that you put into this service are public.

learnr documentation built on Sept. 28, 2023, 9:06 a.m.