feat: use v1 api in get_jobs() (#343)

Co-authored-by: Barret Schloerke <[email protected]>

Author: toph-allen

NEWS.md

@@ -1,5 +1,12 @@
 # connectapi (development version)
 
+## Breaking changes
+
+- `get_jobs()` now uses the public endpoint. The data returned by this function
+  has changed. In particular, the `finalized` column is no longer present,
+  replaced by `status`. `status == 0` corresponds to `isFALSE(finalized)`. See
+  `?get_jobs()` for more details about the new return format. (#340)
+
 ## New features
 
 - `get_users()` now supports filtering users with the `account_status` and

R/connectapi.R

@@ -27,5 +27,6 @@ current_connect_version <- "2024.03.0"
 
 .onLoad <- function(...) {
   vctrs::s3_register("dplyr::collect", "tbl_connect")
+  vctrs::s3_register("vctrs::vec_cast", "character.integer")
   invisible()
 }

R/content.R

@@ -95,9 +95,23 @@ Content <- R6::R6Class(
     },
     #' @description Return the jobs for this content.
     jobs = function() {
-      warn_experimental("jobs")
-      url <- unversioned_url("applications", self$get_content()$guid, "jobs")
-      res <- self$get_connect()$GET(url)
+      res <- self$connect$GET(v1_url("content", self$content$guid, "jobs"), parser = NULL)
+      use_unversioned <- endpoint_does_not_exist(res)
+      if (use_unversioned) {
+        res <- self$connect$GET(unversioned_url("applications", self$content$guid, "jobs"), parser = NULL)
+      }
+      self$connect$raise_error(res)
+      parsed <- httr::content(res, as = "parsed")
+      if (use_unversioned) {
+        # The unversioned endpoint does not contain a `status` field. Its field
+        # `finalized` is `FALSE` corresponds to active jobs. The `finalized`
+        # field is dropped during parsing.
+        parsed <- purrr::modify_if(parsed, ~ isFALSE(.x$finalized), function(x) {
+          x$status <- 0
+          x
+        })
+      }
+      parsed
     },
     #' @description Return a single job for this content.
     #' @param key The job key.
@@ -588,26 +602,73 @@ content_ensure <- function(
 
 #' Get Jobs
 #'
-#' `r lifecycle::badge('experimental')` Retrieve details about jobs associated with a `content_item`.
-#' "Jobs" in Posit Connect are content executions
+#' Retrieve details about server processes associated with a `content_item`,
+#' such as a FastAPI app or a Quarto render.
+#'
+#' Note that Connect versions below 2022.10.0 use a legacy endpoint, and will
+#' not return the complete set of information provided by newer versions.
 #'
 #' @param content A Content object, as returned by `content_item()`
-#' @param key The key for a job
 #'
-#' @rdname jobs
+#' @return A data frame with a row for each job, with the following columns:
+#'
+#' - `id`: The job identifier.
+#' - `ppid`: The job's parent process identifier (see Note 1).
+#' - `pid`: The job's process identifier.
+#' - `key`: The job's unique key identifier.
+#' - `remote_id`: The job's identifier for off-host execution configurations
+#' (see Note 1).
+#' - `app_id`: The job's parent content identifier
+#' - `variant_id`: The identifier of the variant owning this job.
+#' - `bundle_id`: The identifier of a content bundle linked to this job.
+#' - `start_time`: The timestamp (RFC3339) indicating when this job started.
+#' - `end_time`: The timestamp (RFC3339) indicating when this job finished.
+#' - `last_heartbeat_time`: The timestamp (RFC3339) indicating the last time
+#' this job was observed to be running (see Note 1).
+#' - `queued_time`: The timestamp (RFC3339) indicating when this job was added
+#' to the queue to be processed. Only scheduled reports will present a value
+#' for this field (see Note 1).
+#' - `queue_name`: The name of the queue which processes the job. Only
+#' scheduled reports will present a value for this field (see Note 1).
+#' - `tag`: A tag to identify the nature of the job.
+#' - `exit_code`: The job's exit code. Present only when job is finished.
+#' - `status`: The current status of the job. On Connect 2022.10.0 and newer,
+#' one of Active: 0, Finished: 1, Finalized: 2; on earlier versions, Active:
+#' 0, otherwise `NA`.
+#' - `hostname`: The name of the node which processes the job.
+#' - `cluster`: The location where this content runs. Content running on the
+#' same server as Connect will have either a null value or the string Local.
+#' Gives the name of the cluster when run external to the Connect host
+#' (see Note 1).
+#' - `image`: The location where this content runs. Content running on
+#' the same server as Connect will have either a null value or the string
+#' Local. References the name of the target image when content runs in
+#' a clustered environment such as Kubernetes (see Note 1).
+#' - `run_as`: The UNIX user that executed this job.
+#'
+#' @note
+#' 1. On Connect instances earlier than 2022.10.0, these columns will contain `NA` values.
+#'
+#' @family job functions
 #' @family content functions
 #' @export
 get_jobs <- function(content) {
-  warn_experimental("get_jobs")
-  scoped_experimental_silence()
   validate_R6_class(content, "Content")
 
   jobs <- content$jobs()
-  parse_connectapi_typed(jobs, connectapi_ptypes$jobs)
+  parse_connectapi_typed(jobs, connectapi_ptypes$jobs, order_columns = TRUE)
 }
 
 # TODO: Need to test `logged_error` on a real error
-#' @rdname jobs
+#'
+#' Retrieve details about a server process
+#' associated with a `content_item`, such as a FastAPI app or a Quarto render.
+#'
+#' @param content A Content object, as returned by `content_item()`
+#' @param key The key for a job
+#'
+#' @family job functions
+#' @family content functions
 #' @export
 get_job <- function(content, key) {
   warn_experimental("get_job")

R/parse.R

@@ -27,14 +27,19 @@ make_timestamp <- function(input) {
   safe_format(input, "%Y-%m-%dT%H:%M:%SZ", tz = "UTC", usetz = FALSE)
 }
 
-ensure_columns <- function(.data, ptype) {
+ensure_columns <- function(.data, ptype, order_columns = FALSE) {
   # Given a prototype, ensure that all columns are present and cast to the correct type.
   # If a column is missing in .data, it will be created with all missing values of the correct type.
   # If a column is present in both, it will be cast to the correct type.
   # If a column is present in .data but not in ptype, it will be left as is.
   for (i in names(ptype)) {
     .data <- ensure_column(.data, ptype[[i]], i)
   }
+
+  if (order_columns) {
+    .data <- .data[, unique(c(names(ptype), names(.data))), drop = FALSE]
+  }
+
   .data
 }
 
@@ -59,14 +64,14 @@ ensure_column <- function(data, default, name) {
     if (inherits(default, "list") && !inherits(col, "list")) {
       col <- list(col)
     }
-    col <- vctrs::vec_cast(col, default)
+    col <- vctrs::vec_cast(col, default, x_arg = name)
   }
   data[[name]] <- col
   data
 }
 
-parse_connectapi_typed <- function(data, ptype) {
-  ensure_columns(parse_connectapi(data), ptype)
+parse_connectapi_typed <- function(data, ptype, order_columns = FALSE) {
+  ensure_columns(parse_connectapi(data), ptype, order_columns)
 }
 
 parse_connectapi <- function(data) {
@@ -185,6 +190,10 @@ tzone <- function(x) {
   attr(x, "tzone")[[1]] %||% ""
 }
 
+vec_cast.character.integer <- function(x, to, ...) { # nolint: object_name_linter
+  as.character(x)
+}
+
 new_datetime <- function(x = double(), tzone = "") {
   tzone <- tzone %||% ""
   if (is.integer(x)) {

R/ptype.R

@@ -161,20 +161,26 @@ connectapi_ptypes <- list(
     variant_key = NA_character_,
   ),
   jobs = tibble::tibble(
-    id = NA_integer_,
-    pid = NA_integer_,
+    id = NA_character_,
+    ppid = NA_character_,
+    pid = NA_character_,
     key = NA_character_,
-    app_id = NA_integer_,
-    app_guid = NA_character_,
-    variant_id = NA_integer_,
-    bundle_id = NA_integer_,
+    remote_id = NA_character_,
+    app_id = NA_character_,
+    variant_id = NA_character_,
+    bundle_id = NA_character_,
     start_time = NA_datetime_,
     end_time = NA_datetime_,
+    last_heartbeat_time = NA_datetime_,
+    queued_time = NA_datetime_,
+    queue_name = NA_character_,
     tag = NA_character_,
     exit_code = NA_integer_,
-    finalized = NA,
+    status = NA_integer_,
     hostname = NA_character_,
-    variant_key = NA_character_
+    cluster = NA_character_,
+    image = NA_character_,
+    run_as = NA_character_,
   ),
   job = tibble::tibble(
     pid = NA_integer_,

R/utils.R

@@ -186,3 +186,13 @@ token_hex <- function(n) {
   raw <- as.raw(sample(0:255, n, replace = TRUE))
   paste(as.character(raw), collapse = "")
 }
+
+# Checks to see if an http status contains a 404 error, and that that 404
+# response does not contain an error code indicating that an endpoint was called
+# but encountered a 404 for some other reason.
+endpoint_does_not_exist <- function(res) {
+  return(
+    httr::status_code(res) == "404" &&
+      !("code" %in% names(httr::content(res, as = "parsed")))
+  )
+}