From 75ace5d86556ae0dd29389901b88c8c5099921a2 Mon Sep 17 00:00:00 2001
From: Greg Sutcliffe <chatstat@emeraldreverie.org>
Date: Tue, 14 Dec 2021 15:28:22 +0000
Subject: [PATCH 1/2] Add event_id so we can de-duplicate lists if needed

---
 R/utils.R | 1 +
 1 file changed, 1 insertion(+)

diff --git a/R/utils.R b/R/utils.R
index 2cff7ed..1ce949b 100644
--- a/R/utils.R
+++ b/R/utils.R
@@ -15,6 +15,7 @@ api_url <- function(url, host = NULL, port = NULL) {
 process_events <- function(events) {
   tibble::tibble(event = events) |>
     tidyr::hoist(event,
+      id           = "event_id",
       time         = "origin_server_ts",
       type         = "type",
       sender       = "sender",

From c794978bb443f7a648b6da2c59d0080eca60b816 Mon Sep 17 00:00:00 2001
From: Greg Sutcliffe <chatstat@emeraldreverie.org>
Date: Tue, 14 Dec 2021 15:29:50 +0000
Subject: [PATCH 2/2] Make room_history handle multiple IDs

---
 DESCRIPTION                           |  1 +
 NAMESPACE                             |  2 +-
 R/room_history.R                      | 60 +++++++++++++++++++--------
 man/{room_history.Rd => get_rooms.Rd} | 11 +++--
 4 files changed, 51 insertions(+), 23 deletions(-)
 rename man/{room_history.Rd => get_rooms.Rd} (58%)

diff --git a/DESCRIPTION b/DESCRIPTION
index 5742855..4165a57 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -28,6 +28,7 @@ Imports:
     glue,
     httr,
     lubridate,
+    purrr,
     rlog,
     tibble,
     tidyr
diff --git a/NAMESPACE b/NAMESPACE
index 3dbe4e3..8f3126f 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -1,5 +1,5 @@
 # Generated by roxygen2: do not edit by hand
 
 export(get_messages)
-export(room_history)
+export(get_rooms)
 export(sync)
diff --git a/R/room_history.R b/R/room_history.R
index 0c5dfd9..625dba7 100644
--- a/R/room_history.R
+++ b/R/room_history.R
@@ -1,31 +1,25 @@
 #' Get the room events for a given room ID.
 #'
-#' @param room_id The room to get data for.
-#' @param since   Stop paginating when reaching this time.
+#' @param room_id      The room to get data for.
+#' @param since        Stop paginating when reaching this time.
+#' @param initial_sync Result of a prior call to [sync()].
 #'
 #' @return A tibble containing event information.
 #'
-#' @export
-room_history <- function(room_id, since) {
-  # We expect since to be a POSIX datetime, but could be a string.
-  since <- lubridate::as_datetime(since)
+#' @noRd
+room_history <- function(room_id, since, initial_sync) {
+  # TODO: Add better configuration.
+  token    <- Sys.getenv("token")
+  timeline <- initial_sync$rooms$join[[room_id]]$timeline
+  from     <- timeline$prev_batch
+
+  events   <- process_events(timeline$events)
+  rlog::log_debug(glue::glue("Initial sync yielded {nrow(events)} events."))
 
   rlog::log_info(
     glue::glue("Fetching events for room {room_id} since {since}.")
   )
 
-  # TODO: Add better configuration.
-  token <- Sys.getenv("token")
-
-  # Perform an initial sync and get events for the room.
-  initial_sync <- sync()
-  timeline     <- initial_sync$rooms$join[[room_id]]$timeline
-  from         <- timeline$prev_batch
-
-  events <- process_events(timeline$events)
-
-  rlog::log_debug(glue::glue("Initial sync yielded {nrow(events)} events."))
-
   while (TRUE) {
     oldest_time <- events |>
       dplyr::slice_min(time) |>
@@ -55,3 +49,33 @@ room_history <- function(room_id, since) {
 
   events |> dplyr::filter(time >= since)
 }
+
+#' Get the room events for a given room ID.
+#'
+#' @param room_ids Vector of room IDs to iterate over.
+#' @param since    Stop paginating when reaching this time.
+#' @param sync     (Optional) Result of a prior call to [sync()] to save
+#'                 duplication.
+#'
+#' @return A tibble containing event information.
+#'
+#' @export
+get_rooms <- function(room_ids, since, sync = NULL) {
+  # We expect since to be a POSIX datetime, but could be a string.
+  since <- lubridate::as_datetime(since)
+
+  rlog::log_debug(glue::glue("Getting history for {length(room_ids)} rooms."))
+
+  # Perform an initial sync and get events for the room.
+  initial_sync <- if (is.null(sync)) {
+    rlog::log_debug("Initial sync not provided - running sync() now.")
+    sync()
+  } else {
+    rlog::log_debug("Initial sync provided - not running sync().")
+    sync
+  }
+
+  tidyr::tibble(room = room_ids) |>
+    dplyr::group_by(room) |>
+    dplyr::mutate(events = purrr::map(room, room_history, since, initial_sync))
+}
diff --git a/man/room_history.Rd b/man/get_rooms.Rd
similarity index 58%
rename from man/room_history.Rd
rename to man/get_rooms.Rd
index 7977084..ce416c0 100644
--- a/man/room_history.Rd
+++ b/man/get_rooms.Rd
@@ -1,15 +1,18 @@
 % Generated by roxygen2: do not edit by hand
 % Please edit documentation in R/room_history.R
-\name{room_history}
-\alias{room_history}
+\name{get_rooms}
+\alias{get_rooms}
 \title{Get the room events for a given room ID.}
 \usage{
-room_history(room_id, since)
+get_rooms(room_ids, since, sync = NULL)
 }
 \arguments{
-\item{room_id}{The room to get data for.}
+\item{room_ids}{Vector of room IDs to iterate over.}
 
 \item{since}{Stop paginating when reaching this time.}
+
+\item{sync}{(Optional) Result of a prior call to \code{\link[=sync]{sync()}} to save
+duplication.}
 }
 \value{
 A tibble containing event information.