Starting cache for stats with CBS scrape

Author: atungate

R/caching_helpers.R

@@ -93,18 +93,25 @@ ensure_cache_dir_exists = function() {
 
 clear_cache_by_time = function() {
   cache_dir = tools::R_user_dir("ffanalytics", "cache")
+
+  # First checking for all files older than 8 hours
   file_names = list.files(cache_dir, full.names = TRUE)
 
   if(length(file_names) == 0) {
     return(NULL)
   }
-
   file_mtimes = file.mtime(file_names)
-  files_to_clear = difftime(Sys.time(), file_mtimes, units = "hours") > 8
+  all_files_to_clear = difftime(Sys.time(), file_mtimes, units = "hours") > 8
+  scrapes_to_clear = grepl("_scrape", basename(file_names), fixed = TRUE) &
+    difftime(Sys.time(), file_mtimes, units = "hours") > 1
+
+  files_to_clear = as.logical(pmax(all_files_to_clear, scrapes_to_clear))
+
 
   if(any(files_to_clear, na.rm = TRUE)) {
     file.remove(file_names[files_to_clear])
   }
+
 }
 
 cache_file_names = c(
@@ -188,7 +195,8 @@ cache_file_names = c(
   "ecr_draft_db_ppr.rds" = "ECR Draft DB PPR",
   "ecr_weekly_db_std.rds" = "ECR Weekly DB Std",
   "ecr_weekly_db_half.rds" = "ECR Weekly DB Half",
-  "ecr_weekly_db_ppr.rds" = "ECR Weekly DB PPR"
+  "ecr_weekly_db_ppr.rds" = "ECR Weekly DB PPR",
+  "cbs_scrape.rds" = "CBS Scrape"
 )
 
 

R/ffanalytics.R

@@ -19,8 +19,8 @@
 
 .onAttach <- function(libname, pkgname) {
   packageStartupMessage(
-    "Note: the ffanalytics package locally caches ADP & ECR data scrapes. Cached scrapes",
-    "\nolder than 8 hours are dropped (upon checking)",
+    "Note: the ffanalytics package locally caches ADP & ECR data scrapes for 8 hours,
+    and projected stats for 1 hour. Older cached scrapes are dropped (upon checking)",
     "\n  - See ?clear_ffanalytics_cache() for how to manually clear the cache",
     "\n  - Use list_ffanalytics_cache() to see what is currently cached"
     )

R/source_scrapes.R

@@ -17,6 +17,31 @@ scrape_cbs = function(pos = c("QB", "RB", "WR", "TE", "K", "DST"), season = NULL
     scrape_week = week
   }
 
+  curr_cache = list_ffanalytics_cache(quiet = TRUE)
+  is_cached = "CBS Scrape" %in% curr_cache$object
+
+  if(is_cached) {
+    l_pos = get_cached_object("cbs_scrape.rds")
+
+    pos_match = all(toupper(pos) %in% toupper(sort(names(l_pos))))
+
+    if(isTRUE(pos_match)) {
+
+      scrape_message = paste0(
+        "\n",
+        "Using the CBS scrape that was cached ",
+        curr_cache$hr_min_since_cache[curr_cache$object == "CBS Scrape"],
+        " ago:"
+      )
+      message(scrape_message)
+      return(l_pos[pos])
+
+    } else {
+      clear_ffanalytics_cache("CBS Scrape")
+    }
+  }
+
+
   message("\nThe CBS scrape uses a 2 second delay between pages")
 
   base_link = paste0("https://www.cbssports.com/fantasy/football/")
@@ -97,6 +122,8 @@ scrape_cbs = function(pos = c("QB", "RB", "WR", "TE", "K", "DST"), season = NULL
   names(l_pos) = pos
   attr(l_pos, "season") = season
   attr(l_pos, "week") = week
+
+  cache_object(l_pos, "cbs_scrape.rds")
   l_pos
 }