devbuild

R/filter_devbuilds.R

+#' Filters devbuilds.
+#' @description
+#' Uses dplyr. To be used with data.frame or connection object.
+#' returns data.frame with columns:
+#' date, addonName, addonVersion, resource, application, downloadCount, dayssince
+#' @param dataframe or connection object with table
+#' @param mindate included, defaults to 2000-01-01
+#' @param maxdate excluded, defaults to today
+#' @param settingsfile yamlfile with the same format as the default one
+#' @keywords devbuilds filter
+#' @import magrittr RPostgreSQL yaml
+#' @importFrom dplyr filter group_by summarize collect
+#' @export
+
+filter_devbuilds <-
+  function(input,
+           mindate = '2000-01-01',
+           maxdate = as.character(Sys.Date()),
+           redshift = FALSE,
+           settingsfile = system.file("filter_devbuilds_conf.yaml", package = "bbbi", mustWork = TRUE)
+   ) {
+    # focus on most used resources
+    settings <- yaml.load_file(settingsfile)
+    settings$resources_easy <- download_default_easylist_subscriptions()
+    settings$resources <- c(settings$resources_main, settings$resources_easy)
+
+    # we want an operator that will work for dplyr sql as well as for data.frame
+    # sql is already implemented by dplyr with RPostgreSQL
+    `%~%` <- function(x,y) grepl(y,x)
+
+    # input can be dataframe or connection object.
+    # some further code will not work with dates in factor format
+    if ( is.data.frame(input) && is.factor(input$date) ) {
+        input <- input %>% mutate(date = as.character(date))
+    }
+
+    # connection can be to postgresql or redshift
+    # redshift needs double slashes in regexp and only has lowercase table names
+    if ( redshift ) {
+        query_sum <- . %>%
+            group_by(date, addonname, addonversion, resource, application, downloadcount, dayssince) %>%
+            summarize(n = sum(n))
+        query <- . %>%
+            filter(date >= local(mindate) &&
+                   date < local(maxdate) &&
+                   resource %in% local(settings$resources) &&
+                   ((addonname %in% local(settings$addonNames_4dot) &&
+                     addonversion %~% '^[0-9]*\\\\.[0-9]*\\\\.[0-9]*\\\\.[0-9]*(-.*)?$') ||
+                    (addonname == local(settings$addonNames_3dot) &&
+                     addonversion %~% '^[0-9]*\\\\.[0-9]*\\\\.[0-9]*(-.*)?$') ||
+                    (addonname == local(settings$addonNames_all))) &&
+                   !(addonversion %~% '.*signed.*')
+                   ) %>%
+            query_sum
+    } else {
+        query_sum <- . %>%
+            group_by(date, addonName, addonVersion, resource, application, downloadCount, dayssince) %>%
+            summarize(N = sum(N))
+        query <- . %>%
+            filter(date >= local(mindate) &&
+                   date < local(maxdate) &&
+                   resource %in% local(settings$resources) &&
+                   ((addonName %in% local(settings$addonNames_4dot) &&
+                     addonVersion %~% '^[0-9]*\\.[0-9]*\\.[0-9]*\\.[0-9]*(-.*)?$') ||
+                    (addonName == local(settings$addonNames_3dot) &&
+                     addonVersion %~% '^[0-9]*\\.[0-9]*\\.[0-9]*(-.*)?$') ||
+                    (addonName == local(settings$addonNames_all))) &&
+                   !(addonVersion %~% '.*signed.*')
+                   ) %>%
+            query_sum
+    }
+
+    result <- input %>% query %>% collect %>% ungroup  # necessary for further code
+    # focus on most important resources:
+    # one of the easylists is delivered by default, therefore can be combined
+    # however, users can add more lists (for downloadCount > 0)
+    result <- result %>%
+        mutate(resource = ifelse(resource %in% settings$resources_main, resource, "easylist default")) %>%
+        query_sum
+
+    # redshift only has lowercase table names
+    if ( redshift ) {
+        result <- result %>%
+            rename(addonName = addonname) %>%
+            rename(addonVersion = addonversion) %>%
+            rename(downloadCount = downloadcount) %>%
+            rename(N = n)
+    }
+
+    return(result)
+}