Curate Dropbox - List file and folder sizes to identify deletion targets

curate_dropbox.R

library(rdrop2) # devtools::install_github("karthik/rdrop2")
library(dplyr)

top_level_path <- function(x){
  stringr::str_extract(x, "^(.+?)(\\/)")
}

rdrop2::drop_auth()

file_list_raw <- drop_dir(recursive = TRUE)
# saveRDS(file_list_raw, "dropbox-file-listing_2020-09-07.rds")
# file_list_raw <- readRDS("dropbox-file-listing_2020-09-07.rds")
file_list     <- file_list_raw %>% 
  dplyr::filter(.tag == "file") %>%
  dplyr::mutate(size          = size / 1000000, # convert from b to mb
                shared_folder = case_when(
                  !is.na(parent_shared_folder_id) ~ TRUE, 
                  TRUE ~ FALSE)
                ) %>% 
  dplyr::select(c("name", "size", "path_lower", "client_modified",
                  "shared_folder")) %>%
  dplyr::arrange(desc(size)) %>%
  dplyr::mutate(top_folder = top_level_path(path_lower)) %>%
  data.frame()

file_list %>% 
  group_by(top_folder) %>%
  summarise(size = sum(size)) %>%
  arrange(desc(size)) %>%
  data.frame()

# 2000 mb limit on free tier
sum(file_list$size) * 0.001 # mb to gb