Switched to rsqlite, and added process to clean db and fetch new IPs

geolocation

geoip.R

@@ -1,12 +1,32 @@
 library(tidyverse)
+library(RSQLite)
 library(ipapi)
 
-system("grep GraphTV /var/log/nginx/access.log | grep -vi bot | cut -d ' ' -f 1 | sort | uniq", intern = T) -> ips
+dbConnect(SQLite(), "access.db") %>%
+  dbGetQuery("delete from access where ip = '192.168.0.254' or user = 'maxx' or agent like '%bot%' or ip = '164.2.255.244';")
 
-read_delim("/var/log/nginx/access.log", delim = " ", col_names = c("ip", "null", "user", "timestamp", "zone", "req", "status", "size", "referer", "agent")) %>%
-  select(-null, -user, -zone, -size) %>%
-  mutate(timestamp = timestamp %>% str_sub(2) %>% as.POSIXct(format = "%d/%b/%Y:%H:%M:%S")) %>%
-  mutate_at(vars(status), factor) -> access
+dbConnect(SQLite(), "access.db") %>%
+  dbGetQuery("vacuum;")
 
-geolocate(ips) -> geoip
+src_sqlite("access.db") %>%
+  tbl("geoip") %>%
+  select(query) %>%
+  collect %>%
+  pull(query) -> known_ips
 
+src_sqlite("access.db") %>%
+  tbl("access") %>%
+  select(ip) %>%
+  collect %>%
+  pull(ip) %>%
+  unique %>%
+  setdiff(known_ips) -> ips
+
+if (length(ips) > 0)
+{
+  ips %>%
+    geolocate -> geoip
+
+  dbConnect(SQLite(), "access.db") %>%
+    dbWriteTable("geoip", geoip, append = T)
+}