geologist/geoip.R

library(tidyverse)
library(RSQLite)
library(ipapi)
library(leaflet)

dbConnect(SQLite(), "access.db") -> db

read_delim("access.log", delim = " ", col_names = c("ip", "null1", "user", "timestamp", "zone", "req", "status", "size", "referer", "agent", "null2")) %>%
  mutate_all(na_if, "-") %>%
  filter(user %>% is.na) %>%
  filter(!ip %in% c("192.168.0.254", "164.2.255.244")) %>%
  mutate(timestamp = timestamp %>% as.POSIXct(format = "[%d/%b/%Y:%H:%M:%S")) %>%
  separate(req, into = c("method", "url", "version"), sep = " ") %>%
  select(ip, timestamp, url, status, referer, agent) %>%
  dbWriteTable(conn = db, name = "access", value = ., append = T)

known_ips <- character(0)

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)
}

src_sqlite("access.db") %>%
  tbl("access") %>%
  collect %>%
  left_join(geoip, by = c("ip" = "query")) %>%
  select(-null1, -user, -zone, -size, -null2, -as, -countryCode, -org, -region, -regionName, -status.y, -timezone, -zip, -isp) %>%
  mutate(timestamp = timestamp %>% str_sub(2) %>% as.POSIXct(format = "%d/%b/%Y:%H:%M:%S"),
         req = req %>% str_replace("(HEAD|GET|POST) ", ""),
         req = req %>% str_replace("/(.*?/).*", "\\1")) -> geoaccess

leaflet(data = geoaccess) %>%
  addProviderTiles(providers$CartoDB.Positron) %>%
  addMarkers(~lon, ~lat, clusterOptions = markerClusterOptions())