From 03134c4260f724008d6600a2296ccdc86f24c490 Mon Sep 17 00:00:00 2001
From: Maxime Wack <maximewack@free.Fr>
Date: Thu, 11 Jan 2018 09:38:46 +0100
Subject: [PATCH] Prototype (access.log parsing + geolocation)

---
 geoip.R | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/geoip.R b/geoip.R
index e69de29..891970c 100644
--- a/geoip.R
+++ b/geoip.R
@@ -0,0 +1,12 @@
+library(tidyverse)
+library(ipapi)
+
+system("grep GraphTV /var/log/nginx/access.log | grep -vi bot | cut -d ' ' -f 1 | sort | uniq", intern = T) -> ips
+
+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
+
+geolocate(ips) -> geoip
+