From ae9c681c6d37062721ddc0e3b00b0f856adb7c29 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Lauri=20V=C3=B5sandi?= <lauri.vosandi@gmail.com>
Date: Sat, 16 Jun 2018 00:55:57 +0300
Subject: [PATCH] Apache logiparsija parandused

---
 raport/log_parsers.py |  7 +++++--
 raport/main.py        | 23 ++++++++++++++++++++---
 raport/template.html  |  2 +-
 3 files changed, 26 insertions(+), 6 deletions(-)

diff --git a/raport/log_parsers.py b/raport/log_parsers.py
index 03301a2..b1f6a72 100644
--- a/raport/log_parsers.py
+++ b/raport/log_parsers.py
@@ -4,8 +4,9 @@ import sys
 from datetime import datetime
 
 class GenericParser(object):
-    def __init__(self, fh): # siia võid anda sys.stdin, gzip.open, open vms file handle tüüpi obj
+    def __init__(self, fh, errors_only=False): # siia võid anda sys.stdin, gzip.open, open vms file handle tüüpi obj
         self.fh = fh
+        self.errors_only = errors_only
 
     def __iter__(self):
         multiline_message = ""
@@ -21,7 +22,9 @@ class GenericParser(object):
                 continue
             m = re.match(self.RE_LOG_ENTRY, line)
             if m:
-                if log_entry and self.is_serious(log_entry):
+                if log_entry:
+                    if self.errors_only and not self.is_serious(log_entry):
+                        continue
                     stack_trace = "\n".join(multiline_message.split("\n")[1:])
                     event_count += 1
                     row = \
diff --git a/raport/main.py b/raport/main.py
index 136e50f..1355158 100644
--- a/raport/main.py
+++ b/raport/main.py
@@ -4,8 +4,16 @@
 """
 Sõltuvuste paigladamiseks:
 
-  apt install -y python-jinja2 python-lxml python-pygal
-  dnf install -y python-jinja2 python-lxml python-pygal
+  apt install -y python-jinja2 python-lxml python-pygal python-geoip python-numpy python-matplotlib
+  dnf install -y python-jinja2 python-lxml python-pygal python2-GeoIP python2-numpy python2-matplotlib
+
+Käivitamiseks
+
+  (cat /path/to/access.log; zcat /path/to/access.log.1.gz) | python main.py
+  (cat /path/to/access.log; zcat /path/to/access.log.[12].gz) | python main.py
+  (cat /path/to/access.log; zcat /path/to/access.log.[1-5].gz) | python main.py
+  (cat /path/to/access.log; zcat /path/to/access.log.*.gz) | python main.py
+
 """
 
 import GeoIP
@@ -28,7 +36,11 @@ gi = GeoIP.open("/usr/share/GeoIP/GeoIP.dat", GeoIP.GEOIP_MEMORY_CACHE)
 print "Loen standardsisendist..."
 for timestamp, log_entry, stack_trace, byte_count, line_count, event_count in ApacheLogParser(sys.stdin):
     country_code = gi.country_code_by_addr(log_entry.get("remote_addr"))
-    hits_per_path[log_entry.get("path")] += 1
+    if not country_code:
+        # Jäta vahele sisevõrgu aadressid, localhost jms mis ei lahendu riigiks
+        continue
+    if int(log_entry.get("status")) < 400: # 2xx ja 3xx arvestamiseks
+        hits_per_path[log_entry.get("path")] += 1
     hits_per_remote_addr[log_entry.get("remote_addr")] += 1
     hits_per_user_agent[log_entry.get("user_agent")] += 1
     hits_per_country[country_code] += 1
@@ -71,6 +83,11 @@ print "Kaart salvestatud faili top.svg"
 ### Koosta päringute arvu graafik päevade lõikes ###
 ####################################################
 
+# matplotlibi serveris kasutamiseks tuleb välja vahetada vaikimisi joonsitamise backend
+# https://matplotlib.org/tutorials/introductory/usage.html#what-is-a-backend
+import matplotlib
+matplotlib.use("Agg")
+
 import matplotlib.pyplot as plt
 fig = plt.figure(figsize=(10, 7))
 
diff --git a/raport/template.html b/raport/template.html
index 9916b19..51e8552 100644
--- a/raport/template.html
+++ b/raport/template.html
@@ -9,7 +9,7 @@
     <script type="text/javascript" src="http://kozea.github.com/pygal.js/latest/pygal-tooltips.min.js"></script>
     <style>
     svg, img {
-        max-width: 80%;
+        max-width: 100%; /* joonise laius 100% lehe laiusest */
     }
     </style>
   </head>