From 718c0339891a9c87f03fcff62ef223f16829c41d Mon Sep 17 00:00:00 2001
From: Bastian Kleineidam <calvin@debian.org>
Date: Fri, 21 Sep 2012 14:50:33 +0200
Subject: [PATCH] Add alexa test run script.

---
 .gitignore            |  2 ++
 tests/run_alexa_1m.sh | 14 ++++++++++++++
 2 files changed, 16 insertions(+)
 create mode 100755 tests/run_alexa_1m.sh

diff --git a/.gitignore b/.gitignore
index df17cdf4..849376b6 100644
--- a/.gitignore
+++ b/.gitignore
@@ -32,3 +32,5 @@ Changelog.linkchecker*
 /LinkChecker-*.exe
 /LinkChecker.egg-info
 /todo
+/alexa*.log
+
diff --git a/tests/run_alexa_1m.sh b/tests/run_alexa_1m.sh
new file mode 100755
index 00000000..8cd10c1c
--- /dev/null
+++ b/tests/run_alexa_1m.sh
@@ -0,0 +1,14 @@
+#!/bin/sh
+# Run the top 1 million URLs as reported by alexa site monitoring.
+# The logfile should be searched for internal errors first, then
+# all unusual errors and warnings should be inspected.
+#
+# Note that the result is usually depending on the current location
+# from where the tests are run.
+set -u
+logfile=alexa_1m.log
+rm -f $logfile
+for url in $(cat $HOME/src/alexatopsites/top-1m.txt); do
+  ./linkchecker -r1 $url >> $logfile 2>&1
+done
+