From 5c9d080009c22465875a1765ddaa3425501b6f47 Mon Sep 17 00:00:00 2001
From: Will Woods <wwoods@redhat.com>
Date: Fri, 26 Jun 2020 16:08:50 -0400
Subject: [PATCH] roles/awstats/files/combineHttpLogs.sh: accept DATE_STR

This commit tweaks combineHttpLogs.sh so that you can provide an
optional DATE_STR on the CLI to have it run for a date other than the
default (which is currently 2 days ago, despite the comments about it
being somewhere between 5-3 days..). Links in `latest/` are not updated
when running for user-provided dates.

There's some other refactoring involved but AFAICT this doesn't change
its default behavior at all.

This should let us re-run this script manually to combine logs for the
days that we missed during the colo move etc.
---
 roles/awstats/files/combineHttpLogs.sh | 68 +++++++++++++++++++-------
 1 file changed, 50 insertions(+), 18 deletions(-)

diff --git a/roles/awstats/files/combineHttpLogs.sh b/roles/awstats/files/combineHttpLogs.sh
index a8d63f2b2b..fcafc2a0d3 100644
--- a/roles/awstats/files/combineHttpLogs.sh
+++ b/roles/awstats/files/combineHttpLogs.sh
@@ -19,26 +19,58 @@
 # along with Fedora Project Infrastructure Ansible Repository.  If
 # not, see <http://www.gnu.org/licenses/>.
 
-# Because sync-http may not get all logs for 3 days, we only merge
-# things after 4 days. 
-
-# 2019-10-01 Dropped this down to 3 days.. 
-
-NUMDAYS=2
-YEAR=$(/bin/date -d "-${NUMDAYS} days" +%Y)
-MONTH=$(/bin/date -d "-${NUMDAYS} days" +%m)
-DAY=$(/bin/date -d "-${NUMDAYS} days" +%d)
-
+# Some constants / standard paths
 LOGDIR=/var/log/hosts
 NFSDIR=/mnt/fedora_stats/combined-http
-PROXYLOG=${LOGDIR}/proxy*/${YEAR}/${MONTH}/${DAY}/http/
-DL_LOG=${LOGDIR}/dl*/${YEAR}/${MONTH}/${DAY}/http/
-PEOPLE=${LOGDIR}/people*/${YEAR}/${MONTH}/${DAY}/http/
-
-TARGET=${NFSDIR}/${YEAR}/${MONTH}/${DAY}
-
 LOGMERGE=/usr/share/awstats/tools/logresolvemerge.pl
 
+# Because sync-http may not get all logs immediately, we look back
+# a couple days to find the "latest" logs to merge.
+LATEST_LOG_DATE="2 days ago"
+
+# Funtion to parse a DATE_STR and print YYYY/MM/DD
+ymd() { date -d "$*" +%Y/%m/%d; }
+
+# Get YYYY/MM/DD for LATEST_LOG_DATE, for later use
+LATEST_YMD=$(ymd $LATEST_LOG_DATE)
+
+# Prints usage. Also serves as docs for anyone reading the source (hi there!)
+usage() {
+cat <<__USAGE__
+usage: $0 [DATE_STR]
+combine daily logs from $LOGDIR to $NFSDIR.
+
+Default date is '$LATEST_LOG_DATE' (currently $LATEST_YMD).
+DATE_STR may be any date older than that, in any format understood by date(1):
+  "June 9"
+  "2020-06-23 -2weeks"
+__USAGE__
+}
+
+# Check CLI args to set LOG_DATE
+case $# in
+    0) LOG_DATE="$LATEST_LOG_DATE"; UPDATE_LATEST=1 ;;
+    1) [ "$1" == "-h" -o "$1" == "--help" ] && usage && exit 0
+       LOG_DATE="$1" ;;
+    *) usage; exit 2 ;;
+esac
+
+# Parse LOG_DATE
+YMD=$(ymd $LOG_DATE) || exit 2
+
+# Safety check for dates that are too new for us to handle.
+# (Also catches weird "dates" that date(1) allows, like '' or '0' or 'wet')
+if [[ "$YMD" > "$LATEST_YMD" ]]; then
+    echo "$0: error: DATE_STR '$LOG_DATE' ($YMD) newer than LATEST_LOG_DATE ($LATEST_YMD)" >&2
+    exit 3
+fi
+
+# Okay we're good. Set paths, make directories, and do some merging!
+PROXYLOG=${LOGDIR}/proxy*/${YMD}/http/
+DL_LOG=${LOGDIR}/dl*/${YMD}/http/
+PEOPLE=${LOGDIR}/people*/${YMD}/http/
+
+TARGET=${NFSDIR}/${YMD}
 mkdir -p ${TARGET}
 
 ##
@@ -76,10 +108,10 @@ done
 # 3. remove the old links
 # 4. link up all the files we merged over
 
-if [[ -d ${NFSDIR}/latest ]]; then
+if [[ "$UPDATE_LATEST" && -d ${NFSDIR}/latest ]]; then
     pushd ${NFSDIR}/latest &> /dev/null
     /bin/rm -f *
-    for file in ../${YEAR}/${MONTH}/${DAY}/*; do
+    for file in ../${YMD}/*; do
 	ln -s ${file} .
     done
     popd &> /dev/null