Server synchronisation

gen_site.sh

@@ -1,10 +1,91 @@
 #!/bin/sh
 
 BASEURL="localhost"
+SALT_TEAM="connected"
+OUT_TEAM="./teams"
+
+MAX_PARAL=9
+
+DEBUG=0
+
 
 cd `dirname "$0"`
 
+if [ "$UID" = "0" ]
+then
+    SCRIPT=`pwd`/`basename "$0"`
+    su -c "sh -c '$SCRIPT $@'" synchro
+    exit $?
+fi
+
+if [ -f "/tmp/generate_site" ]
+then
+    echo "This script is already running" 1>&2
+    echo "Remove the file /tmp/generate_site if you are sure this is not true" 1>&2
+    exit 1
+fi
+
+touch /tmp/generate_site
+
+WGET_OPT="--no-check-certificate -c"
+
+if [ $DEBUG -ne 1 ]
+then
+    WGET_OPT="-q"
+fi
+
 mkdir -p out
 cd out
 
-wget -c -m https://$BASEURL/ https://$BASEURL/connected/
+# First, remove existing version if any
+rm -rf "$BASEURL" "$OUT_TEAM"
+
+wget $WGET_OPT -m -b "http://$BASEURL/" -o /dev/null
+
+mkdir -p "$BASEURL"
+ln -sf "`pwd`/../files/" "$BASEURL/files"
+
+NB=0
+PIDLIST=
+# Get list of teams and fetch them in parallel
+for l in $(curl -k "http://$BASEURL/$SALT_TEAM/" 2> /dev/null | grep -oE "/[^/]+/[0-9]+/")
+do
+    (
+	wget $WGET_OPT -m "http://$BASEURL/$l"
+
+	for m in $(grep -R "<form " "$BASEURL/$l" | grep -oE "/[^/]+/([^/]+)/([0-9]+)-[^/]+/([a-zA-Z0-9_]+)/submission")
+	do
+	    OUT=`echo "$m" | sed -E 's#/([^/]+)/([^/]+)/([0-9]+)-[^/]+/([a-zA-Z0-9_]+)/submission#\1/\2/submission-\3-\4#'`
+	    wget $WGET_OPT "http://$BASEURL/$m" -O "$BASEURL/$OUT.html"
+	done
+
+	# Remove /connected/XY
+	sed -Ei "s#/[^/]+/([0-9]+)/#/#" "$BASEURL/$l/"*
+	sed -Ei "s#/([0-9]+)-[^/]*/([a-zA-Z0-9_]+)/submission#/submission-\1-\2.html#" "$BASEURL/$l/"*
+    ) &
+
+    PIDLIST="$PIDLIST $!"
+    NB=$(($NB + 1))
+
+    if [ $NB -ge $MAX_PARAL ]
+    then
+	echo "Generating teams ...$PIDLIST"
+	wait $PIDLIST
+	PIDLIST=
+	NB=0
+    fi
+done
+
+echo "Generating teams ...$PIDLIST"
+wait $PIDLIST
+
+rm /tmp/generate_site
+
+# Move connected/ at root
+mv "$BASEURL/$SALT_TEAM/" "$OUT_TEAM"
+
+# Remove all robots.txt
+find . -name robots.txt -exec rm {} \;
+
+# Remove useless symlink
+rm "$BASEURL/files"