#!/bin/bash
dir=/home/webadmin/twiki
USAGE="daily_cleanup
This script is to be run daily under the login of the apache web server
or root to do various maintainance operations on TWiki installations:
   * truncate the debug.txt & warning.txt file that can grow huge, and 
are
     only needed to temporarily debug
   * remove topics & attachments older than N days in some webs (Trash, 
Tmp)
   * Note that it does not remove Trash from today, only yesterday and 
beyond

USAGE: [options] [wiki-dir]
Options:
  -n do nothing, just prints what whould be done
wiki-dir by default is $dir
"

set -a 
#set -xv
unset CDPATH # avoid cd messages
doit=true
#doit=false

while test -z "${1##-*}" -a -n "$1";do case "$1" in
  -n) doit=false;;
  *)  echo "$USAGE"; exit 1;;
  esac;shift
done  
if test -d "$1"; then dir="$1"; fi

# local wiki data dir
cd $dir/data
# truncate debug.txt & warning.txt
#>debug.txt; chmod a+rw debug.txt 2>/dev/null
#>warning.txt; chmod a+rw warning.txt 2>/dev/null

system_web=_default
#cd _default
#system_topics=' '
#for i in `find . -name \*.txt`; do system_topics="$system_topics $i"; 
done
#system_topics="$system_topics "
#echo "The following system topics will be excluded from the sweep: 
'".$system_topics."'"

clean_web () {
  web=$1
  days=$2
  if test -d data/$web; then
# topics
    cd $dir/data/$web
    for fileInWeb in `find . -type f -name \*.txt -mtime +$days`; do
      if [ -f $dir/data/$system_web/$fileInWeb ]; then
        echo "Ignoring system topic ". $fileInWeb
      else
        remove_topic "$fileInWeb"
      fi
    done

# corresponding attachments
    if test -d ../../pub/$web; then
      cd ../../pub/$web
      for topic_and_attachment in `find . -type f -mtime +$days`; do
        topic_and_attachment="${topic_and_attachment#./}"
        if [ -f $dir/pub/$system_web/$topic_and_attachment ]; then
          echo "Ignoring as $topic_and_attachment is a system 
attachment" 
        else
          remove_attachment 
../../data/$web/"${topic_and_attachment%%/*}.txt" 
"${topic_and_attachment#*/}" "$topic_and_attachment"
        fi
      done
    fi

# now delete from the topic any references to attachments that have been 
deleted
    cd $dir/data/$web
    grep '^%META:FILEATTACHMENT[{]' *.txt| while read line; do
      topic="${line%%:*}"
      file="${line#*\"}"; file="${file%%\" *}"
      if test ! -e "../../pub/$web/${topic%.txt}/$file"; then
        remove_attachment $topic "$file" 
      fi
    done
    
  fi
}

# suppose we are in data/web, argument is file name
remove_topic () {
  if $doit; then
    rm -rf "$1"* 
    rm -rf ../../pub/$web/${1%.txt}
  else
    echo "$web: Removing topic ${1#./} & attachments"
  fi
}
# create a new topic file that does not mention the attachments
# $1= topic.txt $2= relative file name $3= complete path of file 
(optional)
remove_attachment () {
  topic="$1"; file="$2"; path="$3"
  if $doit; then
    mv $topic .topic
    fgrep -v "META:FILEATTACHMENT{name=\"$file\"" <.topic >$topic
    chmod a+rw $topic 2>/dev/null
    rm -f .topic
    if test -n "$path"; then rm -rf "$path"; fi
  else
    echo "$web: Removing attachment $file from topic $topic"
  fi
}

cd $dir
clean_web Trash 0
cd $dir
clean_web Tmp 100