urlwatch

#
# urlwatch
#
# v1.4 2002-11-13
#
# Mathieu Lubrano <m.l@compaqnet.fr>
#
# This file is GPL.
#

#!/bin/sh
WRKDIR=~/.urlwatch
LOGFILE=$WRKDIR/urlwatch.log
URLFILE=$WRKDIR/urlwatch.conf
EMAIL=""
URLS=""
DEBUG=""
URLID=""

init() {

  source /etc/profile 1>/dev/null 2>&1

  if [ ! -d $WRKDIR ]; then
    mkdir $WRKDIR 1>/dev/null 2>&1
  fi

  cd $WRKDIR

  if [ ! -e $URLFILE ]; then
    echo "$URLFILE not found."
    exit 1
  fi

  URLS=`cat $URLFILE`

  rm -f $LOGFILE $LOGFILE.body

  if [ "$1" == "--help" ] || [ "$1" == "-h" ]; then
    echo "Usage:"
    echo "$0 [-e user@domain.com] [-d] [-u url id]"
    echo "-e: send email report to user@domain.com"
    echo "-d: debug"
    echo "-u: check only this url id"
    echo " "
    echo "Add urls in $WRKDIR/$URLFILE with this format to check urls:"
    echo "TheProgramName#http://www.site.com/theWebPage.html"
    echo "AnotherOne#http://www.domain.net/rep/"
    echo "..."
    echo ""
    exit 0
  fi

  if [ "$1" == "-d" ] || [ "$2" == "-d" ] || [ "$3" == "-d" ] || [ "$4" == "-d" ]; then
    DEBUG="yes"
  else
    unset DEBUG
  fi

  if [ "$1" == "-e" ]; then
    EMAIL=$2
  fi

  if [ "$2" == "-e" ]; then
    EMAIL=$3
  fi

  if [ "$3" == "-e" ]; then
    EMAIL=$4
  fi

  if [ "$1" == "-u" ]; then
    URLID=$2
  fi

  if [ "$2" == "-e" ]; then
    URLID=$3
  fi

  if [ "$3" == "-u" ]; then
    URLID=$4
  fi 
}

checkURLs() {

  URLS=`cat $URLFILE`

  if [ "$URLID" != "" ]; then
    echo "Checking $URLID only..."
  fi

  for url in $URLS; do

    # if the line is a comment, skip to next line
    commentTest=`echo $url | cut -b 1` 
    if [ "$commentTest" == "#" ]; then
      continue
    fi

    theName=${url%#*}
    theUrl=${url#*#}

    # if URLID is set, just check this one
    if [ "$URLID" != "" ] && [ "$URLID" != "$theName" ]; then
      # continue (skip test) until we found URLID = theName
      continue
    fi

    echo -n "Checking $theName @ $theUrl..."
  
    if [ ! -d "$theName" ]; then
      mkdir $theName 1>/dev/null 2>&1
    fi
    
    if [ ! -e "$theName/$theName" ]; then
      touch $theName/$theName
    fi

    mv $theName/$theName $theName/$theName.old 1>/dev/null 2>&1

    #wget -q $theUrl -O $theName/$theName
    lynx -reload -dump $theUrl 1>$theName/$theName 2>&1

    # execute content filtering if any
    if [ -e $theName/filter ]; then
      cd $theName
      source ./filter
      cd ..
    fi

    # avoid first test false alarm
    if [ ! -e $theName/$theName.old ]; then
      cp $theName/$theName $theName/$theName.old
    fi

    # avoid false alarm if lynx failed
    if [ ! -s $theName/$theName ]; then
      cp $theName/$theName.old $theName/$theName
    fi

    diffs=`diff $theName/$theName $theName/$theName.old`

    if [ -n "$DEBUG" ]; then
      echo " "
      cat $theName/$theName
    fi

    if [ ! "$diffs" == "" ]; then
      echo " <- UPDATED."
      echo "$theName $theUrl" >> $LOGFILE
      echo "$theName $theUrl" >> $LOGFILE.body
      echo " " >> $LOGFILE.body
      echo "$diffs" >> $LOGFILE.body
      echo "======================================================================"  >> $LOGFILE.body
      echo " " >> $LOGFILE.body
    else
      echo "done."
    fi

  done

  if [ -e $LOGFILE ]; then
    echo " " >> $LOGFILE
    echo "======================================================================"  >> $LOGFILE
    echo " " >> $LOGFILE
    cat $LOGFILE.body >> $LOGFILE
  fi

  rm -f $LOGFILE.body
}


emailReport() {
  if [ -e "$LOGFILE" ]; then
    echo -n "Sending email report $$..."
    cat $LOGFILE | nail -s "$0 report-$$ changed urls..." $EMAIL
    echo "done."
  fi
}

main() {
  init $*

  checkURLs

  if [ -n "$EMAIL" ]; then
    emailReport
  fi

  rm -f $LOGFILE $LOGFILE.body

  exit 0
}

main $*