#! /bin/sh
#+
#
# NAME
# ====
#
# twindoi.sh - detection of duplicate DOI
#
# SYNOPSYS
# ========
#
# ::
#
#  $ twindoi.sh -i filein -t type
#
#
# DESCRIPTION
# ===========
#
#
# detection of duplicate DOI
#
# EXAMPLES
# ========
#
# ::
#
#  $ ./twindoi.sh -i data/biball.txt -t raw
#
#
# ::
#
#  $ ./twindoi.sh -i data/biball.xml -t xml
#
# FILES
# =====
#
# original location
# ~~~~~~~~~~~~~~~~~
#
# /usr/home/fplod/src/superbib_ws/twindoi.sh sur aedon.locean-ipsl.upmc.fr
#
#
# EVOLUTIONS
# ==========
#
# ++ option debug
#
# ++ the following command wich is not convinient
# (xml vs txt) did not give any alert and check inside xml comments
# ::
#
#  $ ./twindoi.sh -i data/biball.xml -t raw
#
#
# $Id$
#
# - fplod 2008-05-05T14:26:31Z aedon.locean-ipsl.upmc.fr (Darwin)
#
#   * usage of xml(starlet) for doi extraction in xml file
#
# - fplod 2007-06-20T16:12:22Z aedon.locean-ipsl.upmc.fr (Darwin)
#
#   * consolidation and homogeneisation
#
# - smasson 2007-06-20T16:11:47Z
#
#   * creation
#
#-
#
system=$(uname)
case "${system}" in
 AIX|IRIX64)
  echo " www : no specific posix checking"
 ;;
 *)
  set -o posix
 ;;
esac
command=$(basename ${0})
log_date=$(date -u +"%Y-%m-%dT%H:%M:%SZ")
log=/tmp/$(basename ${command} .sh).${log_date}
#
usage=" Usage : ${command} -i filein -t type"
#
while [ ! -z "${1}" ] # ++ pb bash
do
 case ${1} in
 -i) # filein
  filein=${2}
  shift
 ;;
 -t) # type
  type=${2}
  shift
 ;;
 *) # other choice
  echo "${usage}"
  exit 1
 ;;
 esac
 shift # next flag
done
set -u
#
# check for filein
if [ ! -f ${filein} ]
then
  echo "eee : ${filein} not found"
  exit 1
fi
#
case ${type} in
raw) # file like data/biball.txt
 fileraw=${filein}
;;
xml)  # file like data/biball.xml
 filexml=${filein}
;;
*)
   echo "eee : type should be raw or xml"
   exit 1
;;
esac
#
case ${type} in
     raw)
        grep -i "doi:" ${fileraw} | \
        sed -e "s/^.*doi: *//" | \
        sed -e "s/^\(.*\)\.$/ \1/" | \
	grep -v "???" | \
        sort -d > /tmp/doilist.txt
        ;;
     xml)
        xml sel -N dbk="http://docbook.org/ns/docbook" \
        -t -m "//dbk:biblioid[@class='doi']" -v . -n ${filexml} | \
	grep -v "???" | \
        sort -d > /tmp/doilist.txt
        ;;
     *)
        echo "eee : error unknown file type"
        exit 1
        ;;
esac
#
nl=$( cat /tmp/doilist.txt | wc -l )
if [ ${nl} -eq 0 ]
then
   echo "www : no DOI found in ${filein}"
   rm /tmp/doilist.txt 2> /dev/null
   exit 1
fi
n=1
while [ ${n} -lt ${nl} ]
   do
   l1=$( head -${n} /tmp/doilist.txt | tail -1 )
   l2=$( head -$(( ${n} + 1 )) /tmp/doilist.txt | tail -1 )
   [ "${l1}" == "${l2}" ] && echo "eee : line ${n} : ${l1}"
n=$(( ${n} + 1 ))
done
#
rm /tmp/doilist.txt 2> /dev/null
exit 0