source: trunk/twindoi.sh @ 140

#! /bin/sh
#+
#
# .. program:: twindoi.sh
#
# ==========
# twindoi.sh
# ==========
#
# --------------------------
# detection of duplicate DOI
# --------------------------
#
# SYNOPSIS
# ========
#
# ::
#
#  $ twindoi.sh -i filein -t type
#
#
# DESCRIPTION
# ===========
#
#
# .. option:: -i <filein>
# .. option:: -t <type>
#
# detection of duplicate DOI
#
# EXAMPLES
# ========
#
# ::
#
#  $ ./twindoi.sh -i data/biball.txt -t raw
#
#
# ::
#
#  $ ./twindoi.sh -i data/biball.xml -t xml
#
# TODO
# ====
#
# ++ 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
#
# EVOLUTIONS
# ==========
#
# $Id$
#
# - fplod 20100318T083708Z aedon.locean-ipsl.upmc.fr (Darwin)
#
#   * unset
#
# - 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
unset system
#
set -u
#
command=$(basename ${0})
log_date=$(date -u +"%Y%m%dT%H%M%SZ")
log=/tmp/$(basename ${command} .sh).log.${log_date}
#
usage=" Usage : ${command} -i filein -t type"
#
while [ ${#} -gt 0 ]
do
   case ${1} in
      -i)
         filein=${2}
         shift
      ;;
      -t)
         type=${2}
         shift
      ;;
      *)
        # other choice
        echo "eee : unknown option ${1}"
        echo "${usage}"
        exit 1
      ;;
   esac
   # next flag
   shift
done
unset usage
#
# 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} : ${l1}"
   unset l1
   unet l2
   n=$(( ${n} + 1 ))
done
unset n
#
rm /tmp/doilist.txt 2> /dev/null
exit 0