source: trunk/twindoi.sh

#! /bin/sh
#+
#
# .. program:: twindoi.sh
#
# ==========
# twindoi.sh
# ==========
#
# SYNOPSIS
# ========
#
# .. code-block:: bash
#
#    twindoi.sh -i filein -t type
#
# DESCRIPTION
# ===========
#
# .. option:: -i <filein>
# .. option:: -t <type>
#
# detection of duplicate DOI
#
# EXAMPLES
# ========
#
# To detect duplicate DOI in a raw file:
#
# .. code-block:: bash
#
#    twindoi.sh -i data/biball.txt -t raw
#
# To detect duplicate DOI in a XML/DocBook file:
#
# .. code-block:: bash
#
#    twindoi.sh -i data/biball.xml -t xml
#
# To detect duplicate DOI in a bibtex file:
#
# .. code-block:: bash
#
#    twindoi.sh -i data/biball.xml -t bibtex
#
# TODO
# ====
#
# option debug
#
# the following command which is not convenient
# (xml vs txt) did not give any alert and check inside xml comments
#
# .. code-block:: bash
#
#    twindoi.sh -i data/biball.xml -t raw
#
# EVOLUTIONS
# ==========
#
# $Id$
#
# - fplod 20131010T113730Z callisto.locean-ipsl.upmc.fr (Linux)
#
#   * dynamic xmlstarlet vs xml
#
# - fplod 20120521T080342Z cratos (Linux)
#
#   * rename type variable to ftype to avoid usage of a reserved word
#   * revision of indentation
#   * typo
#   * add bibtex as file type
#
# - 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 homogenisation
#
# - smasson 2007-06-20T16:11:47Z
#
#   * creation
#
#-
#
system=$(uname)
case "${system}" in
    AIX|IRIX64)
        echo "${command} : www : no specific posix checking"
    ;;
    *)
        set -o posix
    ;;
esac
unset system
#
set -u
action=$(basename ${0} .sh)
command=$(basename ${0})
log_date=$(date -u +"%Y%m%dT%H%M%SZ")
log=${PROJECT_LOG}/$(basename ${command} .sh).log.${log_date}
#
tool=${xmlcmd}
type ${tool} 1> /dev/null 2>&1
status=${?}
if [ ${status} -ne 0 ]
then
    echo "${command} : eee : tool ${tool} not found"
    exit 1
fi
unset tool
unset status
#
usage=" Usage : ${command} -i filein -t type"
#
minargcount=4
#echo " narg ${#}"
if [ ${#} -lt ${minargcount} ]
then
    echo "${command} : eee : not enough arguments"
    echo "${usage}"
    exit 1
fi
unset minargcount
#
while [ ${#} -gt 0 ]
do
    case ${1} in
        -i)
            filein=${2}
            shift
        ;;
        -t)
            ftype=${2}
            shift
        ;;
        *)
            # other choice
            echo "${command} : eee : unknown option ${1}"
            echo "${usage}"
            exit 1
        ;;
    esac
    # next flag
    shift
done
unset usage
#
# check for filein
if [ ! -f ${filein} ]
then
    echo "${command} : eee : ${filein} not found"
    exit 1
fi
#
case ${ftype} in
    raw) # file like data/biball.txt
        fileraw=${filein}
    ;;
    xml) # file like data/biball.xml
        filexml=${filein}
    ;;
    bibtex) # file like data/biball.bib
        filebibtex=${filein}
    ;;
    *)
       echo "${command} : eee : type should be raw, xml or bibtex"
       exit 1
    ;;
esac
unset filein
#
case ${ftype} in
    raw)
        grep -i "doi:" ${fileraw} | \
        sed -e "s/^.*doi: *//" | \
        sed -e "s/^\(.*\)\.$/ \1/" | \
        grep -v "???" | \
        sort -d > ${PROJECT_LOG}/${action}${$}.txt
    ;;
    bibtex)
        grep -i "doi *= *" ${filebibtex} | \
        sed -e "s/^.*doi *= *//" | \
        sed -e "s/^\(.*\)\.$/ \1/" | \
        grep -v "???" | \
        sort -d > ${PROJECT_LOG}/${action}${$}.txt
    ;;
    xml)
        ${xmlcmd} sel -N dbk="http://docbook.org/ns/docbook" \
        -t -m "//dbk:biblioid[@class='doi']" -v . -n ${filexml} | \
        grep  -v "???" | \
        sort -d > ${PROJECT_LOG}/${action}${$}.txt
    ;;
    *)
        echo "${command} : eee : error unknown file type ${ftype}"
        exit 1
    ;;
esac
unset ftype
#
nl=$( cat ${PROJECT_LOG}/${action}${$}.txt | wc -l )
if [ ${nl} -eq 0 ]
then
    echo "${command} : www : no DOI found in ${filein}"
    rm ${PROJECT_LOG}/${action}${$}.txt 2> /dev/null
    exit 1
fi
n=1
while [ ${n} -lt ${nl} ]
do
    l1=$( head -${n} ${PROJECT_LOG}/${action}${$}.txt | tail -1 )
    l2=$( head -$(( ${n} + 1 )) ${PROJECT_LOG}/${action}${$}.txt | tail -1 )
    [ "${l1}" == "${l2}" ] && echo "${command} : eee : line ${n} : ${l1}"
    unset l1
    unset l2
    n=$(( ${n} + 1 ))
done
unset n
unset nl
#
rm ${PROJECT_LOG}/${action}${$}.txt 2> /dev/null
unset command
unset log
unset log_date
#
#++set
exit 0