source: trunk/bibopa.sh @ 21

#!/bin/bash
#
#    Behera, S. K., J. Luo, S. Masson, S. Rao, S. Gualdi, P. Delecluse, A.
#    Navarra and T. Yamagata, 2004 : Paramount Impact of the Indian Ocean
#    Dipole on the East African Short Rains: A CGCM Study, J. Climate, In
#    press.
#    
#    donnerait
#    
#    <biblioentry id="behara2004">
#    <authorgroup>
#    <author><surname>Behera</surname> <firstname>S. K.</firstname> </author>
#    <author><firstname>J.</firstname> <surname>Luo</surname></author>
#    <author><firstname>S.</firstname> <surname>Masson</surname></author>
#    <author><firstname>S.</firstname> <surname>Rao</surname></author>
#    <author><firstname>S.</firstname> <surname>Gualdi</surname></author>
#    <author><firstname>P.</firstname> <surname>Delecluse</surname></author>
#    <author><firstname>A.</firstname> <surname>Navara</surname></author>
#    <author><firstname>T.</firstname> <surname>Yamagata</surname></author>
#    </authorgroup>
#    <date>2004</date>
#    <title>Paramount Impact of the Indian Ocean Dipole on the East African
#    Short Rains: A CGCM Study</title>
#    <publishername>J. Climate</publishername>
#    <biblioid class="doi">doi</bibliomisc>
#    <bibliomisc role="pseudoref">In press.</bibliomisc>
#    <bibliomisc role="internalref">from 
#    http://www.lodyc.jussieu.fr/~opatlod/NEMO_v1/6_Menu/2_page/index.html
#    2007-03-29T16:24:31Z fplod by hand</bibliomisc>
#    </biblioentry>
#    
#
# example :
# $ ./bibopa.sh -i /Users/smasson/Bibopa/biball.txt -t raw
# $ ./bibopa.sh -i data/mail2007-04-25T08:58:16Z.txt -t mailbody
#
# see also mailtousernemo.sh
#
# original location :
# /usr/home/fplod/incas/bibnemo/src/bibnemomaf/bibopa.sh sur cerbere.locean-ipsl.upmc.fr
#
# update
# ++ gestion des comments 
# ++ gestion des id existants (cf à la fin)
# ++ option debug
# fplod 2007-06-20T17:18:02Z aedon.locean-ipsl.upmc.fr (Darwin)
# <bibliomisc role="id"> replace by <biblioid class="doi">
# smasson 2007-06-07T16:43:42Z arete.locean-ipsl.upmc.fr (Darwin)
# Add journals
# fplod 2007-05-10T09:17:09Z aedon.locean-ipsl.upmc.fr (Darwin)
# dernières touches
# fplod 2007-04-25T10:59:49Z cerbere.locean-ipsl.upmc.fr (Linux)
# add a filein parameter and an option for mailbody
# comments  (line begininig with #) are now possible
# Sebastien Masson avril 2007 creation
# 

rmbl () {
    echo "$1" | sed -e "s/^ *//" | sed -e "s/ *$//"
}
cleanname () {
    echo "$1" | sed -e "s/^ *//" \
        -e "s/^ *,//" \
        -e "s/^ *;//" \
        -e "s/^ *\.//" \
        -e "s/ *$//" \
        -e "s/, *$//" \
        -e "s/; *$//" \
        -e "s/\. *$//"
}
#
set -o posix
command=$(basename ${0} .sh)
log_date=$(date -u +"%Y-%m-%dT%H:%M:%SZ")
log=/tmp/${command}.${log_date}
#
# test if xsltproc is available
type xsltproc 1> /dev/null 2>&1
status=${?}
if [ ${status} -ne 0 ]
then
 echo " eee : xsltproc not found"
 exit 1
fi
#
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}
;;
mailbody) # file like data/mail2007-04-25T08:58:16Z.txt
 fileraw=/tmp/$(basename ${filein}).raw
 echo "# from ${filein}" > ${fileraw}
 echo "# ${log_date}" >> ${fileraw}
 echo "# corrections" >> ${fileraw}
 grep "correction_.*=" ${filein} | grep -v "correction_.*=$" | \
 sed -e "s/correction_.*=//">> ${fileraw}
 echo "# new references" >> ${fileraw}
 awk "/newreferences=/,/comments=/" ${filein} | sed -e "s/newreferences=//" -e "/comments=/d" >> ${fileraw}
 #more ${fileraw} # ++ if debug
 #read a #++ if debug
;;
*) 
   echo "eee : type should be raw or mailbody"
   exit 1
;;
esac
#
# check for output
fileou=$( basename ${filein} .txt).xml
if [ -f ${fileou} ]
then
   echo "eee : ${fileou} already exist"
   #exit 1 # ++ if not debug
fi
#
jlist[1]="J. Climate"
jlist[2]="Journal of Climate"
jlist[3]="Ocean Modelling"
jlist[4]="Geophys. Res. Lett."
jlist[5]="J. Geophys. Res."
jlist[6]="Tellus A"
jlist[7]="Tellus B"
jlist[8]="J. Phys. Oceanogr."
jlist[9]="Clim. Dyn."
jlist[10]="Climate Dynamics"
jlist[11]="Dyn. Atmos. Oceans"
jlist[12]="Mon. Wea. Rev."
jlist[13]="Global Biogeochem. Cycles"
jlist[14]="Nonlinear Processes in Geophysics"
jlist[15]="Ocean Science"
jlist[16]="J. Mar. Systems"
jlist[17]="J. Atmos. Sc."
jlist[18]="Proc. Royal Soc."
jlist[19]="Bull. Amer. Meteorol. Soc."
jlist[20]="Ocean Dyn."
jlist[21]="Geophysical Monograph Series"
jlist[22]="Paleoceanography"
jlist[23]="Annales Geophysicae"
jlist[24]="Annals of Geophys."
jlist[25]="Deep Sea Research Part II: Topical Studies in Oceanography"
jlist[26]="Deep Sea Res. II"
jlist[27]="Atmospheric Chemistry and Physics"
jlist[28]="Atmospheric Chemistry and Physics Discussions"
jlist[29]="Earth Plan. Sc. Lett."
jlist[30]="J. Mar. Res."
jlist[31]="PCMDI Report Series"
jlist[32]="J. Mar. Syst."
jlist[33]="Note du Pole de Modelisation"
jlist[34]="Calculateurs Paralleles"
jlist[35]="Note Scientifique du Pole de Modelisation"
jlist[36]="Lecture Notes in Computer Science"
jlist[37]="Prog. Oceanogr."
jlist[38]="Deep Sea Research Part I: Oceanographic Research Papers"
jlist[39]="Deep Sea Res. I"
jlist[40]="IEEE Trans. Geosci. Remote Sensing"
jlist[41]="ECMWF Tech. Memorandum"
jlist[42]="J. Environ. Radioactivity"
jlist[43]="Int. WOCE Newsletter"
jlist[44]="C. R. Acad. Sci. Paris"
jlist[45]="C. R. Acad. Sci. Paris, Earth and Planetary Sciences"
jlist[46]="In The mathematics of models for climatology and environment"
jlist[47]="Oceanol. Acta"
jlist[48]="The global atmosphere and ocean system"
jlist[49]="NATO Advanced Study Institute"
jlist[50]="WCRP"
jlist[51]="The Courier"
jlist[52]="Elsevier Oceanographic Series"
jlist[53]="In Conference Proceedings of the 1988 International Conference on Supercomputing"
jlist[54]="In Science and engineering on Cray Supercomputers"
jlist[55]="In Modeling the Earth's Climate and its Variability"
jlist[56]="Fish. Oceanogr."
jlist[57]="Q. J. R. Meteor. Soc."
jlist[58]="In High performance computing in the geosciences"
jlist[59]="Int. J. Numer. Meth. in Fluids"
jlist[60]="Lecture notes in Physics"
jlist[61]="J. Meterol. Soc. Japan"
jlist[62]="Journal of Marine Systems"
jlist[63]="Canadian Journal of Fisheries and Aquatic Sciences"
jlist[64]="J. Atmos. Ocean. Tech."
jlist[65]="EOS"
jlist[66]="Nature"
jlist[67]="Physics and Chemistry of the Earth, Part B: Hydrology, Oceans and Atmosphere"
jlist[68]="Journal of Atmospheric and Oceanic Technology"
jlist[69]="Lecture Notes in Computational Science Engineering"
jlist[70]="In Greenhouse Gas Control Technologies"
jlist[71]="Chemical Geology"
jlist[72]="Marine Geology"
jlist[73]="Developments in Paleoenvironmental Research"
jlist[74]="Science"
jlist[75]="In Mediterranean Climate Variability"
jlist[76]="Global and Planetary Change"
jlist[77]="Physical Review Letters"
#jlist[]=""
#
#
    cat <<EOF > ${fileou}
<?xml version='1.0' encoding='ISO-8859-1'?>
<!DOCTYPE bibliography PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
 "http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd">
<bibliography id="bibrefnemo" lang="en">
<bibliographyinfo>
<date>$( date -u +"%Y-%m-%dT%H:%M:%SZ" )</date>
</bibliographyinfo>
EOF
#
# strip comments
fileraw_strict=/tmp/$(basename ${fileraw})_strict
grep -v "^#" ${fileraw} > ${fileraw_strict}
totlines=$( wc -l ${fileraw_strict} | awk '{print $1}' )
l=1
while [ ${l} -le ${totlines}  ]
do
# extract one line

  line=$( sed -n ${l}p ${fileraw_strict} )
  orgline=$( echo ${line} | sed -e "s/--/- -/g" )
  line=$( echo ${line} | sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" )
# before the first :
  tmp=${line%%:*}
# before the last ,
  auths=${tmp%,*},
# supress and
  auths=$( echo "${auths}" | sed -e "s/ and //g" )
# after the last ,
  year=${tmp##*,}
  year=$( rmbl "${year}" )
## first author before the first .,
  first=${auths%%.,*}.
# its firstname after the last ,
  firstfn=${first##*,}
  firstfn=$( rmbl "${firstfn}" )
# its surname ; before the first ,
  firstsn=${first%%,*}
  firstsn=$( rmbl "${firstsn}" )
## ref id
  refid=$( echo ${firstsn} | tr "[:upper:]" "[:lower:]" | tr -s " " "_"  | tr -s "'" "_" | recode -d -f ISO-8859-1..flat )${year}
  num=$( grep -c "<biblioentry id=\"${refid}_[0-9][0-9]\">" ${fileou} )
  num=$(( ${num} + 1 ))
  [ ${num} -le 9 ] && num=0${num} 
  refid=${refid}_${num}
  
          cat <<EOF >> ${fileou}
<biblioentry id="${refid}">
  <!-- date
  $( date -u +"%Y-%m-%dT%H:%M:%SZ" )
  -->
  <!-- original text
  $orgline
  -->
  <authorgroup>
    <author> <personname> <surname>${firstsn}</surname> <firstname>${firstfn}</firstname> </personname> </author>
EOF
  
## other authors..
  previous=${first},
# next authors...
  next=${auths##*${previous}}
# while the next author is not empty
  while [  "${next}" != "" ]
    do
# get the first next author; before the first ,
    next=${next%%,*}
# its surname ; after the last .
    nextsn=${next##*.}
    nextsn=$( rmbl "${nextsn}" )
# its firstname ; before the last .
    nextfn=${next%.*}.
    nextfn=$( rmbl "${nextfn}" )
#
    echo "    <author> <personname> <surname>${nextsn}</surname> <firstname>${nextfn}</firstname> </personname> </author>" >> ${fileou}
    previous=${next},
    next=${auths##*${previous}}
    
  done
  echo "  </authorgroup>"  >> ${fileou}
  
# end of the line ; after the first :
  endline=${line#*:}
  
## find the journal
  j=1
  jfound=""
  jlistsize=${#jlist[@]}
  while [[ $j -le $jlistsize && "${jfound}" == "" ]]
    do
    ok=$( echo ${endline} | grep -ci "${jlist[j]} *," ) 
    [ $ok -eq 1 ] && jfound="${jlist[j]}"
    j=$(( $j + 1 ))
  done
  if [ "$jfound" == "" ]
      then
      echo "ERROR Journal not found"
      echo "${endline}"
      exit
  fi
## title 
# before the first :
  title=${endline%%${jfound}*}
  title=$( cleanname "${title}" )
  echo "  <title>${title}</title>" >> ${fileou}
## end 
## end of the line ; after the first ${jfound}
  endline=${endline#*${jfound}}
  endline=$( cleanname "${endline}" )
## doi
  endline=$( echo ${endline} | sed -e "s/[dD][oO][iI] *\t* *: *\t* */doi:/" )
  ok=$( echo ${endline} | grep -ic "doi:" )
  if [ $ok -eq 1 ]
      then
      doi=${endline##*doi:}
      echo "  <biblioid class=\"doi\">${doi}</biblioid>" >> ${fileou}
      endline=${endline%doi:*}
      endline=$( cleanname "${endline}" )
  else
      echo "non doi: ${line}"
  fi
  num=$( echo ${endline} |  tr -dc "," | wc -c )
  case ${num} in
      1) 
### echo ${num}: ${endline}
          vol=${endline%,*}
          vol=$( cleanname "${vol}" )
          pag=${endline##*,} 
          pag=$( cleanname "${pag}" )
          cat <<EOF >> ${fileou}
  <biblioset relation="journal">
    <title>${jfound}</title>
    <volumenum>${vol}</volumenum><pagenums>${pag}</pagenums>
    <pubdate>${year}</pubdate>
  </biblioset>
EOF
      ;;
      2) 
          vol=${endline%,*}
          vol=$( cleanname "${vol}" )
          iss=${vol##*,} 
          iss=$( cleanname "${iss}" )
          vol=${vol%,*}
          vol=$( cleanname "${vol}" )
          pag=${endline##*,} 
          pag=$( cleanname "${pag}" )
          cat <<EOF >> ${fileou}
  <biblioset role="journal">
    <title>${jfound}</title>
    <volumenum>${vol}</volumenum><issuenum>${iss}</issuenum><pagenums>${pag}</pagenums>
    <pubdate>${year}</pubdate>
  </biblioset>
EOF
      ;;
      *)
echo ${num}: ${endline}
          cat <<EOF >> ${fileou}
  <biblioset role="journal">
    <title>${jfound}</title>
    <pubdate>${year}</pubdate>
    <bibliomisc>${endline}</bibliomisc>
  </biblioset>
EOF
      ;;
  esac
 
          cat <<EOF >> ${fileou}
</biblioentry>
  
EOF




  
  l=$(( $l + 1 ))
  
done
echo "</bibliography>" >> ${fileou}

xsltproc \
--output $( basename ${filein} .txt).html \
http://docbook.sourceforge.net/release/xsl/current/xhtml/docbook.xsl \
${fileou}

xmlto pdf ${fileou} 2> xml.err
#
# clean
echo "iii : xml.err contains stderr from the following command "
echo "iii : which was done just to check consistence of ${fileou}" 
echo "iii : xmlto pdf ${fileou}"
rm -i xml.err
case ${type} in
raw) 
 echo "iii : ${fileraw_strict} contains a copy of input file without comments"
 rm -i ${fileraw_strict}
;;
mailbody)
 echo "iii : ${fileraw} contains a copy of input file"
 rm -i ${fileraw}
 echo "iii : ${fileraw_strict} contains a copy of input file without comments"
 rm -i ${fileraw_strict}
;;
esac
#
echo " iii : you have a some new or modified bibliographic references in ${fileou} (biblioentry)"
echo " iii : you can add them in bibrefnemo.xml" #++ filename path
echo " iii : modify date in bibrefnemo.xml" #++ filename path
echo " iii : but please take care of id last part (after underscore) that may"
echo " iii : be in conflict with existing ones" #++ pour s'affranchir de ce souci il faudrait donner en entrée complémentaire le fichier à compléter.
#
exit 0