| 1 | #!/bin/bash |
|---|
| 2 | # |
|---|
| 3 | # Behera, S. K., J. Luo, S. Masson, S. Rao, S. Gualdi, P. Delecluse, A. |
|---|
| 4 | # Navarra and T. Yamagata, 2004 : Paramount Impact of the Indian Ocean |
|---|
| 5 | # Dipole on the East African Short Rains: A CGCM Study, J. Climate, In |
|---|
| 6 | # press. |
|---|
| 7 | # |
|---|
| 8 | # donnerait |
|---|
| 9 | # |
|---|
| 10 | # <biblioentry xml:id="behara2004"> |
|---|
| 11 | # <authorgroup> |
|---|
| 12 | # <author><surname>Behera</surname> <firstname>S. K.</firstname> </author> |
|---|
| 13 | # <author><firstname>J.</firstname> <surname>Luo</surname></author> |
|---|
| 14 | # <author><firstname>S.</firstname> <surname>Masson</surname></author> |
|---|
| 15 | # <author><firstname>S.</firstname> <surname>Rao</surname></author> |
|---|
| 16 | # <author><firstname>S.</firstname> <surname>Gualdi</surname></author> |
|---|
| 17 | # <author><firstname>P.</firstname> <surname>Delecluse</surname></author> |
|---|
| 18 | # <author><firstname>A.</firstname> <surname>Navara</surname></author> |
|---|
| 19 | # <author><firstname>T.</firstname> <surname>Yamagata</surname></author> |
|---|
| 20 | # </authorgroup> |
|---|
| 21 | # <date>2004</date> |
|---|
| 22 | # <title>Paramount Impact of the Indian Ocean Dipole on the East African |
|---|
| 23 | # Short Rains: A CGCM Study</title> |
|---|
| 24 | # <publishername>J. Climate</publishername> |
|---|
| 25 | # <biblioid class="doi">doi</biblioid> |
|---|
| 26 | # <bibliomisc role="pseudoref">In press.</bibliomisc> |
|---|
| 27 | # <bibliomisc role="internalref">from |
|---|
| 28 | # http://www.lodyc.jussieu.fr/~opatlod/NEMO_v1/6_Menu/2_page/index.html |
|---|
| 29 | # 2007-03-29T16:24:31Z fplod by hand</bibliomisc> |
|---|
| 30 | # </biblioentry> |
|---|
| 31 | # |
|---|
| 32 | # @book{MetReiCoh2004 , |
|---|
| 33 | # author = {Michael Metcalf and |
|---|
| 34 | # John Reid and Malcolm Cohen}, |
|---|
| 35 | # title = {Fortran 95/2003 explained}, |
|---|
| 36 | # year = {2004}, |
|---|
| 37 | # publisher = {Oxford University Press}, |
|---|
| 38 | # edition = {Third}, |
|---|
| 39 | # ISBN = {0-19-852693-8}, |
|---|
| 40 | # pages = {434} |
|---|
| 41 | # } |
|---|
| 42 | # donnerait |
|---|
| 43 | # <biblioentry xml:id="MetReiCoh2004"> |
|---|
| 44 | # <authorgroup> |
|---|
| 45 | # <author><firstname>Michael</firstname> <surname>Metcalf</surname></author> |
|---|
| 46 | # <author><firstname>John</firstname> <surname>Reid</surname></author> |
|---|
| 47 | # <author><firstname>Malcolm</firstname> <surname>Cohen</surname></author> |
|---|
| 48 | # </authorgroup> |
|---|
| 49 | # <date>2004}</date> |
|---|
| 50 | # <title>Fortran 95/2003 explained</title> |
|---|
| 51 | # <publisher><publishername>Oxford University Press</publishername></publisher> |
|---|
| 52 | # <biblioid class="isbn">0-19-852693-8</biblioid> |
|---|
| 53 | # <pagenums>434</pagenums> |
|---|
| 54 | # <edition>Third</edition> |
|---|
| 55 | # |
|---|
| 56 | # example : |
|---|
| 57 | # $ ./bibopa.sh -p birefnemo -i data/biball.txt -t raw |
|---|
| 58 | # $ ./bibopa.sh -p bibrefnemo -i data/mail2007-05-10T09:01:56Z -t mailbody |
|---|
| 59 | # $ ./bibopa.sh -p polyfortran -i data/petitpolyfp.bib -t bibtex -l fr |
|---|
| 60 | # $ ./bibopa.sh -p demo1 -i data/bibdemo1.txt -t raw -l fr |
|---|
| 61 | # |
|---|
| 62 | # see also mailtouser.sh |
|---|
| 63 | # |
|---|
| 64 | # original location : |
|---|
| 65 | # /usr/home/fplod/incas/bibnemo/src/bibnemomaf/bibopa.sh sur cerbere.locean-ipsl.upmc.fr |
|---|
| 66 | # |
|---|
| 67 | # update |
|---|
| 68 | # ++ gestion des comments |
|---|
| 69 | # ++ gestion des id existants (cf à la fin) |
|---|
| 70 | # ++ option debug |
|---|
| 71 | # $Id$ |
|---|
| 72 | # fplod 2008-04-30T07:11:58Z aedon.locean-ipsl.upmc.fr (Darwin) |
|---|
| 73 | # chgt for dbk5 out |
|---|
| 74 | # xmlto 0.0.18 does'nt like this docbook release. remove |
|---|
| 75 | # usage of xml (xmlstarlet) for validation |
|---|
| 76 | # bug fix |
|---|
| 77 | # add journal names from demo1 |
|---|
| 78 | # fplod 2008-03-12T16:11:07Z aedon.locean-ipsl.upmc.fr (Darwin) |
|---|
| 79 | # add -p option (to overide bibrefnemo in file name and id) |
|---|
| 80 | # add -l option by default en |
|---|
| 81 | # fplod 2008-03-11T11:25:27Z aedon.locean-ipsl.upmc.fr (Darwin) |
|---|
| 82 | # add bibtex entry (not finished)++ |
|---|
| 83 | # fplod 2007-06-20T17:18:02Z aedon.locean-ipsl.upmc.fr (Darwin) |
|---|
| 84 | # <bibliomisc role="id"> replace by <biblioid class="doi"> |
|---|
| 85 | # smasson 2007-06-07T16:43:42Z arete.locean-ipsl.upmc.fr (Darwin) |
|---|
| 86 | # Add journals |
|---|
| 87 | # fplod 2007-05-10T09:17:09Z aedon.locean-ipsl.upmc.fr (Darwin) |
|---|
| 88 | # dernières touches |
|---|
| 89 | # fplod 2007-04-25T10:59:49Z cerbere.locean-ipsl.upmc.fr (Linux) |
|---|
| 90 | # add a filein parameter and an option for mailbody |
|---|
| 91 | # comments (line begininig with #) are now possible |
|---|
| 92 | # Sebastien Masson avril 2007 creation |
|---|
| 93 | # |
|---|
| 94 | |
|---|
| 95 | rmbl () { |
|---|
| 96 | echo "${1}" | sed -e "s/^ *//" | sed -e "s/ *$//" |
|---|
| 97 | } |
|---|
| 98 | cleanname () { |
|---|
| 99 | echo "${1}" | sed -e "s/^ *//" \ |
|---|
| 100 | -e "s/^ *,//" \ |
|---|
| 101 | -e "s/^ *;//" \ |
|---|
| 102 | -e "s/^ *\.//" \ |
|---|
| 103 | -e "s/ *$//" \ |
|---|
| 104 | -e "s/, *$//" \ |
|---|
| 105 | -e "s/; *$//" \ |
|---|
| 106 | -e "s/\. *$//" |
|---|
| 107 | } |
|---|
| 108 | # |
|---|
| 109 | set -o posix |
|---|
| 110 | command=$(basename ${0}) |
|---|
| 111 | log_date=$(date -u +"%Y-%m-%dT%H:%M:%SZ") |
|---|
| 112 | log=/tmp/${command}.${log_date} |
|---|
| 113 | # |
|---|
| 114 | # test if xsltproc is available |
|---|
| 115 | type xsltproc 1> /dev/null 2>&1 |
|---|
| 116 | status=${?} |
|---|
| 117 | if [ ${status} -ne 0 ] |
|---|
| 118 | then |
|---|
| 119 | echo " eee : xsltproc not found" |
|---|
| 120 | exit 1 |
|---|
| 121 | fi |
|---|
| 122 | # |
|---|
| 123 | # test if xml is available |
|---|
| 124 | type xml 1> /dev/null 2>&1 |
|---|
| 125 | status=${?} |
|---|
| 126 | if [ ${status} -ne 0 ] |
|---|
| 127 | then |
|---|
| 128 | echo " eee : xml not found" |
|---|
| 129 | exit 1 |
|---|
| 130 | fi |
|---|
| 131 | # |
|---|
| 132 | usage=" Usage : ${command} -i filein -t type -p project -l lang" |
|---|
| 133 | # |
|---|
| 134 | # default |
|---|
| 135 | lang=en |
|---|
| 136 | # |
|---|
| 137 | minargcount=6 |
|---|
| 138 | #echo " narg ${#}" |
|---|
| 139 | if [ ${#} -lt ${minargcount} ] |
|---|
| 140 | then |
|---|
| 141 | echo "eee : not enought arguments" |
|---|
| 142 | echo "${usage}" |
|---|
| 143 | exit 1 |
|---|
| 144 | fi |
|---|
| 145 | # |
|---|
| 146 | while [ ! -z "${1}" ] |
|---|
| 147 | do |
|---|
| 148 | case ${1} in |
|---|
| 149 | -i) # filein |
|---|
| 150 | filein=${2} |
|---|
| 151 | shift |
|---|
| 152 | ;; |
|---|
| 153 | -t) # type |
|---|
| 154 | type=${2} |
|---|
| 155 | shift |
|---|
| 156 | ;; |
|---|
| 157 | -p) |
|---|
| 158 | project=${2} |
|---|
| 159 | shift |
|---|
| 160 | ;; |
|---|
| 161 | -l) |
|---|
| 162 | lang=${2} |
|---|
| 163 | shift |
|---|
| 164 | ;; |
|---|
| 165 | -h) |
|---|
| 166 | echo "${usage}" |
|---|
| 167 | exit 0 |
|---|
| 168 | ;; |
|---|
| 169 | *) # other choice |
|---|
| 170 | echo "${usage}" |
|---|
| 171 | exit 1 |
|---|
| 172 | ;; |
|---|
| 173 | esac |
|---|
| 174 | shift # next flag |
|---|
| 175 | done |
|---|
| 176 | # |
|---|
| 177 | set -u |
|---|
| 178 | # |
|---|
| 179 | # check for filein |
|---|
| 180 | if [ ! -f ${filein} ] |
|---|
| 181 | then |
|---|
| 182 | echo "eee : ${filein} not found" |
|---|
| 183 | exit 1 |
|---|
| 184 | fi |
|---|
| 185 | # |
|---|
| 186 | case ${type} in |
|---|
| 187 | raw) # file like data/biball.txt |
|---|
| 188 | fileraw=${filein} |
|---|
| 189 | ;; |
|---|
| 190 | mailbody) # file like data/mail2007-04-25T08:58:16Z.txt |
|---|
| 191 | fileraw=/tmp/$(basename ${filein}).raw |
|---|
| 192 | echo "# from ${filein}" > ${fileraw} |
|---|
| 193 | echo "# ${log_date}" >> ${fileraw} |
|---|
| 194 | echo "# corrections" >> ${fileraw} |
|---|
| 195 | grep "correction_.*=" ${filein} | grep -v "correction_.*=$" | \ |
|---|
| 196 | sed -e "s/correction_.*=//">> ${fileraw} |
|---|
| 197 | echo "# new references" >> ${fileraw} |
|---|
| 198 | awk "/newreferences=/,/comments=/" ${filein} | sed -e "s/newreferences=//" -e "/comments=/d" >> ${fileraw} |
|---|
| 199 | #more ${fileraw} # ++ if debug |
|---|
| 200 | #read a #++ if debug |
|---|
| 201 | ;; |
|---|
| 202 | bibtex) # file like data/petitpoly.bib |
|---|
| 203 | fileraw=/tmp/$(basename ${filein}).raw |
|---|
| 204 | awk -f join_endcomma.awk ${filein} >> ${fileraw} |
|---|
| 205 | #more ${fileraw} # ++ if debug |
|---|
| 206 | #read a #++ if debug |
|---|
| 207 | ;; |
|---|
| 208 | *) |
|---|
| 209 | echo "eee : type should be raw, mailbody or bibtex" |
|---|
| 210 | exit 1 |
|---|
| 211 | ;; |
|---|
| 212 | esac |
|---|
| 213 | # |
|---|
| 214 | # check for output |
|---|
| 215 | fileou=$( basename ${filein} .txt).xml |
|---|
| 216 | if [ -f ${fileou} ] |
|---|
| 217 | then |
|---|
| 218 | echo "eee : ${fileou} already exist" |
|---|
| 219 | exit 1 |
|---|
| 220 | fi |
|---|
| 221 | # |
|---|
| 222 | jlist[1]="J. Climate" |
|---|
| 223 | jlist[2]="Journal of Climate" |
|---|
| 224 | jlist[3]="Ocean Modelling" |
|---|
| 225 | jlist[4]="Geophys. Res. Lett." |
|---|
| 226 | jlist[5]="J. Geophys. Res." |
|---|
| 227 | jlist[6]="Tellus A" |
|---|
| 228 | jlist[7]="Tellus B" |
|---|
| 229 | jlist[8]="J. Phys. Oceanogr." |
|---|
| 230 | jlist[9]="Clim. Dyn." |
|---|
| 231 | jlist[10]="Climate Dynamics" |
|---|
| 232 | jlist[11]="Dyn. Atmos. Oceans" |
|---|
| 233 | jlist[12]="Mon. Wea. Rev." |
|---|
| 234 | jlist[13]="Global Biogeochem. Cycles" |
|---|
| 235 | jlist[14]="Nonlinear Processes in Geophysics" |
|---|
| 236 | jlist[15]="Ocean Science" |
|---|
| 237 | jlist[16]="J. Mar. Systems" |
|---|
| 238 | jlist[17]="J. Atmos. Sc." |
|---|
| 239 | jlist[18]="Proc. Royal Soc." |
|---|
| 240 | jlist[19]="Bull. Amer. Meteorol. Soc." |
|---|
| 241 | jlist[20]="Ocean Dyn." |
|---|
| 242 | jlist[21]="Geophysical Monograph Series" |
|---|
| 243 | jlist[22]="Paleoceanography" |
|---|
| 244 | jlist[23]="Annales Geophysicae" |
|---|
| 245 | jlist[24]="Annals of Geophys." |
|---|
| 246 | jlist[25]="Deep Sea Research Part II: Topical Studies in Oceanography" |
|---|
| 247 | jlist[26]="Deep Sea Res. II" |
|---|
| 248 | jlist[27]="Atmospheric Chemistry and Physics" |
|---|
| 249 | jlist[28]="Atmospheric Chemistry and Physics Discussions" |
|---|
| 250 | jlist[29]="Earth Plan. Sc. Lett." |
|---|
| 251 | jlist[30]="J. Mar. Res." |
|---|
| 252 | jlist[31]="PCMDI Report Series" |
|---|
| 253 | jlist[32]="J. Mar. Syst." |
|---|
| 254 | jlist[33]="Note du Pole de Modelisation" |
|---|
| 255 | jlist[34]="Calculateurs Paralleles" |
|---|
| 256 | jlist[35]="Note Scientifique du Pole de Modelisation" |
|---|
| 257 | jlist[36]="Lecture Notes in Computer Science" |
|---|
| 258 | jlist[37]="Prog. Oceanogr." |
|---|
| 259 | jlist[38]="Deep Sea Research Part I: Oceanographic Research Papers" |
|---|
| 260 | jlist[39]="Deep Sea Res. I" |
|---|
| 261 | jlist[40]="IEEE Trans. Geosci. Remote Sensing" |
|---|
| 262 | jlist[41]="ECMWF Tech. Memorandum" |
|---|
| 263 | jlist[42]="J. Environ. Radioactivity" |
|---|
| 264 | jlist[43]="Int. WOCE Newsletter" |
|---|
| 265 | jlist[44]="C. R. Acad. Sci. Paris" |
|---|
| 266 | jlist[45]="C. R. Acad. Sci. Paris, Earth and Planetary Sciences" |
|---|
| 267 | jlist[46]="In The mathematics of models for climatology and environment" |
|---|
| 268 | jlist[47]="Oceanol. Acta" |
|---|
| 269 | jlist[48]="The global atmosphere and ocean system" |
|---|
| 270 | jlist[49]="NATO Advanced Study Institute" |
|---|
| 271 | jlist[50]="WCRP" |
|---|
| 272 | jlist[51]="The Courier" |
|---|
| 273 | jlist[52]="Elsevier Oceanographic Series" |
|---|
| 274 | jlist[53]="In Conference Proceedings of the 1988 International Conference on Supercomputing" |
|---|
| 275 | jlist[54]="In Science and engineering on Cray Supercomputers" |
|---|
| 276 | jlist[55]="In Modeling the Earth's Climate and its Variability" |
|---|
| 277 | jlist[56]="Fish. Oceanogr." |
|---|
| 278 | jlist[57]="Q. J. R. Meteor. Soc." |
|---|
| 279 | jlist[58]="In High performance computing in the geosciences" |
|---|
| 280 | jlist[59]="Int. J. Numer. Meth. in Fluids" |
|---|
| 281 | jlist[60]="Lecture notes in Physics" |
|---|
| 282 | jlist[61]="J. Meterol. Soc. Japan" |
|---|
| 283 | jlist[62]="Journal of Marine Systems" |
|---|
| 284 | jlist[63]="Canadian Journal of Fisheries and Aquatic Sciences" |
|---|
| 285 | jlist[64]="J. Atmos. Ocean. Tech." |
|---|
| 286 | jlist[65]="EOS" |
|---|
| 287 | jlist[66]="Nature" |
|---|
| 288 | jlist[67]="Physics and Chemistry of the Earth, Part B: Hydrology, Oceans and Atmosphere" |
|---|
| 289 | jlist[68]="Journal of Atmospheric and Oceanic Technology" |
|---|
| 290 | jlist[69]="Lecture Notes in Computational Science Engineering" |
|---|
| 291 | jlist[70]="In Greenhouse Gas Control Technologies" |
|---|
| 292 | jlist[71]="Chemical Geology" |
|---|
| 293 | jlist[72]="Marine Geology" |
|---|
| 294 | jlist[73]="Developments in Paleoenvironmental Research" |
|---|
| 295 | jlist[74]="Science" |
|---|
| 296 | jlist[75]="In Mediterranean Climate Variability" |
|---|
| 297 | jlist[76]="Global and Planetary Change" |
|---|
| 298 | jlist[77]="Physical Review Letters" |
|---|
| 299 | # |
|---|
| 300 | # following Journal name are fake ones for demo1 : references extracted from |
|---|
| 301 | # Cantratrix Sopronica L. written by Georges Perec |
|---|
| 302 | #find in http://www-rocq.inria.fr/who/Marc.Thiriet/PsFil/Biblio/cantatrix.ps |
|---|
| 303 | ij=77 |
|---|
| 304 | ij=$(( ${ij} + 1)) |
|---|
| 305 | jlist[$ij]="Res. Proc. neurophysiol. Fanatic Soc" |
|---|
| 306 | ij=$(( ${ij} + 1)) |
|---|
| 307 | jlist[$ij]="New Records Ass. J." |
|---|
| 308 | ij=$(( ${ij} + 1)) |
|---|
| 309 | jlist[$ij]="Am. J. Allegrol." |
|---|
| 310 | ij=$(( ${ij} + 1)) |
|---|
| 311 | jlist[$ij]="J. Neurochem. Neurocytol. Enzymol." |
|---|
| 312 | ij=$(( ${ij} + 1)) |
|---|
| 313 | jlist[$ij]="Hammersmith J." |
|---|
| 314 | ij=$(( ${ij} + 1)) |
|---|
| 315 | jlist[$ij]="Nasa Rept." |
|---|
| 316 | ij=$(( ${ij} + 1)) |
|---|
| 317 | jlist[$ij]="C.r. Assoc. Conc. Lam." |
|---|
| 318 | ij=$(( ${ij} + 1)) |
|---|
| 319 | jlist[$ij]="Gaz. med. franco-rus." |
|---|
| 320 | ij=$(( ${ij} + 1)) |
|---|
| 321 | jlist[$ij]="Amer. J. music. Deciency" |
|---|
| 322 | #jlist[]="" |
|---|
| 323 | # |
|---|
| 324 | # |
|---|
| 325 | cat <<EOF > ${fileou} |
|---|
| 326 | <?xml version='1.0' encoding='ISO-8859-1'?> |
|---|
| 327 | <bibliography |
|---|
| 328 | version="5.0" |
|---|
| 329 | xmlns="http://docbook.org/ns/docbook" |
|---|
| 330 | xml:id="bib${project}ref" |
|---|
| 331 | xml:lang="${lang}"> |
|---|
| 332 | |
|---|
| 333 | <info> |
|---|
| 334 | <date>$( date -u +"%Y-%m-%dT%H:%M:%SZ" )</date> |
|---|
| 335 | </info> |
|---|
| 336 | EOF |
|---|
| 337 | # |
|---|
| 338 | # strip comments |
|---|
| 339 | fileraw_strict=/tmp/$(basename ${fileraw})_strict |
|---|
| 340 | grep -v "^#" ${fileraw} > ${fileraw_strict} |
|---|
| 341 | totlines=$( wc -l ${fileraw_strict} | awk '{print $1}' ) |
|---|
| 342 | l=1 |
|---|
| 343 | while [ ${l} -le ${totlines} ] |
|---|
| 344 | do |
|---|
| 345 | # extract one line |
|---|
| 346 | line=$( sed -n ${l}p ${fileraw_strict} ) |
|---|
| 347 | orgline=$( echo ${line} | sed -e "s/--/- -/g" ) |
|---|
| 348 | line=$( echo ${line} | sed -e "s/</\</g" -e "s/>/\>/g" ) |
|---|
| 349 | # detect if bibtex or not bibtex (starting with @) |
|---|
| 350 | if [ ${line:0:1} != "@" ] |
|---|
| 351 | then |
|---|
| 352 | # parsing non bibtex line |
|---|
| 353 | bibtex=0 |
|---|
| 354 | # before the first : |
|---|
| 355 | tmp=${line%%:*} |
|---|
| 356 | # before the last , |
|---|
| 357 | auths=${tmp%,*}, |
|---|
| 358 | # replace "and" by "," |
|---|
| 359 | auths=$( echo "${auths}" | sed -e "s/ and /,/g" ) |
|---|
| 360 | # after the last , |
|---|
| 361 | year=${tmp##*,} |
|---|
| 362 | year=$( rmbl "${year}" ) |
|---|
| 363 | ## first author before the first ., |
|---|
| 364 | first=${auths%%.,*}. |
|---|
| 365 | # its firstname after the last , |
|---|
| 366 | firstfn=${first##*,} |
|---|
| 367 | firstfn=$( rmbl "${firstfn}" ) |
|---|
| 368 | # its surname ; before the first , |
|---|
| 369 | firstsn=${first%%,*} |
|---|
| 370 | firstsn=$( rmbl "${firstsn}" ) |
|---|
| 371 | ## ref id |
|---|
| 372 | refid=$( echo ${firstsn} | tr "[:upper:]" "[:lower:]" | tr -s " " "_" | tr -s "'" "_" | recode -d -f ISO-8859-1..flat )${year} |
|---|
| 373 | num=$( grep -c "<biblioentry xml:id=\"${refid}_[0-9][0-9]\">" ${fileou} ) |
|---|
| 374 | num=$(( ${num} + 1 )) |
|---|
| 375 | [ ${num} -le 9 ] && num=0${num} |
|---|
| 376 | refid=${refid}_${num} |
|---|
| 377 | hasauthor=1 |
|---|
| 378 | else |
|---|
| 379 | # parsing bibtex line |
|---|
| 380 | bibtex=1 |
|---|
| 381 | # before the first , |
|---|
| 382 | tmp=${line%%,*} |
|---|
| 383 | endline=${line#*,} |
|---|
| 384 | # refid is after { and before the first comma |
|---|
| 385 | refid=${tmp#*{} |
|---|
| 386 | # ++ test if refid already exist |
|---|
| 387 | # check if author field is there |
|---|
| 388 | echo ${endline} | grep -q "author" |
|---|
| 389 | hasauthor=${?} |
|---|
| 390 | if [ ${hasauthor} -eq 0 ] |
|---|
| 391 | then |
|---|
| 392 | auths=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(author * = *{\)\(.*\)\(}\)/\2/") |
|---|
| 393 | #echo "auths $auths" # genre Michael Metcalf and John Reid and Malcolm Cohen |
|---|
| 394 | else |
|---|
| 395 | echo "pas d'auteurs" # ++ |
|---|
| 396 | fi |
|---|
| 397 | endline=${endline#*,} |
|---|
| 398 | title=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(title * = *{\)\(.*\)\(}\)/\2/") |
|---|
| 399 | endline=${endline#*,} |
|---|
| 400 | #echo ${endline} | awk -F "," '{print $1}' #++debug |
|---|
| 401 | year=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(year * = *{\)\(.*\)\(}\)/\2/") |
|---|
| 402 | endline=${endline#*,} |
|---|
| 403 | publisher=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(publisher * = *{\)\(.*\)\(}\)/\2/") |
|---|
| 404 | endline=${endline#*,} |
|---|
| 405 | edition=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(edition * = *{\)\(.*\)\(}\)/\2/") |
|---|
| 406 | endline=${endline#*,} |
|---|
| 407 | isbn=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(isbn * = *{\)\(.*\)\(}\)/\2/") |
|---|
| 408 | endline=${endline#*,} |
|---|
| 409 | pag=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(page * = *{\)\(.*\)\(}\)/\2/") |
|---|
| 410 | |
|---|
| 411 | if [ ${hasauthor} -eq 1 ] |
|---|
| 412 | then |
|---|
| 413 | ## first author before the first and |
|---|
| 414 | first=${auths%%and*} |
|---|
| 415 | # its surname after the last " " ++ pas vrai double nom et pb blanc dans la syntaxe |
|---|
| 416 | # firstsn=${first##* } . ok en interactif . pas ok en script |
|---|
| 417 | firstsn=$( echo ${first} | awk '{print $2}') |
|---|
| 418 | #echo "firstsn $firstsn" |
|---|
| 419 | # its firstname before the first " " |
|---|
| 420 | firstfn=${first%% *} |
|---|
| 421 | fi |
|---|
| 422 | fi |
|---|
| 423 | |
|---|
| 424 | cat <<EOF >> ${fileou} |
|---|
| 425 | <biblioentry xml:id="${refid}"> |
|---|
| 426 | <!-- date |
|---|
| 427 | $( date -u +"%Y-%m-%dT%H:%M:%SZ" ) |
|---|
| 428 | --> |
|---|
| 429 | <!-- original text |
|---|
| 430 | ${orgline} |
|---|
| 431 | --> |
|---|
| 432 | EOF |
|---|
| 433 | if [ ${hasauthor} -eq 1 ] |
|---|
| 434 | then |
|---|
| 435 | cat <<EOF >> ${fileou} |
|---|
| 436 | <authorgroup> |
|---|
| 437 | <author> <personname> <surname>${firstsn}</surname> <firstname>${firstfn}</firstname> </personname> </author> |
|---|
| 438 | EOF |
|---|
| 439 | |
|---|
| 440 | ## other authors.. |
|---|
| 441 | if [ ${bibtex} -eq 0 ] |
|---|
| 442 | then |
|---|
| 443 | previous=${first}, |
|---|
| 444 | # next authors... |
|---|
| 445 | next=${auths##*${previous}} |
|---|
| 446 | fi |
|---|
| 447 | if [ ${bibtex} -eq 1 ] |
|---|
| 448 | then |
|---|
| 449 | previous=${first}"and " |
|---|
| 450 | # next authors... |
|---|
| 451 | #echo "auths ${auths}" |
|---|
| 452 | next=${auths##*${previous}} |
|---|
| 453 | if [ "${next}and " = "${previous}" ] |
|---|
| 454 | then |
|---|
| 455 | next="" |
|---|
| 456 | fi |
|---|
| 457 | fi |
|---|
| 458 | #echo "previous ${previous}" |
|---|
| 459 | #echo "next ${next}" |
|---|
| 460 | # while the next author is not empty |
|---|
| 461 | while [ "${next}" != "" ] |
|---|
| 462 | do |
|---|
| 463 | if [ ${bibtex} -eq 0 ] |
|---|
| 464 | then |
|---|
| 465 | # get the first next author; before the first , |
|---|
| 466 | next=${next%%,*} |
|---|
| 467 | # its surname ; after the last . |
|---|
| 468 | nextsn=${next##*.} |
|---|
| 469 | nextsn=$( rmbl "${nextsn}" ) |
|---|
| 470 | # its firstname ; before the last . |
|---|
| 471 | nextfn=${next%.*}. |
|---|
| 472 | nextfn=$( rmbl "${nextfn}" ) |
|---|
| 473 | fi |
|---|
| 474 | if [ ${bibtex} -eq 1 ] |
|---|
| 475 | then |
|---|
| 476 | # get the first next author; before the first "and" |
|---|
| 477 | next=${next%%and *} |
|---|
| 478 | # its surname after the last " " ++ pas vrai double nom et pb blanc dans la syntaxe |
|---|
| 479 | # nextsn=${next##* } . ok en interactif . pas ok en script |
|---|
| 480 | nextsn=$( echo ${next} | awk '{print $2}') |
|---|
| 481 | #echo "nextsn $nextsn" |
|---|
| 482 | # its nextname before the first " " |
|---|
| 483 | nextfn=${next%% *} |
|---|
| 484 | fi |
|---|
| 485 | # |
|---|
| 486 | echo " <author> <personname> <surname>${nextsn}</surname> <firstname>${nextfn}</firstname> </personname> </author>" >> ${fileou} |
|---|
| 487 | #echo " <author> <personname> <surname>${nextsn}</surname> <firstname>${nextfn}</firstname> </personname> </author>" #++debug |
|---|
| 488 | if [ ${bibtex} -eq 0 ] |
|---|
| 489 | then |
|---|
| 490 | previous=${next}, |
|---|
| 491 | next=${auths##*${previous}} |
|---|
| 492 | fi |
|---|
| 493 | if [ ${bibtex} -eq 1 ] |
|---|
| 494 | then |
|---|
| 495 | previous=${next}"and " |
|---|
| 496 | next=${auths##*${previous}} |
|---|
| 497 | fi |
|---|
| 498 | done |
|---|
| 499 | echo " </authorgroup>" >> ${fileou} |
|---|
| 500 | fi |
|---|
| 501 | |
|---|
| 502 | # end of the line ; after the first : |
|---|
| 503 | endline=${line#*:} |
|---|
| 504 | |
|---|
| 505 | ## find the journal |
|---|
| 506 | j=1 |
|---|
| 507 | jfound="" |
|---|
| 508 | jlistsize=${#jlist[@]} |
|---|
| 509 | while [[ ${j} -le ${jlistsize} && "${jfound}" == "" ]] |
|---|
| 510 | do |
|---|
| 511 | ok=$( echo ${endline} | grep -ci "${jlist[j]} *," ) |
|---|
| 512 | [ $ok -eq 1 ] && jfound="${jlist[j]}" |
|---|
| 513 | j=$(( ${j} + 1 )) |
|---|
| 514 | done |
|---|
| 515 | if [ "${jfound}" == "" ] |
|---|
| 516 | then |
|---|
| 517 | echo "eee: Journal not found " |
|---|
| 518 | echo "${endline}" |
|---|
| 519 | #++exit 1 |
|---|
| 520 | #+++ following lines are not yet validate |
|---|
| 521 | # it might be a book, a manual, a conference, etc. ++ |
|---|
| 522 | echo " <title>${title}</title>" >> ${fileou} |
|---|
| 523 | echo " <biblioid class=\"isbn\">${isbn}</biblioid>" >> ${fileou} |
|---|
| 524 | cat <<EOF >> ${fileou} |
|---|
| 525 | <biblioset relation="nojournal"> |
|---|
| 526 | <title>${title}</title> |
|---|
| 527 | <pagenums>${pag}</pagenums> |
|---|
| 528 | <pubdate>${year}</pubdate> |
|---|
| 529 | </biblioset> |
|---|
| 530 | EOF |
|---|
| 531 | cat <<EOF >> ${fileou} |
|---|
| 532 | </biblioentry> |
|---|
| 533 | |
|---|
| 534 | EOF |
|---|
| 535 | fi # end of if jfound empty (ie not an article) |
|---|
| 536 | if [ "${jfound}" != "" ] |
|---|
| 537 | then |
|---|
| 538 | ## title |
|---|
| 539 | # before the first : |
|---|
| 540 | title=${endline%%${jfound}*} |
|---|
| 541 | title=$( cleanname "${title}" ) |
|---|
| 542 | echo " <title>${title}</title>" >> ${fileou} |
|---|
| 543 | ## end |
|---|
| 544 | ## end of the line ; after the first ${jfound} |
|---|
| 545 | endline=${endline#*${jfound}} |
|---|
| 546 | endline=$( cleanname "${endline}" ) |
|---|
| 547 | ## doi |
|---|
| 548 | endline=$( echo ${endline} | sed -e "s/[dD][oO][iI] *\t* *: *\t* */doi:/" ) |
|---|
| 549 | ok=$( echo ${endline} | grep -ic "doi:" ) |
|---|
| 550 | if [ ${ok} -eq 1 ] |
|---|
| 551 | then |
|---|
| 552 | doi=${endline##*doi:} |
|---|
| 553 | echo " <biblioid class=\"doi\">${doi}</biblioid>" >> ${fileou} |
|---|
| 554 | endline=${endline%doi:*} |
|---|
| 555 | endline=$( cleanname "${endline}" ) |
|---|
| 556 | else |
|---|
| 557 | echo "non doi: ${line}" |
|---|
| 558 | fi |
|---|
| 559 | num=$( echo ${endline} | tr -dc "," | wc -c ) |
|---|
| 560 | case ${num} in |
|---|
| 561 | 1) |
|---|
| 562 | ### echo ${num}: ${endline} |
|---|
| 563 | vol=${endline%,*} |
|---|
| 564 | vol=$( cleanname "${vol}" ) |
|---|
| 565 | pag=${endline##*,} |
|---|
| 566 | pag=$( cleanname "${pag}" ) |
|---|
| 567 | cat <<EOF >> ${fileou} |
|---|
| 568 | <biblioset relation="journal"> |
|---|
| 569 | <title>${jfound}</title> |
|---|
| 570 | <volumenum>${vol}</volumenum><pagenums>${pag}</pagenums> |
|---|
| 571 | <pubdate>${year}</pubdate> |
|---|
| 572 | </biblioset> |
|---|
| 573 | EOF |
|---|
| 574 | ;; |
|---|
| 575 | 2) |
|---|
| 576 | vol=${endline%,*} |
|---|
| 577 | vol=$( cleanname "${vol}" ) |
|---|
| 578 | iss=${vol##*,} |
|---|
| 579 | iss=$( cleanname "${iss}" ) |
|---|
| 580 | vol=${vol%,*} |
|---|
| 581 | vol=$( cleanname "${vol}" ) |
|---|
| 582 | pag=${endline##*,} |
|---|
| 583 | pag=$( cleanname "${pag}" ) |
|---|
| 584 | cat <<EOF >> ${fileou} |
|---|
| 585 | <biblioset role="journal"> |
|---|
| 586 | <title>${jfound}</title> |
|---|
| 587 | <volumenum>${vol}</volumenum><issuenum>${iss}</issuenum><pagenums>${pag}</pagenums> |
|---|
| 588 | <pubdate>${year}</pubdate> |
|---|
| 589 | </biblioset> |
|---|
| 590 | EOF |
|---|
| 591 | ;; |
|---|
| 592 | *) |
|---|
| 593 | #echo ${num}: ${endline} # ++debug |
|---|
| 594 | cat <<EOF >> ${fileou} |
|---|
| 595 | <biblioset role="journal"> |
|---|
| 596 | <title>${jfound}</title> |
|---|
| 597 | <pubdate>${year}</pubdate> |
|---|
| 598 | <bibliomisc>${endline}</bibliomisc> |
|---|
| 599 | </biblioset> |
|---|
| 600 | EOF |
|---|
| 601 | ;; |
|---|
| 602 | esac |
|---|
| 603 | |
|---|
| 604 | cat <<EOF >> ${fileou} |
|---|
| 605 | </biblioentry> |
|---|
| 606 | |
|---|
| 607 | EOF |
|---|
| 608 | |
|---|
| 609 | |
|---|
| 610 | |
|---|
| 611 | |
|---|
| 612 | fi # end of if jfound not empty |
|---|
| 613 | l=$(( ${l} + 1 )) |
|---|
| 614 | |
|---|
| 615 | done |
|---|
| 616 | echo "</bibliography>" >> ${fileou} |
|---|
| 617 | |
|---|
| 618 | xml val --err \ |
|---|
| 619 | --xsd http://www.docbook.org/xml/5.0/xsd/docbook.xsd \ |
|---|
| 620 | ${fileou} 1> xmlstarlet.log 2>&1 |
|---|
| 621 | |
|---|
| 622 | # clean |
|---|
| 623 | echo "iii : xslstarlet.log contains sdtout and stderr from xml command on ${fileou}" |
|---|
| 624 | echo "iii : which was done just to check consistence of ${fileou}" |
|---|
| 625 | more xmlstarlet.log |
|---|
| 626 | rm -i xmlstarlet.log |
|---|
| 627 | # |
|---|
| 628 | case ${type} in |
|---|
| 629 | raw) |
|---|
| 630 | echo "iii : ${fileraw_strict} contains a copy of input file without comments" |
|---|
| 631 | rm -i ${fileraw_strict} |
|---|
| 632 | ;; |
|---|
| 633 | mailbody) |
|---|
| 634 | echo "iii : ${fileraw} contains a copy of input file" |
|---|
| 635 | rm -i ${fileraw} |
|---|
| 636 | echo "iii : ${fileraw_strict} contains a copy of input file without comments" |
|---|
| 637 | rm -i ${fileraw_strict} |
|---|
| 638 | ;; |
|---|
| 639 | esac |
|---|
| 640 | # |
|---|
| 641 | echo " iii : you have a some new or modified bibliographic references in ${fileou} (biblioentry)" |
|---|
| 642 | echo " iii : you can add them in ${fileou}" #++ filename path |
|---|
| 643 | echo " iii : modify date in ${fileou}" #++ filename path |
|---|
| 644 | echo " iii : but please take care of id last part (after underscore) that may" |
|---|
| 645 | echo " iii : be in conflict with existing ones" #++ pour s'affranchir de ce souci il faudrait donner en entrée complémentaire le fichier à compléter. |
|---|
| 646 | # |
|---|
| 647 | exit 0 |
|---|
