Changeset 46 for trunk

Timestamp:

03/12/08 14:35:41 (16 years ago)

Author:

pinsard

Message:

add bibtex input (continuation)

Location:

trunk

Files:

• bibopa.sh (modified) (4 diffs)
• data/petitpolyfp.bib (modified) (1 diff)

trunk/bibopa.sh

@@ -55 +55 @@
 #
 # example :
-# $ ./bibopa.sh -i /Users/smasson/Bibopa/biball.txt -t raw
+# $ ./bibopa.sh -i data/biball.txt -t raw
 # $ ./bibopa.sh -i data/mail2007-04-25T08:58:16Z.txt -t mailbody
 # $ ./bibopa.sh -i data/petitpolyfp.bib -t bibtex
@@ -175 +175 @@
  #more ${fileraw} # ++ if debug
  #read a #++ if debug
- echo "not yet implemented. sorry."
- exit 1
 ;;
 *)
@@ -293 +291 @@
   orgline=$( 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}
+  # detect if bibtex or not bibtex (starting with @)
+  if [ ${line:0:1} != "@" ]
+  then
+     # parsing non bibtex line
+     bibtex=0
+     # before the first :
+       tmp=${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}
+   else
+      # parsing bibtex line
+      bibtex=1
+# ++
+      # before the first ,
+      tmp=${line%%,*}
+      endline=${line#*,}
+      # refid is after { and before the first comma
+      refid=${tmp#*{}
+      # ++ test if refid already exist
+      auths=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(author * = *{\)\(.*\)\(}\)/\2/")
+      echo "auths $auths" # genre  Michael Metcalf and John Reid and Malcolm Cohen
+      endline=${endline#*,}
+      title=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(title * = *{\)\(.*\)\(}\)/\2/")
+      endline=${endline#*,}
+echo ${endline} | awk -F "," '{print $1}'
+      year=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(year * = *{\)\(.*\)\(}\)/\2/")
+      endline=${endline#*,}
+      publisher=$( echo ${endline} | awk -F "," '{print $1}' | sed -e "s/\(publisher * = *{\)\(.*\)\(}\)/\2/")
+
+     ## first author before the first and
+       first=${auths%%and*}
+     # its surname after the last " "  ++ pas vrai double nom et pb blanc dans la syntaxe
+     # firstsn=${first##* } . ok en interactif . pas ok en script
+       firstsn=$( echo ${first}  | awk '{print $2}')
+       echo "firstsn $firstsn"
+     # its firstname before the first " "
+       firstfn=${first%% *}
+   fi
 
           cat <<EOF >> ${fileou}
@@ -330 +362 @@
 
 ## other authors..
-  previous=${first},
-# next authors...
-  next=${auths##*${previous}}
+  if [ ${bibtex} -eq 0 ]
+  then
+   previous=${first},
+   # next authors...
+   next=${auths##*${previous}}
+  fi
+  if [ ${bibtex} -eq 1 ]
+  then
+   previous=${first}"and "
+   # next authors...
+   echo "auths ${auths}"
+   next=${auths##*${previous}}
+   if [ "${next}and " = "${previous}" ]
+   then
+     next=""
+   fi
+   read a
+  fi
+  echo "previous ${previous}"
+  echo "next ${next}"
 # while the next author is not empty
   while [  "${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}" )
+      if [ ${bibtex} -eq 0 ]
+      then
+         # get the first next author; before the first ,
+         next=${next%%,*}
+         # its surname ; after the last .
+         nextsn=${next##*.}
+         nextsn=$( rmbl "${nextsn}" )
+         # its firstname ; before the last .
+         nextfn=${next%.*}.
+         nextfn=$( rmbl "${nextfn}" )
+       fi
+      if [ ${bibtex} -eq 1 ]
+      then
+         # get the first next author; before the first "and"
+         next=${next%%and *}
+  echo "next 2= ${next}"
+  read a
+
+         # its surname after the last " "  ++ pas vrai double nom et pb blanc dans la syntaxe
+         # nextsn=${next##* } . ok en interactif . pas ok en script
+         nextsn=$( echo ${next}  | awk '{print $2}')
+         echo "nextsn $nextsn"
+         # its nextname before the first " "
+         nextfn=${next%% *}
+      fi
 #
     echo "    <author> <personname> <surname>${nextsn}</surname> <firstname>${nextfn}</firstname> </personname> </author>" >> ${fileou}
     echo "    <author> <personname> <surname>${nextsn}</surname> <firstname>${nextfn}</firstname> </personname> </author>"  #++debug
-    previous=${next},
-    next=${auths##*${previous}}
-
+    if [ ${bibtex} -eq 0 ]
+    then
+      previous=${next},
+      next=${auths##*${previous}}
+    fi
+    if [ ${bibtex} -eq 1 ]
+    then
+      previous=${next}"and "
+      next=${auths##*${previous}}
+    fi
   done
   echo "  </authorgroup>"  >> ${fileou}

trunk/data/petitpolyfp.bib

@@ -1 +1 @@
 @book{MetReiCoh2004 ,
-   author    = {Michael Metcalf and
-           John Reid and Malcolm Cohen},
+   author    = {Michael Metcalf and John Reid and Malcolm Cohen},
    title     = {Fortran 95/2003 explained},
    year      = {2004},