staden-lg/src/indexseqlibs/pirfreetext.script

70 lines
1.5 KiB
Tcsh

#! /bin/csh -f
#
# script to make a free text index for the pir library
#
echo "pirfreetext.script Version 1.0"
echo ""
set RM = "/bin/rm -f"
set SORT = sort
set UNIQ = uniq
#
# LIBFILE is the name youve given to the library file
#
set LIBFILE1 = pir1.dat
set LIBFILE2 = pir2.dat
set LIBFILE3 = pir3.dat
set WORDS = stopwords
pirfreetext $LIBFILE1 pir1.list
if ( $status ) goto abort
pirfreetext $LIBFILE2 pir2.list
if ( $status ) goto abort
pirfreetext $LIBFILE3 pir3.list
if ( $status ) goto abort
# sort on words, remove dumplicates, remove stopwords
${SORT} -b +1 pir1.list | ${UNIQ} | excludewords ${WORDS} >! freetext.presort
if ( $status ) goto abort
${RM} pir1.list
${SORT} -b +1 pir2.list | ${UNIQ} | excludewords ${WORDS} >> freetext.presort
if ( $status ) goto abort
${RM} pir2.list
${SORT} -b +1 pir3.list | ${UNIQ} | excludewords ${WORDS} >> freetext.presort
if ( $status ) goto abort
${RM} pir3.list
# resort on entry name
${SORT} freetext.presort >! freetext.sorted
if ( $status ) goto abort
${RM} freetext.presort
# include entry numbers in the file
freetext2 freetext.sorted freetext.entry
if ( $status ) goto abort
${RM} freetext.sorted
# sort on text
${SORT} -b +2 freetext.entry >! freetext.sorted2
if ( $status ) goto abort
${RM} freetext.entry
# create the index
freetext4 freetext.sorted2 freetext.trg freetext.hit
if ( $status ) goto abort
${RM} freetext.sorted2
echo "Completion successful"
echo ""
exit 0
#
# Abort
#
abort:
echo "pirfreetext.script: aborting due to serious error"
exit 2