#!/bin/ksh
#This script takes the new Serial Solutions marc file and extracts
#the 001 tags and matches them to the 001's within the catalogue database to 
#determine matches. (for "overlay") and also gets their ckeys
#
#It outputs matching ckeys found via seltext
# 
#  
#N. Lall, July 2002 
#
#Modified: N. Lall, Dec 2003 , Aug 2010 for overlay of matched records


IDENT=${0##*/}
USAGE="$IDENT <input name of new ss.mrc file to get 001s from>"
#if [[ $# -ne 2 || ! -s $1 || -s $2 ]]
if [[ $# -ne 1 || ! -s $1 ]]
 then
 echo "ERROR: $USAGE"
 exit 1
fi

#Use Damon's better way of calling all the settings from Config/environ:
. /sirsi/s/sirsi/Unicorn/Config/environ
for i in `cut -f1 -d= /sirsi/s/sirsi/Unicorn/Config/environ`
        do
                export $i
        done

######

INPUT=$1
INPUTfile=`echo $INPUT | sed -e 's/\/..*\///g'`
#OUTPUT=$2
#OUTPUTfile=`echo $OUTPUT | sed -e 's/\/..*\///g'`
temp=`getpathname rpttemp`/findmatchss$$
today=`date +"%Y%m"`
DEBUG=YES
workdir=/sirsi/s/sirsi/Unicorn/Work/NDL/Serialsolutions
OUTPUT=${workdir}/matchedckeys

#There should be no $OUTPUT.sort file if the remove program ran last month!
if [[ -s $OUTPUT.sort ]]
 then
   #report this out and stop the program
   print "Error, there already is a $OUTPUT.sort file" > ${temp}msg
   print "Did remove program run last month? Investigate!" >> ${temp}msg
   mailx -s"Problem with findmatchv3 SS program" ndlall@ucalgary.ca < ${temp}msg
   mailx -s"Problem with findmatchv3 SS program" dkbrown@ucalgary.ca < ${temp}msg
   exit 1
fi

#Check that LOCAL tags in SERIAL format have not changed since last time:
cat /s/sirsi/Unicorn/Custom/Format/ENTRY.7 |\
selascii -iR -oF1 -r 25 "L" 2>/dev/null |\
sort > ${temp}testlocal

sdiff -s ${temp}testlocal ${workdir}/SERIALlocaltags.bak > ${temp}testlocal.out
if [[ -s ${temp}testlocal.out ]]
 then
  #report this out and stop the program
  print "LOCAL tags in SERIAL format have changed since last run." > ${temp}msg
  print "You will need to change the remlocal.dat file to reflect the changes." >> ${temp}msg
  print "Here is the list of current LOCAL tags in SERIAL format:" >> ${temp}msg
  cat ${temp}testlocal >> ${temp}msg
   mailx -s"Problem with findmatchv3 SS program" ndlall@ucalgary.ca < ${temp}msg
   mailx -s"Problem with findmatchv3 SS program" dkbrown@ucalgary.ca < ${temp}msg
   exit 1
fi
  

cat $INPUT | marcprint -t"001" > ${temp}a 2>/dev/null

if [[ ! -s ${temp}a ]]
 then
   print "Error, no tag 001s generated, check input file name." > ${temp}msg
   mailx -s"INPUT name wrong for findmatch" ndlall@ucalgary.ca < ${temp}msg
  mailx -s"INPUT name wrong for findmatch" dkbrown@ucalgary.ca < ${temp}msg
   exit 1
else
 cat ${temp}a | sed -e 's/^num=001   //g' > ${temp}b
 if [[ -s ${temp}b ]]
  then
   cat ${temp}b |\
   while read PT1
   do
    sirsiecho "$PT1 {001}" | seltext -lBOTH >> ${temp}c 2>${temp}log
    TEST=`grep -c "Search time limit exceeded" ${temp}log`
    #print "In the first sirsiecho and TEST being set to: $TEST"
    until [ $TEST -eq 0 ]
     do
#      print "In the second while loop because TEST is not zero it is: $TEST"
      sirsiecho "$PT1 {001}" | seltext -lBOTH >> ${temp}c 2>${temp}log
      TEST=`grep -c "Search time limit exceeded" ${temp}log`
#      print "Test is reset in the loop to: $TEST"
     done

   done
 fi
fi

if [[ -s ${temp}c ]]
 then
  #go back and get the 001s for these ckeys
  cat ${temp}c | selcatalog -iC -e"001" -oe > $OUTPUT.matched001s 2>/dev/null
  print "The matched 001s are in $OUTPUT.matched001s." > ${temp}msg

  #Ensure that all matched ckeys are old loaded Serial Solutions records
  cat ${temp}c | sort > $OUTPUT.sort
  comm -23 $OUTPUT.sort ${workdir}/ALLckeys.sort > ${temp}problems
  print "" >> ${temp}msg
  print "The matched ckeys are in $OUTPUT.sort." >> ${temp}msg
  countmatch=`cat $OUTPUT.sort | wc -l | sed -e 's/^  *//g'`
  print "     There were $countmatch records matched in this process." >> ${temp}msg
  print "" >> ${temp}msg
  
  if [[ -s ${temp}problems ]]
   then
    print "" >> ${temp}msg
    print "***NOT all matched ckeys were old Serial Solutions records!" >> ${temp}msg
    print "Here are the problem ckeys to investigate" >> ${temp}msg
    cat ${temp}problems >> ${temp}msg
  else
    print "All matched ckeys were old Serial Solutions records!" >> ${temp}msg
  fi

  #Any records with more than one copy or less than one copy?
  #These must be reported out and fixed before the remove program will 
  #run. Feb 2009--group records now so report out but remove program
  #WILL remove them.
  cat $OUTPUT.sort | selitem -iC -oC 2>/dev/null |\
  sort > ${temp}ckcopies
  comm -23 $OUTPUT.sort ${temp}ckcopies > ${temp}nocopy
  comm -23 ${temp}ckcopies $OUTPUT.sort | sort -u > ${temp}morecopy

  #Report out any of these oddball copy problems and send to CMB
  if [[ -s ${temp}nocopy ]]
   then
    print "Some matched ckeys have NO copy! Report to Bib.Services" >> ${temp}msg
    print "The remove old program will not run until these are fixed!" >> ${temp}msg
    cat ${temp}nocopy >> ${temp}msg
  fi
#  if [[ -s ${temp}morecopy ]]
#   then
#    print "Some matched ckeys have MORE than one copy! Report to Bib.Services" >> ${temp}msg
#   print "The remove old program WILL run AND remove these now that we have GROUP records!" >> ${temp}msg
#   cat ${temp}morecopy >> ${temp}msg
#  fi

  #Do the steps of removing LOCAL tags and changing the call number to AUTO XX call num
  # for all the MATCHED cases and remove all copies before overlaying them
  cat $OUTPUT.sort |\
  editmarc -f"${workdir}/remlocal.dat" -u > ${temp}remlocal 2> ${temp}remlocal.log

  cat $OUTPUT.sort | selitem -iC -oBC 2>/dev/null |\
  sort -t"|" -u +1 -2 | ${workdir}/makeapifile.pl |\
  apiserver -e"${workdir}/api.err" -h > ${temp}apiout 2> ${temp}apiout.err

  if [[ -s ${workdir}api.err ]]
   then
    print "There were some api.errs to investigate!" >> ${temp}msg
  fi

  cat $OUTPUT.sort | selitem -iC -oI 2>/dev/null |\
  remitem -l"ADMIN|VT100" > ${temp}remitem 2> ${temp}remitem.err

  #now to handle the full delete cases...may try catalogload -Dd first
  #otherwise get the 001s of the delete cases, match them and delete with delitem
#  cat $INPUT |\
#  flatskip -a"SERIAL" -d"Y" -im -om 2>/dev/null |\
#  marcprint -t"001" > ${temp}delete001s


  #Create the file for load into the catalogue with the 
  #delete records (use to use to ignore the delete records, now need them -dN)
  cat $INPUT |\
  flatskip -a"SERIAL" -im -om -e"/sirsi/s/sirsi/Unicorn/Work/NDL/Serialsolutions/flat.err" > ${workdir}/$today.mrc 2> ${temp}flatlog

  #Find out how many records TOTAL are in the SS file for webpage stats
  #Put into a file for use at end of removeold program so all stats given once
  grep "marc. ..1305.$" ${temp}flatlog | translate | sed -e 's/[^0-9][^0-9]*//g' > ${workdir}/sstotalrecords


  #Create a stamp file for this variable to be stored for use in loadNEWcron.ksh
  sirsiecho $today > ${workdir}/stamp

  print "" >> ${temp}msg
  print "Here is the log of converting marc through flatskip for load" >> ${temp}msg
  cat ${temp}flatlog | translate >> ${temp}msg
  print "" >> ${temp}msg

  if [[ -s /s/sirsi/Unicorn/Work/NDL/Serialsolutions/flat.err ]]
   then
    print "" >> ${temp}msg
    print "Error records are in /s/sirsi/Unicorn/Work/NDL/Serialsolutions/flat.err" >> ${temp}msg
    print "These were created in the flatskip step of taking the marc records" >> ${temp}msg
   print "and converting them to records to load." >> ${temp}msg
   print "The flat records that did not error out are in ${workdir}/$today.flat" >> ${temp}msg
  else
   print "The marc records for load are in ${workdir}/$today.flat" >> ${temp}msg
   print "Setup the crontab to load these with loadNEWcronv3.ksh" >> ${temp}msg
   print "" >> ${temp}msg
  fi

else
#Create the flat file for load into the catalogue with the
  #delete records (use to use to ignore the delete records, now need them -dN)
  cat $INPUT |\
  flatskip -a"SERIAL" -im -om -e"/sirsi/s/sirsi/Unicorn/Work/NDL/Serialsolutions/flat.err" > ${workdir}/$today.mrc 2> ${temp}flatlog
  #Find out how many records TOTAL are in the SS file for webpage stats
  #Put into a file for use at end of removeold program so all stats given once
  grep "marc. ..1305.$" ${temp}flatlog | translate | sed -e 's/[^0-9][^0-9]*//g' > ${workdir}/sstotalrecords

 #Create a stamp file for this variable to be stored for use in loadNEWcron.ksh
  sirsiecho $today > ${workdir}/stamp

  print "" >> ${temp}msg
  print "Here is the log of converting marc through flatskip for load" >> ${temp}msg
  cat ${temp}flatlog | translate >> ${temp}msg
  print "" >> ${temp}msg

  if [[ -s /s/sirsi/Unicorn/Work/NDL/Serialsolutions/flat.err ]]
   then
    print "" >> ${temp}msg
    print "Error records are in /s/sirsi/Unicorn/Work/NDL/Serialsolutions/flat.err" >> ${temp}msg
    print "These were created in the flatskip step of taking the marc records" >> ${temp}msg
   print "and converting them to records to load." >> ${temp}msg
   print "The flat records that did not error out are in ${workdir}/$today.flat" >> ${temp}msg
  else
   print "The marc records for load are in ${workdir}/$today.flat" >> ${temp}msg
   print "Setup the crontab to load these with loadNEWcronv3.ksh" >> ${temp}msg
   print "" >> ${temp}msg
  fi


  print "There were no matching 001 tags! Does that seem right?" >> ${temp}msg
  print "If it seems right, do catalogload with only CREATE mode, no Update." >> ${temp}msg
fi

mailx -s"Findmatch results" ndlall@ucalgary.ca < ${temp}msg
#mailx -s"Findmatch results" dkbrown@ucalgary.ca < ${temp}msg
cat ${temp}msg > ${workdir}/findmatchrpt.$today

if [[ $DEBUG = "NO" ]]
 then
  rm -f ${temp}*
fi

exit 0