#!/bin/bash

#./Gen_ACR ABC_bc.dat
#
# input file: ABC_bc.dat is list of bibicode, citation (per line) for author "ABC"
#
# output files:
# bc.dat      ... list of bibcodes of author with citations>0
# ref_bc.dat  ... list of bibcodes of papers referred to be the author
# cite_bc.dat ... list of bibcodes of papers citing the author
# ABC_coauthors.dat .. list of co-authors of the author


  if [[ -z $1 ]]; then
    echo "exit: need file listing bibcodes, no. of citations"
    exit -1
  fi

infile=$1
outf=${infile%_*}
echo infile, $outfile

Token=$(cat ADS_Token)

          #list of bibcodes for publications with non zero publications
awk -F, '$2>0{print $1}' $infile  > bc.dat
	#number of papers and number of citations	
npap=$(wc -l < bc.dat); npap=$((npap+0))

ncites= $(awk -F"," '{c+=$2}END{print "citations: " c}' $infile)
echo "number of papers with cites>0: " $npap " number of cites: " $ncites

cat cmd1 > cmd
echo -n "$'bibcode" >> cmd
awk '{printf "\\n%s", $1}' bc.dat >> cmd
echo -n "'" >> cmd
source cmd >  Astruct    #the output is stored in structure "Astruct"

jq -r '.response.docs[] | .author[]' Astruct | sed 's/&/%26/g' > ${outf}_coauthors.dat

                #replace "&" by URL encoding
jq -r '.response.docs[] | .citation[]'  Astruct | sed 's/&/%26/g'> cite_bc.dat
nc=$(wc -l < cite_bc.dat)
echo "generated cite_bc.dat number of cites:" $nc

jq -r '.response.docs[] | .reference[]'  Astruct | sed 's/&/%26/g'> ref_bc.dat
nr=$(wc -l < ref_bc.dat)
echo "generated ref_bc.dat  number of refs:" $nr