peck-intersect

#!/bin/bash
#
# Combine the output of two peck searches
#
# Written by FFS
#
# Changelog:
#
#       2014-10-04 Intersection of two searches
#
# -----------------------------------------------------------

# Help screen
if [ "$1" = "-h" -o "$1" = "--help" -o "$1" = "help" ] ; then
  echo -e "\n\t\t* * * Red Hen Commandline Search Intersection * * *"
  echo -e "\n\tGenerate the intersection of two peck search results."
  echo -e "\n\tSyntax:"
  echo -e "\n\t\t`basename $0` <input file #1> <input file #2> <output file> [clobber]"
  echo -e "\n\tExamples (clobber to overwrite existing output file):"
  echo -e "\n\t\t`basename $0` ~/MyResults1.csv ~/MyResults2.csv ~/Intersection.csv"
  echo -e "\n\tThe script also handles iterative intersections.\n"
  echo -e "\n\tSee also peck, peck-clip, peck-filter, and peck-seg.\n"
   exit
fi

# First search result file
if [ -z "$1" ]
  then echo -e "\nPlease include two filenames of peck search results to intersect and an output file; see `basename $0` -h for usage.\n" ; exit
  else FIL1="$1"
fi

# Second search result file
if [ -z "$2" ]
  then echo -e "\nPlease include a second file to intersect with; see `basename $0` -h for usage.\n" ; exit
  else FIL2="$2"
fi

# Intersection output file
if [ -z "$3" ]
  then echo -e "\nPlease include an output filename; see `basename $0` -h for usage.\n" ; exit
  else FIL3="$3"
fi

# Clobber?
if [ -n "$4" ] ; then
  if [ "$4" = "clobber" ] ; then
    if [ -f "$FIL3" ] ; then rm "$FIL3" ; fi
  fi
fi

# Existing file
if [ -f "$FIL3" ] ; then echo -e "The output file $FIL3 already exists -- add clobber to overwrite.\n" ; exit ; fi

# Welcome
echo -e "\n\tRed Hen generates the intersection of $FIL1 and $FIL2 search results in `pwd` ...\n"

# Line number
LEN="$( cat $FIL1 | wc -l )"

# Loop through the lines, skipping the header
for NUM in `seq 2 $LEN` ; do unset LIN2
  LIN1="$( sed -n "$NUM p" $FIL1 )"
  LIN2="$( grep -m1 "${LIN1%|CC*}" $FIL2 )"

  # Write to file
  if [ "$LIN2" ] ; then echo "$LIN1" >> $FIL3
    if [ "$LIN2" != "$LIN1" ] ; then echo "$LIN2" >> $FIL3 ; fi
  fi

done

# Exit if empty intersection
if [ ! -f $FIL3 ] ; then echo -e "\n\tThere are no shared search results in these two files.\n" ; exit ; fi

# Remove duplicate lines
sort -u $FIL3 | sponge $FIL3

# Add field header for R
LIN="$( sed -n "1 p" $FIL1 )" ; sed -i "1i$LIN" $FIL3

# Display the results
cat $FIL3

# Receipt
echo -e "\n\tRed Hen generated the intersection of $FIL1 and $FIL2 in $FIL3.\n"

# EOF