cutcols

#!/usr/bin/bash


if [ -z $2 ]; then
	echo "Usage: cutcols [col1] [col2] [...] [filename]"
	echo 
	echo "Extracts columns from a delimited text file by case insensitive header names "
	echo
	exit 
fi


# Last arg is filename; others are requested column names
infile="${!#}"

fileroot=$(echo "${infile}" | sed 's/\.....\?$//')
outfile="${fileroot}_fixed.tsv"

# Collect column names (all args except last)
COLS=("$@")
unset 'COLS[${#COLS[@]}-1]' # remove filename element

awk -v FS="\t" -v OFS="\t" -v want_cols="$(printf "%s\n" "${COLS[@]}")" -v outfile="${outfile}" '
BEGIN {
  n_want = 0
  # Load requested column names
  split(want_cols, lines, /\n/)
  for (i in lines) {
    if (length(lines[i]) == 0) continue
    ++n_want
    wname = lines[i]
    want_order[n_want] = wname
    key = tolower(wname)
    want[key] = n_want  # store position in request order
  }
}

NR == 1 {
  # Map header names to indices
  for (i = 1; i <= NF; i++) {
    header = $i
    key = tolower(header)
    h2idx[key] = i
    original[key] = header
  }

  # Resolve requested columns to indices; warn for missing
  n_out = 0
  for (k = 1; k <= n_want; k++) {
    wname = want_order[k]
    key = tolower(wname)
    if (key in h2idx) {
      ++n_out
      out_idx[n_out] = h2idx[key]
      out_name[n_out] = original[key]
    } else {
      missing[++missing_n] = wname
    }
  }

  # Emit warnings for any missing columns to stderr
  if (missing_n > 0) {
    for (m = 1; m <= missing_n; m++) {
      printf("Warning: column not found: %s\n", missing[m]) > "/dev/stderr"
    }
  }

  # If none found, exit non-zero
  if (n_out == 0) {
    print "Error: none of the requested columns were found in header." > "/dev/stderr"
    exit 3
  }

  # Print header row for selected columns
  for (j = 1; j <= n_out; j++) {
    sel[j] = out_idx[j]
    hdr[j] = out_name[j]
  }
  print_row = ""
  for (j = 1; j <= n_out; j++) {
    print_row = (j == 1) ? hdr[j] : print_row OFS hdr[j]
  }
  print print_row > outfile
  next
}

# Print data rows
{
  print_row = ""
  for (j = 1; j <= n_out; j++) {
    val = (sel[j] <= NF) ? $(sel[j]) : ""
    print_row = (j == 1) ? val : print_row OFS val
  }
  print print_row > outfile
}
' "${infile}"
echo
echo "The desired columns were extracted to ${outfile}"
echo