#!/bin/sh

# NAME
#
#   letter_freq.sh - analyze letter frequency in a string
#
# SYNOPSIS
#
#   letter_freq.sh -f filename
#   letter_freq.sh "string to analyze"
#
# INSTALLATION
#
#   Just copy this into a file called letter_freq.sh, and then:
#
#   $ chmod +x ./letter_freq.sh
#
# CAVEATS
#
#   Shell is terrible and Brennen is terrible at shell, so this
#   is doubly likely to contain traps for the unwary.
#
# AUTHORS
#
#   https://mspsocial.net/@earthtopus
#   https://mastodon.social/@brennen

# A function for squawking at the user with some helptext:
print_usage () {
  cat <<END

USAGE

  # Examine a string directly:
  $ ./letter_freq.sh "Quick zephyrs blow, vexing daft Jim."

  # Examine a string from a file:
  $ ./letter_freq.sh -f pangram.txt

END
}

# Encapsulate the sort operation in a function for less cruft below:
do_sort () {
  sort | uniq -c | sort -rn;
}

# The script really starts here.

# First, we'll handle arguments, hackily.  $1 is the first parameter to the
# command, $2 is the second, and so on.  There's a trap here - if you say
# something like
#
#   ./letter_freq.sh foo bar
#
# ...then $1 will be 'foo' and $2 will be 'bar'.  To have an entire string with
# spaces come in as the first param, use quotes like so:
#
#   ./letter_freq.sh "foo bar"
#
# The same sort of thing applies all over the place.  When in doubt, put double
# quotes around variables to avoid weirdness.

if [ "$1" = '-f' ]; then
  if [ -z "$2" ]; then
    # Zero-length string for second param - didn't get a filename
    echo "Expected a filename to read string from."
    print_usage
    # These arbitrary-seeming exit codes are taken from:
    # https://www.freebsd.org/cgi/man.cgi?query=sysexits&sektion=3
    # (I don't actually know whether this is a good idea.)
    exit 64
  fi
  if [ ! -f "$2" ]; then
    echo "No such file: $2"
    print_usage
    exit 64
  fi

  # Read the file into a variable:
  string=$(cat "$2")
elif [ -z "$1" ]; then
  echo "Expected a string to analyze."
  print_usage
  exit 66
else
  # Use the provided string:
  string="$1"
fi

# Next, we actually Do the Stuff:

count=$(echo "$string" | tr -cd '[:alpha:]' | wc -m)
echo "$count total chars"

echo "$string" | grep -oE '[[:alpha:]]' | \
  tr '[:upper:]' '[:lower:]' | \
  do_sort | \
  awk -v count="$count" '{ print ($1 / count), $1, $2; }'

# Success!
exit 0