parsemail

#!/bin/sh

# TODO:
# - multiple 'to' addresses: parse only until first comma (first address)
# - print usage

# Usage:
# `parsemail [PATH_TO_MAIL_FILE]
# Returns tab-separated result in following order:
#   TO, FROM, SUBJECT, PRIORITY

# awk commands:
# line 1: replace first colon by tab to split into 2 columns by first colon.
# line 6: return tab-separated results

DECODE_HEADER_PYTHON='import email.header; import sys; bytes, enc = email.header.decode_header(sys.stdin.read())[0]; print(bytes.decode(enc) if enc else bytes.replace("\n", ""))'

cat "${1--}" | sed 's/\t/ /g' | # replace tabs with whitespaces first
  awk -F '\t' '
{ sub(/: /, "\t") }
$1 == "To" { to = $2 }
$1 == "From" { from = $2 }
$1 == "Subject" { subject = $2 }
$1 == "X-Priority" { priority = $2 }
END { print to "\t" from "\t" subject "\t" priority }
' | while IFS="$(printf '\t')" read -r TO FROM SUBJECT PRIORITY; do
  printf "%s\t%s\t%s\t%s\n" \
    "$TO" \
    "$FROM" \
    "$(printf "%s" "$SUBJECT" | python -c "$DECODE_HEADER_PYTHON")" \
    "$PRIORITY"
done