#!/bin/sh
#
# Given a large set of MH folders, in multiple MH trees, generate a list of
# tests hit by each message.   Output the test lists into "spam.local" and
# "nonspam.local".
#
# This then gives us a statistical base to test weight changes against.

# the directories to search.
MAILDIRS="$HOME/Mail $HOME/Mail.old"

# the names of folders that usually contain only spam messages.  All
# other folders are assumed to contain non-spam.  Refile messages
# to ensure this is the case, where possible, to keep your output
# consistent!
ARE_SPAM="Sapm Spam MissedSpam spam"

# skipped folders.
SKIPPED="drafts outbox SA"

: > commands.sh

for maildir in $MAILDIRS ; do
  [ -d "$maildir" ] || continue

  for folder in $maildir/* ; do
    [ -d "$folder" ] || continue
    [ -f "$folder/1" ] || continue

    if perl -e '@pats = qw('"$SKIPPED"'); $_ = q('"$folder"');
		foreach $pat (@pats) { exit 0 if /\/$pat$/; } exit 1;'
    then
      continue
    fi

    echo "echo Checking $folder:" >> commands.sh
    if perl -e '@pats = qw('"$ARE_SPAM"'); $_ = q('"$folder"');
		foreach $pat (@pats) { exit 0 if /\/$pat$/; } exit 1;'
    then
      echo "./mass-phrase-freq $folder >> spam.local" >> commands.sh
    else
      echo "./mass-phrase-freq $folder >> nonspam.local" >> commands.sh
    fi
  done
done

echo "Now run commands.sh to check all folders."