#!/bin/sh

# Goal: Show total versioned and unversioned size of local tree.
#
# Current status: shows total unversioned size in bytes in SVN trees
#
# TODO: 
#   - This should work with more VCSs than just SVN.
#   - Should show the largest files, or all files over size N
#   - Should print output human-readably

if [ ! -d ".git" -a ! -d ".svn" ]; then
  echo "ERROR: must be run from the top of a version-controlled tree" >&2
  exit 1
fi


if [ ! -d ".svn" ]; then
  echo "ERROR: er, actually, this only supports SVN right now, sorry" >&2
  exit 1
fi

echo "Gathering sizes..."
TOTAL_BYTES=`du -s --block-size=1 | cut -f1`

SAVED_IFS="${IFS}"
IFS='
'
for name in `svn status | grep -E "^?       " | cut -c9-`; do
  # ls --block-size=1 -s "${name}"
  stat -c "%s" "${name}"
done > vsize-tmp-$$
IFS="${SAVED_IFS}"

UNVERSIONED_BYTES=`awk '{ sum += $1 } END { print sum }' vsize-tmp-$$`
VERSIONED_BYTES=`dc -e "${TOTAL_BYTES} ${UNVERSIONED_BYTES} - p"`

# It would be interesting to compare $VERSIONED_BYTES as obtained via
# the above method with calculating it from 'svn ls -vR' output (or
# with whatever the equivalent in git is).

BYTES_PER_MB=1048576  # old-style megabytes are (1024 * 1024) bytes

TOTAL_MB=`dc -e "${TOTAL_BYTES} ${BYTES_PER_MB} / p"`
UNVERSIONED_MB=`dc -e "${UNVERSIONED_BYTES} ${BYTES_PER_MB} / p"`
VERSIONED_MB=`dc -e "${VERSIONED_BYTES} ${BYTES_PER_MB} / p"`

echo ""
echo "Sizes (MB == 1024x1024 bytes):"
echo ""
echo "Total size:         ${TOTAL_MB} MB"
echo "Versioned size:     ${VERSIONED_MB} MB"
echo "Unversioned size:   ${UNVERSIONED_MB} MB"
echo ""

rm vsize-tmp-$$