path: root/scripts/mageia-reset-and-import.sh

#!/bin/bash

NEWSVNURL="file:///home/colin/svn-git/soft"
SOFTWARE=
if [ -d ".git" -a -f ".git/config" -a -f ".git/svn/.metadata" ]; then
  SOFTWARE=$(git config svn-remote.svn.fetch | cut -d':' -f1 | sed 's,/trunk,,' | trim)
fi
NEWSOFTWARE=$SOFTWARE
if [ -n "$1" ]; then
  NEWSOFTWARE=$1
fi
echo
echo "Identified software name: $SOFTWARE (New name: $NEWSOFTWARE)"

if [ -z "$SOFTWARE" ]; then
  echo "You must run this in a git-svn clone folder." >&2
  exit 1
fi

if [ -d ../$SOFTWARE.git ]; then
  echo "It seems the converted git repository already exists (../$SOFTWARE.git)" >&2
  exit 1
fi

AUTHORMAP=$(realpath $(dirname $(realpath $0))/../metadata/mageia-user-map.txt)
if [ ! -f "$AUTHORMAP" ]; then
  echo "Cannot find author map." >&2
  exit 1
fi

REPLAY=
if [ -f ../svn-import.log ]; then
  REPLAYDATA=$(grep ^$SOFTWARE, ../svn-import.log | tail -n 1)
  if [ -n "$REPLAYDATA" ]; then
    resetsha1=$(echo $REPLAYDATA | cut -d',' -f 2)
    revision=$(echo $REPLAYDATA | cut -d',' -f 3)
    morerevisionsx=$(echo $REPLAYDATA | cut -d',' -f 4)
    skiprevisionsx=$(echo $REPLAYDATA | cut -d',' -f 5)
    synthesized=$(echo $REPLAYDATA | cut -d',' -f 6)

    echo
    echo "I have found data from a previous run:"
    echo
    echo "Reset SHA1:                    $resetsha1"
    echo "Mageia Import SVN Revision:    $revision"
    echo "Extra SVN Revisions to squash:$morerevisionsx"
    echo "SVN Revisions to skip:        $skiprevisionsx"
    echo
    echo "Do you want to replay it? [Y/n]"
    read REPLAY
    if [ -z "$REPLAY" -o "Y" = "$REPLAY" ]; then
      REPLAY=y
    else
      REPLAY=
    fi
  fi
fi

remotes=$(git remote -v | cut -f1 | sort -u | xargs)
if [ -n "$remotes" ]; then
  echo -n "Removing old remotes '$remotes'..."
  for remote in $remotes; do
    git remote rm $remote
  done
  echo "done"
fi

echo -n "Finding UUID for SVN Repository $NEWSVNURL... "
NEWSVNUUID=$(svn info $NEWSVNURL | grep "^Repository UUID:" | sed 's/Repository UUID: //')
echo $NEWSVNUUID

echo -n "Updating SVN Remote config (old value: $(git config svn-remote.svn.url))... "
git config svn-remote.svn.url $NEWSVNURL
git config svn-remote.svn.fetch "$NEWSOFTWARE/trunk:refs/remotes/trunk"
git config svn-remote.svn.branches "$NEWSOFTWARE/branches/*:refs/remotes/*"
git config svn-remote.svn.tags "$NEWSOFTWARE/tags/*:refs/remotes/tags/*"
echo "done"

echo -n "Updating Internal SVN metadata... "
OLDSVNUUID=$(grep "uuid = " .git/svn/.metadata | cut -d= -f2 | trim)
sed -i "s,reposRoot = .*$,reposRoot = $NEWSVNURL,;s,uuid = .*$,uuid = $NEWSVNUUID," .git/svn/.metadata
echo "done"


echo
echo "OK now the fun stuff beings. You need to find the git commit at which you want"
echo "to start the magiea import. This will typically require some investigation but"
echo "the date will be around February 2011."
echo "I will drop you to a shell. When you are done, simply copy the SHA1 sum and"
echo "exit."
echo "The commands you like want are 'git log trunk' and 'git show SHA1'"
echo
if [ -z "$REPLAY" ]; then
  echo "Press any key to contiue."
  read -n 1
  bash
  echo
  echo "OK you're back :)"
  echo "What was the SHA1?"
  read resetsha1
fi

if [ -z "$resetsha1" ]; then
  echo "You did not give any SHA1 sum. I will not reset the git repository and carry"
  echo "on importing from it's current state. You can quit this script and start"
  echo "again if this is incorrect."
else
  echo "I got SHA1 '$resetsha1'. I hope that's correct".
  echo
  echo "I will now reset your git trunk and master branches."
  if [ -z "$REPLAY" ]; then
    echo "Is this OK? (press ctrl+c if not)"
    read -n 1
  fi
  git update-ref -m "Rewind SVN Git to pre-Magiea import" refs/remotes/trunk $resetsha1
  git checkout master
  git reset --hard trunk

  # Remove any heads that are newer than this (except trunk)
  resetreldate=$(git show -s --pretty='format:%ar' $resetsha1)
  resettimestamp=$(git show -s --pretty='format:%at' $resetsha1)
  echo "Removing any branches/tags newer than '$resetsha1' ($resetreldate)..."
  git for-each-ref --format='%(refname)' refs/remotes/tags/* refs/remotes/branches/* | \
    while read ref; do
      # As svn tags show as new commits, we mustn't parse $ref directly, but
      # traverse back up the tag tree to find the first commit that *do not*
      # result in any changes (i.e. not an svn cp) and use that as our reference
      # comparison point.
      reftreesha1=$(git rev-parse "$ref":)
      parentref=$ref
      while [ "$(git rev-parse --quiet --verify "$parentref"^: )" = "$reftreesha1" ]; do
        parentref="$parentref"^
      done

      reftimestamp=$(git show -s --pretty='format:%at' $parentref)
      if [ $reftimestamp -gt $resettimestamp ]; then
        echo "  Removing '$ref'"
        git update-ref -d $ref

        # Also trash the old revmap
        rm -f .git/svn/$ref/.rev_map.*
      fi
    done
  echo "... done"
fi


echo
echo "I will now run 'svn log' on the software. You should be able to guess the right"
echo "revision to base the import on. When done, simply copy (or remember) the"
echo "revision number and enter it into this script. You will also be able to"
echo "(optional) select several other commits which you wish to squash into this"
echo "initial commit."
echo
if [ -z "$REPLAY" ]; then
  echo "Press any key to contiue."
  read -n 1

  svn log $NEWSVNURL/$NEWSOFTWARE | less

  echo
  echo "OK you're back :)"
  echo "What was the initial revision number?"
  read revision
fi
revision=$(echo $revision | sed 's/r//g')
revision=$(( $revision + 0 ))
echo "I got revision '$revision'. I hope that's correct".
echo

if [ -z "$REPLAY" ]; then
  echo "Now enter any further commits you wish to squash into our initial commit:"
  read morerevisionsx
fi
morerevisionsx=$(echo $morerevisionsx | trim)
if [ -n "$morerevisionsx" ]; then
  for rev in $morerevisionsx; do
    rev=$(echo $rev | sed 's/r//g')
    rev=$(( $rev + 0 ))
    if [ $rev -gt 0 ]; then
      morerevisions="$morerevisions $rev"
    fi
  done
  echo
  echo "OK, you said you also wanted to squash the following commits:"
  echo "  $morerevisions"
  echo
fi

if [ -z "$REPLAY" ]; then
  echo "Now enter any commits you wish to skip completely (e.g. for commits+reverts)"
  read skiprevisionsx
fi
skiprevisionsx=$(echo $skiprevisionsx | trim)
if [ -n "$skiprevisionsx" ]; then
  for rev in $skiprevisionsx; do
    rev=$(echo $rev | sed 's/r//g')
    rev=$(( $rev + 0 ))
    if [ $rev -gt 0 ]; then
      skiprevisions="$skiprevisions $rev"
    fi
  done
  echo
  echo "OK, you said you wanted to skip the following commits:"
  echo "  $skiprevisions"
  echo
fi

echo
echo "The next step is the dangerous bit. We will clear out all files in the current"
echo "working copy, run an svn export and then drop you to a shell to investigate the"
echo "differences."
echo "In that shell you must git add any new files that appear and ensure you are"
echo "happy with the general state of things."
echo
echo "Continue? (press ctrl+c if not)"
read -n 1
git reset --hard trunk
resetsha1=$(git rev-list --max-count=1 HEAD)
# The below is a "safer" verion of rm -rf * that also deletes hidden files
find . -depth -not -iwholename '.' -not -iwholename "./.git" -not -iwholename "./.git/*" -not -iwholename "./.git/.*" -delete

# We have to create a subversion repository here for the import as we need a repository
# for the svn merging below... svn sucks.
SVNPATH=$(realpath ../$SOFTWARE.svn)
rm -rf $SVNPATH
mkdir -p $SVNPATH/export
cd $SVNPATH/export
svn export --force --ignore-keywords --ignore-externals -r $revision $NEWSVNURL/$NEWSOFTWARE/trunk .
cd -
svnadmin create $SVNPATH/repo
svnrepo=file://$SVNPATH/repo
svn import -m "Import" $SVNPATH/export $svnrepo
svn checkout $svnrepo $SVNPATH/checkout
cd $SVNPATH/checkout
if [ -n "$morerevisions" ]; then
  for rev in $morerevisions; do
    svn merge -c $rev --ignore-ancestry --accept theirs-full --allow-mixed-revisions $NEWSVNURL/$NEWSOFTWARE/trunk .
    svn commit -m "Merged r$rev"
    if [ $? -ne 0 ]; then
      echo
      echo "Merging in svn revision $rev failed. You need to rethink your import strategy."
      echo
      exit 1
    fi
  done
fi
cd -
svn export --force $svnrepo .
rm -rf $SVNPATH
echo
echo "Dropping you to a shell. Use 'git status', 'git diff' and 'git add'."
echo "DO NOT run 'git commit', I'll do that later for you."
bash

echo
echo "OK you're back :)"
echo "If everything is OK, and you're ready to continue, then hit return."
read -n 1

COMMITTXT=$(mktemp /tmp/commitmsg.XXXXXX)
cat >$COMMITTXT <<EOF
Synthesized commit during git-svn import combining previous Mandriva history with Magiea.

This commit consitsts of the following subversion commits:

EOF
svn log -r 1:$revision $NEWSVNURL/$NEWSOFTWARE/trunk >>$COMMITTXT
squashedrevs=$(svn log -r 1:$revision $NEWSVNURL/$NEWSOFTWARE/trunk | grep -E "^r[0-9]+ \| " | cut -d' ' -f1 | xargs | sed 's/r//g')

rev=0
if [ -n "$morerevisions" ]; then
  for rev in $morerevisions; do
    svn log -r $rev $NEWSVNURL/$NEWSOFTWARE | tail -n +2 >>$COMMITTXT
  done
fi

# Import any documentation from the CLEANUP_PROGRESS.txt file
[ "$rev" = "0" ] && rev=$revision
CLEANUP_PROGRESS=$(mktemp /tmp/cleanupprogress.XXXXXX)
svn cat $NEWSVNURL/$NEWSOFTWARE/CLEANUP_PROGRESS.txt@$rev 2>/dev/null >$CLEANUP_PROGRESS
if [ -s $CLEANUP_PROGRESS ]; then
  echo >>$COMMITTXT
  echo "In addition to the above commits, the following cleaning work has been recorded:" >>$COMMITTXT
  echo >>$COMMITTXT
  cat $CLEANUP_PROGRESS >>$COMMITTXT
fi
rm -f $CLEANUP_PROGRESS

COMMITDATE=$(svn log -r $revision $NEWSVNURL/$NEWSOFTWARE | head -n2 | tail -n1 | cut -d'|' -f3 | cut -d'(' -f1 | trim)
env GIT_COMMITTER_NAME="Mageia SVN-Git Migration" \
    GIT_COMMITTER_EMAIL="svn-git-migration@mageia.org" \
    GIT_COMMITTER_DATE="$COMMITDATE" \
    GIT_AUTHOR_NAME="Mageia SVN-Git Migration" \
    GIT_AUTHOR_EMAIL="svn-git-migration@mageia.org" \
    GIT_AUTHOR_DATE="$COMMITDATE" \
    git commit -a --file=$COMMITTXT
rm -f $COMMITTXT

echo
echo "OK, looks like we're good to go :)"
echo "Press any key to continue with the subversion import process".
read -n 1

git svn gc
fakesha1=$(git rev-list --max-count=1 HEAD)
echo
echo -n "Updating trunk branch to include synthesized commit $fakesha1... "
git update-ref -m "Reset to synthesized commit representing Magiea cleaned import (SVN Git)" refs/remotes/trunk $fakesha1
echo "done"

echo -n "Faking git-svn metadata to allow continuation... "
printf "0000000: %08x%s\t................" $revision $fakesha1 | xxd -r -c24 >.git/svn/refs/remotes/trunk/.rev_map.$NEWSVNUUID
echo "done"

# Keep a small log of the sha1's etc.
if [ -z "$REPLAY" ]; then
  if [ ! -f ../svn-import.log ]; then
    echo "software,startingcommit,startingrevision,otherrevisions,skiprevisions,fakecommit" >../svn-import.log
  fi
  echo "$SOFTWARE,$resetsha1,$revision,$morerevisions,$skiprevisions,$fakesha1" >>../svn-import.log
else
  sed -i "s/,$synthesized\$/,$fakesha1/" ../svn-import.log
fi

echo "Continuing svn fetch"
if [ -n "$morerevisions" ]; then
  for rev in $morerevisions; do
    git svn fetch -A $AUTHORMAP -r $(( $revision + 1 )):$(( $rev - 1 ))
    revision=$rev
  done
fi

# Note the revisions to skip have to be after any squashed revisions which we
# should really check for....
if [ -n "$skiprevisions" ]; then
  for rev in $skiprevisions; do
    git svn fetch -A $AUTHORMAP -r $(( $revision + 1 )):$(( $rev - 1 ))
    revision=$rev
  done
fi

git svn fetch -A $AUTHORMAP -r $(( $revision + 1 )):HEAD
git reset --hard trunk
echo
echo "done"

echo
echo "Now creating bare git repository"
git init --bare ../$SOFTWARE.git
git remote rm origin
git remote add origin ../$SOFTWARE.git
#git push --set-upstream master
git push origin master 'refs/remotes/*:refs/heads/*'

pushd ../$SOFTWARE.git
git branch -D trunk
git branch -D origin/master 2>/dev/null

# Tagging logic inspired by https://github.com/nothingmuch/git-svn-abandon/blob/master/git-svn-abandon-fix-refs
# Keep a map of tags and their original sha1 for the SQL database cerated below
declare -A tagmap
tags=$(git for-each-ref --format='%(refname)' refs/heads/tags | cut -d / -f 4)
for tag in $tags; do
  ref="refs/heads/tags/$tag"
  refsha1=$(git rev-parse "$ref")
  reftreesha1=$(git rev-parse "$ref":)

  # Find the oldest ancestor for which the tree is the same
  parentref="$ref"
  while [ "$(git rev-parse --quiet --verify "$parentref"^:)" = "$reftreesha1" ]; do
    parentref="$parentref"^
  done
  parent=$(git rev-parse "$parentref")

  # If this ancestor is in master then we can just tag it
  # otherwise the tag has diverged from master and it's actually more like a
  # branch than a tag
  merge=$(git merge-base master $parent)
  if [ "$merge" = "$parent" ]; then
    targetref=$parent
  else
    targetref=$refsha1
  fi

  tagmap[$refsha1]=$tag

  # create an annotated tag based on the last commit in the tag, and delete the "branchy" ref for the tag
  git show -s --pretty='format:%s%n%n%b' "$ref" | \
    env GIT_COMMITTER_NAME="$(git show -s --pretty='format:%an' "$ref")" \
        GIT_COMMITTER_EMAIL="$(git show -s --pretty='format:%ae' "$ref")" \
        GIT_COMMITTER_DATE="$(git show -s --pretty='format:%ad' "$ref")" \
        GIT_AUTHOR_NAME="$(git show -s --pretty='format:%an' "$ref")" \
        GIT_AUTHOR_EMAIL="$(git show -s --pretty='format:%ae' "$ref")" \
        GIT_AUTHOR_DATE="$(git show -s --pretty='format:%ad' "$ref")" \
        git tag -a -F - "$tag" "$targetref"

  git update-ref -d "$ref"
done

# Implement a branch name policy
#   1. Branches matching ([1-3]) are renamed to distro/mga\1
#   2. Branches matching (20[0-9\.]+) are renamed to distro/mdv\1
#   3. Branches matching (mes[0-9].*) are renamed to distro/\1
#   4. All other branches are renamed to topic/\1 (excl. master)

branches=$(git for-each-ref --format='%(refname)' refs/heads | cut -d / -f 3-)
for branch in $branches; do
  ref="refs/heads/$branch"
  refsha1=$(git rev-parse "$ref")
  reftreesha1=$(git rev-parse "$ref":)

  # Find the oldest ancestor for which the tree is the same
  parentref="$ref"
  while [ "$(git rev-parse --quiet --verify "$parentref"^:)" = "$reftreesha1" ]; do
    parentref="$parentref"^
  done
  parent=$(git rev-parse "$parentref")

  # If this ancestor is in master then we can just squash it
  # otherwise the branch has diverged from master and it's actually a proper
  # branch
  merge=$(git merge-base master $parent)
  if [ "$merge" = "$parent" ]; then
    git update-ref "$ref" $parent $refsha1
  fi

  if (echo $branch | grep -qE '^[0-3]$'); then
    git branch -m $branch distro/mga$branch
  elif (echo $branch | grep -qE '^20[0-9\.]+$'); then
    git branch -m $branch distro/mdv$branch
  elif (echo $branch | grep -qE '^mes[0-9].*$'); then
    git branch -m $branch distro/$branch
  elif [ "$branch" != "master" ]; then
    git branch -m $branch topic/$branch
  fi
done

git gc --aggressive
popd


# And finally we create some SQL for creating a nice revision map database

# CREATE TABLE refs (distro char(3) NOT NULL, soft varchar(255) NOT NULL, revision int(10) unsigned NOT NULL,  sha1 char(40) NOT NULL, head varchar(255) NOT NULL);

# Parse a git-svn revmap file into SQL
# $1 = Name of revmap file
# $2 = Distro ('mdv' or 'mga')
# $3 = head name i.e. branch or tag name (only used when not processing trunk/master)
# $4 = Special SHA1 (last sha1 to look for if 'mdv', fake sha1 commit if 'mga')
parsesvnrevmap()
{
  if [ ! -f "$1" ]; then
    echo "No such file '$1'" >&2
    exit 1
  fi
  for map in $(cat "$1" | xxd -c24 -g24  | cut -b 10-57); do
    sha1=$(echo $map | cut -b 9-)
    if [ "$sha1" = "0000000000000000000000000000000000000000" ]; then
      continue
    fi
    if [ -n "$4" -a "$2" = "mga" -a "$sha1" = "$4" ]; then
      continue
    fi
    rev=$(printf "%d" 0x$(echo $map | cut -b 1-8))

    if [ -n "${tagmap[$sha1]}" ]; then
      echo "INSERT INTO refs VALUES('$2', '$SOFTWARE', $rev, '', '${tagmap[$sha1]}');"
    else
      branch=$3
      if [ -n "$branch" ]; then
        if (echo $branch | grep -qE '^[0-3]$'); then
          branch=distro/mga$branch
        elif (echo $branch | grep -qE '^20[0-9\.]+$'); then
          branch=distro/mdv$branch
        elif (echo $branch | grep -qE '^mes[0-9].*$'); then
          branch=distro/$branch
        elif [ "$branch" != "master" ]; then
          branch=topic/$branch
        fi
      fi
      echo "INSERT INTO refs VALUES('$2', '$SOFTWARE', $rev, '$sha1', '$branch');"
    fi

    # Exit if we've reached our reset sha1 as any future revisions are not used
    # by Mageia
    if [ -n "$4" -a "$2" = "mdv" -a "$sha1" = "$4" ]; then
      break
    fi
  done
}

echo
echo -n "Creating revision -> sha1 map SQL... "

sql="../$SOFTWARE-revmap.sql"
rm -f "$sql" "$sql".xz

# First to the old (i.e. mdv) ones
revmap=".git/svn/refs/remotes/trunk/.rev_map.$OLDSVNUUID"
if [ -f "$revmap" ]; then
  parsesvnrevmap "$revmap" "mdv" "" $resetsha1 >>"$sql"
fi
for revmap in $(find .git/svn/refs/remotes/tags -name .rev_map.$OLDSVNUUID 2>/dev/null); do
  # Note, still pass in the tag name in case the tag is really more of a branch...
  parsesvnrevmap "$revmap" "mdv" "$(echo $revmap | cut -d'/' -f6)" >>"$sql"
done
for revmap in $(find .git/svn/refs/remotes -maxdepth 2 -not -iwholename ".git/svn/refs/remotes/trunk/.rev_map.*" -name .rev_map.$OLDSVNUUID 2>/dev/null); do
  parsesvnrevmap "$revmap" "mdv" "$(echo $revmap | cut -d'/' -f5)" >>"$sql"
done


# Inject the squashed commits into the map
for rev in $squashedrevs $morerevisions; do
  echo "INSERT INTO refs VALUES('mga', '$SOFTWARE', $rev, '$fakesha1', '');" >>"$sql"
done
if [ -n "$skiprevisions" ]; then
  for rev in $skiprevisions; do
    echo "INSERT INTO refs VALUES('mga', '$SOFTWARE', $rev, '', '');" >>"$sql"
  done
fi

# And the rest of the Mageia commits
revmap=".git/svn/refs/remotes/trunk/.rev_map.$NEWSVNUUID"
if [ -f "$revmap" ]; then
  parsesvnrevmap "$revmap" "mga" "" $fakesha1 >>"$sql"
fi
for revmap in $(find .git/svn/refs/remotes/tags -name .rev_map.$NEWSVNUUID 2>/dev/null); do
  # Note, still pass in the tag name in case the tag is really more of a branch...
  parsesvnrevmap "$revmap" "mga" "$(echo $revmap | cut -d'/' -f6)" >>"$sql"
done
for revmap in $(find .git/svn/refs/remotes -maxdepth 2 -not -iwholename ".git/svn/refs/remotes/trunk/.rev_map.*" -name .rev_map.$NEWSVNUUID 2>/dev/null); do
  parsesvnrevmap "$revmap" "mga" "$(echo $revmap | cut -d'/' -f5)" >>"$sql"
done

xz "$sql"

echo "done"