git-subtree.sh

   1 #!/bin/bash
   2 #
   3 # git-subtree.sh: split/join git repositories in subdirectories of this one
   4 #
   5 # Copyright (C) 2009 Avery Pennarun <apenwarr@gmail.com>
   6 #
   7 if [ $# -eq 0 ]; then
   8     set -- -h
   9 fi
  10 OPTS_SPEC="\
  11 git subtree add --prefix=<prefix> <commit>
  12 git subtree split [options...] --prefix=<prefix> <commit...>
  13 git subtree merge --prefix=<prefix> <commit>
  14 git subtree pull  --prefix=<prefix> <repository> <refspec...>
  15 --
  16 h,help        show the help
  17 q             quiet
  18 prefix=       the name of the subdir to split out
  19  options for 'split'
  20 onto=         try connecting new tree to an existing one
  21 rejoin        merge the new branch back into HEAD
  22 ignore-joins  ignore prior --rejoin commits
  23 "
  24 eval $(echo "$OPTS_SPEC" | git rev-parse --parseopt -- "$@" || echo exit $?)
  25 . git-sh-setup
  26 require_work_tree
  27
  28 quiet=
  29 command=
  30 onto=
  31 rejoin=
  32 ignore_joins=
  33
  34 debug()
  35 {
  36         if [ -z "$quiet" ]; then
  37                 echo "$@" >&2
  38         fi
  39 }
  40
  41 assert()
  42 {
  43         if "$@"; then
  44                 :
  45         else
  46                 die "assertion failed: " "$@"
  47         fi
  48 }
  49
  50
  51 #echo "Options: $*"
  52
  53 while [ $# -gt 0 ]; do
  54         opt="$1"
  55         shift
  56         case "$opt" in
  57                 -q) quiet=1 ;;
  58                 --prefix) prefix="$1"; shift ;;
  59                 --no-prefix) prefix= ;;
  60                 --onto) onto="$1"; shift ;;
  61                 --no-onto) onto= ;;
  62                 --rejoin) rejoin=1 ;;
  63                 --no-rejoin) rejoin= ;;
  64                 --ignore-joins) ignore_joins=1 ;;
  65                 --no-ignore-joins) ignore_joins= ;;
  66                 --) break ;;
  67         esac
  68 done
  69
  70 command="$1"
  71 shift
  72 case "$command" in
  73         add|merge|pull) default= ;;
  74         split) default="--default HEAD" ;;
  75         *) die "Unknown command '$command'" ;;
  76 esac
  77
  78 if [ -z "$prefix" ]; then
  79         die "You must provide the --prefix option."
  80 fi
  81 dir="$prefix"
  82
  83 if [ "$command" != "pull" ]; then
  84         revs=$(git rev-parse $default --revs-only "$@") || exit $?
  85         dirs="$(git rev-parse --no-revs --no-flags "$@")" || exit $?
  86         if [ -n "$dirs" ]; then
  87                 die "Error: Use --prefix instead of bare filenames."
  88         fi
  89 fi
  90
  91 debug "command: {$command}"
  92 debug "quiet: {$quiet}"
  93 debug "revs: {$revs}"
  94 debug "dir: {$dir}"
  95 debug "opts: {$*}"
  96 debug
  97
  98 cache_setup()
  99 {
 100         cachedir="$GIT_DIR/subtree-cache/$$"
 101         rm -rf "$cachedir" || die "Can't delete old cachedir: $cachedir"
 102         mkdir -p "$cachedir" || die "Can't create new cachedir: $cachedir"
 103         debug "Using cachedir: $cachedir" >&2
 104 }
 105
 106 cache_get()
 107 {
 108         for oldrev in $*; do
 109                 if [ -r "$cachedir/$oldrev" ]; then
 110                         read newrev <"$cachedir/$oldrev"
 111                         echo $newrev
 112                 fi
 113         done
 114 }
 115
 116 cache_set()
 117 {
 118         oldrev="$1"
 119         newrev="$2"
 120         if [ "$oldrev" != "latest_old" \
 121              -a "$oldrev" != "latest_new" \
 122              -a -e "$cachedir/$oldrev" ]; then
 123                 die "cache for $oldrev already exists!"
 124         fi
 125         echo "$newrev" >"$cachedir/$oldrev"
 126 }
 127
 128 # if a commit doesn't have a parent, this might not work.  But we only want
 129 # to remove the parent from the rev-list, and since it doesn't exist, it won't
 130 # be there anyway, so do nothing in that case.
 131 try_remove_previous()
 132 {
 133         if git rev-parse "$1^" >/dev/null 2>&1; then
 134                 echo "^$1^"
 135         fi
 136 }
 137
 138 find_existing_splits()
 139 {
 140         debug "Looking for prior splits..."
 141         dir="$1"
 142         revs="$2"
 143         git log --grep="^git-subtree-dir: $dir\$" \
 144                 --pretty=format:'%s%n%n%b%nEND' "$revs" |
 145         while read a b junk; do
 146                 case "$a" in
 147                         git-subtree-mainline:) main="$b" ;;
 148                         git-subtree-split:) sub="$b" ;;
 149                         *)
 150                                 if [ -n "$main" -a -n "$sub" ]; then
 151                                         debug "  Prior: $main -> $sub"
 152                                         cache_set $main $sub
 153                                         try_remove_previous "$main"
 154                                         try_remove_previous "$sub"
 155                                         main=
 156                                         sub=
 157                                 fi
 158                                 ;;
 159                 esac
 160         done
 161 }
 162
 163 copy_commit()
 164 {
 165         # We're doing to set some environment vars here, so
 166         # do it in a subshell to get rid of them safely later
 167         git log -1 --pretty=format:'%an%n%ae%n%ad%n%cn%n%ce%n%cd%n%s%n%n%b' "$1" |
 168         (
 169                 read GIT_AUTHOR_NAME
 170                 read GIT_AUTHOR_EMAIL
 171                 read GIT_AUTHOR_DATE
 172                 read GIT_COMMITTER_NAME
 173                 read GIT_COMMITTER_EMAIL
 174                 read GIT_COMMITTER_DATE
 175                 export  GIT_AUTHOR_NAME \
 176                         GIT_AUTHOR_EMAIL \
 177                         GIT_AUTHOR_DATE \
 178                         GIT_COMMITTER_NAME \
 179                         GIT_COMMITTER_EMAIL \
 180                         GIT_COMMITTER_DATE
 181                 (echo -n '*'; cat ) |  # FIXME
 182                 git commit-tree "$2" $3  # reads the rest of stdin
 183         ) || die "Can't copy commit $1"
 184 }
 185
 186 add_msg()
 187 {
 188         dir="$1"
 189         latest_old="$2"
 190         latest_new="$3"
 191         cat <<-EOF
 192                 Add '$dir/' from commit '$latest_new'
 193
 194                 git-subtree-dir: $dir
 195                 git-subtree-mainline: $latest_old
 196                 git-subtree-split: $latest_new
 197         EOF
 198 }
 199
 200 merge_msg()
 201 {
 202         dir="$1"
 203         latest_old="$2"
 204         latest_new="$3"
 205         cat <<-EOF
 206                 Split '$dir/' into commit '$latest_new'
 207
 208                 git-subtree-dir: $dir
 209                 git-subtree-mainline: $latest_old
 210                 git-subtree-split: $latest_new
 211         EOF
 212 }
 213
 214 toptree_for_commit()
 215 {
 216         commit="$1"
 217         git log -1 --pretty=format:'%T' "$commit" -- || exit $?
 218 }
 219
 220 subtree_for_commit()
 221 {
 222         commit="$1"
 223         dir="$2"
 224         git ls-tree "$commit" -- "$dir" |
 225         while read mode type tree name; do
 226                 assert [ "$name" = "$dir" ]
 227                 echo $tree
 228                 break
 229         done
 230 }
 231
 232 tree_changed()
 233 {
 234         tree=$1
 235         shift
 236         if [ $# -ne 1 ]; then
 237                 return 0   # weird parents, consider it changed
 238         else
 239                 ptree=$(toptree_for_commit $1)
 240                 if [ "$ptree" != "$tree" ]; then
 241                         return 0   # changed
 242                 else
 243                         return 1   # not changed
 244                 fi
 245         fi
 246 }
 247
 248 copy_or_skip()
 249 {
 250         rev="$1"
 251         tree="$2"
 252         newparents="$3"
 253         assert [ -n "$tree" ]
 254
 255         identical=
 256         p=
 257         for parent in $newparents; do
 258                 ptree=$(toptree_for_commit $parent) || exit $?
 259                 if [ "$ptree" = "$tree" ]; then
 260                         # an identical parent could be used in place of this rev.
 261                         identical="$parent"
 262                 fi
 263                 if [ -n "$ptree" ]; then
 264                         parentmatch="$parentmatch$parent"
 265                         p="$p -p $parent"
 266                 fi
 267         done
 268
 269         if [ -n "$identical" -a "$parentmatch" = "$identical" ]; then
 270                 echo $identical
 271         else
 272                 copy_commit $rev $tree "$p" || exit $?
 273         fi
 274 }
 275
 276 ensure_clean()
 277 {
 278         if ! git diff-index HEAD --exit-code --quiet; then
 279                 die "Working tree has modifications.  Cannot add."
 280         fi
 281         if ! git diff-index --cached HEAD --exit-code --quiet; then
 282                 die "Index has modifications.  Cannot add."
 283         fi
 284 }
 285
 286 cmd_add()
 287 {
 288         if [ -e "$dir" ]; then
 289                 die "'$dir' already exists.  Cannot add."
 290         fi
 291         ensure_clean
 292
 293         set -- $revs
 294         if [ $# -ne 1 ]; then
 295                 die "You must provide exactly one revision.  Got: '$revs'"
 296         fi
 297         rev="$1"
 298
 299         debug "Adding $dir as '$rev'..."
 300         git read-tree --prefix="$dir" $rev || exit $?
 301         git checkout "$dir" || exit $?
 302         tree=$(git write-tree) || exit $?
 303
 304         headrev=$(git rev-parse HEAD) || exit $?
 305         if [ -n "$headrev" -a "$headrev" != "$rev" ]; then
 306                 headp="-p $headrev"
 307         else
 308                 headp=
 309         fi
 310         commit=$(add_msg "$dir" "$headrev" "$rev" |
 311                  git commit-tree $tree $headp -p "$rev") || exit $?
 312         git reset "$commit" || exit $?
 313 }
 314
 315 cmd_split()
 316 {
 317         debug "Splitting $dir..."
 318         cache_setup || exit $?
 319
 320         if [ -n "$onto" ]; then
 321                 debug "Reading history for --onto=$onto..."
 322                 git rev-list $onto |
 323                 while read rev; do
 324                         # the 'onto' history is already just the subdir, so
 325                         # any parent we find there can be used verbatim
 326                         debug "  cache: $rev"
 327                         cache_set $rev $rev
 328                 done
 329         fi
 330
 331         if [ -n "$ignore_joins" ]; then
 332                 unrevs=
 333         else
 334                 unrevs="$(find_existing_splits "$dir" "$revs")"
 335         fi
 336
 337         # We can't restrict rev-list to only "$dir" here, because that leaves out
 338         # critical information about commit parents.
 339         debug "git rev-list --reverse --parents $revs $unrevs"
 340         git rev-list --reverse --parents $revs $unrevs |
 341         while read rev parents; do
 342                 debug
 343                 debug "Processing commit: $rev"
 344                 exists=$(cache_get $rev)
 345                 if [ -n "$exists" ]; then
 346                         debug "  prior: $exists"
 347                         continue
 348                 fi
 349                 debug "  parents: $parents"
 350                 newparents=$(cache_get $parents)
 351                 debug "  newparents: $newparents"
 352
 353                 tree=$(subtree_for_commit $rev "$dir")
 354                 debug "  tree is: $tree"
 355                 [ -z $tree ] && continue
 356
 357                 newrev=$(copy_or_skip "$rev" "$tree" "$newparents") || exit $?
 358                 debug "  newrev is: $newrev"
 359                 cache_set $rev $newrev
 360                 cache_set latest_new $newrev
 361                 cache_set latest_old $rev
 362         done || exit $?
 363         latest_new=$(cache_get latest_new)
 364         if [ -z "$latest_new" ]; then
 365                 die "No new revisions were found"
 366         fi
 367
 368         if [ -n "$rejoin" ]; then
 369                 debug "Merging split branch into HEAD..."
 370                 latest_old=$(cache_get latest_old)
 371                 git merge -s ours \
 372                         -m "$(merge_msg $dir $latest_old $latest_new)" \
 373                         $latest_new >&2
 374         fi
 375         echo $latest_new
 376         exit 0
 377 }
 378
 379 cmd_merge()
 380 {
 381         ensure_clean
 382
 383         set -- $revs
 384         if [ $# -ne 1 ]; then
 385                 die "You must provide exactly one revision.  Got: '$revs'"
 386         fi
 387         rev="$1"
 388
 389         git merge -s subtree $rev
 390 }
 391
 392 cmd_pull()
 393 {
 394         ensure_clean
 395         set -x
 396         git pull -s subtree "$@"
 397 }
 398
 399 "cmd_$command" "$@"