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 merge --prefix=<prefix> <commit>
  13 git subtree pull  --prefix=<prefix> <repository> <refspec...>
  14 git subtree split --prefix=<prefix> <commit...>
  15 --
  16 h,help        show the help
  17 q             quiet
  18 d             show debug messages
  19 prefix=       the name of the subdir to split out
  20  options for 'split'
  21 annotate=     add a prefix to commit message of new commits
  22 b,branch=     create a new branch from the split subtree
  23 ignore-joins  ignore prior --rejoin commits
  24 onto=         try connecting new tree to an existing one
  25 rejoin        merge the new branch back into HEAD
  26  options for 'add', 'merge', and 'pull'
  27 squash        merge subtree changes as a single commit
  28 "
  29 eval $(echo "$OPTS_SPEC" | git rev-parse --parseopt -- "$@" || echo exit $?)
  30 PATH=$(git --exec-path):$PATH
  31 . git-sh-setup
  32 require_work_tree
  33
  34 quiet=
  35 branch=
  36 debug=
  37 command=
  38 onto=
  39 rejoin=
  40 ignore_joins=
  41 annotate=
  42 squash=
  43
  44 debug()
  45 {
  46         if [ -n "$debug" ]; then
  47                 echo "$@" >&2
  48         fi
  49 }
  50
  51 say()
  52 {
  53         if [ -z "$quiet" ]; then
  54                 echo "$@" >&2
  55         fi
  56 }
  57
  58 assert()
  59 {
  60         if "$@"; then
  61                 :
  62         else
  63                 die "assertion failed: " "$@"
  64         fi
  65 }
  66
  67
  68 #echo "Options: $*"
  69
  70 while [ $# -gt 0 ]; do
  71         opt="$1"
  72         shift
  73         case "$opt" in
  74                 -q) quiet=1 ;;
  75                 -d) debug=1 ;;
  76                 --annotate) annotate="$1"; shift ;;
  77                 --no-annotate) annotate= ;;
  78                 -b) branch="$1"; shift ;;
  79                 --prefix) prefix="$1"; shift ;;
  80                 --no-prefix) prefix= ;;
  81                 --onto) onto="$1"; shift ;;
  82                 --no-onto) onto= ;;
  83                 --rejoin) rejoin=1 ;;
  84                 --no-rejoin) rejoin= ;;
  85                 --ignore-joins) ignore_joins=1 ;;
  86                 --no-ignore-joins) ignore_joins= ;;
  87                 --squash) squash=1 ;;
  88                 --no-squash) squash= ;;
  89                 --) break ;;
  90                 *) die "Unexpected option: $opt" ;;
  91         esac
  92 done
  93
  94 command="$1"
  95 shift
  96 case "$command" in
  97         add|merge|pull) default= ;;
  98         split) default="--default HEAD" ;;
  99         *) die "Unknown command '$command'" ;;
 100 esac
 101
 102 if [ -z "$prefix" ]; then
 103         die "You must provide the --prefix option."
 104 fi
 105 dir="$prefix"
 106
 107 if [ "$command" != "pull" ]; then
 108         revs=$(git rev-parse $default --revs-only "$@") || exit $?
 109         dirs="$(git rev-parse --no-revs --no-flags "$@")" || exit $?
 110         if [ -n "$dirs" ]; then
 111                 die "Error: Use --prefix instead of bare filenames."
 112         fi
 113 fi
 114
 115 debug "command: {$command}"
 116 debug "quiet: {$quiet}"
 117 debug "revs: {$revs}"
 118 debug "dir: {$dir}"
 119 debug "opts: {$*}"
 120 debug
 121
 122 cache_setup()
 123 {
 124         cachedir="$GIT_DIR/subtree-cache/$$"
 125         rm -rf "$cachedir" || die "Can't delete old cachedir: $cachedir"
 126         mkdir -p "$cachedir" || die "Can't create new cachedir: $cachedir"
 127         debug "Using cachedir: $cachedir" >&2
 128 }
 129
 130 cache_get()
 131 {
 132         for oldrev in $*; do
 133                 if [ -r "$cachedir/$oldrev" ]; then
 134                         read newrev <"$cachedir/$oldrev"
 135                         echo $newrev
 136                 fi
 137         done
 138 }
 139
 140 cache_set()
 141 {
 142         oldrev="$1"
 143         newrev="$2"
 144         if [ "$oldrev" != "latest_old" \
 145              -a "$oldrev" != "latest_new" \
 146              -a -e "$cachedir/$oldrev" ]; then
 147                 die "cache for $oldrev already exists!"
 148         fi
 149         echo "$newrev" >"$cachedir/$oldrev"
 150 }
 151
 152 rev_exists()
 153 {
 154         if git rev-parse "$1" >/dev/null 2>&1; then
 155                 return 0
 156         else
 157                 return 1
 158         fi
 159 }
 160
 161 # if a commit doesn't have a parent, this might not work.  But we only want
 162 # to remove the parent from the rev-list, and since it doesn't exist, it won't
 163 # be there anyway, so do nothing in that case.
 164 try_remove_previous()
 165 {
 166         if rev_exists "$1^"; then
 167                 echo "^$1^"
 168         fi
 169 }
 170
 171 find_latest_squash()
 172 {
 173         debug "Looking for latest squash ($dir)..."
 174         dir="$1"
 175         sq=
 176         main=
 177         sub=
 178         git log --grep="^git-subtree-dir: $dir\$" \
 179                 --pretty=format:'START %H%n%s%n%n%b%nEND%n' HEAD |
 180         while read a b junk; do
 181                 debug "$a $b $junk"
 182                 debug "{{$sq/$main/$sub}}"
 183                 case "$a" in
 184                         START) sq="$b" ;;
 185                         git-subtree-mainline:) main="$b" ;;
 186                         git-subtree-split:) sub="$b" ;;
 187                         END)
 188                                 if [ -n "$sub" ]; then
 189                                         if [ -n "$main" ]; then
 190                                                 # a rejoin commit?
 191                                                 # Pretend its sub was a squash.
 192                                                 sq="$sub"
 193                                         fi
 194                                         debug "Squash found: $sq $sub"
 195                                         echo "$sq" "$sub"
 196                                         break
 197                                 fi
 198                                 sq=
 199                                 main=
 200                                 sub=
 201                                 ;;
 202                 esac
 203         done
 204 }
 205
 206 find_existing_splits()
 207 {
 208         debug "Looking for prior splits..."
 209         dir="$1"
 210         revs="$2"
 211         main=
 212         sub=
 213         git log --grep="^git-subtree-dir: $dir\$" \
 214                 --pretty=format:'START %H%n%s%n%n%b%nEND%n' $revs |
 215         while read a b junk; do
 216                 case "$a" in
 217                         START) main="$b"; sq="$b" ;;
 218                         git-subtree-mainline:) main="$b" ;;
 219                         git-subtree-split:) sub="$b" ;;
 220                         END)
 221                                 if [ -z "$main" -a -n "$sub" ]; then
 222                                         # squash commits refer to a subtree
 223                                         cache_set "$sq" "$sub"
 224                                 fi
 225                                 if [ -n "$main" -a -n "$sub" ]; then
 226                                         debug "  Prior: $main -> $sub"
 227                                         cache_set $main $sub
 228                                         try_remove_previous "$main"
 229                                         try_remove_previous "$sub"
 230                                 fi
 231                                 main=
 232                                 sub=
 233                                 ;;
 234                 esac
 235         done
 236 }
 237
 238 copy_commit()
 239 {
 240         # We're going to set some environment vars here, so
 241         # do it in a subshell to get rid of them safely later
 242         debug copy_commit "{$1}" "{$2}" "{$3}"
 243         git log -1 --pretty=format:'%an%n%ae%n%ad%n%cn%n%ce%n%cd%n%s%n%n%b' "$1" |
 244         (
 245                 read GIT_AUTHOR_NAME
 246                 read GIT_AUTHOR_EMAIL
 247                 read GIT_AUTHOR_DATE
 248                 read GIT_COMMITTER_NAME
 249                 read GIT_COMMITTER_EMAIL
 250                 read GIT_COMMITTER_DATE
 251                 export  GIT_AUTHOR_NAME \
 252                         GIT_AUTHOR_EMAIL \
 253                         GIT_AUTHOR_DATE \
 254                         GIT_COMMITTER_NAME \
 255                         GIT_COMMITTER_EMAIL \
 256                         GIT_COMMITTER_DATE
 257                 (echo -n "$annotate"; cat ) |
 258                 git commit-tree "$2" $3  # reads the rest of stdin
 259         ) || die "Can't copy commit $1"
 260 }
 261
 262 add_msg()
 263 {
 264         dir="$1"
 265         latest_old="$2"
 266         latest_new="$3"
 267         cat <<-EOF
 268                 Add '$dir/' from commit '$latest_new'
 269
 270                 git-subtree-dir: $dir
 271                 git-subtree-mainline: $latest_old
 272                 git-subtree-split: $latest_new
 273         EOF
 274 }
 275
 276 rejoin_msg()
 277 {
 278         dir="$1"
 279         latest_old="$2"
 280         latest_new="$3"
 281         cat <<-EOF
 282                 Split '$dir/' into commit '$latest_new'
 283
 284                 git-subtree-dir: $dir
 285                 git-subtree-mainline: $latest_old
 286                 git-subtree-split: $latest_new
 287         EOF
 288 }
 289
 290 squash_msg()
 291 {
 292         dir="$1"
 293         oldsub="$2"
 294         newsub="$3"
 295         newsub_short=$(git rev-parse --short "$newsub")
 296
 297         if [ -n "$oldsub" ]; then
 298                 oldsub_short=$(git rev-parse --short "$oldsub")
 299                 echo "Squashed '$dir/' changes from $oldsub_short..$newsub_short"
 300                 echo
 301                 git log --pretty=tformat:'%h %s' "$oldsub..$newsub"
 302                 git log --pretty=tformat:'REVERT: %h %s' "$newsub..$oldsub"
 303         else
 304                 echo "Squashed '$dir/' content from commit $newsub_short"
 305         fi
 306
 307         echo
 308         echo "git-subtree-dir: $dir"
 309         echo "git-subtree-split: $newsub"
 310 }
 311
 312 toptree_for_commit()
 313 {
 314         commit="$1"
 315         git log -1 --pretty=format:'%T' "$commit" -- || exit $?
 316 }
 317
 318 subtree_for_commit()
 319 {
 320         commit="$1"
 321         dir="$2"
 322         git ls-tree "$commit" -- "$dir" |
 323         while read mode type tree name; do
 324                 assert [ "$name" = "$dir" ]
 325                 echo $tree
 326                 break
 327         done
 328 }
 329
 330 tree_changed()
 331 {
 332         tree=$1
 333         shift
 334         if [ $# -ne 1 ]; then
 335                 return 0   # weird parents, consider it changed
 336         else
 337                 ptree=$(toptree_for_commit $1)
 338                 if [ "$ptree" != "$tree" ]; then
 339                         return 0   # changed
 340                 else
 341                         return 1   # not changed
 342                 fi
 343         fi
 344 }
 345
 346 new_squash_commit()
 347 {
 348         old="$1"
 349         oldsub="$2"
 350         newsub="$3"
 351         tree=$(toptree_for_commit $newsub) || exit $?
 352         if [ -n "$old" ]; then
 353                 squash_msg "$dir" "$oldsub" "$newsub" |
 354                         git commit-tree "$tree" -p "$old" || exit $?
 355         else
 356                 squash_msg "$dir" "" "$newsub" |
 357                         git commit-tree "$tree" || exit $?
 358         fi
 359 }
 360
 361 copy_or_skip()
 362 {
 363         rev="$1"
 364         tree="$2"
 365         newparents="$3"
 366         assert [ -n "$tree" ]
 367
 368         identical=
 369         nonidentical=
 370         p=
 371         gotparents=
 372         for parent in $newparents; do
 373                 ptree=$(toptree_for_commit $parent) || exit $?
 374                 [ -z "$ptree" ] && continue
 375                 if [ "$ptree" = "$tree" ]; then
 376                         # an identical parent could be used in place of this rev.
 377                         identical="$parent"
 378                 else
 379                         nonidentical="$parent"
 380                 fi
 381
 382                 # sometimes both old parents map to the same newparent;
 383                 # eliminate duplicates
 384                 is_new=1
 385                 for gp in $gotparents; do
 386                         if [ "$gp" = "$parent" ]; then
 387                                 is_new=
 388                                 break
 389                         fi
 390                 done
 391                 if [ -n "$is_new" ]; then
 392                         gotparents="$gotparents $parent"
 393                         p="$p -p $parent"
 394                 fi
 395         done
 396
 397         if [ -n "$identical" ]; then
 398                 echo $identical
 399         else
 400                 copy_commit $rev $tree "$p" || exit $?
 401         fi
 402 }
 403
 404 ensure_clean()
 405 {
 406         if ! git diff-index HEAD --exit-code --quiet; then
 407                 die "Working tree has modifications.  Cannot add."
 408         fi
 409         if ! git diff-index --cached HEAD --exit-code --quiet; then
 410                 die "Index has modifications.  Cannot add."
 411         fi
 412 }
 413
 414 cmd_add()
 415 {
 416         if [ -e "$dir" ]; then
 417                 die "'$dir' already exists.  Cannot add."
 418         fi
 419         ensure_clean
 420
 421         set -- $revs
 422         if [ $# -ne 1 ]; then
 423                 die "You must provide exactly one revision.  Got: '$revs'"
 424         fi
 425         rev="$1"
 426
 427         debug "Adding $dir as '$rev'..."
 428         git read-tree --prefix="$dir" $rev || exit $?
 429         git checkout "$dir" || exit $?
 430         tree=$(git write-tree) || exit $?
 431
 432         headrev=$(git rev-parse HEAD) || exit $?
 433         if [ -n "$headrev" -a "$headrev" != "$rev" ]; then
 434                 headp="-p $headrev"
 435         else
 436                 headp=
 437         fi
 438
 439         if [ -n "$squash" ]; then
 440                 rev=$(new_squash_commit "" "" "$rev") || exit $?
 441                 commit=$(echo "Merge commit '$rev' as '$dir'" |
 442                          git commit-tree $tree $headp -p "$rev") || exit $?
 443         else
 444                 commit=$(add_msg "$dir" "$headrev" "$rev" |
 445                          git commit-tree $tree $headp -p "$rev") || exit $?
 446         fi
 447         git reset "$commit" || exit $?
 448
 449         say "Added dir '$dir'"
 450 }
 451
 452 cmd_split()
 453 {
 454         if [ -n "$branch" ] && rev_exists "refs/heads/$branch"; then
 455                 die "Branch '$branch' already exists."
 456         fi
 457
 458         debug "Splitting $dir..."
 459         cache_setup || exit $?
 460
 461         if [ -n "$onto" ]; then
 462                 debug "Reading history for --onto=$onto..."
 463                 git rev-list $onto |
 464                 while read rev; do
 465                         # the 'onto' history is already just the subdir, so
 466                         # any parent we find there can be used verbatim
 467                         debug "  cache: $rev"
 468                         cache_set $rev $rev
 469                 done
 470         fi
 471
 472         if [ -n "$ignore_joins" ]; then
 473                 unrevs=
 474         else
 475                 unrevs="$(find_existing_splits "$dir" "$revs")"
 476         fi
 477
 478         # We can't restrict rev-list to only $dir here, because some of our
 479         # parents have the $dir contents the root, and those won't match.
 480         # (and rev-list --follow doesn't seem to solve this)
 481         grl='git rev-list --reverse --parents $revs $unrevs'
 482         revmax=$(eval "$grl" | wc -l)
 483         revcount=0
 484         createcount=0
 485         eval "$grl" |
 486         while read rev parents; do
 487                 revcount=$(($revcount + 1))
 488                 say -n "$revcount/$revmax ($createcount)\r"
 489                 debug "Processing commit: $rev"
 490                 exists=$(cache_get $rev)
 491                 if [ -n "$exists" ]; then
 492                         debug "  prior: $exists"
 493                         continue
 494                 fi
 495                 createcount=$(($createcount + 1))
 496                 debug "  parents: $parents"
 497                 newparents=$(cache_get $parents)
 498                 debug "  newparents: $newparents"
 499
 500                 tree=$(subtree_for_commit $rev "$dir")
 501                 debug "  tree is: $tree"
 502
 503                 # ugly.  is there no better way to tell if this is a subtree
 504                 # vs. a mainline commit?  Does it matter?
 505                 [ -z $tree ] && continue
 506
 507                 newrev=$(copy_or_skip "$rev" "$tree" "$newparents") || exit $?
 508                 debug "  newrev is: $newrev"
 509                 cache_set $rev $newrev
 510                 cache_set latest_new $newrev
 511                 cache_set latest_old $rev
 512         done || exit $?
 513         latest_new=$(cache_get latest_new)
 514         if [ -z "$latest_new" ]; then
 515                 die "No new revisions were found"
 516         fi
 517
 518         if [ -n "$rejoin" ]; then
 519                 debug "Merging split branch into HEAD..."
 520                 latest_old=$(cache_get latest_old)
 521                 git merge -s ours \
 522                         -m "$(rejoin_msg $dir $latest_old $latest_new)" \
 523                         $latest_new >&2 || exit $?
 524         fi
 525         if [ -n "$branch" ]; then
 526                 git update-ref -m 'subtree split' "refs/heads/$branch" \
 527                         $latest_new "" || exit $?
 528                 say "Created branch '$branch'"
 529         fi
 530         echo $latest_new
 531         exit 0
 532 }
 533
 534 cmd_merge()
 535 {
 536         ensure_clean
 537
 538         set -- $revs
 539         if [ $# -ne 1 ]; then
 540                 die "You must provide exactly one revision.  Got: '$revs'"
 541         fi
 542         rev="$1"
 543
 544         if [ -n "$squash" ]; then
 545                 first_split="$(find_latest_squash "$dir")"
 546                 if [ -z "$first_split" ]; then
 547                         die "Can't squash-merge: '$dir' was never added."
 548                 fi
 549                 set $first_split
 550                 old=$1
 551                 sub=$2
 552                 if [ "$sub" = "$rev" ]; then
 553                         say "Subtree is already at commit $rev."
 554                         exit 0
 555                 fi
 556                 new=$(new_squash_commit "$old" "$sub" "$rev") || exit $?
 557                 debug "New squash commit: $new"
 558                 rev="$new"
 559         fi
 560
 561         git merge -s subtree $rev
 562 }
 563
 564 cmd_pull()
 565 {
 566         ensure_clean
 567         set -x
 568         git pull -s subtree "$@"
 569 }
 570
 571 "cmd_$command" "$@"