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