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_existing_splits()
 171 {
 172         debug "Looking for prior splits..."
 173         dir="$1"
 174         revs="$2"
 175         git log --grep="^git-subtree-dir: $dir\$" \
 176                 --pretty=format:'%s%n%n%b%nEND' $revs |
 177         while read a b junk; do
 178                 case "$a" in
 179                         git-subtree-mainline:) main="$b" ;;
 180                         git-subtree-split:) sub="$b" ;;
 181                         END)
 182                                 if [ -n "$main" -a -n "$sub" ]; then
 183                                         debug "  Prior: $main -> $sub"
 184                                         cache_set $main $sub
 185                                         try_remove_previous "$main"
 186                                         try_remove_previous "$sub"
 187                                 fi
 188                                 main=
 189                                 sub=
 190                                 ;;
 191                 esac
 192         done
 193 }
 194
 195 copy_commit()
 196 {
 197         # We're going to set some environment vars here, so
 198         # do it in a subshell to get rid of them safely later
 199         debug copy_commit "{$1}" "{$2}" "{$3}"
 200         git log -1 --pretty=format:'%an%n%ae%n%ad%n%cn%n%ce%n%cd%n%s%n%n%b' "$1" |
 201         (
 202                 read GIT_AUTHOR_NAME
 203                 read GIT_AUTHOR_EMAIL
 204                 read GIT_AUTHOR_DATE
 205                 read GIT_COMMITTER_NAME
 206                 read GIT_COMMITTER_EMAIL
 207                 read GIT_COMMITTER_DATE
 208                 export  GIT_AUTHOR_NAME \
 209                         GIT_AUTHOR_EMAIL \
 210                         GIT_AUTHOR_DATE \
 211                         GIT_COMMITTER_NAME \
 212                         GIT_COMMITTER_EMAIL \
 213                         GIT_COMMITTER_DATE
 214                 (echo -n "$annotate"; cat ) |
 215                 git commit-tree "$2" $3  # reads the rest of stdin
 216         ) || die "Can't copy commit $1"
 217 }
 218
 219 add_msg()
 220 {
 221         dir="$1"
 222         latest_old="$2"
 223         latest_new="$3"
 224         cat <<-EOF
 225                 Add '$dir/' from commit '$latest_new'
 226
 227                 git-subtree-dir: $dir
 228                 git-subtree-mainline: $latest_old
 229                 git-subtree-split: $latest_new
 230         EOF
 231 }
 232
 233 rejoin_msg()
 234 {
 235         dir="$1"
 236         latest_old="$2"
 237         latest_new="$3"
 238         cat <<-EOF
 239                 Split '$dir/' into commit '$latest_new'
 240
 241                 git-subtree-dir: $dir
 242                 git-subtree-mainline: $latest_old
 243                 git-subtree-split: $latest_new
 244         EOF
 245 }
 246
 247 toptree_for_commit()
 248 {
 249         commit="$1"
 250         git log -1 --pretty=format:'%T' "$commit" -- || exit $?
 251 }
 252
 253 subtree_for_commit()
 254 {
 255         commit="$1"
 256         dir="$2"
 257         git ls-tree "$commit" -- "$dir" |
 258         while read mode type tree name; do
 259                 assert [ "$name" = "$dir" ]
 260                 echo $tree
 261                 break
 262         done
 263 }
 264
 265 tree_changed()
 266 {
 267         tree=$1
 268         shift
 269         if [ $# -ne 1 ]; then
 270                 return 0   # weird parents, consider it changed
 271         else
 272                 ptree=$(toptree_for_commit $1)
 273                 if [ "$ptree" != "$tree" ]; then
 274                         return 0   # changed
 275                 else
 276                         return 1   # not changed
 277                 fi
 278         fi
 279 }
 280
 281 copy_or_skip()
 282 {
 283         rev="$1"
 284         tree="$2"
 285         newparents="$3"
 286         assert [ -n "$tree" ]
 287
 288         identical=
 289         nonidentical=
 290         p=
 291         gotparents=
 292         for parent in $newparents; do
 293                 ptree=$(toptree_for_commit $parent) || exit $?
 294                 [ -z "$ptree" ] && continue
 295                 if [ "$ptree" = "$tree" ]; then
 296                         # an identical parent could be used in place of this rev.
 297                         identical="$parent"
 298                 else
 299                         nonidentical="$parent"
 300                 fi
 301
 302                 # sometimes both old parents map to the same newparent;
 303                 # eliminate duplicates
 304                 is_new=1
 305                 for gp in $gotparents; do
 306                         if [ "$gp" = "$parent" ]; then
 307                                 is_new=
 308                                 break
 309                         fi
 310                 done
 311                 if [ -n "$is_new" ]; then
 312                         gotparents="$gotparents $parent"
 313                         p="$p -p $parent"
 314                 fi
 315         done
 316
 317         if [ -n "$identical" ]; then
 318                 echo $identical
 319         else
 320                 copy_commit $rev $tree "$p" || exit $?
 321         fi
 322 }
 323
 324 ensure_clean()
 325 {
 326         if ! git diff-index HEAD --exit-code --quiet; then
 327                 die "Working tree has modifications.  Cannot add."
 328         fi
 329         if ! git diff-index --cached HEAD --exit-code --quiet; then
 330                 die "Index has modifications.  Cannot add."
 331         fi
 332 }
 333
 334 cmd_add()
 335 {
 336         if [ -e "$dir" ]; then
 337                 die "'$dir' already exists.  Cannot add."
 338         fi
 339         ensure_clean
 340
 341         set -- $revs
 342         if [ $# -ne 1 ]; then
 343                 die "You must provide exactly one revision.  Got: '$revs'"
 344         fi
 345         rev="$1"
 346
 347         debug "Adding $dir as '$rev'..."
 348         git read-tree --prefix="$dir" $rev || exit $?
 349         git checkout "$dir" || exit $?
 350         tree=$(git write-tree) || exit $?
 351
 352         headrev=$(git rev-parse HEAD) || exit $?
 353         if [ -n "$headrev" -a "$headrev" != "$rev" ]; then
 354                 headp="-p $headrev"
 355         else
 356                 headp=
 357         fi
 358         commit=$(add_msg "$dir" "$headrev" "$rev" |
 359                  git commit-tree $tree $headp -p "$rev") || exit $?
 360         git reset "$commit" || exit $?
 361 }
 362
 363 cmd_split()
 364 {
 365         if [ -n "$branch" ] && rev_exists "refs/heads/$branch"; then
 366                 die "Branch '$branch' already exists."
 367         fi
 368
 369         debug "Splitting $dir..."
 370         cache_setup || exit $?
 371
 372         if [ -n "$onto" ]; then
 373                 debug "Reading history for --onto=$onto..."
 374                 git rev-list $onto |
 375                 while read rev; do
 376                         # the 'onto' history is already just the subdir, so
 377                         # any parent we find there can be used verbatim
 378                         debug "  cache: $rev"
 379                         cache_set $rev $rev
 380                 done
 381         fi
 382
 383         if [ -n "$ignore_joins" ]; then
 384                 unrevs=
 385         else
 386                 unrevs="$(find_existing_splits "$dir" "$revs")"
 387         fi
 388
 389         # We can't restrict rev-list to only $dir here, because some of our
 390         # parents have the $dir contents the root, and those won't match.
 391         # (and rev-list --follow doesn't seem to solve this)
 392         grl='git rev-list --reverse --parents $revs $unrevs'
 393         revmax=$(eval "$grl" | wc -l)
 394         revcount=0
 395         createcount=0
 396         eval "$grl" |
 397         while read rev parents; do
 398                 revcount=$(($revcount + 1))
 399                 say -n "$revcount/$revmax ($createcount)\r"
 400                 debug "Processing commit: $rev"
 401                 exists=$(cache_get $rev)
 402                 if [ -n "$exists" ]; then
 403                         debug "  prior: $exists"
 404                         continue
 405                 fi
 406                 createcount=$(($createcount + 1))
 407                 debug "  parents: $parents"
 408                 newparents=$(cache_get $parents)
 409                 debug "  newparents: $newparents"
 410
 411                 tree=$(subtree_for_commit $rev "$dir")
 412                 debug "  tree is: $tree"
 413
 414                 # ugly.  is there no better way to tell if this is a subtree
 415                 # vs. a mainline commit?  Does it matter?
 416                 [ -z $tree ] && continue
 417
 418                 newrev=$(copy_or_skip "$rev" "$tree" "$newparents") || exit $?
 419                 debug "  newrev is: $newrev"
 420                 cache_set $rev $newrev
 421                 cache_set latest_new $newrev
 422                 cache_set latest_old $rev
 423         done || exit $?
 424         latest_new=$(cache_get latest_new)
 425         if [ -z "$latest_new" ]; then
 426                 die "No new revisions were found"
 427         fi
 428
 429         if [ -n "$rejoin" ]; then
 430                 debug "Merging split branch into HEAD..."
 431                 latest_old=$(cache_get latest_old)
 432                 git merge -s ours \
 433                         -m "$(rejoin_msg $dir $latest_old $latest_new)" \
 434                         $latest_new >&2 || exit $?
 435         fi
 436         if [ -n "$branch" ]; then
 437                 git update-ref -m 'subtree split' "refs/heads/$branch" \
 438                         $latest_new "" || exit $?
 439                 say "Created branch '$branch'"
 440         fi
 441         echo $latest_new
 442         exit 0
 443 }
 444
 445 cmd_merge()
 446 {
 447         ensure_clean
 448
 449         set -- $revs
 450         if [ $# -ne 1 ]; then
 451                 die "You must provide exactly one revision.  Got: '$revs'"
 452         fi
 453         rev="$1"
 454
 455         git merge -s subtree $rev
 456 }
 457
 458 cmd_pull()
 459 {
 460         ensure_clean
 461         set -x
 462         git pull -s subtree "$@"
 463 }
 464
 465 "cmd_$command" "$@"