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 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 onto=         try connecting new tree to an existing one
  23 rejoin        merge the new branch back into HEAD
  24 ignore-joins  ignore prior --rejoin commits
  25 "
  26 eval $(echo "$OPTS_SPEC" | git rev-parse --parseopt -- "$@" || echo exit $?)
  27 . git-sh-setup
  28 require_work_tree
  29
  30 quiet=
  31 debug=
  32 command=
  33 onto=
  34 rejoin=
  35 ignore_joins=
  36 annotate=
  37
  38 debug()
  39 {
  40         if [ -n "$debug" ]; then
  41                 echo "$@" >&2
  42         fi
  43 }
  44
  45 say()
  46 {
  47         if [ -z "$quiet" ]; then
  48                 echo "$@" >&2
  49         fi
  50 }
  51
  52 assert()
  53 {
  54         if "$@"; then
  55                 :
  56         else
  57                 die "assertion failed: " "$@"
  58         fi
  59 }
  60
  61
  62 #echo "Options: $*"
  63
  64 while [ $# -gt 0 ]; do
  65         opt="$1"
  66         shift
  67         case "$opt" in
  68                 -q) quiet=1 ;;
  69                 -d) debug=1 ;;
  70                 --annotate) annotate="$1"; shift ;;
  71                 --no-annotate) annotate= ;;
  72                 --prefix) prefix="$1"; shift ;;
  73                 --no-prefix) prefix= ;;
  74                 --onto) onto="$1"; shift ;;
  75                 --no-onto) onto= ;;
  76                 --rejoin) rejoin=1 ;;
  77                 --no-rejoin) rejoin= ;;
  78                 --ignore-joins) ignore_joins=1 ;;
  79                 --no-ignore-joins) ignore_joins= ;;
  80                 --) break ;;
  81         esac
  82 done
  83
  84 command="$1"
  85 shift
  86 case "$command" in
  87         add|merge|pull) default= ;;
  88         split) default="--default HEAD" ;;
  89         *) die "Unknown command '$command'" ;;
  90 esac
  91
  92 if [ -z "$prefix" ]; then
  93         die "You must provide the --prefix option."
  94 fi
  95 dir="$prefix"
  96
  97 if [ "$command" != "pull" ]; then
  98         revs=$(git rev-parse $default --revs-only "$@") || exit $?
  99         dirs="$(git rev-parse --no-revs --no-flags "$@")" || exit $?
 100         if [ -n "$dirs" ]; then
 101                 die "Error: Use --prefix instead of bare filenames."
 102         fi
 103 fi
 104
 105 debug "command: {$command}"
 106 debug "quiet: {$quiet}"
 107 debug "revs: {$revs}"
 108 debug "dir: {$dir}"
 109 debug "opts: {$*}"
 110 debug
 111
 112 cache_setup()
 113 {
 114         cachedir="$GIT_DIR/subtree-cache/$$"
 115         rm -rf "$cachedir" || die "Can't delete old cachedir: $cachedir"
 116         mkdir -p "$cachedir" || die "Can't create new cachedir: $cachedir"
 117         debug "Using cachedir: $cachedir" >&2
 118 }
 119
 120 cache_get()
 121 {
 122         for oldrev in $*; do
 123                 if [ -r "$cachedir/$oldrev" ]; then
 124                         read newrev <"$cachedir/$oldrev"
 125                         echo $newrev
 126                 fi
 127         done
 128 }
 129
 130 cache_set()
 131 {
 132         oldrev="$1"
 133         newrev="$2"
 134         if [ "$oldrev" != "latest_old" \
 135              -a "$oldrev" != "latest_new" \
 136              -a -e "$cachedir/$oldrev" ]; then
 137                 die "cache for $oldrev already exists!"
 138         fi
 139         echo "$newrev" >"$cachedir/$oldrev"
 140 }
 141
 142 # if a commit doesn't have a parent, this might not work.  But we only want
 143 # to remove the parent from the rev-list, and since it doesn't exist, it won't
 144 # be there anyway, so do nothing in that case.
 145 try_remove_previous()
 146 {
 147         if git rev-parse "$1^" >/dev/null 2>&1; then
 148                 echo "^$1^"
 149         fi
 150 }
 151
 152 find_existing_splits()
 153 {
 154         debug "Looking for prior splits..."
 155         dir="$1"
 156         revs="$2"
 157         git log --grep="^git-subtree-dir: $dir\$" \
 158                 --pretty=format:'%s%n%n%b%nEND' $revs |
 159         while read a b junk; do
 160                 case "$a" in
 161                         git-subtree-mainline:) main="$b" ;;
 162                         git-subtree-split:) sub="$b" ;;
 163                         *)
 164                                 if [ -n "$main" -a -n "$sub" ]; then
 165                                         debug "  Prior: $main -> $sub"
 166                                         cache_set $main $sub
 167                                         try_remove_previous "$main"
 168                                         try_remove_previous "$sub"
 169                                         main=
 170                                         sub=
 171                                 fi
 172                                 ;;
 173                 esac
 174         done
 175 }
 176
 177 copy_commit()
 178 {
 179         # We're doing to set some environment vars here, so
 180         # do it in a subshell to get rid of them safely later
 181         debug copy_commit "{$1}" "{$2}" "{$3}"
 182         git log -1 --pretty=format:'%an%n%ae%n%ad%n%cn%n%ce%n%cd%n%s%n%n%b' "$1" |
 183         (
 184                 read GIT_AUTHOR_NAME
 185                 read GIT_AUTHOR_EMAIL
 186                 read GIT_AUTHOR_DATE
 187                 read GIT_COMMITTER_NAME
 188                 read GIT_COMMITTER_EMAIL
 189                 read GIT_COMMITTER_DATE
 190                 export  GIT_AUTHOR_NAME \
 191                         GIT_AUTHOR_EMAIL \
 192                         GIT_AUTHOR_DATE \
 193                         GIT_COMMITTER_NAME \
 194                         GIT_COMMITTER_EMAIL \
 195                         GIT_COMMITTER_DATE
 196                 (echo -n "$annotate"; cat ) |
 197                 git commit-tree "$2" $3  # reads the rest of stdin
 198         ) || die "Can't copy commit $1"
 199 }
 200
 201 add_msg()
 202 {
 203         dir="$1"
 204         latest_old="$2"
 205         latest_new="$3"
 206         cat <<-EOF
 207                 Add '$dir/' from commit '$latest_new'
 208
 209                 git-subtree-dir: $dir
 210                 git-subtree-mainline: $latest_old
 211                 git-subtree-split: $latest_new
 212         EOF
 213 }
 214
 215 merge_msg()
 216 {
 217         dir="$1"
 218         latest_old="$2"
 219         latest_new="$3"
 220         cat <<-EOF
 221                 Split '$dir/' into commit '$latest_new'
 222
 223                 git-subtree-dir: $dir
 224                 git-subtree-mainline: $latest_old
 225                 git-subtree-split: $latest_new
 226         EOF
 227 }
 228
 229 toptree_for_commit()
 230 {
 231         commit="$1"
 232         git log -1 --pretty=format:'%T' "$commit" -- || exit $?
 233 }
 234
 235 subtree_for_commit()
 236 {
 237         commit="$1"
 238         dir="$2"
 239         git ls-tree "$commit" -- "$dir" |
 240         while read mode type tree name; do
 241                 assert [ "$name" = "$dir" ]
 242                 echo $tree
 243                 break
 244         done
 245 }
 246
 247 tree_changed()
 248 {
 249         tree=$1
 250         shift
 251         if [ $# -ne 1 ]; then
 252                 return 0   # weird parents, consider it changed
 253         else
 254                 ptree=$(toptree_for_commit $1)
 255                 if [ "$ptree" != "$tree" ]; then
 256                         return 0   # changed
 257                 else
 258                         return 1   # not changed
 259                 fi
 260         fi
 261 }
 262
 263 copy_or_skip()
 264 {
 265         rev="$1"
 266         tree="$2"
 267         newparents="$3"
 268         assert [ -n "$tree" ]
 269
 270         identical=
 271         nonidentical=
 272         p=
 273         gotparents=
 274         for parent in $newparents; do
 275                 ptree=$(toptree_for_commit $parent) || exit $?
 276                 [ -z "$ptree" ] && continue
 277                 if [ "$ptree" = "$tree" ]; then
 278                         # an identical parent could be used in place of this rev.
 279                         identical="$parent"
 280                 else
 281                         nonidentical="$parent"
 282                 fi
 283
 284                 # sometimes both old parents map to the same newparent;
 285                 # eliminate duplicates
 286                 is_new=1
 287                 for gp in $gotparents; do
 288                         if [ "$gp" = "$parent" ]; then
 289                                 is_new=
 290                                 break
 291                         fi
 292                 done
 293                 if [ -n "$is_new" ]; then
 294                         gotparents="$gotparents $parent"
 295                         p="$p -p $parent"
 296                 fi
 297         done
 298
 299         if [ -n "$identical" ]; then
 300                 echo $identical
 301         else
 302                 copy_commit $rev $tree "$p" || exit $?
 303         fi
 304 }
 305
 306 ensure_clean()
 307 {
 308         if ! git diff-index HEAD --exit-code --quiet; then
 309                 die "Working tree has modifications.  Cannot add."
 310         fi
 311         if ! git diff-index --cached HEAD --exit-code --quiet; then
 312                 die "Index has modifications.  Cannot add."
 313         fi
 314 }
 315
 316 cmd_add()
 317 {
 318         if [ -e "$dir" ]; then
 319                 die "'$dir' already exists.  Cannot add."
 320         fi
 321         ensure_clean
 322
 323         set -- $revs
 324         if [ $# -ne 1 ]; then
 325                 die "You must provide exactly one revision.  Got: '$revs'"
 326         fi
 327         rev="$1"
 328
 329         debug "Adding $dir as '$rev'..."
 330         git read-tree --prefix="$dir" $rev || exit $?
 331         git checkout "$dir" || exit $?
 332         tree=$(git write-tree) || exit $?
 333
 334         headrev=$(git rev-parse HEAD) || exit $?
 335         if [ -n "$headrev" -a "$headrev" != "$rev" ]; then
 336                 headp="-p $headrev"
 337         else
 338                 headp=
 339         fi
 340         commit=$(add_msg "$dir" "$headrev" "$rev" |
 341                  git commit-tree $tree $headp -p "$rev") || exit $?
 342         git reset "$commit" || exit $?
 343 }
 344
 345 cmd_split()
 346 {
 347         debug "Splitting $dir..."
 348         cache_setup || exit $?
 349
 350         if [ -n "$onto" ]; then
 351                 debug "Reading history for --onto=$onto..."
 352                 git rev-list $onto |
 353                 while read rev; do
 354                         # the 'onto' history is already just the subdir, so
 355                         # any parent we find there can be used verbatim
 356                         debug "  cache: $rev"
 357                         cache_set $rev $rev
 358                 done
 359         fi
 360
 361         if [ -n "$ignore_joins" ]; then
 362                 unrevs=
 363         else
 364                 unrevs="$(find_existing_splits "$dir" "$revs")"
 365         fi
 366
 367         # We can't restrict rev-list to only $dir here, because some of our
 368         # parents have the $dir contents the root, and those won't match.
 369         # (and rev-list --follow doesn't seem to solve this)
 370         grl='git rev-list --reverse --parents $revs $unrevs'
 371         revmax=$(eval "$grl" | wc -l)
 372         revcount=0
 373         createcount=0
 374         eval "$grl" |
 375         while read rev parents; do
 376                 revcount=$(($revcount + 1))
 377                 say -n "$revcount/$revmax ($createcount)\r"
 378                 debug "Processing commit: $rev"
 379                 exists=$(cache_get $rev)
 380                 if [ -n "$exists" ]; then
 381                         debug "  prior: $exists"
 382                         continue
 383                 fi
 384                 createcount=$(($createcount + 1))
 385                 debug "  parents: $parents"
 386                 newparents=$(cache_get $parents)
 387                 debug "  newparents: $newparents"
 388
 389                 tree=$(subtree_for_commit $rev "$dir")
 390                 debug "  tree is: $tree"
 391                 [ -z $tree ] && continue
 392
 393                 newrev=$(copy_or_skip "$rev" "$tree" "$newparents") || exit $?
 394                 debug "  newrev is: $newrev"
 395                 cache_set $rev $newrev
 396                 cache_set latest_new $newrev
 397                 cache_set latest_old $rev
 398         done || exit $?
 399         latest_new=$(cache_get latest_new)
 400         if [ -z "$latest_new" ]; then
 401                 die "No new revisions were found"
 402         fi
 403
 404         if [ -n "$rejoin" ]; then
 405                 debug "Merging split branch into HEAD..."
 406                 latest_old=$(cache_get latest_old)
 407                 git merge -s ours \
 408                         -m "$(merge_msg $dir $latest_old $latest_new)" \
 409                         $latest_new >&2 || exit $?
 410         fi
 411         echo $latest_new
 412         exit 0
 413 }
 414
 415 cmd_merge()
 416 {
 417         ensure_clean
 418
 419         set -- $revs
 420         if [ $# -ne 1 ]; then
 421                 die "You must provide exactly one revision.  Got: '$revs'"
 422         fi
 423         rev="$1"
 424
 425         git merge -s subtree $rev
 426 }
 427
 428 cmd_pull()
 429 {
 430         ensure_clean
 431         set -x
 432         git pull -s subtree "$@"
 433 }
 434
 435 "cmd_$command" "$@"