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