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
  14 --
  15 h,help        show the help
  16 q             quiet
  17 prefix=       the name of the subdir to split out
  18 onto=         try connecting new tree to an existing one
  19 rejoin        merge the new branch back into HEAD
  20 ignore-joins  ignore prior --rejoin commits
  21 "
  22 eval $(echo "$OPTS_SPEC" | git rev-parse --parseopt -- "$@" || echo exit $?)
  23 . git-sh-setup
  24 require_work_tree
  25
  26 quiet=
  27 command=
  28 onto=
  29 rejoin=
  30 ignore_joins=
  31
  32 debug()
  33 {
  34         if [ -z "$quiet" ]; then
  35                 echo "$@" >&2
  36         fi
  37 }
  38
  39 assert()
  40 {
  41         if "$@"; then
  42                 :
  43         else
  44                 die "assertion failed: " "$@"
  45         fi
  46 }
  47
  48
  49 #echo "Options: $*"
  50
  51 while [ $# -gt 0 ]; do
  52         opt="$1"
  53         shift
  54         case "$opt" in
  55                 -q) quiet=1 ;;
  56                 --prefix) prefix="$1"; shift ;;
  57                 --no-prefix) prefix= ;;
  58                 --onto) onto="$1"; shift ;;
  59                 --no-onto) onto= ;;
  60                 --rejoin) rejoin=1 ;;
  61                 --no-rejoin) rejoin= ;;
  62                 --ignore-joins) ignore_joins=1 ;;
  63                 --no-ignore-joins) ignore_joins= ;;
  64                 --) break ;;
  65         esac
  66 done
  67
  68 command="$1"
  69 shift
  70 case "$command" in
  71         add|merge) default= ;;
  72         split) default="--default HEAD" ;;
  73         *) die "Unknown command '$command'" ;;
  74 esac
  75
  76 revs=$(git rev-parse $default --revs-only "$@") || exit $?
  77
  78 if [ -z "$prefix" ]; then
  79         die "You must provide the --prefix option."
  80 fi
  81 dir="$prefix"
  82
  83 dirs="$(git rev-parse --no-revs --no-flags "$@")" || exit $?
  84 if [ -n "$dirs" ]; then
  85         die "Error: Use --prefix instead of bare filenames."
  86 fi
  87
  88 debug "command: {$command}"
  89 debug "quiet: {$quiet}"
  90 debug "revs: {$revs}"
  91 debug "dir: {$dir}"
  92 debug
  93
  94 cache_setup()
  95 {
  96         cachedir="$GIT_DIR/subtree-cache/$$"
  97         rm -rf "$cachedir" || die "Can't delete old cachedir: $cachedir"
  98         mkdir -p "$cachedir" || die "Can't create new cachedir: $cachedir"
  99         debug "Using cachedir: $cachedir" >&2
 100 }
 101
 102 cache_get()
 103 {
 104         for oldrev in $*; do
 105                 if [ -r "$cachedir/$oldrev" ]; then
 106                         read newrev <"$cachedir/$oldrev"
 107                         echo $newrev
 108                 fi
 109         done
 110 }
 111
 112 cache_set()
 113 {
 114         oldrev="$1"
 115         newrev="$2"
 116         if [ "$oldrev" != "latest_old" \
 117              -a "$oldrev" != "latest_new" \
 118              -a -e "$cachedir/$oldrev" ]; then
 119                 die "cache for $oldrev already exists!"
 120         fi
 121         echo "$newrev" >"$cachedir/$oldrev"
 122 }
 123
 124 find_existing_splits()
 125 {
 126         debug "Looking for prior splits..."
 127         dir="$1"
 128         revs="$2"
 129         git log --grep="^git-subtree-dir: $dir\$" \
 130                 --pretty=format:'%s%n%n%b%nEND' "$revs" |
 131         while read a b junk; do
 132                 case "$a" in
 133                         git-subtree-mainline:) main="$b" ;;
 134                         git-subtree-split:) sub="$b" ;;
 135                         *)
 136                                 if [ -n "$main" -a -n "$sub" ]; then
 137                                         debug "  Prior: $main -> $sub"
 138                                         cache_set $main $sub
 139                                         echo "^$main^ ^$sub^"
 140                                         main=
 141                                         sub=
 142                                 fi
 143                                 ;;
 144                 esac
 145         done
 146 }
 147
 148 copy_commit()
 149 {
 150         # We're doing to set some environment vars here, so
 151         # do it in a subshell to get rid of them safely later
 152         git log -1 --pretty=format:'%an%n%ae%n%ad%n%cn%n%ce%n%cd%n%s%n%n%b' "$1" |
 153         (
 154                 read GIT_AUTHOR_NAME
 155                 read GIT_AUTHOR_EMAIL
 156                 read GIT_AUTHOR_DATE
 157                 read GIT_COMMITTER_NAME
 158                 read GIT_COMMITTER_EMAIL
 159                 read GIT_COMMITTER_DATE
 160                 export  GIT_AUTHOR_NAME \
 161                         GIT_AUTHOR_EMAIL \
 162                         GIT_AUTHOR_DATE \
 163                         GIT_COMMITTER_NAME \
 164                         GIT_COMMITTER_EMAIL \
 165                         GIT_COMMITTER_DATE
 166                 (echo -n '*'; cat ) |  # FIXME
 167                 git commit-tree "$2" $3  # reads the rest of stdin
 168         ) || die "Can't copy commit $1"
 169 }
 170
 171 add_msg()
 172 {
 173         dir="$1"
 174         latest_old="$2"
 175         latest_new="$3"
 176         cat <<-EOF
 177                 Add '$dir/' from commit '$latest_new'
 178
 179                 git-subtree-dir: $dir
 180                 git-subtree-mainline: $latest_old
 181                 git-subtree-split: $latest_new
 182         EOF
 183 }
 184
 185 merge_msg()
 186 {
 187         dir="$1"
 188         latest_old="$2"
 189         latest_new="$3"
 190         cat <<-EOF
 191                 Split '$dir/' into commit '$latest_new'
 192
 193                 git-subtree-dir: $dir
 194                 git-subtree-mainline: $latest_old
 195                 git-subtree-split: $latest_new
 196         EOF
 197 }
 198
 199 toptree_for_commit()
 200 {
 201         commit="$1"
 202         git log -1 --pretty=format:'%T' "$commit" -- || exit $?
 203 }
 204
 205 subtree_for_commit()
 206 {
 207         commit="$1"
 208         dir="$2"
 209         git ls-tree "$commit" -- "$dir" |
 210         while read mode type tree name; do
 211                 assert [ "$name" = "$dir" ]
 212                 echo $tree
 213                 break
 214         done
 215 }
 216
 217 tree_changed()
 218 {
 219         tree=$1
 220         shift
 221         if [ $# -ne 1 ]; then
 222                 return 0   # weird parents, consider it changed
 223         else
 224                 ptree=$(toptree_for_commit $1)
 225                 if [ "$ptree" != "$tree" ]; then
 226                         return 0   # changed
 227                 else
 228                         return 1   # not changed
 229                 fi
 230         fi
 231 }
 232
 233 copy_or_skip()
 234 {
 235         rev="$1"
 236         tree="$2"
 237         newparents="$3"
 238         assert [ -n "$tree" ]
 239
 240         identical=
 241         p=
 242         for parent in $newparents; do
 243                 ptree=$(toptree_for_commit $parent) || exit $?
 244                 if [ "$ptree" = "$tree" ]; then
 245                         # an identical parent could be used in place of this rev.
 246                         identical="$parent"
 247                 fi
 248                 if [ -n "$ptree" ]; then
 249                         parentmatch="$parentmatch$parent"
 250                         p="$p -p $parent"
 251                 fi
 252         done
 253
 254         if [ -n "$identical" -a "$parentmatch" = "$identical" ]; then
 255                 echo $identical
 256         else
 257                 copy_commit $rev $tree "$p" || exit $?
 258         fi
 259 }
 260
 261 cmd_add()
 262 {
 263         if [ -e "$dir" ]; then
 264                 die "'$dir' already exists.  Cannot add."
 265         fi
 266         if ! git diff-index HEAD --exit-code --quiet; then
 267                 die "Working tree has modifications.  Cannot add."
 268         fi
 269         if ! git diff-index --cached HEAD --exit-code --quiet; then
 270                 die "Index has modifications.  Cannot add."
 271         fi
 272         set -- $revs
 273         if [ $# -ne 1 ]; then
 274                 die "You must provide exactly one revision.  Got: '$revs'"
 275         fi
 276         rev="$1"
 277
 278         debug "Adding $dir as '$rev'..."
 279         git read-tree --prefix="$dir" $rev || exit $?
 280         git checkout "$dir" || exit $?
 281         tree=$(git write-tree) || exit $?
 282
 283         headrev=$(git rev-parse HEAD) || exit $?
 284         if [ -n "$headrev" -a "$headrev" != "$rev" ]; then
 285                 headp="-p $headrev"
 286         else
 287                 headp=
 288         fi
 289         commit=$(add_msg "$dir" "$headrev" "$rev" |
 290                  git commit-tree $tree $headp -p "$rev") || exit $?
 291         git reset "$commit" || exit $?
 292 }
 293
 294 cmd_split()
 295 {
 296         debug "Splitting $dir..."
 297         cache_setup || exit $?
 298
 299         if [ -n "$onto" ]; then
 300                 debug "Reading history for --onto=$onto..."
 301                 git rev-list $onto |
 302                 while read rev; do
 303                         # the 'onto' history is already just the subdir, so
 304                         # any parent we find there can be used verbatim
 305                         debug "  cache: $rev"
 306                         cache_set $rev $rev
 307                 done
 308         fi
 309
 310         if [ -n "$ignore_joins" ]; then
 311                 unrevs=
 312         else
 313                 unrevs="$(find_existing_splits "$dir" "$revs")"
 314         fi
 315
 316         # We can't restrict rev-list to only "$dir" here, because that leaves out
 317         # critical information about commit parents.
 318         debug "git rev-list --reverse --parents $revs $unrevs"
 319         git rev-list --reverse --parents $revs $unrevs |
 320         while read rev parents; do
 321                 debug
 322                 debug "Processing commit: $rev"
 323                 exists=$(cache_get $rev)
 324                 if [ -n "$exists" ]; then
 325                         debug "  prior: $exists"
 326                         continue
 327                 fi
 328                 debug "  parents: $parents"
 329                 newparents=$(cache_get $parents)
 330                 debug "  newparents: $newparents"
 331
 332                 tree=$(subtree_for_commit $rev "$dir")
 333                 debug "  tree is: $tree"
 334                 [ -z $tree ] && continue
 335
 336                 newrev=$(copy_or_skip "$rev" "$tree" "$newparents") || exit $?
 337                 debug "  newrev is: $newrev"
 338                 cache_set $rev $newrev
 339                 cache_set latest_new $newrev
 340                 cache_set latest_old $rev
 341         done || exit $?
 342         latest_new=$(cache_get latest_new)
 343         if [ -z "$latest_new" ]; then
 344                 die "No new revisions were found"
 345         fi
 346
 347         if [ -n "$rejoin" ]; then
 348                 debug "Merging split branch into HEAD..."
 349                 latest_old=$(cache_get latest_old)
 350                 git merge -s ours \
 351                         -m "$(merge_msg $dir $latest_old $latest_new)" \
 352                         $latest_new >&2
 353         fi
 354         echo $latest_new
 355         exit 0
 356 }
 357
 358 cmd_merge()
 359 {
 360         die "merge command not implemented yet"
 361 }
 362
 363 "cmd_$command"