svn-merge2git.sh

   1 #!/bin/sh
   2 # Copyright (c) 2008  Benoit Sigoure <tsuna@lrde.epita.fr>
   3 #
   4 # This program is free software: you can redistribute it and/or modify
   5 # it under the terms of the GNU General Public License as published by
   6 # the Free Software Foundation, either version 3 of the License, or
   7 # (at your option) any later version.
   8 #
   9 # This program is distributed in the hope that it will be useful,
  10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12 # GNU General Public License for more details.
  13 #
  14 # You should have received a copy of the GNU General Public License
  15 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16
  17 OPTIONS_SPEC="\
  18 svn-merge2git [options] [refspec]
  19 --
  20 a,all         Do the work on all local the branches
  21 d,debug       Turn on debug mode (useful if you're hacking the script)
  22 H,no-rewrite  Do not rewrite the entire history (leave the grafts)
  23 n,dry-run     Do the entire processing without actually changing anything
  24 p,prune       Clean up everything after rewriting the history (not undo-able!)
  25 P,no-repack   Do not repack the resulting repository (implied by -H)
  26 r,remote      Include remote branches (create a local branch for each of them)
  27 v,verbose     Be more verbose
  28 "
  29 SUBDIRECTORY_OK=Yes
  30 . git-sh-setup
  31 cd_to_toplevel
  32
  33 : ${TMPDIR=/tmp}
  34 export TMPDIR
  35
  36 # BRE (Basic RegExp) compatible with `git rev-list --grep' and `sed'.  The RE
  37 # *must* capture the revision merged in its first group.
  38 merge_pattern='[Mm]erge.*[0-9][0-9]*:\([0-9][0-9]*\)'
  39
  40 # BRE which is used to exclude commits whose line that matches of
  41 # $merge_pattern also match this pattern.
  42 exclude_pattern='Finish'
  43
  44 # BRE which is used to exclude matches in the commit log of potential merge
  45 # commits.
  46 log_exclude_pattern='Finish.*merge'
  47
  48 # extract_svn_branch_name <string>
  49 # --------------------------------
  50 # Find the string the name of a SVN branch.  Put the result in
  51 # $svn_branch_name.  Assumes SVN "stdlayout".
  52 extract_svn_branch_name()
  53 {
  54   case $1 in
  55     '')
  56       fatal 'extract_svn_branch_name called with empty argument';;
  57     */branches/*)
  58       extract_svn_branch_name_ 'branches' "$1";;
  59     */tags/*)
  60       extract_svn_branch_name_ 'tags' "$1"
  61       warn "found a merge from tag '$svn_branch_name'";;
  62     */trunk*)
  63       svn_branch_name='trunk';;
  64   esac
  65 }
  66
  67 # extract_svn_branch_name_ <kind> <string>
  68 # ----------------------------------------
  69 # Helper of extract_svn_branch_name below to factor some code.
  70 # <kind> is probably either 'branches' or 'tags' (for SVN "stdlayout").
  71 # Put the result in $svn_branch_name.
  72 extract_svn_branch_name_()
  73 {
  74   # XXX: Assumes that a branch name does contain a whitespace.  Fragile.
  75   sed_tmp="s|.*/\\($1/[^        ]*\\).*|\\1|"
  76   svn_branch_name=`echo "$2" | sed "$sed_tmp"`
  77 }
  78
  79 me=`basename "$0"`
  80
  81 # fatal <msg>
  82 # -----------
  83 # print <msg> on stderr and exit 1
  84 fatal()
  85 {
  86   die "$me: error: $*"
  87 }
  88
  89 warnings=0
  90 # warn <msg>
  91 # ----------
  92 # print <msg> on stderr
  93 warn()
  94 {
  95   echo "$me: warning: $*" >&2
  96   warnings=$(($warnings + 1))
  97 }
  98
  99 # verb <msg>
 100 # ----------
 101 # Print <msg> when verbose mode is enabled.
 102 verb()
 103 {
 104   $opt_verbose && echo "$*"
 105 }
 106
 107 # debug <msg>
 108 # -----------
 109 # Print <msg> when debug mode is enabled.
 110 debug()
 111 {
 112   $opt_debug && echo "$*"
 113 }
 114
 115 # find_merge_parent <ref> <merge-line>
 116 # ------------------------------------
 117 # Return (in $merge_parent) the sha1 of the commit that has been merged in by
 118 # <ref>.  <merge-line> must be a line extracted from the commit message of
 119 # <ref> and will be used to extract the SVN revision merged.  For instance, if
 120 # <ref> is a SVN merge of merge-line='Merge -r42:51 in branch foo', this
 121 # function puts the sha1 of the first commit the revision of which is <= 51
 122 # which happens to be in branch foo in $merge_parent.
 123 # If the name of the branch being merged couldn't be found, $merge_parent
 124 # contains 'unknown'.
 125 find_merge_parent()
 126 {
 127   # Find the first line that matches $merge_pattern, do the substitution and
 128   # quit.  Ignore all the other lines.
 129   sed_tmp="s/.*$merge_pattern.*/\\1/"
 130   svn_merge_to=`echo "$2" | sed "$sed_tmp"`
 131   case $svn_merge_to in #(
 132     '' | *[^0-9]*)
 133       fatal "invalid SVN revision '$svn_merge_to' found in $1";;
 134   esac
 135   # Now $svn_merge_to is not necessarily a commit that took part of the
 136   # merge.  For instance, you can merge -r42:51 https://.../branches/foo
 137   # even if the last commit in branch foo is at r46.  So it's utterly
 138   # important that we find the last commit on the branch being merged the
 139   # revision of which must be <= $svn_merge_to (which is 51 in this example).
 140   extract_svn_branch_name "$2"
 141   if test -z "$svn_branch_name"; then
 142     merge_parent='unknown'
 143     return 0
 144   fi
 145   # Create a range to intelligently limit the match of rev-list.  This will
 146   # produce a RE that rules out all the impossible revision numbers (that is,
 147   # the revisions >TO).  e.g:
 148   # 7    -> ([0-7])
 149   # 42   -> (4[0-2]|[0-3][0-9]|[1-9])
 150   # 123  -> (12[0-3]|1[0-1][0-9]|0[0-9][0-9]|[1-9][0-9]{0,1})
 151   # 6951 -> (695[0-1]|69[0-4][0-9]|6[0-8][0-9][0-9]|[0-5][0-9][0-9][0-9]|[1-9][0-9]{0,2})
 152   perl_tmp='$_ = "'"$svn_merge_to"'";
 153             my $l = length($_);
 154             my @r;
 155             foreach my $i (0 .. $l - 1) {
 156               /^(\d*)(\d)(\d{$i})$/;
 157               my ($a, $b, $c) = ($1, int($2), $3);
 158               if ($i != 0) {
 159                 # Avoid pitfalls e.g. 10[0-9] or 0[0-9][0-9] for 101
 160                 next if $b == 0 or ($b == 1 and $a eq "");
 161                 --$b;
 162               }
 163               $b = "[0-$b]" if $b;
 164               $c =~ s/./[0-9]/g;
 165               push(@r, "$a$b$c");
 166             }
 167             push(@r, "[1-9]" . ($l - 2 ? "[0-9]{0," . ($l - 2) . "}" : ""))
 168               if $l > 1;
 169             print "(" . join("|", @r) . ")";'
 170   rev_range=`perl -we "$perl_tmp"`
 171   sed_tmp='s/^ *git-svn-id: .*@\([0-9]*\) [-0-9a-f]*$/\1/p'
 172   svn_merge_parent=`git rev-list --all -1 --header -E \
 173        --grep="^ *git-svn-id: .*/$svn_branch_name@$rev_range [-0-9a-f]*\\$" \
 174        | sed -n "$sed_tmp"`
 175   case $svn_merge_parent in #(
 176     '' | *[^0-9]*) fatal "invalid svn_merge_parent: '$svn_merge_parent'";;
 177   esac
 178   rv=$?
 179   test $rv -eq 0 || fatal "perl returned $rv"
 180   if $opt_verbose; then
 181     if test "$svn_merge_to" -eq "$svn_merge_parent"; then
 182       verb_tmp=
 183     else
 184       verb_tmp=" (in fact r$svn_merge_parent)"
 185     fi
 186   fi
 187   verb "  $1 is merging SVN r$svn_merge_to$verb_tmp from branch $svn_branch_name"
 188   # Now find the sha1 of the merge parent.
 189   merge_parent=`git rev-list --all \
 190                  --grep="^ *git-svn-id: .*@$svn_merge_parent [-0-9a-f]*\\$"`
 191   rv=$?
 192   test $rv -eq 0 || fatal "git rev-list returned $rv"
 193 }
 194
 195 # create_graft <ref> <merge-parent>
 196 # ---------------------------------
 197 # Add <merge-parent> as 2nd parent of the commit designated by <ref>.
 198 create_graft()
 199 {
 200   # --parents will print $1 along with its current parents.
 201   grafted_commit=`git rev-list --no-walk --parents "$1"`
 202   rv=$?
 203   test $rv -eq 0 || fatal "git rev-list returned $rv"
 204   graft_merge_parent=$2
 205
 206   case $grafted_commit in #(
 207     *"$graft_merge_parent"*)
 208       debug "  not grafting commit $1: $graft_merge_parent is already a parent ($grafted_commit)"
 209       return 0;;
 210   esac
 211
 212   graft="$grafted_commit $graft_merge_parent"
 213   existing_graft=`grep "^$1" "$graft_file"`
 214   if test $? -eq 0; then
 215     if test x"$existing_graft" != x"$graft"; then
 216       fatal "$1 is already graft ($existing_graft)\
 217  and the graft is different than what I was going to graft ($graft)"
 218     fi
 219     debug "  not grafting commit $1: already properly grafted"
 220     return 0
 221   fi
 222   debug "  grafting commit $1: add parent $graft_merge_parent"
 223   $opt_dryrun && return 0
 224   nconverted=$(($nconverted + 1))
 225   echo >>"$graft_file" "$graft" \
 226   || fatal "Failed to add a graft in $graft_file"
 227 }
 228
 229 # rm_original_refs
 230 # ----------------
 231 # Remove all the refs under refs/original.
 232 rm_original_refs()
 233 {
 234   if test -f "$GIT_DIR/packed-refs"; then
 235     sed -i '/refs\/original\//d' "$GIT_DIR/packed-refs" \
 236       || warn "Failed to edit '$GIT_DIR/packed-refs' (sed returned $?)"
 237   fi
 238   rm -rf "$GIT_DIR/refs/original"
 239 }
 240
 241 # rewrite_history
 242 # ---------------
 243 # Make *all* the grafts part of the actual history.
 244 rewrite_history()
 245 {
 246   $opt_dryrun && return 0
 247   filter_branch='git filter-branch --parent-filter cat -- --all'
 248   if $opt_rewrite; then
 249     # Refresh all the timestamps.  I don't know why, they always change with
 250     # me (only the timestamps!) and git filter-branch will complain because
 251     # git diff-files will return differences (due to the timestamp change).
 252     # FIXME: Investigate why this script seems to touch the entire WC.
 253     git status >/dev/null
 254     if test -f "$graft_file"; then
 255       $filter_branch || fatal "git filter-branch returned $?"
 256       rm "$graft_file" || warn "Failed to rm $graft_file"
 257     else
 258       warn "No history rewriting necessary"
 259     fi
 260     # FIXME: Is it really necessary to repack if we didn't go through the
 261     # previous `if'?
 262     if $opt_repack; then
 263       if $opt_prune; then
 264         prune='--prune'
 265         rm_original_refs
 266       else
 267         prune=
 268       fi
 269       git gc $prune || warn "git gc $prune returned $?"
 270     fi
 271   else
 272     echo "$me: use '$filter_branch' to rewrite the entire history"
 273   fi
 274 }
 275
 276 # doit <REF>
 277 # ----------
 278 # Find all the merge mentionned in the commit messages and make them become
 279 # real Git merges.
 280 doit()
 281 {
 282   refspec=$1
 283   verb " >> Processing merges in the history of $refspec"
 284
 285   git rev-list --no-walk "$refspec" >/dev/null \
 286     || fatal "'$refspec' does not seem to be a valid refspec"
 287
 288   git rev-list --grep="$merge_pattern" "$refspec" >"$tmp_buf"
 289   rv=$?
 290   test $rv -eq 0 || fatal "git rev-list failed and returned $rv"
 291   while read commit; do
 292     merge_log=`git log --no-walk "$commit"`
 293     rv=$?
 294     test $rv -eq 0 || fatal "git log returned $rv"
 295     merge_line=`echo "$merge_log" | sed "/$merge_pattern/!d;//q"`
 296
 297     # Maybe skip the commit if it matches $exclude_pattern or
 298     # $log_exclude_pattern (in which case it's not a merge)
 299     if $has_exclude \
 300     && echo "$merge_line" | grep -- "$exclude_pattern" >/dev/null; then
 301       verb "  skipping $commit whose log merge-line is: $merge_line"
 302       continue
 303     fi
 304     if $has_log_exclude \
 305     && echo "$merge_log" | grep -- "$log_exclude_pattern" >/dev/null; then
 306       verb "  skipping $commit whose log is:"
 307       $opt_verbose && echo "$merge_log" | sed 's/^/  | /'
 308       continue
 309     fi
 310
 311     nmerge=$(($nmerge + 1))
 312     verb "  $commit is a merge commit, log says:
 313   | $merge_line"
 314
 315     find_merge_parent "$commit" "$merge_line"
 316
 317     case $merge_parent in #(
 318       unknown)
 319         warn "could not find the merge parent of $commit"
 320         continue;; #(
 321       '' | *[^0-9a-f]*)
 322         fatal "invalid merge_parent: '$merge_parent'";;
 323     esac
 324
 325     create_graft "$commit" "$merge_parent"
 326     test $? -eq 0 || fatal "failed to create a graft for commit $commit"
 327   done <"$tmp_buf"
 328 }
 329
 330 # ------------------ #
 331 # `main' starts here #
 332 # ------------------ #
 333
 334 test -d "$TMPDIR" || fatal "TMPDIR='$TMPDIR' is not a directory"
 335 test -w "$TMPDIR" || fatal "TMPDIR='$TMPDIR' is not writable"
 336 tmp_buf=`mktemp "$TMPDIR/$me.XXXXXX"`
 337 # Clean up temp file upon exit.
 338 trap "exit_status=$?; rm -f $tmp_buf; exit \$exit_status" 0
 339
 340 # Parse the options passed to the script.
 341 # Initialize the defaults
 342 opt_all=false
 343 opt_debug=false
 344 opt_dryrun=false
 345 opt_prune=false
 346 opt_remote=false
 347 opt_repack=:
 348 opt_rewrite=:
 349 opt_verbose=false
 350
 351 # -------- #
 352 # `getopt' #
 353 # -------- #
 354 while test $# != 0
 355 do
 356   case $1 in #(
 357     -a | --all)
 358       opt_all=:;; #(
 359     -d | --debug)
 360       opt_debug=:;; #(
 361     -H | --no-rewrite)
 362       opt_rewrite=false;; #(
 363     -n | --dry-run)
 364       opt_dryrun=:;; #(
 365     -p | --prune)
 366       opt_prune=:;; #(
 367     -P | --no-repack)
 368       opt_repack=false;; #(
 369     -r | --remote)
 370       opt_remote=:;; #(
 371     -v | --verbose)
 372       opt_verbose=:;; #(
 373     --)
 374       shift; break;; #(
 375     -*)
 376       usage;; #(
 377   esac
 378   shift
 379 done
 380
 381 # We use rev-list --all a lot.  When we finish, git filter-tree saves all the
 382 # original refs under refs/original.  These will be selected by rev-list --all
 383 # which is something we want to avoid.  So we bail out when refs/original
 384 # exists.
 385 if test -n "`git for-each-ref refs/original`"; then
 386   if $opt_prune; then
 387     rm_original_refs
 388   else
 389     fatal "There are some refs under refs/original which could be
 390 the refs saved by a previous run of myself.  This can also occur if you used
 391 git filter-branch (which I personally do).  Please get rid of them if you want
 392 to re-run me or re-run me with the --prune options and I'll do it for you."
 393   fi
 394 fi
 395
 396 if $opt_all; then
 397   git for-each-ref --shell --format='ref=%(refname)' refs/heads >"$tmp_buf"
 398   rv=$?
 399   test $rv -eq 0 || fatal "git for-each-ref failed and returned $rv"
 400   while read line
 401   do
 402     eval "$line"
 403     set "$@" "$ref"
 404   done <"$tmp_buf"
 405 fi
 406
 407 if $opt_remote; then
 408   git for-each-ref --shell --format='ref=%(refname)' refs/remotes >"$tmp_buf"
 409   rv=$?
 410   test $rv -eq 0 || fatal "git for-each-ref failed and returned $rv"
 411   while read line
 412   do
 413     eval "$line"
 414     branch=`basename "$ref"`
 415     case $branch in #(
 416       HEAD) # Skip branches named `HEAD' (which does happen)
 417         continue;; # because they create ambiguities.
 418     esac
 419     # if the local $branch does not already exist, we create one
 420     exists=`git rev-list --no-walk refs/heads/"$branch" 2>/dev/null`
 421     if test -z "$exists"; then # the $branch does not locally exist
 422       verb "creating branch '$branch' from '$ref'"
 423       git branch "$branch" "$ref" \
 424         || fatal "could not create branch '$branch' from '$ref'"
 425     else # there already is a local $branch
 426       sha1=`git rev-list --no-walk "$ref"`
 427       # Maybe the existing local $branch is identical to the remote $ref?
 428       if test "$sha1" = "$exists"; then # OK, local = remote
 429         verb "branch '$branch' is already properly initialized to '$ref'"
 430         set "$@" "refs/heads/$branch"
 431       else # KO, local != remote
 432         warn "there already exists a local branch '$branch'
 433 and it is at $exists whereas the remote branch '$ref'
 434 is at $sha1 so I'm skipping it..."
 435       fi
 436     fi
 437   done <"$tmp_buf"
 438 fi
 439
 440 # No refspec given => work on HEAD
 441 test -z "$*" && set HEAD
 442 graft_file="$GIT_DIR/info/grafts"
 443
 444 if test -z "$exclude_pattern"; then
 445   has_exclude=false
 446 else
 447   has_exclude=:
 448 fi
 449
 450 if test -z "$log_exclude_pattern"; then
 451   has_log_exclude=false
 452 else
 453   has_log_exclude=:
 454 fi
 455
 456 totalmerge=0
 457 totalconverted=0
 458 for refspec
 459 do
 460   nconverted=0
 461   nmerge=0
 462   doit "$refspec"
 463   echo ">>> processed $nconverted/$nmerge merges in $refspec"
 464   totalmerge=$(($totalmerge + $nmerge))
 465   totalconverted=$(($totalconverted + $nconverted))
 466 done
 467
 468 rewrite_history
 469
 470 echo "Done.  Processed $totalconverted/$totalmerge merges"
 471 test $warnings -eq 0 || warn "job completed with $warnings warnings"