]>
Commit | Line | Data |
---|---|---|
1 | #!/usr/bin/env perl | |
2 | # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net> | |
3 | # License: GPL v2 or later | |
4 | use warnings; | |
5 | use strict; | |
6 | use vars qw/ $AUTHOR $VERSION | |
7 | $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID | |
8 | $GIT_SVN_INDEX $GIT_SVN | |
9 | $GIT_DIR $REV_DIR $GIT_SVN_DIR/; | |
10 | $AUTHOR = 'Eric Wong <normalperson@yhbt.net>'; | |
11 | $VERSION = '1.1.0-pre'; | |
12 | ||
13 | use Cwd qw/abs_path/; | |
14 | $GIT_DIR = abs_path($ENV{GIT_DIR} || '.git'); | |
15 | $ENV{GIT_DIR} = $GIT_DIR; | |
16 | ||
17 | my $LC_ALL = $ENV{LC_ALL}; | |
18 | my $TZ = $ENV{TZ}; | |
19 | # make sure the svn binary gives consistent output between locales and TZs: | |
20 | $ENV{TZ} = 'UTC'; | |
21 | $ENV{LC_ALL} = 'C'; | |
22 | ||
23 | # If SVN:: library support is added, please make the dependencies | |
24 | # optional and preserve the capability to use the command-line client. | |
25 | # use eval { require SVN::... } to make it lazy load | |
26 | # We don't use any modules not in the standard Perl distribution: | |
27 | use Carp qw/croak/; | |
28 | use IO::File qw//; | |
29 | use File::Basename qw/dirname basename/; | |
30 | use File::Path qw/mkpath/; | |
31 | use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/; | |
32 | use File::Spec qw//; | |
33 | use POSIX qw/strftime/; | |
34 | my $sha1 = qr/[a-f\d]{40}/; | |
35 | my $sha1_short = qr/[a-f\d]{4,40}/; | |
36 | my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit, | |
37 | $_find_copies_harder, $_l, $_cp_similarity, | |
38 | $_repack, $_repack_nr, $_repack_flags, | |
39 | $_template, $_shared, $_no_default_regex, $_no_graft_copy, | |
40 | $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit, | |
41 | $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m); | |
42 | my (@_branch_from, %tree_map, %users, %rusers); | |
43 | my ($_svn_co_url_revs, $_svn_pg_peg_revs); | |
44 | my @repo_path_split_cache; | |
45 | ||
46 | my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext, | |
47 | 'branch|b=s' => \@_branch_from, | |
48 | 'branch-all-refs|B' => \$_branch_all_refs, | |
49 | 'authors-file|A=s' => \$_authors, | |
50 | 'repack:i' => \$_repack, | |
51 | 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags); | |
52 | ||
53 | my ($_trunk, $_tags, $_branches); | |
54 | my %multi_opts = ( 'trunk|T=s' => \$_trunk, | |
55 | 'tags|t=s' => \$_tags, | |
56 | 'branches|b=s' => \$_branches ); | |
57 | my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared ); | |
58 | ||
59 | # yes, 'native' sets "\n". Patches to fix this for non-*nix systems welcome: | |
60 | my %EOL = ( CR => "\015", LF => "\012", CRLF => "\015\012", native => "\012" ); | |
61 | ||
62 | my %cmd = ( | |
63 | fetch => [ \&fetch, "Download new revisions from SVN", | |
64 | { 'revision|r=s' => \$_revision, %fc_opts } ], | |
65 | init => [ \&init, "Initialize a repo for tracking" . | |
66 | " (requires URL argument)", | |
67 | \%init_opts ], | |
68 | commit => [ \&commit, "Commit git revisions to SVN", | |
69 | { 'stdin|' => \$_stdin, | |
70 | 'edit|e' => \$_edit, | |
71 | 'rmdir' => \$_rmdir, | |
72 | 'find-copies-harder' => \$_find_copies_harder, | |
73 | 'l=i' => \$_l, | |
74 | 'copy-similarity|C=i'=> \$_cp_similarity, | |
75 | %fc_opts, | |
76 | } ], | |
77 | 'show-ignore' => [ \&show_ignore, "Show svn:ignore listings", { } ], | |
78 | rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)", | |
79 | { 'no-ignore-externals' => \$_no_ignore_ext, | |
80 | 'upgrade' => \$_upgrade } ], | |
81 | 'graft-branches' => [ \&graft_branches, | |
82 | 'Detect merges/branches from already imported history', | |
83 | { 'merge-rx|m' => \@_opt_m, | |
84 | 'no-default-regex' => \$_no_default_regex, | |
85 | 'no-graft-copy' => \$_no_graft_copy } ], | |
86 | 'multi-init' => [ \&multi_init, | |
87 | 'Initialize multiple trees (like git-svnimport)', | |
88 | { %multi_opts, %fc_opts } ], | |
89 | 'multi-fetch' => [ \&multi_fetch, | |
90 | 'Fetch multiple trees (like git-svnimport)', | |
91 | \%fc_opts ], | |
92 | 'log' => [ \&show_log, 'Show commit logs', | |
93 | { 'limit=i' => \$_limit, | |
94 | 'revision|r=s' => \$_revision, | |
95 | 'verbose|v' => \$_verbose, | |
96 | 'incremental' => \$_incremental, | |
97 | 'oneline' => \$_oneline, | |
98 | 'show-commit' => \$_show_commit, | |
99 | 'authors-file|A=s' => \$_authors, | |
100 | } ], | |
101 | ); | |
102 | ||
103 | my $cmd; | |
104 | for (my $i = 0; $i < @ARGV; $i++) { | |
105 | if (defined $cmd{$ARGV[$i]}) { | |
106 | $cmd = $ARGV[$i]; | |
107 | splice @ARGV, $i, 1; | |
108 | last; | |
109 | } | |
110 | }; | |
111 | ||
112 | my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd); | |
113 | ||
114 | read_repo_config(\%opts); | |
115 | my $rv = GetOptions(%opts, 'help|H|h' => \$_help, | |
116 | 'version|V' => \$_version, | |
117 | 'id|i=s' => \$GIT_SVN); | |
118 | exit 1 if (!$rv && $cmd ne 'log'); | |
119 | ||
120 | set_default_vals(); | |
121 | usage(0) if $_help; | |
122 | version() if $_version; | |
123 | usage(1) unless defined $cmd; | |
124 | init_vars(); | |
125 | load_authors() if $_authors; | |
126 | load_all_refs() if $_branch_all_refs; | |
127 | svn_compat_check(); | |
128 | migration_check() unless $cmd =~ /^(?:init|multi-init)$/; | |
129 | $cmd{$cmd}->[0]->(@ARGV); | |
130 | exit 0; | |
131 | ||
132 | ####################### primary functions ###################### | |
133 | sub usage { | |
134 | my $exit = shift || 0; | |
135 | my $fd = $exit ? \*STDERR : \*STDOUT; | |
136 | print $fd <<""; | |
137 | git-svn - bidirectional operations between a single Subversion tree and git | |
138 | Usage: $0 <command> [options] [arguments]\n | |
139 | ||
140 | print $fd "Available commands:\n" unless $cmd; | |
141 | ||
142 | foreach (sort keys %cmd) { | |
143 | next if $cmd && $cmd ne $_; | |
144 | print $fd ' ',pack('A13',$_),$cmd{$_}->[1],"\n"; | |
145 | foreach (keys %{$cmd{$_}->[2]}) { | |
146 | # prints out arguments as they should be passed: | |
147 | my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : ''; | |
148 | print $fd ' ' x 17, join(', ', map { length $_ > 1 ? | |
149 | "--$_" : "-$_" } | |
150 | split /\|/,$_)," $x\n"; | |
151 | } | |
152 | } | |
153 | print $fd <<""; | |
154 | \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an | |
155 | arbitrary identifier if you're tracking multiple SVN branches/repositories in | |
156 | one git repository and want to keep them separate. See git-svn(1) for more | |
157 | information. | |
158 | ||
159 | exit $exit; | |
160 | } | |
161 | ||
162 | sub version { | |
163 | print "git-svn version $VERSION\n"; | |
164 | exit 0; | |
165 | } | |
166 | ||
167 | sub rebuild { | |
168 | $SVN_URL = shift or undef; | |
169 | my $newest_rev = 0; | |
170 | if ($_upgrade) { | |
171 | sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD"); | |
172 | } else { | |
173 | check_upgrade_needed(); | |
174 | } | |
175 | ||
176 | my $pid = open(my $rev_list,'-|'); | |
177 | defined $pid or croak $!; | |
178 | if ($pid == 0) { | |
179 | exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!; | |
180 | } | |
181 | my $latest; | |
182 | while (<$rev_list>) { | |
183 | chomp; | |
184 | my $c = $_; | |
185 | croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o; | |
186 | my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`); | |
187 | next if (!@commit); # skip merges | |
188 | my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]); | |
189 | if (!$rev || !$uuid) { | |
190 | croak "Unable to extract revision or UUID from ", | |
191 | "$c, $commit[$#commit]\n"; | |
192 | } | |
193 | ||
194 | # if we merged or otherwise started elsewhere, this is | |
195 | # how we break out of it | |
196 | next if (defined $SVN_UUID && ($uuid ne $SVN_UUID)); | |
197 | next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL)); | |
198 | ||
199 | print "r$rev = $c\n"; | |
200 | unless (defined $latest) { | |
201 | if (!$SVN_URL && !$url) { | |
202 | croak "SVN repository location required: $url\n"; | |
203 | } | |
204 | $SVN_URL ||= $url; | |
205 | $SVN_UUID ||= $uuid; | |
206 | setup_git_svn(); | |
207 | $latest = $rev; | |
208 | } | |
209 | assert_revision_eq_or_unknown($rev, $c); | |
210 | sys('git-update-ref',"svn/$GIT_SVN/revs/$rev",$c); | |
211 | $newest_rev = $rev if ($rev > $newest_rev); | |
212 | } | |
213 | close $rev_list or croak $?; | |
214 | if (!chdir $SVN_WC) { | |
215 | svn_cmd_checkout($SVN_URL, $latest, $SVN_WC); | |
216 | chdir $SVN_WC or croak $!; | |
217 | } | |
218 | ||
219 | $pid = fork; | |
220 | defined $pid or croak $!; | |
221 | if ($pid == 0) { | |
222 | my @svn_up = qw(svn up); | |
223 | push @svn_up, '--ignore-externals' unless $_no_ignore_ext; | |
224 | sys(@svn_up,"-r$newest_rev"); | |
225 | $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX; | |
226 | index_changes(); | |
227 | exec('git-write-tree') or croak $!; | |
228 | } | |
229 | waitpid $pid, 0; | |
230 | croak $? if $?; | |
231 | ||
232 | if ($_upgrade) { | |
233 | print STDERR <<""; | |
234 | Keeping deprecated refs/head/$GIT_SVN-HEAD for now. Please remove it | |
235 | when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN | |
236 | ||
237 | } | |
238 | } | |
239 | ||
240 | sub init { | |
241 | $SVN_URL = shift or die "SVN repository location required " . | |
242 | "as a command-line argument\n"; | |
243 | $SVN_URL =~ s!/+$!!; # strip trailing slash | |
244 | unless (-d $GIT_DIR) { | |
245 | my @init_db = ('git-init-db'); | |
246 | push @init_db, "--template=$_template" if defined $_template; | |
247 | push @init_db, "--shared" if defined $_shared; | |
248 | sys(@init_db); | |
249 | } | |
250 | setup_git_svn(); | |
251 | } | |
252 | ||
253 | sub fetch { | |
254 | my (@parents) = @_; | |
255 | check_upgrade_needed(); | |
256 | $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url"); | |
257 | my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL); | |
258 | unless ($_revision) { | |
259 | $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD'; | |
260 | } | |
261 | push @log_args, "-r$_revision"; | |
262 | push @log_args, '--stop-on-copy' unless $_no_stop_copy; | |
263 | ||
264 | my $svn_log = svn_log_raw(@log_args); | |
265 | ||
266 | my $base = next_log_entry($svn_log) or croak "No base revision!\n"; | |
267 | my $last_commit = undef; | |
268 | unless (-d $SVN_WC) { | |
269 | svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC); | |
270 | chdir $SVN_WC or croak $!; | |
271 | read_uuid(); | |
272 | $last_commit = git_commit($base, @parents); | |
273 | assert_tree($last_commit); | |
274 | } else { | |
275 | chdir $SVN_WC or croak $!; | |
276 | read_uuid(); | |
277 | eval { $last_commit = file_to_s("$REV_DIR/$base->{revision}") }; | |
278 | # looks like a user manually cp'd and svn switch'ed | |
279 | unless ($last_commit) { | |
280 | sys(qw/svn revert -R ./); | |
281 | assert_svn_wc_clean($base->{revision}); | |
282 | $last_commit = git_commit($base, @parents); | |
283 | assert_tree($last_commit); | |
284 | } | |
285 | } | |
286 | my @svn_up = qw(svn up); | |
287 | push @svn_up, '--ignore-externals' unless $_no_ignore_ext; | |
288 | my $last = $base; | |
289 | while (my $log_msg = next_log_entry($svn_log)) { | |
290 | assert_tree($last_commit); | |
291 | if ($last->{revision} >= $log_msg->{revision}) { | |
292 | croak "Out of order: last >= current: ", | |
293 | "$last->{revision} >= $log_msg->{revision}\n"; | |
294 | } | |
295 | # Revert is needed for cases like: | |
296 | # https://svn.musicpd.org/Jamming/trunk (r166:167), but | |
297 | # I can't seem to reproduce something like that on a test... | |
298 | sys(qw/svn revert -R ./); | |
299 | assert_svn_wc_clean($last->{revision}); | |
300 | sys(@svn_up,"-r$log_msg->{revision}"); | |
301 | $last_commit = git_commit($log_msg, $last_commit, @parents); | |
302 | $last = $log_msg; | |
303 | } | |
304 | unless (-e "$GIT_DIR/refs/heads/master") { | |
305 | sys(qw(git-update-ref refs/heads/master),$last_commit); | |
306 | } | |
307 | close $svn_log->{fh}; | |
308 | return $last; | |
309 | } | |
310 | ||
311 | sub commit { | |
312 | my (@commits) = @_; | |
313 | check_upgrade_needed(); | |
314 | if ($_stdin || !@commits) { | |
315 | print "Reading from stdin...\n"; | |
316 | @commits = (); | |
317 | while (<STDIN>) { | |
318 | if (/\b($sha1_short)\b/o) { | |
319 | unshift @commits, $1; | |
320 | } | |
321 | } | |
322 | } | |
323 | my @revs; | |
324 | foreach my $c (@commits) { | |
325 | chomp(my @tmp = safe_qx('git-rev-parse',$c)); | |
326 | if (scalar @tmp == 1) { | |
327 | push @revs, $tmp[0]; | |
328 | } elsif (scalar @tmp > 1) { | |
329 | push @revs, reverse (safe_qx('git-rev-list',@tmp)); | |
330 | } else { | |
331 | die "Failed to rev-parse $c\n"; | |
332 | } | |
333 | } | |
334 | chomp @revs; | |
335 | ||
336 | chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n"; | |
337 | my $info = svn_info('.'); | |
338 | my $fetched = fetch(); | |
339 | if ($info->{Revision} != $fetched->{revision}) { | |
340 | print STDERR "There are new revisions that were fetched ", | |
341 | "and need to be merged (or acknowledged) ", | |
342 | "before committing.\n"; | |
343 | exit 1; | |
344 | } | |
345 | $info = svn_info('.'); | |
346 | read_uuid($info); | |
347 | my $svn_current_rev = $info->{'Last Changed Rev'}; | |
348 | foreach my $c (@revs) { | |
349 | my $mods = svn_checkout_tree($svn_current_rev, $c); | |
350 | if (scalar @$mods == 0) { | |
351 | print "Skipping, no changes detected\n"; | |
352 | next; | |
353 | } | |
354 | $svn_current_rev = svn_commit_tree($svn_current_rev, $c); | |
355 | } | |
356 | print "Done committing ",scalar @revs," revisions to SVN\n"; | |
357 | } | |
358 | ||
359 | sub show_ignore { | |
360 | require File::Find or die $!; | |
361 | my $exclude_file = "$GIT_DIR/info/exclude"; | |
362 | open my $fh, '<', $exclude_file or croak $!; | |
363 | chomp(my @excludes = (<$fh>)); | |
364 | close $fh or croak $!; | |
365 | ||
366 | $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url"); | |
367 | chdir $SVN_WC or croak $!; | |
368 | my %ign; | |
369 | File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){ | |
370 | s#^\./##; | |
371 | @{$ign{$_}} = svn_propget_base('svn:ignore', $_); | |
372 | }}, no_chdir=>1},'.'); | |
373 | ||
374 | print "\n# /\n"; | |
375 | foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ } | |
376 | delete $ign{'.'}; | |
377 | foreach my $i (sort keys %ign) { | |
378 | print "\n# ",$i,"\n"; | |
379 | foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ } | |
380 | } | |
381 | } | |
382 | ||
383 | sub graft_branches { | |
384 | my $gr_file = "$GIT_DIR/info/grafts"; | |
385 | my ($grafts, $comments) = read_grafts($gr_file); | |
386 | my $gr_sha1; | |
387 | ||
388 | if (%$grafts) { | |
389 | # temporarily disable our grafts file to make this idempotent | |
390 | chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file)); | |
391 | rename $gr_file, "$gr_file~$gr_sha1" or croak $!; | |
392 | } | |
393 | ||
394 | my $l_map = read_url_paths(); | |
395 | my @re = map { qr/$_/is } @_opt_m if @_opt_m; | |
396 | unless ($_no_default_regex) { | |
397 | push @re, ( qr/\b(?:merge|merging|merged)\s+(\S.+)/is, | |
398 | qr/\b(?:from|of)\s+(\S.+)/is ); | |
399 | } | |
400 | foreach my $u (keys %$l_map) { | |
401 | if (@re) { | |
402 | foreach my $p (keys %{$l_map->{$u}}) { | |
403 | graft_merge_msg($grafts,$l_map,$u,$p); | |
404 | } | |
405 | } | |
406 | graft_file_copy($grafts,$l_map,$u) unless $_no_graft_copy; | |
407 | } | |
408 | ||
409 | write_grafts($grafts, $comments, $gr_file); | |
410 | unlink "$gr_file~$gr_sha1" if $gr_sha1; | |
411 | } | |
412 | ||
413 | sub multi_init { | |
414 | my $url = shift; | |
415 | $_trunk ||= 'trunk'; | |
416 | $_trunk =~ s#/+$##; | |
417 | $url =~ s#/+$## if $url; | |
418 | if ($_trunk !~ m#^[a-z\+]+://#) { | |
419 | $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#); | |
420 | unless ($url) { | |
421 | print STDERR "E: '$_trunk' is not a complete URL ", | |
422 | "and a separate URL is not specified\n"; | |
423 | exit 1; | |
424 | } | |
425 | $_trunk = $url . $_trunk; | |
426 | } | |
427 | if ($GIT_SVN eq 'git-svn') { | |
428 | print "GIT_SVN_ID set to 'trunk' for $_trunk\n"; | |
429 | $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk'; | |
430 | } | |
431 | init_vars(); | |
432 | init($_trunk); | |
433 | complete_url_ls_init($url, $_branches, '--branches/-b', ''); | |
434 | complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/'); | |
435 | } | |
436 | ||
437 | sub multi_fetch { | |
438 | # try to do trunk first, since branches/tags | |
439 | # may be descended from it. | |
440 | if (-d "$GIT_DIR/svn/trunk") { | |
441 | print "Fetching trunk\n"; | |
442 | defined(my $pid = fork) or croak $!; | |
443 | if (!$pid) { | |
444 | $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk'; | |
445 | init_vars(); | |
446 | fetch(@_); | |
447 | exit 0; | |
448 | } | |
449 | waitpid $pid, 0; | |
450 | croak $? if $?; | |
451 | } | |
452 | rec_fetch('', "$GIT_DIR/svn", @_); | |
453 | } | |
454 | ||
455 | sub show_log { | |
456 | my (@args) = @_; | |
457 | my ($r_min, $r_max); | |
458 | my $r_last = -1; # prevent dupes | |
459 | rload_authors() if $_authors; | |
460 | if (defined $TZ) { | |
461 | $ENV{TZ} = $TZ; | |
462 | } else { | |
463 | delete $ENV{TZ}; | |
464 | } | |
465 | if (defined $_revision) { | |
466 | if ($_revision =~ /^(\d+):(\d+)$/) { | |
467 | ($r_min, $r_max) = ($1, $2); | |
468 | } elsif ($_revision =~ /^\d+$/) { | |
469 | $r_min = $r_max = $_revision; | |
470 | } else { | |
471 | print STDERR "-r$_revision is not supported, use ", | |
472 | "standard \'git log\' arguments instead\n"; | |
473 | exit 1; | |
474 | } | |
475 | } | |
476 | ||
477 | my $pid = open(my $log,'-|'); | |
478 | defined $pid or croak $!; | |
479 | if (!$pid) { | |
480 | my @rl = (qw/git-log --abbrev-commit --pretty=raw | |
481 | --default/, "remotes/$GIT_SVN"); | |
482 | push @rl, '--raw' if $_verbose; | |
483 | exec(@rl, @args) or croak $!; | |
484 | } | |
485 | setup_pager(); | |
486 | my (@k, $c, $d); | |
487 | while (<$log>) { | |
488 | if (/^commit ($sha1_short)/o) { | |
489 | my $cmt = $1; | |
490 | if ($c && defined $c->{r} && $c->{r} != $r_last) { | |
491 | $r_last = $c->{r}; | |
492 | process_commit($c, $r_min, $r_max, \@k) or | |
493 | goto out; | |
494 | } | |
495 | $d = undef; | |
496 | $c = { c => $cmt }; | |
497 | } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) { | |
498 | get_author_info($c, $1, $2, $3); | |
499 | } elsif (/^(?:tree|parent|committer) /) { | |
500 | # ignore | |
501 | } elsif (/^:\d{6} \d{6} $sha1_short/o) { | |
502 | push @{$c->{raw}}, $_; | |
503 | } elsif (/^diff /) { | |
504 | $d = 1; | |
505 | push @{$c->{diff}}, $_; | |
506 | } elsif ($d) { | |
507 | push @{$c->{diff}}, $_; | |
508 | } elsif (/^ (git-svn-id:.+)$/) { | |
509 | my ($url, $rev, $uuid) = extract_metadata($1); | |
510 | $c->{r} = $rev; | |
511 | } elsif (s/^ //) { | |
512 | push @{$c->{l}}, $_; | |
513 | } | |
514 | } | |
515 | if ($c && defined $c->{r} && $c->{r} != $r_last) { | |
516 | $r_last = $c->{r}; | |
517 | process_commit($c, $r_min, $r_max, \@k); | |
518 | } | |
519 | if (@k) { | |
520 | my $swap = $r_max; | |
521 | $r_max = $r_min; | |
522 | $r_min = $swap; | |
523 | process_commit($_, $r_min, $r_max) foreach reverse @k; | |
524 | } | |
525 | out: | |
526 | close $log; | |
527 | print '-' x72,"\n" unless $_incremental || $_oneline; | |
528 | } | |
529 | ||
530 | ########################### utility functions ######################### | |
531 | ||
532 | sub rec_fetch { | |
533 | my ($pfx, $p, @args) = @_; | |
534 | my @dir; | |
535 | foreach (sort <$p/*>) { | |
536 | if (-r "$_/info/url") { | |
537 | $pfx .= '/' if $pfx && $pfx !~ m!/$!; | |
538 | my $id = $pfx . basename $_; | |
539 | next if $id eq 'trunk'; | |
540 | print "Fetching $id\n"; | |
541 | defined(my $pid = fork) or croak $!; | |
542 | if (!$pid) { | |
543 | $GIT_SVN = $ENV{GIT_SVN_ID} = $id; | |
544 | init_vars(); | |
545 | fetch(@args); | |
546 | exit 0; | |
547 | } | |
548 | waitpid $pid, 0; | |
549 | croak $? if $?; | |
550 | } elsif (-d $_) { | |
551 | push @dir, $_; | |
552 | } | |
553 | } | |
554 | foreach (@dir) { | |
555 | my $x = $_; | |
556 | $x =~ s!^\Q$GIT_DIR\E/svn/!!; | |
557 | rec_fetch($x, $_); | |
558 | } | |
559 | } | |
560 | ||
561 | sub complete_url_ls_init { | |
562 | my ($url, $var, $switch, $pfx) = @_; | |
563 | unless ($var) { | |
564 | print STDERR "W: $switch not specified\n"; | |
565 | return; | |
566 | } | |
567 | $var =~ s#/+$##; | |
568 | if ($var !~ m#^[a-z\+]+://#) { | |
569 | $var = '/' . $var if ($var !~ m#^/#); | |
570 | unless ($url) { | |
571 | print STDERR "E: '$var' is not a complete URL ", | |
572 | "and a separate URL is not specified\n"; | |
573 | exit 1; | |
574 | } | |
575 | $var = $url . $var; | |
576 | } | |
577 | chomp(my @ls = safe_qx(qw/svn ls --non-interactive/, $var)); | |
578 | my $old = $GIT_SVN; | |
579 | defined(my $pid = fork) or croak $!; | |
580 | if (!$pid) { | |
581 | foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) { | |
582 | $u =~ s#/+$##; | |
583 | if ($u !~ m!\Q$var\E/(.+)$!) { | |
584 | print STDERR "W: Unrecognized URL: $u\n"; | |
585 | die "This should never happen\n"; | |
586 | } | |
587 | my $id = $pfx.$1; | |
588 | print "init $u => $id\n"; | |
589 | $GIT_SVN = $ENV{GIT_SVN_ID} = $id; | |
590 | init_vars(); | |
591 | init($u); | |
592 | } | |
593 | exit 0; | |
594 | } | |
595 | waitpid $pid, 0; | |
596 | croak $? if $?; | |
597 | } | |
598 | ||
599 | sub common_prefix { | |
600 | my $paths = shift; | |
601 | my %common; | |
602 | foreach (@$paths) { | |
603 | my @tmp = split m#/#, $_; | |
604 | my $p = ''; | |
605 | while (my $x = shift @tmp) { | |
606 | $p .= "/$x"; | |
607 | $common{$p} ||= 0; | |
608 | $common{$p}++; | |
609 | } | |
610 | } | |
611 | foreach (sort {length $b <=> length $a} keys %common) { | |
612 | if ($common{$_} == @$paths) { | |
613 | return $_; | |
614 | } | |
615 | } | |
616 | return ''; | |
617 | } | |
618 | ||
619 | # this isn't funky-filename safe, but good enough for now... | |
620 | sub graft_file_copy { | |
621 | my ($grafts, $l_map, $u) = @_; | |
622 | my $paths = $l_map->{$u}; | |
623 | my $pfx = common_prefix([keys %$paths]); | |
624 | ||
625 | my $pid = open my $fh, '-|'; | |
626 | defined $pid or croak $!; | |
627 | unless ($pid) { | |
628 | exec(qw/svn log -v/, $u.$pfx) or croak $!; | |
629 | } | |
630 | my ($r, $mp) = (undef, undef); | |
631 | while (<$fh>) { | |
632 | chomp; | |
633 | if (/^\-{72}$/) { | |
634 | $mp = $r = undef; | |
635 | } elsif (/^r(\d+) \| /) { | |
636 | $r = $1 unless defined $r; | |
637 | } elsif (/^Changed paths:/) { | |
638 | $mp = 1; | |
639 | } elsif ($mp && m#^ [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) { | |
640 | my $dbg = "r$r | $_"; | |
641 | my ($p1, $p0, $r0) = ($1, $2, $3); | |
642 | my $c; | |
643 | foreach my $x (keys %$paths) { | |
644 | next unless ($p1 =~ /^\Q$x\E/); | |
645 | my $i = $paths->{$x}; | |
646 | my $f = "$GIT_DIR/svn/$i/revs/$r"; | |
647 | unless (-r $f) { | |
648 | print STDERR "r$r of $i not imported,", | |
649 | " $dbg\n"; | |
650 | next; | |
651 | } | |
652 | $c = file_to_s($f); | |
653 | } | |
654 | next unless $c; | |
655 | foreach my $x (keys %$paths) { | |
656 | next unless ($p0 =~ /^\Q$x\E/); | |
657 | my $i = $paths->{$x}; | |
658 | my $f = "$GIT_DIR/svn/$i/revs/$r0"; | |
659 | while ($r0 && !-r $f) { | |
660 | # could be an older revision, too... | |
661 | $r0--; | |
662 | $f = "$GIT_DIR/svn/$i/revs/$r0"; | |
663 | } | |
664 | unless (-r $f) { | |
665 | print STDERR "r$r0 of $i not imported,", | |
666 | " $dbg\n"; | |
667 | next; | |
668 | } | |
669 | my $r1 = file_to_s($f); | |
670 | $grafts->{$c}->{$r1} = 1; | |
671 | } | |
672 | } | |
673 | } | |
674 | } | |
675 | ||
676 | sub process_merge_msg_matches { | |
677 | my ($grafts, $l_map, $u, $p, $c, @matches) = @_; | |
678 | my (@strong, @weak); | |
679 | foreach (@matches) { | |
680 | # merging with ourselves is not interesting | |
681 | next if $_ eq $p; | |
682 | if ($l_map->{$u}->{$_}) { | |
683 | push @strong, $_; | |
684 | } else { | |
685 | push @weak, $_; | |
686 | } | |
687 | } | |
688 | foreach my $w (@weak) { | |
689 | last if @strong; | |
690 | # no exact match, use branch name as regexp. | |
691 | my $re = qr/\Q$w\E/i; | |
692 | foreach (keys %{$l_map->{$u}}) { | |
693 | if (/$re/) { | |
694 | push @strong, $_; | |
695 | last; | |
696 | } | |
697 | } | |
698 | last if @strong; | |
699 | $w = basename($w); | |
700 | $re = qr/\Q$w\E/i; | |
701 | foreach (keys %{$l_map->{$u}}) { | |
702 | if (/$re/) { | |
703 | push @strong, $_; | |
704 | last; | |
705 | } | |
706 | } | |
707 | } | |
708 | my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+) | |
709 | \s(?:[a-f\d\-]+)$/xsm); | |
710 | unless (defined $rev) { | |
711 | ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+) | |
712 | \@(?:[a-f\d\-]+)/xsm); | |
713 | return unless defined $rev; | |
714 | } | |
715 | foreach my $m (@strong) { | |
716 | my ($r0, $s0) = find_rev_before($rev, $m); | |
717 | $grafts->{$c->{c}}->{$s0} = 1 if defined $s0; | |
718 | } | |
719 | } | |
720 | ||
721 | sub graft_merge_msg { | |
722 | my ($grafts, $l_map, $u, $p, @re) = @_; | |
723 | ||
724 | my $x = $l_map->{$u}->{$p}; | |
725 | my $rl = rev_list_raw($x); | |
726 | while (my $c = next_rev_list_entry($rl)) { | |
727 | foreach my $re (@re) { | |
728 | my (@br) = ($c->{m} =~ /$re/g); | |
729 | next unless @br; | |
730 | process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br); | |
731 | } | |
732 | } | |
733 | } | |
734 | ||
735 | sub read_uuid { | |
736 | return if $SVN_UUID; | |
737 | my $info = shift || svn_info('.'); | |
738 | $SVN_UUID = $info->{'Repository UUID'} or | |
739 | croak "Repository UUID unreadable\n"; | |
740 | s_to_file($SVN_UUID,"$GIT_SVN_DIR/info/uuid"); | |
741 | } | |
742 | ||
743 | sub quiet_run { | |
744 | my $pid = fork; | |
745 | defined $pid or croak $!; | |
746 | if (!$pid) { | |
747 | open my $null, '>', '/dev/null' or croak $!; | |
748 | open STDERR, '>&', $null or croak $!; | |
749 | open STDOUT, '>&', $null or croak $!; | |
750 | exec @_ or croak $!; | |
751 | } | |
752 | waitpid $pid, 0; | |
753 | return $?; | |
754 | } | |
755 | ||
756 | sub repo_path_split { | |
757 | my $full_url = shift; | |
758 | $full_url =~ s#/+$##; | |
759 | ||
760 | foreach (@repo_path_split_cache) { | |
761 | if ($full_url =~ s#$_##) { | |
762 | my $u = $1; | |
763 | $full_url =~ s#^/+##; | |
764 | return ($u, $full_url); | |
765 | } | |
766 | } | |
767 | ||
768 | my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i); | |
769 | $path =~ s#^/+##; | |
770 | my @paths = split(m#/+#, $path); | |
771 | ||
772 | while (quiet_run(qw/svn ls --non-interactive/, $url)) { | |
773 | my $n = shift @paths || last; | |
774 | $url .= "/$n"; | |
775 | } | |
776 | push @repo_path_split_cache, qr/^(\Q$url\E)/; | |
777 | $path = join('/',@paths); | |
778 | return ($url, $path); | |
779 | } | |
780 | ||
781 | sub setup_git_svn { | |
782 | defined $SVN_URL or croak "SVN repository location required\n"; | |
783 | unless (-d $GIT_DIR) { | |
784 | croak "GIT_DIR=$GIT_DIR does not exist!\n"; | |
785 | } | |
786 | mkpath([$GIT_SVN_DIR]); | |
787 | mkpath(["$GIT_SVN_DIR/info"]); | |
788 | mkpath([$REV_DIR]); | |
789 | s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url"); | |
790 | ||
791 | open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!; | |
792 | print $fd '.svn',"\n"; | |
793 | close $fd or croak $!; | |
794 | my ($url, $path) = repo_path_split($SVN_URL); | |
795 | s_to_file($url, "$GIT_SVN_DIR/info/repo_url"); | |
796 | s_to_file($path, "$GIT_SVN_DIR/info/repo_path"); | |
797 | } | |
798 | ||
799 | sub assert_svn_wc_clean { | |
800 | my ($svn_rev) = @_; | |
801 | croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/); | |
802 | my $lcr = svn_info('.')->{'Last Changed Rev'}; | |
803 | if ($svn_rev != $lcr) { | |
804 | print STDERR "Checking for copy-tree ... "; | |
805 | my @diff = grep(/^Index: /,(safe_qx(qw(svn diff), | |
806 | "-r$lcr:$svn_rev"))); | |
807 | if (@diff) { | |
808 | croak "Nope! Expected r$svn_rev, got r$lcr\n"; | |
809 | } else { | |
810 | print STDERR "OK!\n"; | |
811 | } | |
812 | } | |
813 | my @status = grep(!/^Performing status on external/,(`svn status`)); | |
814 | @status = grep(!/^\s*$/,@status); | |
815 | if (scalar @status) { | |
816 | print STDERR "Tree ($SVN_WC) is not clean:\n"; | |
817 | print STDERR $_ foreach @status; | |
818 | croak; | |
819 | } | |
820 | } | |
821 | ||
822 | sub assert_tree { | |
823 | my ($treeish) = @_; | |
824 | croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o; | |
825 | chomp(my $type = `git-cat-file -t $treeish`); | |
826 | my $expected; | |
827 | while ($type eq 'tag') { | |
828 | chomp(($treeish, $type) = `git-cat-file tag $treeish`); | |
829 | } | |
830 | if ($type eq 'commit') { | |
831 | $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0]; | |
832 | ($expected) = ($expected =~ /^tree ($sha1)$/); | |
833 | die "Unable to get tree from $treeish\n" unless $expected; | |
834 | } elsif ($type eq 'tree') { | |
835 | $expected = $treeish; | |
836 | } else { | |
837 | die "$treeish is a $type, expected tree, tag or commit\n"; | |
838 | } | |
839 | ||
840 | my $old_index = $ENV{GIT_INDEX_FILE}; | |
841 | my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp'; | |
842 | if (-e $tmpindex) { | |
843 | unlink $tmpindex or croak $!; | |
844 | } | |
845 | $ENV{GIT_INDEX_FILE} = $tmpindex; | |
846 | index_changes(1); | |
847 | chomp(my $tree = `git-write-tree`); | |
848 | if ($old_index) { | |
849 | $ENV{GIT_INDEX_FILE} = $old_index; | |
850 | } else { | |
851 | delete $ENV{GIT_INDEX_FILE}; | |
852 | } | |
853 | if ($tree ne $expected) { | |
854 | croak "Tree mismatch, Got: $tree, Expected: $expected\n"; | |
855 | } | |
856 | unlink $tmpindex; | |
857 | } | |
858 | ||
859 | sub parse_diff_tree { | |
860 | my $diff_fh = shift; | |
861 | local $/ = "\0"; | |
862 | my $state = 'meta'; | |
863 | my @mods; | |
864 | while (<$diff_fh>) { | |
865 | chomp $_; # this gets rid of the trailing "\0" | |
866 | if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s | |
867 | $sha1\s($sha1)\s([MTCRAD])\d*$/xo) { | |
868 | push @mods, { mode_a => $1, mode_b => $2, | |
869 | sha1_b => $3, chg => $4 }; | |
870 | if ($4 =~ /^(?:C|R)$/) { | |
871 | $state = 'file_a'; | |
872 | } else { | |
873 | $state = 'file_b'; | |
874 | } | |
875 | } elsif ($state eq 'file_a') { | |
876 | my $x = $mods[$#mods] or croak "Empty array\n"; | |
877 | if ($x->{chg} !~ /^(?:C|R)$/) { | |
878 | croak "Error parsing $_, $x->{chg}\n"; | |
879 | } | |
880 | $x->{file_a} = $_; | |
881 | $state = 'file_b'; | |
882 | } elsif ($state eq 'file_b') { | |
883 | my $x = $mods[$#mods] or croak "Empty array\n"; | |
884 | if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) { | |
885 | croak "Error parsing $_, $x->{chg}\n"; | |
886 | } | |
887 | if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) { | |
888 | croak "Error parsing $_, $x->{chg}\n"; | |
889 | } | |
890 | $x->{file_b} = $_; | |
891 | $state = 'meta'; | |
892 | } else { | |
893 | croak "Error parsing $_\n"; | |
894 | } | |
895 | } | |
896 | close $diff_fh or croak $!; | |
897 | ||
898 | return \@mods; | |
899 | } | |
900 | ||
901 | sub svn_check_prop_executable { | |
902 | my $m = shift; | |
903 | return if -l $m->{file_b}; | |
904 | if ($m->{mode_b} =~ /755$/) { | |
905 | chmod((0755 &~ umask),$m->{file_b}) or croak $!; | |
906 | if ($m->{mode_a} !~ /755$/) { | |
907 | sys(qw(svn propset svn:executable 1), $m->{file_b}); | |
908 | } | |
909 | -x $m->{file_b} or croak "$m->{file_b} is not executable!\n"; | |
910 | } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) { | |
911 | sys(qw(svn propdel svn:executable), $m->{file_b}); | |
912 | chmod((0644 &~ umask),$m->{file_b}) or croak $!; | |
913 | -x $m->{file_b} and croak "$m->{file_b} is executable!\n"; | |
914 | } | |
915 | } | |
916 | ||
917 | sub svn_ensure_parent_path { | |
918 | my $dir_b = dirname(shift); | |
919 | svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir); | |
920 | mkpath([$dir_b]) unless (-d $dir_b); | |
921 | sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn"); | |
922 | } | |
923 | ||
924 | sub precommit_check { | |
925 | my $mods = shift; | |
926 | my (%rm_file, %rmdir_check, %added_check); | |
927 | ||
928 | my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 ); | |
929 | foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) { | |
930 | if ($m->{chg} eq 'R') { | |
931 | if (-d $m->{file_b}) { | |
932 | err_dir_to_file("$m->{file_a} => $m->{file_b}"); | |
933 | } | |
934 | # dir/$file => dir/file/$file | |
935 | my $dirname = dirname($m->{file_b}); | |
936 | while ($dirname ne File::Spec->curdir) { | |
937 | if ($dirname ne $m->{file_a}) { | |
938 | $dirname = dirname($dirname); | |
939 | next; | |
940 | } | |
941 | err_file_to_dir("$m->{file_a} => $m->{file_b}"); | |
942 | } | |
943 | # baz/zzz => baz (baz is a file) | |
944 | $dirname = dirname($m->{file_a}); | |
945 | while ($dirname ne File::Spec->curdir) { | |
946 | if ($dirname ne $m->{file_b}) { | |
947 | $dirname = dirname($dirname); | |
948 | next; | |
949 | } | |
950 | err_dir_to_file("$m->{file_a} => $m->{file_b}"); | |
951 | } | |
952 | } | |
953 | if ($m->{chg} =~ /^(D|R)$/) { | |
954 | my $t = $1 eq 'D' ? 'file_b' : 'file_a'; | |
955 | $rm_file{ $m->{$t} } = 1; | |
956 | my $dirname = dirname( $m->{$t} ); | |
957 | my $basename = basename( $m->{$t} ); | |
958 | $rmdir_check{$dirname}->{$basename} = 1; | |
959 | } elsif ($m->{chg} =~ /^(?:A|C)$/) { | |
960 | if (-d $m->{file_b}) { | |
961 | err_dir_to_file($m->{file_b}); | |
962 | } | |
963 | my $dirname = dirname( $m->{file_b} ); | |
964 | my $basename = basename( $m->{file_b} ); | |
965 | $added_check{$dirname}->{$basename} = 1; | |
966 | while ($dirname ne File::Spec->curdir) { | |
967 | if ($rm_file{$dirname}) { | |
968 | err_file_to_dir($m->{file_b}); | |
969 | } | |
970 | $dirname = dirname $dirname; | |
971 | } | |
972 | } | |
973 | } | |
974 | return (\%rmdir_check, \%added_check); | |
975 | ||
976 | sub err_dir_to_file { | |
977 | my $file = shift; | |
978 | print STDERR "Node change from directory to file ", | |
979 | "is not supported by Subversion: ",$file,"\n"; | |
980 | exit 1; | |
981 | } | |
982 | sub err_file_to_dir { | |
983 | my $file = shift; | |
984 | print STDERR "Node change from file to directory ", | |
985 | "is not supported by Subversion: ",$file,"\n"; | |
986 | exit 1; | |
987 | } | |
988 | } | |
989 | ||
990 | sub svn_checkout_tree { | |
991 | my ($svn_rev, $treeish) = @_; | |
992 | my $from = file_to_s("$REV_DIR/$svn_rev"); | |
993 | assert_tree($from); | |
994 | print "diff-tree $from $treeish\n"; | |
995 | my $pid = open my $diff_fh, '-|'; | |
996 | defined $pid or croak $!; | |
997 | if ($pid == 0) { | |
998 | my @diff_tree = qw(git-diff-tree -z -r); | |
999 | if ($_cp_similarity) { | |
1000 | push @diff_tree, "-C$_cp_similarity"; | |
1001 | } else { | |
1002 | push @diff_tree, '-C'; | |
1003 | } | |
1004 | push @diff_tree, '--find-copies-harder' if $_find_copies_harder; | |
1005 | push @diff_tree, "-l$_l" if defined $_l; | |
1006 | exec(@diff_tree, $from, $treeish) or croak $!; | |
1007 | } | |
1008 | my $mods = parse_diff_tree($diff_fh); | |
1009 | unless (@$mods) { | |
1010 | # git can do empty commits, but SVN doesn't allow it... | |
1011 | return $mods; | |
1012 | } | |
1013 | my ($rm, $add) = precommit_check($mods); | |
1014 | ||
1015 | my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 ); | |
1016 | foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) { | |
1017 | if ($m->{chg} eq 'C') { | |
1018 | svn_ensure_parent_path( $m->{file_b} ); | |
1019 | sys(qw(svn cp), $m->{file_a}, $m->{file_b}); | |
1020 | apply_mod_line_blob($m); | |
1021 | svn_check_prop_executable($m); | |
1022 | } elsif ($m->{chg} eq 'D') { | |
1023 | sys(qw(svn rm --force), $m->{file_b}); | |
1024 | } elsif ($m->{chg} eq 'R') { | |
1025 | svn_ensure_parent_path( $m->{file_b} ); | |
1026 | sys(qw(svn mv --force), $m->{file_a}, $m->{file_b}); | |
1027 | apply_mod_line_blob($m); | |
1028 | svn_check_prop_executable($m); | |
1029 | } elsif ($m->{chg} eq 'M') { | |
1030 | apply_mod_line_blob($m); | |
1031 | svn_check_prop_executable($m); | |
1032 | } elsif ($m->{chg} eq 'T') { | |
1033 | sys(qw(svn rm --force),$m->{file_b}); | |
1034 | apply_mod_line_blob($m); | |
1035 | sys(qw(svn add --force), $m->{file_b}); | |
1036 | svn_check_prop_executable($m); | |
1037 | } elsif ($m->{chg} eq 'A') { | |
1038 | svn_ensure_parent_path( $m->{file_b} ); | |
1039 | apply_mod_line_blob($m); | |
1040 | sys(qw(svn add --force), $m->{file_b}); | |
1041 | svn_check_prop_executable($m); | |
1042 | } else { | |
1043 | croak "Invalid chg: $m->{chg}\n"; | |
1044 | } | |
1045 | } | |
1046 | ||
1047 | assert_tree($treeish); | |
1048 | if ($_rmdir) { # remove empty directories | |
1049 | handle_rmdir($rm, $add); | |
1050 | } | |
1051 | assert_tree($treeish); | |
1052 | return $mods; | |
1053 | } | |
1054 | ||
1055 | # svn ls doesn't work with respect to the current working tree, but what's | |
1056 | # in the repository. There's not even an option for it... *sigh* | |
1057 | # (added files don't show up and removed files remain in the ls listing) | |
1058 | sub svn_ls_current { | |
1059 | my ($dir, $rm, $add) = @_; | |
1060 | chomp(my @ls = safe_qx('svn','ls',$dir)); | |
1061 | my @ret = (); | |
1062 | foreach (@ls) { | |
1063 | s#/$##; # trailing slashes are evil | |
1064 | push @ret, $_ unless $rm->{$dir}->{$_}; | |
1065 | } | |
1066 | if (exists $add->{$dir}) { | |
1067 | push @ret, keys %{$add->{$dir}}; | |
1068 | } | |
1069 | return \@ret; | |
1070 | } | |
1071 | ||
1072 | sub handle_rmdir { | |
1073 | my ($rm, $add) = @_; | |
1074 | ||
1075 | foreach my $dir (sort {length $b <=> length $a} keys %$rm) { | |
1076 | my $ls = svn_ls_current($dir, $rm, $add); | |
1077 | next if (scalar @$ls); | |
1078 | sys(qw(svn rm --force),$dir); | |
1079 | ||
1080 | my $dn = dirname $dir; | |
1081 | $rm->{ $dn }->{ basename $dir } = 1; | |
1082 | $ls = svn_ls_current($dn, $rm, $add); | |
1083 | while (scalar @$ls == 0 && $dn ne File::Spec->curdir) { | |
1084 | sys(qw(svn rm --force),$dn); | |
1085 | $dir = basename $dn; | |
1086 | $dn = dirname $dn; | |
1087 | $rm->{ $dn }->{ $dir } = 1; | |
1088 | $ls = svn_ls_current($dn, $rm, $add); | |
1089 | } | |
1090 | } | |
1091 | } | |
1092 | ||
1093 | sub svn_commit_tree { | |
1094 | my ($svn_rev, $commit) = @_; | |
1095 | my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$"; | |
1096 | my %log_msg = ( msg => '' ); | |
1097 | open my $msg, '>', $commit_msg or croak $!; | |
1098 | ||
1099 | chomp(my $type = `git-cat-file -t $commit`); | |
1100 | if ($type eq 'commit') { | |
1101 | my $pid = open my $msg_fh, '-|'; | |
1102 | defined $pid or croak $!; | |
1103 | ||
1104 | if ($pid == 0) { | |
1105 | exec(qw(git-cat-file commit), $commit) or croak $!; | |
1106 | } | |
1107 | my $in_msg = 0; | |
1108 | while (<$msg_fh>) { | |
1109 | if (!$in_msg) { | |
1110 | $in_msg = 1 if (/^\s*$/); | |
1111 | } elsif (/^git-svn-id: /) { | |
1112 | # skip this, we regenerate the correct one | |
1113 | # on re-fetch anyways | |
1114 | } else { | |
1115 | print $msg $_ or croak $!; | |
1116 | } | |
1117 | } | |
1118 | close $msg_fh or croak $!; | |
1119 | } | |
1120 | close $msg or croak $!; | |
1121 | ||
1122 | if ($_edit || ($type eq 'tree')) { | |
1123 | my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi'; | |
1124 | system($editor, $commit_msg); | |
1125 | } | |
1126 | ||
1127 | # file_to_s removes all trailing newlines, so just use chomp() here: | |
1128 | open $msg, '<', $commit_msg or croak $!; | |
1129 | { local $/; chomp($log_msg{msg} = <$msg>); } | |
1130 | close $msg or croak $!; | |
1131 | ||
1132 | my ($oneline) = ($log_msg{msg} =~ /([^\n\r]+)/); | |
1133 | print "Committing $commit: $oneline\n"; | |
1134 | ||
1135 | if (defined $LC_ALL) { | |
1136 | $ENV{LC_ALL} = $LC_ALL; | |
1137 | } else { | |
1138 | delete $ENV{LC_ALL}; | |
1139 | } | |
1140 | my @ci_output = safe_qx(qw(svn commit -F),$commit_msg); | |
1141 | $ENV{LC_ALL} = 'C'; | |
1142 | unlink $commit_msg; | |
1143 | my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/); | |
1144 | if (!defined $committed) { | |
1145 | my $out = join("\n",@ci_output); | |
1146 | print STDERR "W: Trouble parsing \`svn commit' output:\n\n", | |
1147 | $out, "\n\nAssuming English locale..."; | |
1148 | ($committed) = ($out =~ /^Committed revision \d+\./sm); | |
1149 | defined $committed or die " FAILED!\n", | |
1150 | "Commit output failed to parse committed revision!\n", | |
1151 | print STDERR " OK\n"; | |
1152 | } | |
1153 | ||
1154 | my @svn_up = qw(svn up); | |
1155 | push @svn_up, '--ignore-externals' unless $_no_ignore_ext; | |
1156 | if ($committed == ($svn_rev + 1)) { | |
1157 | push @svn_up, "-r$committed"; | |
1158 | sys(@svn_up); | |
1159 | my $info = svn_info('.'); | |
1160 | my $date = $info->{'Last Changed Date'} or die "Missing date\n"; | |
1161 | if ($info->{'Last Changed Rev'} != $committed) { | |
1162 | croak "$info->{'Last Changed Rev'} != $committed\n" | |
1163 | } | |
1164 | my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~ | |
1165 | /(\d{4})\-(\d\d)\-(\d\d)\s | |
1166 | (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x) | |
1167 | or croak "Failed to parse date: $date\n"; | |
1168 | $log_msg{date} = "$tz $Y-$m-$d $H:$M:$S"; | |
1169 | $log_msg{author} = $info->{'Last Changed Author'}; | |
1170 | $log_msg{revision} = $committed; | |
1171 | $log_msg{msg} .= "\n"; | |
1172 | my $parent = file_to_s("$REV_DIR/$svn_rev"); | |
1173 | git_commit(\%log_msg, $parent, $commit); | |
1174 | return $committed; | |
1175 | } | |
1176 | # resync immediately | |
1177 | push @svn_up, "-r$svn_rev"; | |
1178 | sys(@svn_up); | |
1179 | return fetch("$committed=$commit")->{revision}; | |
1180 | } | |
1181 | ||
1182 | sub rev_list_raw { | |
1183 | my (@args) = @_; | |
1184 | my $pid = open my $fh, '-|'; | |
1185 | defined $pid or croak $!; | |
1186 | if (!$pid) { | |
1187 | exec(qw/git-rev-list --pretty=raw/, @args) or croak $!; | |
1188 | } | |
1189 | return { fh => $fh, t => { } }; | |
1190 | } | |
1191 | ||
1192 | sub next_rev_list_entry { | |
1193 | my $rl = shift; | |
1194 | my $fh = $rl->{fh}; | |
1195 | my $x = $rl->{t}; | |
1196 | while (<$fh>) { | |
1197 | if (/^commit ($sha1)$/o) { | |
1198 | if ($x->{c}) { | |
1199 | $rl->{t} = { c => $1 }; | |
1200 | return $x; | |
1201 | } else { | |
1202 | $x->{c} = $1; | |
1203 | } | |
1204 | } elsif (/^parent ($sha1)$/o) { | |
1205 | $x->{p}->{$1} = 1; | |
1206 | } elsif (s/^ //) { | |
1207 | $x->{m} ||= ''; | |
1208 | $x->{m} .= $_; | |
1209 | } | |
1210 | } | |
1211 | return ($x != $rl->{t}) ? $x : undef; | |
1212 | } | |
1213 | ||
1214 | # read the entire log into a temporary file (which is removed ASAP) | |
1215 | # and store the file handle + parser state | |
1216 | sub svn_log_raw { | |
1217 | my (@log_args) = @_; | |
1218 | my $log_fh = IO::File->new_tmpfile or croak $!; | |
1219 | my $pid = fork; | |
1220 | defined $pid or croak $!; | |
1221 | if (!$pid) { | |
1222 | open STDOUT, '>&', $log_fh or croak $!; | |
1223 | exec (qw(svn log), @log_args) or croak $! | |
1224 | } | |
1225 | waitpid $pid, 0; | |
1226 | croak $? if $?; | |
1227 | seek $log_fh, 0, 0 or croak $!; | |
1228 | return { state => 'sep', fh => $log_fh }; | |
1229 | } | |
1230 | ||
1231 | sub next_log_entry { | |
1232 | my $log = shift; # retval of svn_log_raw() | |
1233 | my $ret = undef; | |
1234 | my $fh = $log->{fh}; | |
1235 | ||
1236 | while (<$fh>) { | |
1237 | chomp; | |
1238 | if (/^\-{72}$/) { | |
1239 | if ($log->{state} eq 'msg') { | |
1240 | if ($ret->{lines}) { | |
1241 | $ret->{msg} .= $_."\n"; | |
1242 | unless(--$ret->{lines}) { | |
1243 | $log->{state} = 'sep'; | |
1244 | } | |
1245 | } else { | |
1246 | croak "Log parse error at: $_\n", | |
1247 | $ret->{revision}, | |
1248 | "\n"; | |
1249 | } | |
1250 | next; | |
1251 | } | |
1252 | if ($log->{state} ne 'sep') { | |
1253 | croak "Log parse error at: $_\n", | |
1254 | "state: $log->{state}\n", | |
1255 | $ret->{revision}, | |
1256 | "\n"; | |
1257 | } | |
1258 | $log->{state} = 'rev'; | |
1259 | ||
1260 | # if we have an empty log message, put something there: | |
1261 | if ($ret) { | |
1262 | $ret->{msg} ||= "\n"; | |
1263 | delete $ret->{lines}; | |
1264 | return $ret; | |
1265 | } | |
1266 | next; | |
1267 | } | |
1268 | if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) { | |
1269 | my $rev = $1; | |
1270 | my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3); | |
1271 | ($lines) = ($lines =~ /(\d+)/); | |
1272 | my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~ | |
1273 | /(\d{4})\-(\d\d)\-(\d\d)\s | |
1274 | (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x) | |
1275 | or croak "Failed to parse date: $date\n"; | |
1276 | $ret = { revision => $rev, | |
1277 | date => "$tz $Y-$m-$d $H:$M:$S", | |
1278 | author => $author, | |
1279 | lines => $lines, | |
1280 | msg => '' }; | |
1281 | if (defined $_authors && ! defined $users{$author}) { | |
1282 | die "Author: $author not defined in ", | |
1283 | "$_authors file\n"; | |
1284 | } | |
1285 | $log->{state} = 'msg_start'; | |
1286 | next; | |
1287 | } | |
1288 | # skip the first blank line of the message: | |
1289 | if ($log->{state} eq 'msg_start' && /^$/) { | |
1290 | $log->{state} = 'msg'; | |
1291 | } elsif ($log->{state} eq 'msg') { | |
1292 | if ($ret->{lines}) { | |
1293 | $ret->{msg} .= $_."\n"; | |
1294 | unless (--$ret->{lines}) { | |
1295 | $log->{state} = 'sep'; | |
1296 | } | |
1297 | } else { | |
1298 | croak "Log parse error at: $_\n", | |
1299 | $ret->{revision},"\n"; | |
1300 | } | |
1301 | } | |
1302 | } | |
1303 | return $ret; | |
1304 | } | |
1305 | ||
1306 | sub svn_info { | |
1307 | my $url = shift || $SVN_URL; | |
1308 | ||
1309 | my $pid = open my $info_fh, '-|'; | |
1310 | defined $pid or croak $!; | |
1311 | ||
1312 | if ($pid == 0) { | |
1313 | exec(qw(svn info),$url) or croak $!; | |
1314 | } | |
1315 | ||
1316 | my $ret = {}; | |
1317 | # only single-lines seem to exist in svn info output | |
1318 | while (<$info_fh>) { | |
1319 | chomp $_; | |
1320 | if (m#^([^:]+)\s*:\s*(\S.*)$#) { | |
1321 | $ret->{$1} = $2; | |
1322 | push @{$ret->{-order}}, $1; | |
1323 | } | |
1324 | } | |
1325 | close $info_fh or croak $!; | |
1326 | return $ret; | |
1327 | } | |
1328 | ||
1329 | sub sys { system(@_) == 0 or croak $? } | |
1330 | ||
1331 | sub eol_cp { | |
1332 | my ($from, $to) = @_; | |
1333 | my $es = svn_propget_base('svn:eol-style', $to); | |
1334 | open my $rfd, '<', $from or croak $!; | |
1335 | binmode $rfd or croak $!; | |
1336 | open my $wfd, '>', $to or croak $!; | |
1337 | binmode $wfd or croak $!; | |
1338 | ||
1339 | my $eol = $EOL{$es} or undef; | |
1340 | my $buf; | |
1341 | use bytes; | |
1342 | while (1) { | |
1343 | my ($r, $w, $t); | |
1344 | defined($r = sysread($rfd, $buf, 4096)) or croak $!; | |
1345 | return unless $r; | |
1346 | if ($eol) { | |
1347 | if ($buf =~ /\015$/) { | |
1348 | my $c; | |
1349 | defined($r = sysread($rfd,$c,1)) or croak $!; | |
1350 | $buf .= $c if $r > 0; | |
1351 | } | |
1352 | $buf =~ s/(?:\015\012|\015|\012)/$eol/gs; | |
1353 | $r = length($buf); | |
1354 | } | |
1355 | for ($w = 0; $w < $r; $w += $t) { | |
1356 | $t = syswrite($wfd, $buf, $r - $w, $w) or croak $!; | |
1357 | } | |
1358 | } | |
1359 | no bytes; | |
1360 | } | |
1361 | ||
1362 | sub do_update_index { | |
1363 | my ($z_cmd, $cmd, $no_text_base) = @_; | |
1364 | ||
1365 | my $z = open my $p, '-|'; | |
1366 | defined $z or croak $!; | |
1367 | unless ($z) { exec @$z_cmd or croak $! } | |
1368 | ||
1369 | my $pid = open my $ui, '|-'; | |
1370 | defined $pid or croak $!; | |
1371 | unless ($pid) { | |
1372 | exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!; | |
1373 | } | |
1374 | local $/ = "\0"; | |
1375 | while (my $x = <$p>) { | |
1376 | chomp $x; | |
1377 | if (!$no_text_base && lstat $x && ! -l _ && | |
1378 | svn_propget_base('svn:keywords', $x)) { | |
1379 | my $mode = -x _ ? 0755 : 0644; | |
1380 | my ($v,$d,$f) = File::Spec->splitpath($x); | |
1381 | my $tb = File::Spec->catfile($d, '.svn', 'tmp', | |
1382 | 'text-base',"$f.svn-base"); | |
1383 | $tb =~ s#^/##; | |
1384 | unless (-f $tb) { | |
1385 | $tb = File::Spec->catfile($d, '.svn', | |
1386 | 'text-base',"$f.svn-base"); | |
1387 | $tb =~ s#^/##; | |
1388 | } | |
1389 | unlink $x or croak $!; | |
1390 | eol_cp($tb, $x); | |
1391 | chmod(($mode &~ umask), $x) or croak $!; | |
1392 | } | |
1393 | print $ui $x,"\0"; | |
1394 | } | |
1395 | close $ui or croak $!; | |
1396 | } | |
1397 | ||
1398 | sub index_changes { | |
1399 | my $no_text_base = shift; | |
1400 | do_update_index([qw/git-diff-files --name-only -z/], | |
1401 | 'remove', | |
1402 | $no_text_base); | |
1403 | do_update_index([qw/git-ls-files -z --others/, | |
1404 | "--exclude-from=$GIT_SVN_DIR/info/exclude"], | |
1405 | 'add', | |
1406 | $no_text_base); | |
1407 | } | |
1408 | ||
1409 | sub s_to_file { | |
1410 | my ($str, $file, $mode) = @_; | |
1411 | open my $fd,'>',$file or croak $!; | |
1412 | print $fd $str,"\n" or croak $!; | |
1413 | close $fd or croak $!; | |
1414 | chmod ($mode &~ umask, $file) if (defined $mode); | |
1415 | } | |
1416 | ||
1417 | sub file_to_s { | |
1418 | my $file = shift; | |
1419 | open my $fd,'<',$file or croak "$!: file: $file\n"; | |
1420 | local $/; | |
1421 | my $ret = <$fd>; | |
1422 | close $fd or croak $!; | |
1423 | $ret =~ s/\s*$//s; | |
1424 | return $ret; | |
1425 | } | |
1426 | ||
1427 | sub assert_revision_unknown { | |
1428 | my $revno = shift; | |
1429 | if (-f "$REV_DIR/$revno") { | |
1430 | croak "$REV_DIR/$revno already exists! ", | |
1431 | "Why are we refetching it?"; | |
1432 | } | |
1433 | } | |
1434 | ||
1435 | sub trees_eq { | |
1436 | my ($x, $y) = @_; | |
1437 | my @x = safe_qx('git-cat-file','commit',$x); | |
1438 | my @y = safe_qx('git-cat-file','commit',$y); | |
1439 | if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/ | |
1440 | || $y[0] !~ /^tree $sha1\n$/) { | |
1441 | print STDERR "Trees not equal: $y[0] != $x[0]\n"; | |
1442 | return 0 | |
1443 | } | |
1444 | return 1; | |
1445 | } | |
1446 | ||
1447 | sub assert_revision_eq_or_unknown { | |
1448 | my ($revno, $commit) = @_; | |
1449 | if (-f "$REV_DIR/$revno") { | |
1450 | my $current = file_to_s("$REV_DIR/$revno"); | |
1451 | if (($commit ne $current) && !trees_eq($commit, $current)) { | |
1452 | croak "$REV_DIR/$revno already exists!\n", | |
1453 | "current: $current\nexpected: $commit\n"; | |
1454 | } | |
1455 | return; | |
1456 | } | |
1457 | } | |
1458 | ||
1459 | sub git_commit { | |
1460 | my ($log_msg, @parents) = @_; | |
1461 | assert_revision_unknown($log_msg->{revision}); | |
1462 | my $out_fh = IO::File->new_tmpfile or croak $!; | |
1463 | ||
1464 | map_tree_joins() if (@_branch_from && !%tree_map); | |
1465 | ||
1466 | # commit parents can be conditionally bound to a particular | |
1467 | # svn revision via: "svn_revno=commit_sha1", filter them out here: | |
1468 | my @exec_parents; | |
1469 | foreach my $p (@parents) { | |
1470 | next unless defined $p; | |
1471 | if ($p =~ /^(\d+)=($sha1_short)$/o) { | |
1472 | if ($1 == $log_msg->{revision}) { | |
1473 | push @exec_parents, $2; | |
1474 | } | |
1475 | } else { | |
1476 | push @exec_parents, $p if $p =~ /$sha1_short/o; | |
1477 | } | |
1478 | } | |
1479 | ||
1480 | my $pid = fork; | |
1481 | defined $pid or croak $!; | |
1482 | if ($pid == 0) { | |
1483 | $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX; | |
1484 | index_changes(); | |
1485 | chomp(my $tree = `git-write-tree`); | |
1486 | croak $? if $?; | |
1487 | if (exists $tree_map{$tree}) { | |
1488 | my %seen_parent = map { $_ => 1 } @exec_parents; | |
1489 | foreach (@{$tree_map{$tree}}) { | |
1490 | # MAXPARENT is defined to 16 in commit-tree.c: | |
1491 | if ($seen_parent{$_} || @exec_parents > 16) { | |
1492 | next; | |
1493 | } | |
1494 | push @exec_parents, $_; | |
1495 | $seen_parent{$_} = 1; | |
1496 | } | |
1497 | } | |
1498 | my $msg_fh = IO::File->new_tmpfile or croak $!; | |
1499 | print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ", | |
1500 | "$SVN_URL\@$log_msg->{revision}", | |
1501 | " $SVN_UUID\n" or croak $!; | |
1502 | $msg_fh->flush == 0 or croak $!; | |
1503 | seek $msg_fh, 0, 0 or croak $!; | |
1504 | ||
1505 | set_commit_env($log_msg); | |
1506 | ||
1507 | my @exec = ('git-commit-tree',$tree); | |
1508 | push @exec, '-p', $_ foreach @exec_parents; | |
1509 | open STDIN, '<&', $msg_fh or croak $!; | |
1510 | open STDOUT, '>&', $out_fh or croak $!; | |
1511 | exec @exec or croak $!; | |
1512 | } | |
1513 | waitpid($pid,0); | |
1514 | croak $? if $?; | |
1515 | ||
1516 | $out_fh->flush == 0 or croak $!; | |
1517 | seek $out_fh, 0, 0 or croak $!; | |
1518 | chomp(my $commit = do { local $/; <$out_fh> }); | |
1519 | if ($commit !~ /^$sha1$/o) { | |
1520 | croak "Failed to commit, invalid sha1: $commit\n"; | |
1521 | } | |
1522 | my @update_ref = ('git-update-ref',"refs/remotes/$GIT_SVN",$commit); | |
1523 | if (my $primary_parent = shift @exec_parents) { | |
1524 | $pid = fork; | |
1525 | defined $pid or croak $!; | |
1526 | if (!$pid) { | |
1527 | close STDERR; | |
1528 | close STDOUT; | |
1529 | exec 'git-rev-parse','--verify', | |
1530 | "refs/remotes/$GIT_SVN^0" or croak $!; | |
1531 | } | |
1532 | waitpid $pid, 0; | |
1533 | push @update_ref, $primary_parent unless $?; | |
1534 | } | |
1535 | sys(@update_ref); | |
1536 | sys('git-update-ref',"svn/$GIT_SVN/revs/$log_msg->{revision}",$commit); | |
1537 | print "r$log_msg->{revision} = $commit\n"; | |
1538 | if ($_repack && (--$_repack_nr == 0)) { | |
1539 | $_repack_nr = $_repack; | |
1540 | sys("git repack $_repack_flags"); | |
1541 | } | |
1542 | return $commit; | |
1543 | } | |
1544 | ||
1545 | sub set_commit_env { | |
1546 | my ($log_msg) = @_; | |
1547 | my $author = $log_msg->{author}; | |
1548 | my ($name,$email) = defined $users{$author} ? @{$users{$author}} | |
1549 | : ($author,"$author\@$SVN_UUID"); | |
1550 | $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name; | |
1551 | $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email; | |
1552 | $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date}; | |
1553 | } | |
1554 | ||
1555 | sub apply_mod_line_blob { | |
1556 | my $m = shift; | |
1557 | if ($m->{mode_b} =~ /^120/) { | |
1558 | blob_to_symlink($m->{sha1_b}, $m->{file_b}); | |
1559 | } else { | |
1560 | blob_to_file($m->{sha1_b}, $m->{file_b}); | |
1561 | } | |
1562 | } | |
1563 | ||
1564 | sub blob_to_symlink { | |
1565 | my ($blob, $link) = @_; | |
1566 | defined $link or croak "\$link not defined!\n"; | |
1567 | croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o; | |
1568 | if (-l $link || -f _) { | |
1569 | unlink $link or croak $!; | |
1570 | } | |
1571 | ||
1572 | my $dest = `git-cat-file blob $blob`; # no newline, so no chomp | |
1573 | symlink $dest, $link or croak $!; | |
1574 | } | |
1575 | ||
1576 | sub blob_to_file { | |
1577 | my ($blob, $file) = @_; | |
1578 | defined $file or croak "\$file not defined!\n"; | |
1579 | croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o; | |
1580 | if (-l $file || -f _) { | |
1581 | unlink $file or croak $!; | |
1582 | } | |
1583 | ||
1584 | open my $blob_fh, '>', $file or croak "$!: $file\n"; | |
1585 | my $pid = fork; | |
1586 | defined $pid or croak $!; | |
1587 | ||
1588 | if ($pid == 0) { | |
1589 | open STDOUT, '>&', $blob_fh or croak $!; | |
1590 | exec('git-cat-file','blob',$blob) or croak $!; | |
1591 | } | |
1592 | waitpid $pid, 0; | |
1593 | croak $? if $?; | |
1594 | ||
1595 | close $blob_fh or croak $!; | |
1596 | } | |
1597 | ||
1598 | sub safe_qx { | |
1599 | my $pid = open my $child, '-|'; | |
1600 | defined $pid or croak $!; | |
1601 | if ($pid == 0) { | |
1602 | exec(@_) or croak $!; | |
1603 | } | |
1604 | my @ret = (<$child>); | |
1605 | close $child or croak $?; | |
1606 | die $? if $?; # just in case close didn't error out | |
1607 | return wantarray ? @ret : join('',@ret); | |
1608 | } | |
1609 | ||
1610 | sub svn_compat_check { | |
1611 | my @co_help = safe_qx(qw(svn co -h)); | |
1612 | unless (grep /ignore-externals/,@co_help) { | |
1613 | print STDERR "W: Installed svn version does not support ", | |
1614 | "--ignore-externals\n"; | |
1615 | $_no_ignore_ext = 1; | |
1616 | } | |
1617 | if (grep /usage: checkout URL\[\@REV\]/,@co_help) { | |
1618 | $_svn_co_url_revs = 1; | |
1619 | } | |
1620 | if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) { | |
1621 | $_svn_pg_peg_revs = 1; | |
1622 | } | |
1623 | ||
1624 | # I really, really hope nobody hits this... | |
1625 | unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) { | |
1626 | print STDERR <<''; | |
1627 | W: The installed svn version does not support the --stop-on-copy flag in | |
1628 | the log command. | |
1629 | Lets hope the directory you're tracking is not a branch or tag | |
1630 | and was never moved within the repository... | |
1631 | ||
1632 | $_no_stop_copy = 1; | |
1633 | } | |
1634 | } | |
1635 | ||
1636 | # *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>, | |
1637 | # (and they won't honor URL@<rev> without -r<rev>, too!) | |
1638 | sub svn_cmd_checkout { | |
1639 | my ($url, $rev, $dir) = @_; | |
1640 | my @cmd = ('svn','co', "-r$rev"); | |
1641 | push @cmd, '--ignore-externals' unless $_no_ignore_ext; | |
1642 | $url .= "\@$rev" if $_svn_co_url_revs; | |
1643 | sys(@cmd, $url, $dir); | |
1644 | } | |
1645 | ||
1646 | sub check_upgrade_needed { | |
1647 | my $old = eval { | |
1648 | my $pid = open my $child, '-|'; | |
1649 | defined $pid or croak $!; | |
1650 | if ($pid == 0) { | |
1651 | close STDERR; | |
1652 | exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!; | |
1653 | } | |
1654 | my @ret = (<$child>); | |
1655 | close $child or croak $?; | |
1656 | die $? if $?; # just in case close didn't error out | |
1657 | return wantarray ? @ret : join('',@ret); | |
1658 | }; | |
1659 | return unless $old; | |
1660 | my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") }; | |
1661 | if ($@ || !$head) { | |
1662 | print STDERR "Please run: $0 rebuild --upgrade\n"; | |
1663 | exit 1; | |
1664 | } | |
1665 | } | |
1666 | ||
1667 | # fills %tree_map with a reverse mapping of trees to commits. Useful | |
1668 | # for finding parents to commit on. | |
1669 | sub map_tree_joins { | |
1670 | my %seen; | |
1671 | foreach my $br (@_branch_from) { | |
1672 | my $pid = open my $pipe, '-|'; | |
1673 | defined $pid or croak $!; | |
1674 | if ($pid == 0) { | |
1675 | exec(qw(git-rev-list --topo-order --pretty=raw), $br) | |
1676 | or croak $!; | |
1677 | } | |
1678 | while (<$pipe>) { | |
1679 | if (/^commit ($sha1)$/o) { | |
1680 | my $commit = $1; | |
1681 | ||
1682 | # if we've seen a commit, | |
1683 | # we've seen its parents | |
1684 | last if $seen{$commit}; | |
1685 | my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o); | |
1686 | unless (defined $tree) { | |
1687 | die "Failed to parse commit $commit\n"; | |
1688 | } | |
1689 | push @{$tree_map{$tree}}, $commit; | |
1690 | $seen{$commit} = 1; | |
1691 | } | |
1692 | } | |
1693 | close $pipe; # we could be breaking the pipe early | |
1694 | } | |
1695 | } | |
1696 | ||
1697 | sub load_all_refs { | |
1698 | if (@_branch_from) { | |
1699 | print STDERR '--branch|-b parameters are ignored when ', | |
1700 | "--branch-all-refs|-B is passed\n"; | |
1701 | } | |
1702 | ||
1703 | # don't worry about rev-list on non-commit objects/tags, | |
1704 | # it shouldn't blow up if a ref is a blob or tree... | |
1705 | chomp(@_branch_from = `git-rev-parse --symbolic --all`); | |
1706 | } | |
1707 | ||
1708 | # '<svn username> = real-name <email address>' mapping based on git-svnimport: | |
1709 | sub load_authors { | |
1710 | open my $authors, '<', $_authors or die "Can't open $_authors $!\n"; | |
1711 | while (<$authors>) { | |
1712 | chomp; | |
1713 | next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/; | |
1714 | my ($user, $name, $email) = ($1, $2, $3); | |
1715 | $users{$user} = [$name, $email]; | |
1716 | } | |
1717 | close $authors or croak $!; | |
1718 | } | |
1719 | ||
1720 | sub rload_authors { | |
1721 | open my $authors, '<', $_authors or die "Can't open $_authors $!\n"; | |
1722 | while (<$authors>) { | |
1723 | chomp; | |
1724 | next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/; | |
1725 | my ($user, $name, $email) = ($1, $2, $3); | |
1726 | $rusers{"$name <$email>"} = $user; | |
1727 | } | |
1728 | close $authors or croak $!; | |
1729 | } | |
1730 | ||
1731 | sub svn_propget_base { | |
1732 | my ($p, $f) = @_; | |
1733 | $f .= '@BASE' if $_svn_pg_peg_revs; | |
1734 | return safe_qx(qw/svn propget/, $p, $f); | |
1735 | } | |
1736 | ||
1737 | sub git_svn_each { | |
1738 | my $sub = shift; | |
1739 | foreach (`git-rev-parse --symbolic --all`) { | |
1740 | next unless s#^refs/remotes/##; | |
1741 | chomp $_; | |
1742 | next unless -f "$GIT_DIR/svn/$_/info/url"; | |
1743 | &$sub($_); | |
1744 | } | |
1745 | } | |
1746 | ||
1747 | sub migration_check { | |
1748 | return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR); | |
1749 | print "Upgrading repository...\n"; | |
1750 | unless (-d "$GIT_DIR/svn") { | |
1751 | mkdir "$GIT_DIR/svn" or croak $!; | |
1752 | } | |
1753 | print "Data from a previous version of git-svn exists, but\n\t", | |
1754 | "$GIT_SVN_DIR\n\t(required for this version ", | |
1755 | "($VERSION) of git-svn) does not.\n"; | |
1756 | ||
1757 | foreach my $x (`git-rev-parse --symbolic --all`) { | |
1758 | next unless $x =~ s#^refs/remotes/##; | |
1759 | chomp $x; | |
1760 | next unless -f "$GIT_DIR/$x/info/url"; | |
1761 | my $u = eval { file_to_s("$GIT_DIR/$x/info/url") }; | |
1762 | next unless $u; | |
1763 | my $dn = dirname("$GIT_DIR/svn/$x"); | |
1764 | mkpath([$dn]) unless -d $dn; | |
1765 | rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x"; | |
1766 | my ($url, $path) = repo_path_split($u); | |
1767 | s_to_file($url, "$GIT_DIR/svn/$x/info/repo_url"); | |
1768 | s_to_file($path, "$GIT_DIR/svn/$x/info/repo_path"); | |
1769 | } | |
1770 | print "Done upgrading.\n"; | |
1771 | } | |
1772 | ||
1773 | sub find_rev_before { | |
1774 | my ($r, $git_svn_id) = @_; | |
1775 | my @revs = map { basename $_ } <$GIT_DIR/svn/$git_svn_id/revs/*>; | |
1776 | foreach my $r0 (sort { $b <=> $a } @revs) { | |
1777 | next if $r0 >= $r; | |
1778 | return ($r0, file_to_s("$GIT_DIR/svn/$git_svn_id/revs/$r0")); | |
1779 | } | |
1780 | return (undef, undef); | |
1781 | } | |
1782 | ||
1783 | sub init_vars { | |
1784 | $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn'; | |
1785 | $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN"; | |
1786 | $GIT_SVN_INDEX = "$GIT_SVN_DIR/index"; | |
1787 | $SVN_URL = undef; | |
1788 | $REV_DIR = "$GIT_SVN_DIR/revs"; | |
1789 | $SVN_WC = "$GIT_SVN_DIR/tree"; | |
1790 | } | |
1791 | ||
1792 | # convert GetOpt::Long specs for use by git-repo-config | |
1793 | sub read_repo_config { | |
1794 | return unless -d $GIT_DIR; | |
1795 | my $opts = shift; | |
1796 | foreach my $o (keys %$opts) { | |
1797 | my $v = $opts->{$o}; | |
1798 | my ($key) = ($o =~ /^([a-z\-]+)/); | |
1799 | $key =~ s/-//g; | |
1800 | my $arg = 'git-repo-config'; | |
1801 | $arg .= ' --int' if ($o =~ /[:=]i$/); | |
1802 | $arg .= ' --bool' if ($o !~ /[:=][sfi]$/); | |
1803 | if (ref $v eq 'ARRAY') { | |
1804 | chomp(my @tmp = `$arg --get-all svn.$key`); | |
1805 | @$v = @tmp if @tmp; | |
1806 | } else { | |
1807 | chomp(my $tmp = `$arg --get svn.$key`); | |
1808 | if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) { | |
1809 | $$v = $tmp; | |
1810 | } | |
1811 | } | |
1812 | } | |
1813 | } | |
1814 | ||
1815 | sub set_default_vals { | |
1816 | if (defined $_repack) { | |
1817 | $_repack = 1000 if ($_repack <= 0); | |
1818 | $_repack_nr = $_repack; | |
1819 | $_repack_flags ||= ''; | |
1820 | } | |
1821 | } | |
1822 | ||
1823 | sub read_grafts { | |
1824 | my $gr_file = shift; | |
1825 | my ($grafts, $comments) = ({}, {}); | |
1826 | if (open my $fh, '<', $gr_file) { | |
1827 | my @tmp; | |
1828 | while (<$fh>) { | |
1829 | if (/^($sha1)\s+/) { | |
1830 | my $c = $1; | |
1831 | if (@tmp) { | |
1832 | @{$comments->{$c}} = @tmp; | |
1833 | @tmp = (); | |
1834 | } | |
1835 | foreach my $p (split /\s+/, $_) { | |
1836 | $grafts->{$c}->{$p} = 1; | |
1837 | } | |
1838 | } else { | |
1839 | push @tmp, $_; | |
1840 | } | |
1841 | } | |
1842 | close $fh or croak $!; | |
1843 | @{$comments->{'END'}} = @tmp if @tmp; | |
1844 | } | |
1845 | return ($grafts, $comments); | |
1846 | } | |
1847 | ||
1848 | sub write_grafts { | |
1849 | my ($grafts, $comments, $gr_file) = @_; | |
1850 | ||
1851 | open my $fh, '>', $gr_file or croak $!; | |
1852 | foreach my $c (sort keys %$grafts) { | |
1853 | if ($comments->{$c}) { | |
1854 | print $fh $_ foreach @{$comments->{$c}}; | |
1855 | } | |
1856 | my $p = $grafts->{$c}; | |
1857 | delete $p->{$c}; # commits are not self-reproducing... | |
1858 | my $pid = open my $ch, '-|'; | |
1859 | defined $pid or croak $!; | |
1860 | if (!$pid) { | |
1861 | exec(qw/git-cat-file commit/, $c) or croak $!; | |
1862 | } | |
1863 | while (<$ch>) { | |
1864 | if (/^parent ([a-f\d]{40})/) { | |
1865 | $p->{$1} = 1; | |
1866 | } else { | |
1867 | last unless /^\S/i; | |
1868 | } | |
1869 | } | |
1870 | close $ch; # breaking the pipe | |
1871 | print $fh $c, ' ', join(' ', sort keys %$p),"\n"; | |
1872 | } | |
1873 | if ($comments->{'END'}) { | |
1874 | print $fh $_ foreach @{$comments->{'END'}}; | |
1875 | } | |
1876 | close $fh or croak $!; | |
1877 | } | |
1878 | ||
1879 | sub read_url_paths { | |
1880 | my $l_map = {}; | |
1881 | git_svn_each(sub { my $x = shift; | |
1882 | my $u = file_to_s("$GIT_DIR/svn/$x/info/repo_url"); | |
1883 | my $p = file_to_s("$GIT_DIR/svn/$x/info/repo_path"); | |
1884 | # we hate trailing slashes | |
1885 | if ($u =~ s#(?:^\/+|\/+$)##g) { | |
1886 | s_to_file($u,"$GIT_DIR/svn/$x/info/repo_url"); | |
1887 | } | |
1888 | if ($p =~ s#(?:^\/+|\/+$)##g) { | |
1889 | s_to_file($p,"$GIT_DIR/svn/$x/info/repo_path"); | |
1890 | } | |
1891 | $l_map->{$u}->{$p} = $x; | |
1892 | }); | |
1893 | return $l_map; | |
1894 | } | |
1895 | ||
1896 | sub extract_metadata { | |
1897 | my $id = shift; | |
1898 | my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+) | |
1899 | \s([a-f\d\-]+)$/x); | |
1900 | if (!$rev || !$uuid || !$url) { | |
1901 | # some of the original repositories I made had | |
1902 | # indentifiers like this: | |
1903 | ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/); | |
1904 | } | |
1905 | return ($url, $rev, $uuid); | |
1906 | } | |
1907 | ||
1908 | sub tz_to_s_offset { | |
1909 | my ($tz) = @_; | |
1910 | $tz =~ s/(\d\d)$//; | |
1911 | return ($1 * 60) + ($tz * 3600); | |
1912 | } | |
1913 | ||
1914 | sub setup_pager { # translated to Perl from pager.c | |
1915 | return unless (-t *STDOUT); | |
1916 | my $pager = $ENV{PAGER}; | |
1917 | if (!defined $pager) { | |
1918 | $pager = 'less'; | |
1919 | } elsif (length $pager == 0 || $pager eq 'cat') { | |
1920 | return; | |
1921 | } | |
1922 | pipe my $rfd, my $wfd or return; | |
1923 | defined(my $pid = fork) or croak $!; | |
1924 | if (!$pid) { | |
1925 | open STDOUT, '>&', $wfd or croak $!; | |
1926 | return; | |
1927 | } | |
1928 | open STDIN, '<&', $rfd or croak $!; | |
1929 | $ENV{LESS} ||= '-S'; | |
1930 | exec $pager or croak "Can't run pager: $!\n";; | |
1931 | } | |
1932 | ||
1933 | sub get_author_info { | |
1934 | my ($dest, $author, $t, $tz) = @_; | |
1935 | $author =~ s/(?:^\s*|\s*$)//g; | |
1936 | my $_a; | |
1937 | if ($_authors) { | |
1938 | $_a = $rusers{$author} || undef; | |
1939 | } | |
1940 | if (!$_a) { | |
1941 | ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/); | |
1942 | } | |
1943 | $dest->{t} = $t; | |
1944 | $dest->{tz} = $tz; | |
1945 | $dest->{a} = $_a; | |
1946 | # Date::Parse isn't in the standard Perl distro :( | |
1947 | if ($tz =~ s/^\+//) { | |
1948 | $t += tz_to_s_offset($tz); | |
1949 | } elsif ($tz =~ s/^\-//) { | |
1950 | $t -= tz_to_s_offset($tz); | |
1951 | } | |
1952 | $dest->{t_utc} = $t; | |
1953 | } | |
1954 | ||
1955 | sub process_commit { | |
1956 | my ($c, $r_min, $r_max, $defer) = @_; | |
1957 | if (defined $r_min && defined $r_max) { | |
1958 | if ($r_min == $c->{r} && $r_min == $r_max) { | |
1959 | show_commit($c); | |
1960 | return 0; | |
1961 | } | |
1962 | return 1 if $r_min == $r_max; | |
1963 | if ($r_min < $r_max) { | |
1964 | # we need to reverse the print order | |
1965 | return 0 if (defined $_limit && --$_limit < 0); | |
1966 | push @$defer, $c; | |
1967 | return 1; | |
1968 | } | |
1969 | if ($r_min != $r_max) { | |
1970 | return 1 if ($r_min < $c->{r}); | |
1971 | return 1 if ($r_max > $c->{r}); | |
1972 | } | |
1973 | } | |
1974 | return 0 if (defined $_limit && --$_limit < 0); | |
1975 | show_commit($c); | |
1976 | return 1; | |
1977 | } | |
1978 | ||
1979 | sub show_commit { | |
1980 | my $c = shift; | |
1981 | if ($_oneline) { | |
1982 | my $x = "\n"; | |
1983 | if (my $l = $c->{l}) { | |
1984 | while ($l->[0] =~ /^\s*$/) { shift @$l } | |
1985 | $x = $l->[0]; | |
1986 | } | |
1987 | $_l_fmt ||= 'A' . length($c->{r}); | |
1988 | print 'r',pack($_l_fmt, $c->{r}),' | '; | |
1989 | print "$c->{c} | " if $_show_commit; | |
1990 | print $x; | |
1991 | } else { | |
1992 | show_commit_normal($c); | |
1993 | } | |
1994 | } | |
1995 | ||
1996 | sub show_commit_normal { | |
1997 | my ($c) = @_; | |
1998 | print '-' x72, "\nr$c->{r} | "; | |
1999 | print "$c->{c} | " if $_show_commit; | |
2000 | print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", | |
2001 | localtime($c->{t_utc})), ' | '; | |
2002 | my $nr_line = 0; | |
2003 | ||
2004 | if (my $l = $c->{l}) { | |
2005 | while ($l->[$#$l] eq "\n" && $l->[($#$l - 1)] eq "\n") { | |
2006 | pop @$l; | |
2007 | } | |
2008 | $nr_line = scalar @$l; | |
2009 | if (!$nr_line) { | |
2010 | print "1 line\n\n\n"; | |
2011 | } else { | |
2012 | if ($nr_line == 1) { | |
2013 | $nr_line = '1 line'; | |
2014 | } else { | |
2015 | $nr_line .= ' lines'; | |
2016 | } | |
2017 | print $nr_line, "\n\n"; | |
2018 | print $_ foreach @$l; | |
2019 | } | |
2020 | } else { | |
2021 | print "1 line\n\n"; | |
2022 | ||
2023 | } | |
2024 | foreach my $x (qw/raw diff/) { | |
2025 | if ($c->{$x}) { | |
2026 | print "\n"; | |
2027 | print $_ foreach @{$c->{$x}} | |
2028 | } | |
2029 | } | |
2030 | } | |
2031 | ||
2032 | __END__ | |
2033 | ||
2034 | Data structures: | |
2035 | ||
2036 | $svn_log hashref (as returned by svn_log_raw) | |
2037 | { | |
2038 | fh => file handle of the log file, | |
2039 | state => state of the log file parser (sep/msg/rev/msg_start...) | |
2040 | } | |
2041 | ||
2042 | $log_msg hashref as returned by next_log_entry($svn_log) | |
2043 | { | |
2044 | msg => 'whitespace-formatted log entry | |
2045 | ', # trailing newline is preserved | |
2046 | revision => '8', # integer | |
2047 | date => '2004-02-24T17:01:44.108345Z', # commit date | |
2048 | author => 'committer name' | |
2049 | }; | |
2050 | ||
2051 | ||
2052 | @mods = array of diff-index line hashes, each element represents one line | |
2053 | of diff-index output | |
2054 | ||
2055 | diff-index line ($m hash) | |
2056 | { | |
2057 | mode_a => first column of diff-index output, no leading ':', | |
2058 | mode_b => second column of diff-index output, | |
2059 | sha1_b => sha1sum of the final blob, | |
2060 | chg => change type [MCRADT], | |
2061 | file_a => original file name of a file (iff chg is 'C' or 'R') | |
2062 | file_b => new/current file name of a file (any chg) | |
2063 | } | |
2064 | ; |