#!/usr/bin/perl # Copyright (C) 2018 The Qt Company Ltd. # Copyright (C) 2019 Oswald Buddenhagen # Contact: http://www.qt.io/licensing/ # # You may use this file under the terms of the 3-clause BSD license. # See the file LICENSE from this package for details. # use v5.14; use strict; use warnings; our ($script, $script_path); BEGIN { use Cwd qw(abs_path); if ($^O eq "msys") { $0 =~ s,\\,/,g; $0 =~ s,^(.):/,/$1/,g; } $script_path = $script = abs_path($0); $script_path =~ s,/[^/]+$,,; unshift @INC, $script_path; } use git_gpush; use List::Util qw(first); use Digest::SHA qw(sha1_hex); use JSON; sub usage { print << "EOM"; Usage: git gpick [options] {[\@parent] [+]Changes[{date}][\@] | /Changes[\@]}... Updates local commits with the specified PatchSets from Gerrit. Description: This program fetches the specified PatchSets of the specified Changes from Gerrit, and updates the local commits with them. Conflicting local modifications are reported and require an override. Remote modifications to series structure are followed by adding, dropping, and re-ordering local commits as necessary; conflicts are also handled. Commits are updated in place, unless a parent to rebase them onto is specified. Changes with no corresponding local commit need to be prefixed with a plus sign, and are picked on top of HEAD by default. If a Change is suffixed with an at-sign, it is made the parent of the subsequent Change specification. Remote commits may be specified only by complete Gerrit Change-Id. Local commits may be specified as either SHA1s or Gerrit Change-Ids, possibly abbreviated. Git rev-spec suffixes like '~2' are allowed; if only a suffix is specified, it is understood to be relative to 'HEAD'. Every Change specification denotes a range, either explicitly when written as .. or :, or implicitly as a part of a series (as determined by its previous push). An empty means 'HEAD'. Similarly, an empty means the merge base with the upstream branch; this also works in the parent specification. This program uses the most recent PatchSets which are not newer than the reference date, which defaults to 'now'. See the git-rev-parse documentation for supported date formats. Specifying PatchSet numbers directly is not supported, as these are meaningless for ranges with multiple Changes. An exception to that is the 1st PatchSet, which may be specified with a literal 1. Local commits may be dropped by prefixing them with a slash. This program uses 'git rebase' as a workhorse. If the rebase aborts, follow the usual instructions on screen. Options: -a, --all Update all commits on the local branch. -c, --check Rather than replacing any commits, only show what would be done. A side effect is that the state information is synchronized with Gerrit, which is necessary when commits were previously pushed or picked without (or with an older version of) gpush or gpick, respectively. Use with --quiet to show only "interesting" Changes. -m, --merge Attempt to merge concurrent local and remote modifications to the same Changes. Use only once you are sure that there are no logical conflicts. This currently works only when each side touches different parts of the Change (author, commit message, or diff). -f, --force Replace or merge the local commits even in case of conflicting modifications. Drop local commits even in case of local modifications. -fs, --force-struct Follow remote structural changes even when conflicting local changes are found. Note that --force is still necessary if this would result in dropping locally modified Changes. -fm, --force-merged Pick Changes even if they are MERGED and were already pulled. This makes sense when re-applying reverted commits. -fc, --force-closed The Changes in a series are categorized by their "degree of openness" (open, deferred, abandoned). Usually, only the ones in the "most open" category are picked, while all "more closed" ones are omitted. Specifying this option causes the Changes in the next actually present category to be picked as well. May be specified twice. -i, --ignore Ignore the new PatchSets that appeared on Gerrit since the previous push from this clone. The local Changes will subsequently appear modified, and the following push will go through without overwrite warning. For Changes that were locally modified since the previous push, this option can be considered complementary to --force. -is, --ignore-struct Ignore structural changes to series on Gerrit. --author, --auth --message, --msg --diff Pick only the named part(s) of the specified commits. This is applicable only to Changes which are already present locally. Note that remote modifications to the omitted parts are forgotten about, as if the --ignore option was supplied. Note also that use of --author and/or --message without --diff disables all actions that modify the structure of a series and implies --ignore-struct. -r, --remote Specify the remote used to determine the Gerrit server to fetch PatchSets from. The fallback behavior matches git-gpush. -b, --branch Specify the branch for which the Changes to download were pushed to Gerrit. In case of ambiguity, the default is the upstram branch of the current local branch. --move/--copy/--hide Deal with cherry-picks of Changes between local branches. This works the same as documented for git-gpush, except that the range may also be a plus-sign prefixed addition specification. Unlike local ranges, these DO get picked without being specified redundantly, because otherwise there would be nothing to operate on in the first place. -n, --dry-run Do everything except actually replacing any commits and updating state. -v, --verbose Show additional progress output. -q, --quiet Suppress the usual progress output. --debug Print debug information. Console Output: During operation, output following this pattern will be produced: () [] and describe the Change on Gerrit. describes what will happen with the local commit. Uppercase denotes actions that lose local modifications, while parenthesization denotes the need to supply additional options. has the format " PS: ". PS is the PatchSet that was created for the Change by the last push from this repository. and indicate the parts of the Change that were modified on the respective side relative to PS, while indicate divergent concurrent modifications; convergent concurrent modifications are completely omitted for brevity. When is unknown, it is shown as 'x' and all differences are shown as conflicts. Modifications of commit parts which are excluded on the command line are shown in parentheses. Examples: git gpick ~1:1 Replace the last-but-one local commit with the latest PatchSet from Gerrit. git gpick ~1 Replace the series identified by the last-but-one local commit with the latest PatchSets from Gerrit. git gpick +I21f8ef385d1793757149dfa5cc69e4e907cb1c04 Add a series from Gerrit on top of the local branch. git gpick +Idc2d0ac4f7d95a5f3bad24e82114e23ada79a542{yesterday} The same, using the most recent PatchSets as of one day ago. git gpick /HEAD:1 \@ ~1:2\@ +I1ad95db7c99018e92d2b2556e4789951b51b2fff:1 Drop the top commit, move the next two commits to the start of the local branch and replace them with the latest PatchSets, and add another Change right on top of them. The other commits in the local branch are rebased on top of these three commits. git gpick --check --all Bootstrap git-gpush before using it the first time with pre-existing pending Changes. git gpick --check :1 Check whether a new PatchSet was pushed to the Change corresponding with the HEAD commit. Copyright: Copyright (C) 2018 The Qt Company Ltd. Copyright (C) 2019 Oswald Buddenhagen Contact: http://www.qt.io/licensing/ License: You may use this file under the terms of the 3-clause BSD license. EOM } use constant { NO_PARTS => 0, AUTHOR_PART => 1, MESSAGE_PART => 2, DIFF_PART => 4, MERGE_PART => 8, # This cannot be picked separately ALL_PARTS => 15 }; my $pick_all = 0; my $pick_parts; my $branch; my $upstream_branch; my $check = 0; my $merge = 0; my $force = 0; my $force_struct = 0; my $force_merged = 0; my $force_closed = 0; my $ignore = 0; my $ignore_struct = 0; my @commit_specs; sub parse_arguments(@) { while (scalar @_) { my $arg = shift @_; if ($arg eq "-v" || $arg eq "--verbose") { $verbose = 1; } elsif ($arg eq "-q" || $arg eq "--quiet") { $quiet = 1; } elsif ($arg eq "--debug") { $debug = 1; $verbose = 1; } elsif ($arg eq "-n" || $arg eq "--dry-run") { $dry_run = 1; } elsif ($arg eq "-r" || $arg eq "--remote") { fail("--remote needs an argument.\n") if (!@_ || ($_[0] =~ /^-/)); $remote = shift @_; } elsif ($arg eq "-b" || $arg eq "--branch") { fail("--branch needs an argument.\n") if (!@_ || ($_[0] =~ /^-/)); $branch = shift @_; } elsif ($arg eq "-a" || $arg eq "--all") { $pick_all = 1; } elsif ($arg eq "-c" || $arg eq "--check") { $check = 1; } elsif ($arg eq "-m" || $arg eq "--merge") { $merge = 1; } elsif ($arg eq "-f" || $arg eq "--force") { $force = 1; } elsif ($arg eq "-fs" || $arg eq "--force-struct") { $force_struct = 1; } elsif ($arg eq "-fm" || $arg eq "--force-merged") { $force_merged = 1; } elsif ($arg eq "-fc" || $arg eq "--force-closed") { $force_closed++; } elsif ($arg eq "-i" || $arg eq "--ignore") { $ignore = 1; } elsif ($arg eq "-is" || $arg eq "--ignore-struct") { $ignore_struct = 1; } elsif ($arg eq "--author" || $arg eq "--auth") { $pick_parts |= AUTHOR_PART; } elsif ($arg eq "--message" || $arg eq "--msg") { $pick_parts |= MESSAGE_PART; } elsif ($arg eq "--diff") { $pick_parts |= DIFF_PART; } elsif ($arg eq "-?" || $arg eq "--?" || $arg eq "-h" || $arg eq "--help") { usage(); exit 0; } elsif (parse_source_option($arg, 1, @_)) { # Nothing } elsif ($arg !~ /^-/) { push @commit_specs, $arg; } else { fail("Invalid option '$arg'.\n"); } } fail("--quiet and --verbose/--debug are mutually exclusive.\n") if ($quiet && $verbose); if ($check) { # Note that --{force,ignore}-struct are intentionally omitted, # because one may want to preview the effects of (not) following # structural changes. wfail("--check and --merge/--force/--force-merged/--force-closed/--ignore" ." are mutually exclusive.\n") if ($merge || $force || $force_merged || $force_closed || $ignore); } fail("--all is mutually exclusive with specifying ranges.\n") if ($pick_all && @commit_specs); fail("--merge/--force and --ignore are mutually exclusive.\n") if (($merge || $force) && $ignore); fail("--force-struct and --ignore-struct are mutually exclusive.\n") if ($force_struct && $ignore_struct); $pick_parts = ALL_PARTS if (!$pick_parts); # We don't follow structural changes to series when omitting # diffs, as that is likely to cause inconsistencies. if (!($pick_parts & DIFF_PART)) { fail("--force-struct is mutually exclusive with omitting diffs.\n") if ($force_struct); $ignore_struct = 1 } } sub determine_local_branch() { my $branches = open_cmd_pipe(0, "git", "branch", "--points-at", "HEAD"); while (read_process($branches)) { if (/^\* $/) { # Unlike in gpush, it makes no sense to work with mid-rebase states # (as we are going to rewrite the branch ourselves). fail("Cannot proceed, a rebase is currently in progress.\n") if (/^\* \(no branch, rebasing (.*)$$/); # This does not work, because the SHA1 is truncated, # and --no-abbrev appears to have no effect on that. #if (/^\* $HEAD detached at (.*)$$/) { # # This potentially saves a `git rev-parse HEAD` call. # $local_tip = $1; #} last; } elsif (/^\* (.*)$/) { $local_branch = $1; $local_tip = $local_refs{$local_branch}; last; } } close_process($branches); if (defined($local_branch)) { $upstream_branch = git_config("branch.$local_branch.merge"); $upstream_branch =~ s,^refs/heads/,, if (defined($upstream_branch)); } setup_remotes($local_branch // 'HEAD'); set_gerrit_config($remote); source_map_validate(); } # Get the list of local commits (which can be replaced). sub get_changes() { return get_commits_free($local_tip) if (analyze_local_branch('HEAD')); # We get here if we have no local Changes ... if (!defined($local_tip)) { # ...when working on a detached HEAD. $local_tip = read_cmd_line(0, 'git', 'rev-parse', 'HEAD'); } # The tip is also the branch base. $local_base = $local_tip; return []; } # Deduce series grouping from picked Changes. # The grouping on Gerrit is considered authoritative, which may result # in previously assigned local series being split or joined. This # affects even Changes which are not being updated, as long as they # are part of a series which any Change of is being updated; this is # done to avoid that partial picks tear apart series. Locally assigned # Changes which are not claimed by an applied remote series remain # assigned to their previous series, unless they lie between Changes # which are claimed by separate remote series. # # The algorithm sequentially interates through all Changes passed to it. # The "prefix" state prepresents a local series which may still overlap # with a remote series. The "suffix" state represents a local series # which extends beyond an overlap with a remote series. A "hold" state # is entered from the respective "plain" state when the local series # runs into the start of a remote series which does not overlap a local # series (yet). The state table can be viewed at # https://docs.google.com/spreadsheets/d/1z2BkqggoYvAe3guPrhPf96Y7lx8JJqB9iUmBx1bISgs/edit?usp=sharing use constant { DSS_DEFAULT => 0, DSS_PREFIX => 1, DSS_PREFIX_HOLD => 2, DSS_SUFFIX => 3, DSS_SUFFIX_HOLD => 4 }; use constant { DSC_LCL_NONE => 0, # Neither current nor previous Change have gid. DSC_LCL_START => 1, # Current Change has gid while previous didn't. DSC_LCL_RESUME => 2, # Same, but the gid equals the last seen one. DSC_LCL_CONT => 3, # Current and previous Change have same gid. DSC_LCL_SWITCH => 4, # Current and previous Change have different gids. DSC_LCL_END => 5, # Current Change has no gid while previous did. DSC_RMT_NONE => 0x00, DSC_RMT_START => 0x10, DSC_RMT_RESUME => 0x20, DSC_RMT_CONT => 0x30, DSC_RMT_SWITCH => 0x40, DSC_RMT_END => 0x50 }; sub deduce_series($;$) { my ($changes, $all) = @_; my %groups; # { group-id => [ change, ... ] } my $state = DSS_DEFAULT; my @backlog; local *drop = sub { my ($sts) = @_; print "... dropping backlog.\n" if ($debug); @backlog = (); $state = $sts // DSS_DEFAULT; }; local *flush = sub { my ($grp, $sts) = @_; print "... committing backlog to $grp.\n" if ($debug); push @{$groups{$grp}}, @backlog; @backlog = (); $state = $sts // DSS_DEFAULT; }; print "Deducing series from picked Changes ...\n" if ($debug); my ($good_lcl_grp, $good_rmt_grp) = (0, ""); # The last non-null value. my ($prev_lcl_grp, $prev_rmt_grp) = (0, ""); # From the immediately preceding Change. foreach my $change (@$changes, undef) { my ($curr_lcl_grp, $curr_rmt_grp) = (0, ""); if ($change) { $curr_lcl_grp = $$change{grp} // 0; my $ginfo = $$change{gerrit}; # Ungrouped Changes which also are not being picked are # completely ignored, unless forced. next if (!$all && !$curr_lcl_grp && !$ginfo); $curr_rmt_grp = $$ginfo{pick_commit}{pgrp} if ($ginfo); } else { # When we run out of Changes, we still synthetize a pair of # commands, so a pending backlog is flushed if necessary. } my $lcl_cmd = $curr_lcl_grp ? $prev_lcl_grp ? ($curr_lcl_grp == $prev_lcl_grp) ? DSC_LCL_CONT : DSC_LCL_SWITCH : ($curr_lcl_grp == $good_lcl_grp) ? DSC_LCL_RESUME : DSC_LCL_START : $prev_lcl_grp ? DSC_LCL_END : DSC_LCL_NONE; my $rmt_cmd = length($curr_rmt_grp) ? length($prev_rmt_grp) ? ($curr_rmt_grp eq $prev_rmt_grp) ? DSC_RMT_CONT : DSC_RMT_SWITCH : ($curr_rmt_grp eq $good_rmt_grp) ? DSC_RMT_RESUME : DSC_RMT_START : length($prev_rmt_grp) ? DSC_RMT_END : DSC_RMT_NONE; printf("sts=%d lcl=%d rmt=%d chg=%s\n", $state, $lcl_cmd, $rmt_cmd >> 4, $change ? $$change{id} : "") if ($debug); my $cmd = $lcl_cmd | $rmt_cmd; if ($state == DSS_PREFIX) { if ($cmd == (DSC_LCL_CONT | DSC_RMT_START) || $cmd == (DSC_LCL_CONT | DSC_RMT_RESUME)) { flush($curr_rmt_grp); } elsif ($cmd == (DSC_LCL_SWITCH | DSC_RMT_START) || $cmd == (DSC_LCL_SWITCH | DSC_RMT_RESUME) || $cmd == (DSC_LCL_END | DSC_RMT_NONE)) { drop(); } elsif ($cmd == (DSC_LCL_SWITCH | DSC_RMT_NONE)) { drop(DSS_PREFIX); } elsif ($cmd == (DSC_LCL_END | DSC_RMT_START) || $cmd == (DSC_LCL_END | DSC_RMT_RESUME)) { $state = DSS_PREFIX_HOLD; } } elsif ($state == DSS_PREFIX_HOLD) { if ($cmd == (DSC_LCL_RESUME | DSC_RMT_CONT)) { flush($curr_rmt_grp); } elsif ($cmd == (DSC_LCL_NONE | DSC_RMT_SWITCH) || $cmd == (DSC_LCL_NONE | DSC_RMT_END) || $cmd == (DSC_LCL_START | DSC_RMT_CONT) || $cmd == (DSC_LCL_START | DSC_RMT_SWITCH) || $cmd == (DSC_LCL_RESUME | DSC_RMT_SWITCH)) { drop(); } elsif ($cmd == (DSC_LCL_START | DSC_RMT_END)) { drop(DSS_PREFIX); } elsif ($cmd == (DSC_LCL_RESUME | DSC_RMT_END)) { $state = DSS_PREFIX; } } elsif ($state == DSS_SUFFIX) { if ($cmd == (DSC_LCL_CONT | DSC_RMT_RESUME) || $cmd == (DSC_LCL_SWITCH | DSC_RMT_START) || $cmd == (DSC_LCL_SWITCH | DSC_RMT_RESUME) || $cmd == (DSC_LCL_END | DSC_RMT_NONE) || $cmd == (DSC_LCL_END | DSC_RMT_RESUME)) { flush($good_rmt_grp); } elsif ($cmd == (DSC_LCL_SWITCH | DSC_RMT_NONE)) { flush($good_rmt_grp, DSS_PREFIX); } elsif ($cmd == (DSC_LCL_CONT | DSC_RMT_START)) { drop(); } elsif ($cmd == (DSC_LCL_END | DSC_RMT_START)) { $state = DSS_SUFFIX_HOLD; } } elsif ($state == DSS_SUFFIX_HOLD) { if ($cmd == (DSC_LCL_NONE | DSC_RMT_SWITCH) || $cmd == (DSC_LCL_NONE | DSC_RMT_END) || $cmd == (DSC_LCL_START | DSC_RMT_CONT) || $cmd == (DSC_LCL_START | DSC_RMT_SWITCH) || $cmd == (DSC_LCL_RESUME | DSC_RMT_SWITCH)) { flush($good_rmt_grp); } elsif ($cmd == (DSC_LCL_START | DSC_RMT_END) || $cmd == (DSC_LCL_RESUME | DSC_RMT_END)) { flush($good_rmt_grp, DSS_PREFIX); } elsif ($cmd == (DSC_LCL_RESUME | DSC_RMT_CONT)) { drop(); } } else { # DSS_DEFAULT if ($cmd == (DSC_LCL_START | DSC_RMT_NONE) || $cmd == (DSC_LCL_START | DSC_RMT_END) || $cmd == (DSC_LCL_SWITCH | DSC_RMT_END)) { $state = DSS_PREFIX; } elsif ($cmd == (DSC_LCL_RESUME | DSC_RMT_END) || $cmd == (DSC_LCL_CONT | DSC_RMT_END)) { $state = DSS_SUFFIX; } } last if (!$change); if (length($curr_rmt_grp)) { print "... comitting Change to $curr_rmt_grp.\n" if ($debug); push @{$groups{$curr_rmt_grp}}, $change; } else { print "... backlogging Change.\n" if ($debug); push @backlog, $change; } ($prev_lcl_grp, $prev_rmt_grp) = ($curr_lcl_grp, $curr_rmt_grp); $good_lcl_grp = $curr_lcl_grp if ($curr_lcl_grp); $good_rmt_grp = $curr_rmt_grp if (length($curr_rmt_grp)); } foreach my $group (values %groups) { assign_series($group); } } sub is_closed_status($) { my ($sts) = @_; return ($sts eq "MERGED" || $sts eq "DEFERRED" || $sts eq "ABANDONED"); } # Report alternative sources for a Change. sub report_extra($$$) { my ($reports, $title, $extra) = @_; return if (!@$extra); my @osrc; foreach my $oth (@$extra) { my $oann = $$oth{branch}; my $osts = $$oth{status}; $oann .= '/'.$osts if (is_closed_status($osts)); push @osrc, $oann; } report_flowed($reports, " $title sources: ".join(' ', @osrc)); } # Report the chosen source for a Change, plus possible alternatives. sub report_source($$$$$$) { my ($reports, $prefix, $annot, $ginfo, $id, $subject) = @_; my @extra; my $suffix = ""; if ($ginfo) { my ($br, $sts, $better, $other) = ($$ginfo{branch}, $$ginfo{status}, $$ginfo{better} // [], $$ginfo{other} // []); my @tags; push @tags, $br if (@$better || @$other || ($br ne ($upstream_branch // ""))); push @tags, $sts if (is_closed_status($sts)); $suffix = ' # '.join('/', @tags) if (@tags); report_extra(\@extra, "Possibly better", $better); report_extra(\@extra, "Other", $other); } push @$reports, { type => "change", id => $id, subject => $subject, prefix => $prefix, suffix => $suffix, annotation => length($annot) ? " [$annot]" : "" }, @extra; } # Report source using the remote Change's metadata for identification. sub report_remote($$$$) { my ($reports, $prefix, $annot, $ginfo) = @_; report_source($reports, $prefix, $annot, $ginfo, $$ginfo{id}, $$ginfo{subject}); } # Report source using the local commit's metadata for identification. # Alternatives are still provided by the remote Change. sub report_update($$$$$) { my ($reports, $prefix, $annot, $commit, $ginfo) = @_; report_source($reports, $prefix, $annot, $ginfo, $$commit{changeid}, $$commit{subject}); } # Report source using the local commit's metadata for identification. # No alternatives are listed. sub report_local($$$$) { my ($reports, $prefix, $annot, $commit) = @_; report_update($reports, $prefix, $annot, $commit, undef); } sub parse_date($) { my ($stamp) = @_; # PS1 may be specified directly. It's the same as specifying # a really old date. return 0 if ($stamp eq "1"); # Unfortunately, Git for Windows doesn't ship Time::ParseDate. my $parsed = read_cmd_line(0, 'git', 'rev-parse', '--before='.$stamp); return int($parsed =~ s/^.*=//r); } use constant { HOLD => 0, INSERT => 1, UPDATE => 2, DELETE => 3 }; # Process the parent and Change specifications from the command line # into an array of "modifier" objects. sub parse_specs($) { my ($raw_specs) = @_; print "Parsing commit specs ...\n" if ($debug); my $parent; my @specs; foreach my $cs (@$raw_specs) { if ($cs =~ s,^\@(?!\{),,) { $cs = '@{u}' if (!length($cs)); wfail("Cannot specify adjacent parents (2nd is $cs).\n") if (defined($parent)); my $sha1 = parse_local_rev($cs, SPEC_PARENT); my $commit = $commit_by_id{$sha1}; $parent = $commit ? $$commit{changeid} : '*'; } else { my %spec; $spec{parent} = $parent; my $isparent = ($cs =~ s,\@$,,); if ($cs =~ s,\{(.*)\}$,,) { $spec{stamp} = parse_date($1); } $spec{orig} = $cs; my $action; if ($cs =~ s,^\+,,) { fail("Cannot specify an addition when checking.\n") if ($check); fail("Cannot specify an addition when omitting diffs.\n") if (!($pick_parts & DIFF_PART)); fail("Missing specification of remote Change.\n") if (!length($cs)); $action = INSERT; } else { if ($cs =~ s,^/,,) { fail("Cannot specify a removal when checking.\n") if ($check); fail("Cannot specify a removal when omitting diffs.\n") if (!($pick_parts & DIFF_PART)); wfail("Cannot specify a removal (/$cs) with a parent.\n") if (defined($parent)); $action = DELETE; } else { $action = UPDATE; } } $spec{action} = $action; if ($cs =~ s,^(.*)\.\.,,) { $spec{base} = $1; } elsif ($cs =~ s,:(\d+)$,,) { $spec{count} = $1; } $spec{tip} = $cs; push @specs, \%spec; # Note that it's actually possible to use Changes we delete as parents. $parent = $isparent ? '@' : undef; } fail("Cannot specify a parent when checking.\n") if (defined($parent) && $check); } fail("Cannot specify a parent without a subsequent commit.\n") if (defined($parent)); fail("No Changes specified.\n") if (!@specs); return \@specs; } # Resolve the local specs into series. sub resolve_specs($) { my ($specs) = @_; print "Resolving commit specs ...\n" if ($debug); foreach my $spec (@$specs) { my $action = $$spec{action}; next if ($action == INSERT); my ($raw_tip, $count, $raw_base) = ($$spec{tip}, $$spec{count}, $$spec{base}); $raw_tip = 'HEAD' if (!length($raw_tip)); my $tip = parse_local_rev($raw_tip, SPEC_TIP); my $range; if (defined($count)) { $range = changes_from_commits(get_commits_count($tip, $count, $raw_tip)); } elsif (defined($raw_base)) { $raw_base = '@{u}' if (!length($raw_base)); my $base = parse_local_rev($raw_base, SPEC_BASE); $range = changes_from_commits(get_commits_base($base, $tip, $raw_base, $raw_tip)); } else { my $gid; my $pivot = $commit_by_id{$tip}{change}; ($range, $gid, undef, undef) = do_determine_series($pivot); wfail("Spec $$spec{orig} points at loose Change(s)." ." Please specify an exact range.\n") if (!defined($gid)); } wfail("Range $$spec{orig} is empty.\n") if (!@$range); foreach my $change (@$range) { my $changeid = $$change{id}; my $ospec = $$change{lspec}; wfail("Specs $$spec{orig} and $$ospec{orig} intersect (change $changeid).\n") if ($ospec); $$change{lspec} = $spec; } $$spec{range} = $range; if ($action == DELETE) { $$spec{new_range} = []; } elsif ($ignore_struct) { $$spec{new_range} = $range; } } } sub select_patchset($$) { my ($ginfo, $stamp) = @_; my $revs = $$ginfo{revs}; my $revidx = $#$revs; if (defined($stamp)) { while ($revidx > 0 && $$revs[$revidx]{ts} > $stamp) { $revidx--; } print "Chose revision $revidx/$#$revs for $$ginfo{id}.\n" if ($debug); } return $revidx; } use constant { MAP_SUCCESS => 0, MAP_MISSING => 1, MAP_AMBIGUOUS => 2 }; # Choose the correct Gerrit Change object from a set of options, # trying to resolve ambiguity. sub map_remote_change($$$$$) { my ($changeid, $br, $ginfos, $spec, $fetches) = @_; # If a specific branch is requested (or implied by the # Change's already set target branch), insist on that. # As we track remote branch changes, this effectively # pins a local Change to the correct remote one even # though we do not insist on matching SHA1s here. As a # consequence, we let it slip when the original remote # Change gets nuked, but a new one with the same id pops # up in its place - which seems desirable. my ($pref_br, $br_reqd) = ($branch // $br, 1); # Otherwise, the possibly present upstream is a preference. ($pref_br, $br_reqd) = ($upstream_branch // "", 0) if (!defined($pref_br)); my $match; my @other; foreach my $gi (@$ginfos) { my ($br, $sts) = ($$gi{branch}, $$gi{status}); if ($br eq $pref_br) { $match = $gi; } else { push @other, $gi; } $$gi{closedness} = ($sts eq "ABANDONED") ? 2 : ($sts eq "DEFERRED" || $sts eq "MERGED") ? 1 : 0; } @other = sort { $$a{closedness} <=> $$b{closedness} || $$a{branch} cmp $$b{branch}} @other; my $fallback; if (!$match) { # This implies that Changes on other branches were found, # as otherwise we wouldn't call the function to start with. if ($br_reqd) { # The request cannot be complied with. return (MAP_MISSING, { id => $changeid, other => \@other, missing => $pref_br }); } $match = shift @other; $fallback = 1; } my (@better, @equal, @worse); my $closedness = $$match{closedness}; foreach my $gi (@other) { my $cmp = $$gi{closedness} <=> $closedness; if ($cmp < 0) { push @better, $gi; } elsif ($cmp > 0) { push @worse, $gi; } else { push @equal, $gi; } } if ($br_reqd) { # The request was complied with. # Still, inform the user about better alternatives if the # branch comes from the Change, in case something happened # on the server since the source was initially determined. # If the branch was given explicitly, we presume the user # already knows the alternatives (they most probably reacted # to a previous error). $$match{better} = \@better if (!defined($branch)); } else { if (!$fallback) { # We have a match for the upstream branch, so prefer it. if (@better) { # Changes which are on the "wrong" branch but are more # open are equally good candidates, so report ambiguity. return (MAP_AMBIGUOUS, { id => $changeid, other => [ $match, @other ] }); } } else { # No upstream or no matching Change found. # @better is empty, as the fallback is the best option. if (@equal) { # Equally open Changes are equally good candidates, # so report ambiguity. return (MAP_AMBIGUOUS, { id => $changeid, other => [ $match, @other ] }); } } # Use the upstream branch; inform about all alternatives. $$match{other} = \@other; } $$match{cross_branch} = !defined($br) && defined($upstream_branch) && ($$match{branch} ne $upstream_branch); $$match{pick_idx} = select_patchset($match, $$spec{stamp}); push @$fetches, $match; return (MAP_SUCCESS, { id => $changeid, match => $match }); } # Print error message about failure to resolve a source specification, # including all available context. sub print_failure($@) { my ($spec, @results) = @_; my @reports; report_flowed(\@reports, "Resolving $$spec{orig} against '$remote' failed:"); my $any_ambiguous; foreach my $result (@results) { my $ginfo = $$result{match}; if ($ginfo) { report_remote(\@reports, " Using ", "", $ginfo); next; } my $other = $$result{other}; if (!$other) { # This can happen only for local Changes, as otherwise we would # have already errored out. report_local(\@reports, " Missing ", "", $$result{local}); next; } my $miss = $$result{missing}; if (defined($miss)) { report_flowed(\@reports, " Change $$result{id} is not on '$miss'. Possible sources:"); } else { report_flowed(\@reports, " Changes $$result{id} have unclear precedence:"); $any_ambiguous = 1; } foreach my $oth (@$other) { my $oann = $$oth{branch}; my $osts = $$oth{status}; $oann .= '/'.$osts if (is_closed_status($osts)); push @reports, { type => "change", subject => $$oth{subject}, prefix => " ", suffix => " # $oann" }; } } report_flowed(\@reports, " Please use --branch to specify the source.") if ($any_ambiguous); print STDERR format_reports(\@reports); } sub map_insertion_spec($$$) { my ($spec, $fetches, $fails) = @_; # We could support partial Change-Ids here, but that seems # pointless: Ids will be almost inevitably copy-and-pasted # directly from Gerrit, and thus complete. Similarly, # rev-spec suffixes like ~2 are not supported, as we expect # the user to do any necessary traversal on Gerrit anyway. my ($tip, $base) = ($$spec{tip}, $$spec{base}); my $gis = $gerrit_infos_by_id{$tip}; if (!$gis) { werr("Change $tip was not found on '$remote'.\n"); $$fails = 1; return; } if (defined($base)) { my $bgis = $gerrit_infos_by_id{$base}; if (!$bgis) { werr("Change $base was not found on '$remote'.\n"); $$fails = 1; return; } } my ($ret, $rslt) = map_remote_change($tip, undef, $gis, $spec, $fetches); if ($ret != MAP_SUCCESS) { print_failure($spec, $rslt); $$fails = 1; return; } $$spec{tip_result} = $$rslt{match}; } sub map_update_spec($$$) { my ($spec, $fetches, $fails) = @_; my @results; my ($found, $ambiguous) = (0, 0); foreach my $change (@{$$spec{range}}) { if ($$change{gerrit}) { # Matched up series will have at least one Change already mapped. $found = 1; next; } my $gis = $gerrit_infos_by_id{$$change{id}}; if (!$gis) { print "No results for $$change{id}.\n" if ($debug); push @results, { local => $$change{local} }; } else { my ($ret, $rslt) = map_remote_change( $$change{id}, $$change{tgt}, $gis, $spec, $fetches); if ($ret != MAP_MISSING) { $found++; if ($ret == MAP_AMBIGUOUS) { # This can happen only for Changes which were not # previously mapped to remote ones. # It would be possible to verify_commit() against # *all* candidates and complain only when none # match, but that seems too much trouble. $ambiguous++; } else { $$change{gerrit} = $$rslt{match}; } } push @results, $rslt; } } # We permit some Changes being missing, as otherwise refreshing # a series which has new local Changes would be rather annoying. # For deletions it is OK to have no hits at all. Likewise in # --check mode. if (($$spec{action} == UPDATE && !$found && !$check) || $ambiguous) { print_failure($spec, @results); $$fails = 1; } } # Map queried SHA1s to the corresponding remote Change objects. sub map_queries($$$) { my ($queries, $field, $fetches) = @_; foreach my $sha1 (keys %$queries) { my $ginfo = $gerrit_info_by_sha1{$sha1}; if ($ginfo) { $$ginfo{$field} = select_patchset($ginfo, $$queries{$sha1}); $$fetches{$$ginfo{key}} = $ginfo; } } } sub add_patchset($$$) { my ($ginfo, $cat, $idxes) = @_; my $idx = $$ginfo{"${cat}_idx"}; return if (!defined($idx) || defined($$ginfo{"${cat}_commit"})); $$idxes{$idx} = 1; my $revs = $$ginfo{revs}; $$idxes{$idx - 1} = 1 if (($idx > 0) && ($idx == $#$revs) && ($$ginfo{status} eq 'MERGED')); } # Fetch the latest PatchSets for the specified Changes. # We don't re-fetch PatchSets which we already have. sub fetch_patchsets($$;$) { my ($ginfos, $visits, $realm) = @_; my @refs; foreach my $ginfo (@$ginfos) { my %idxes; add_patchset($ginfo, 'pick', \%idxes); add_patchset($ginfo, 'push', \%idxes); my $revs = $$ginfo{revs}; foreach my $idx (sort keys %idxes) { my $rev = $$revs[$idx]; my ($rev_id, $rev_ps) = ($$rev{id}, $$rev{ps}); if (defined($$ginfo{fetched}{$rev_ps})) { print "Already have PatchSet $rev_ps for $$ginfo{id}.\n" if ($debug); $$visits{$rev_id} = 1; } elsif ($commit_by_id{$rev_id}) { # When we are fetching the dependencies of Changes we already fetched, # it is quite likely that we already have the corresponding commits. print "Already have PatchSet $rev_ps for $$ginfo{id} (aliasing fetched).\n" if ($debug); $$ginfo{fetched}{$rev_ps} = $rev_id; } elsif (defined($$visits{$rev_id})) { # We obviously already have the PatchSets we have pushed ourselves, # but we might have only scheduled them for visiting at this point. print "Already have PatchSet $rev_ps for $$ginfo{id} (aliasing pushed).\n" if ($debug); $$ginfo{fetched}{$rev_ps} = $rev_id; } else { push @refs, "+$$rev{ref}:refs/gpush/g$$ginfo{key}_$rev_ps"; $$visits{$rev_id} = 1; } } } if (@refs) { state $fetched_upstream; if (!$fetched_upstream) { # We need to fetch the current upstream to be able to exclude it. # See visit_local_commits() for explanation for fetching all branches. push @refs, "+refs/heads/*:refs/remotes/$remote/*"; } print "Fetching ".($realm // "")."PatchSets ...\n" if (!$quiet); my @gitcmd = ("git", "fetch"); # The git-fetch output is quite noisy and unhelpful here, unless debugging. push @gitcmd, '-q' if (!$debug); push @gitcmd, $gerrit_url, @refs; run_process(FWD_OUTPUT, @gitcmd); if (!$fetched_upstream) { $fetched_upstream = 1; print "Re-enumerating upstream branches ...\n" if ($debug); load_refs("refs/remotes/$remote/"); update_excludes(); } } else { print "No ".($realm // "")."PatchSets need fetching.\n" if ($debug); } } sub check_upstreamed($) { my ($ginfos) = @_; # For merged Changes check whether the local branch excludes the final # commit, i.e., whether it was already pulled. # This specifically ignores the choice of older PatchSets - when the Change # is already in our upstream, it will typically not matter that it is a newer # version - we're still likely to get a conflict, at least a logical one. my %upstream; foreach my $ginfo (@$ginfos) { next if (defined($$ginfo{pending})); if ($$ginfo{status} eq 'MERGED') { $upstream{$$ginfo{revs}[-1]{id}} = $ginfo; $$ginfo{pending} = 0; } else { $$ginfo{pending} = 2; } } if (%upstream) { print "Visiting our upstream ...\n" if ($debug); my $revs = open_process(USE_STDIN | USE_STDOUT | FWD_STDERR, 'git', 'rev-list', '--stdin', '--not', $local_base); write_process($revs, map { "$_\n" } keys %upstream); while (read_process($revs)) { my $ginfo = $upstream{$_}; # The listing includes commits we did not request, because we # cannot use --no-walk due to ^$local_base implying a range. next if (!$ginfo); $$ginfo{pending} = 1; } close_process($revs); } } # Verify that the 2nd commit is a cherry-pick of the 1st one. sub verify_cherrypick($$) { my ($prev, $curr) = @_; return 0 if ("@{$$curr{author}}" ne "@{$$prev{author}}"); my @curr_msg = split(/$/m, $$curr{message}); my @prev_msg = split(/$/m, $$prev{message}); # Cherry-picking may add various footers to the commit message. # Therefore, we verify that the current message is a strict # superset of the previous one. while (@curr_msg && @prev_msg) { my $line = pop @curr_msg; # If this was an added line, we will re-sync at some point. # If it wasn't, unconsumed lines will remain in @prev_msg. pop @prev_msg if ($line eq $prev_msg[-1]); } return 0 if (@curr_msg || @prev_msg); my $tree; my $base = get_1st_parent($curr); if (get_1st_parent($prev) eq $base) { # If the pick didn't rebase, we can just compare the trees. $tree = $$prev{tree}; } else { # Otherwise the diff needs rebasing. # The rebase direction is chosen such that the diff may be re-used # in verify_update(). That's also why we don't clear the diff cache. ($tree, undef) = apply_diff($prev, $base, NUL_STDERR); return 0 if (!defined($tree)); } return 0 if ($tree ne $$curr{tree}); return 1; } sub drop_merged_patchsets($$) { my ($ginfos, $cmt_by_id) = @_; foreach my $ginfo (@$ginfos) { my $revs = $$ginfo{revs}; my $curr_commit = $$cmt_by_id{$$revs[-1]{id}}; next if (!$curr_commit); # The Change doesn't need fixup. my $prev_commit = $commit_by_id{$$revs[-2]{id}}; wfail("Last-but-one PatchSet of Change $$ginfo{key} is ALSO already upstream?!\n") if (!$prev_commit); # Actually seen in the wild. Qt Gerrit bug ... # This can happen for Changes which were not cherry-picked - # for example merges, or commits which were amended and then # direct-pushed. wfail("Last PatchSet of Change $$ginfo{key} is already upstream," ." and is not a cherry-pick of the previous one.\n") if (!verify_cherrypick($prev_commit, $curr_commit)); print "Dropping upstreamed last PatchSet of $$ginfo{id}.\n" if ($debug); pop @$revs; } } # Traverse a series until we hit the upstream. On the way back, # set each visited commit's push base to the found merge base. sub set_base_deduced($); sub set_base_deduced($) { my ($sha1) = @_; my $commit = $commit_by_id{$sha1}; return $sha1 if (!$commit); my $base = $$commit{base}; return $base if (defined($base)); $base = set_base_deduced(get_1st_parent($commit)); $$commit{base} = $base; return $base; } # Traverse a series until we hit the upstream. On the way, # set each visited commit's push base to the named base. sub set_base_imposed($$) { my ($sha1, $base) = @_; while (1) { my $commit = $commit_by_id{$sha1}; last if (!$commit); last if (defined($$commit{base})); $$commit{base} = $base; $sha1 = get_1st_parent($commit); last if ($base eq $sha1); } } sub analyze_patchset($$) { my ($ginfo, $cat) = @_; my $idx = $$ginfo{"${cat}_idx"}; return if (!defined($idx) || defined($$ginfo{"${cat}_commit"})); my $revs = $$ginfo{revs}; $$ginfo{"${cat}_idx"} = --$idx if ($idx == @$revs); my $rev = $$revs[$idx]; my ($rev_id, $base) = ($$rev{id}, $$rev{base}); my $commit = $commit_by_id{$rev_id}; wfail("PatchSet $$rev{ps} of Change $$ginfo{key} is apparently upstream?!\n") if (!$commit); $$ginfo{"${cat}_commit"} = $commit; if (defined($base)) { # Series which are pushed on top of other pending Changes have their base # stored in an extra field on Gerrit, as otherwise it's indeterminable. set_base_imposed($rev_id, $base); print "Have push-base $base for $rev_id ($cat).\n" if ($debug); } else { # For the remaining PatchSets, the merge-base with upstream is the base. $base = set_base_deduced($rev_id); print "Inferred push-base $base for $rev_id ($cat).\n" if ($debug); } } # Extract additional information from the fetched PatchSets. sub analyze_patchsets($) { my ($ginfos) = @_; # Note: this must be done before dropping the merged PatchSets! check_upstreamed($ginfos) if (!$force_merged); print "Analyzing fetched PatchSets ...\n" if ($debug); my @upstream; foreach my $ginfo (@$ginfos) { next if ($$ginfo{status} ne 'MERGED'); my $revs = $$ginfo{revs}; next if (($$ginfo{pick_idx} // -1) != $#$revs && ($$ginfo{push_idx} // -1) != $#$revs); # Gerrit creates a new PatchSet when cherry-picking into the target # branch. Obviously, this commit is excluded by ^@{upstream}. It also # has the wrong parent, making it unsuitable for series traversal. my $rev_id = $$revs[-1]{id}; die("Change $$ginfo{key} is already merged, but not excluded by upstream?!\n") if ($debug && $commit_by_id{$rev_id}); # This can happen for Changes which were not cherry-picked - # for example merges. fail("Change $$ginfo{key} has only one PatchSet, and that is already upstream.\n") if (@$revs == 1); push @upstream, $rev_id; } if (@upstream) { my $commits = visit_commits_raw(\@upstream, ['--no-walk']); my %cmt_by_id = map { $$_{id} => $_ } @$commits; with_local_git_index(\&drop_merged_patchsets, $ginfos, \%cmt_by_id); } foreach my $ginfo (@$ginfos) { analyze_patchset($ginfo, 'pick'); analyze_patchset($ginfo, 'push'); } } # Attach a linear series (segment) to an anchor object within # an object map. This potentially builds a tree. sub attach_series($$$) { my ($series, $anchor, $result) = @_; my $prev = $anchor ? $$anchor{id} : "base"; foreach my $commit (@$series) { push @{$$result{$prev}}, $commit; $prev = $$commit{id}; } } sub assemble_series($$$$$@) { my ($changeid, $sha1, $base, $seen, $callback, @args) = @_; my $commit = $commit_by_id{$sha1}; if (!$commit) { print "$changeid ($sha1) got upstreamed.\n" if ($debug); return []; } my @series; while (1) { if ($seen) { if (defined($$seen{$sha1})) { print "Already seen $changeid ($sha1).\n" if ($debug); last; } $$seen{$sha1} = 1; } print "Adding $changeid ($sha1).\n" if ($debug); unshift @series, $commit; $sha1 = get_1st_parent($commit); if (defined($base) && ($sha1 eq $base)) { print "Hit base.\n" if ($debug); $commit = undef; last; } if (!($commit = $commit_by_id{$sha1})) { print "Hit upstream.\n" if ($debug); last; } my ($nsha1, $stop) = $callback->($commit, \$base, @args); if ($stop) { $commit = undef; last; } $changeid = $$commit{changeid}; if (defined($nsha1) && ($nsha1 ne $sha1)) { # The commit corresponds with an outdated PatchSet. Use the # latest one instead, so we get the current dependencies. print "Upgrading $changeid ($sha1 => $nsha1).\n" if ($debug); $sha1 = $nsha1; $commit = $commit_by_id{$sha1}; } } return (\@series, $commit); } sub advance_pushed_series($$@) { my ($commit, $base, $stamp, $pmap, $bmap, $missing, $fails) = @_; my ($sha1, $changeid) = ($$commit{id}, $$commit{changeid}); my $ginfo = $gerrit_info_by_sha1{$sha1}; # If the Change is in the initial set, the commit is given. my $nsha1 = $$pmap{$changeid}; if (defined($nsha1)) { #print "Deduced $changeid ($sha1) is also initial.\n" if ($debug); if (!$ginfo) { # The outer loop will print a proper error message. print "Deduced+initial Change $changeid ($sha1) disappeared.\n" if ($debug); return (undef, 1); } my $rev = $$ginfo{rev_by_id}{$nsha1}; if (!$rev) { # The outer loop will print a proper error message. print "PatchSet $nsha1 disappeared from deduced+initial $changeid.\n" if ($debug); return (undef, 1); } # The ancestors must be younger. See below. $$stamp = $$rev{ts}; my $nbase = $$rev{base} // $$bmap{$changeid}; $$base = $nbase if (defined($nbase)); return $nsha1; } #print "$changeid ($sha1) is only deduced.\n" if ($debug); # Otherwise we have to guess the commit by timestamp. # We assume that the newest PatchSet which is not newer than # the child is the correct one. An actual push might have been # done on top of older PatchSets, but we assume that this would # have been done only if the grandparent was still correct, and # that is all we care about here. if (!$ginfo || !$$ginfo{push_commit}) { if (!$ginfo && $$commit{did_query}) { werr("Deduced Change $changeid ($sha1) disappeared from '$remote';" ." breaking off pushed series.\n"); return (undef, 1); } # We did not query/fetch the corresponding Change yet. print "Missing $changeid ($sha1).\n" if ($debug); $$missing{$sha1} = $stamp; # We'll retry this later. Continue speculatively, to save roundtrips. $$fails = 1; return undef; } my $rev = $$ginfo{revs}[$$ginfo{push_idx}]; my $nbase = $$rev{base}; $$base = $nbase if (defined($nbase)); return $$rev{id}; } # To a given set of Changes, add any missing Changes which were part of # the previous push from this repository of the Changes in the set. # This essentially builds a remote series, but constrained by the commits # (and their timestamps) that were pushed (or picked). sub complete_pushed_series($$) { my ($spec, $missing) = @_; return if ($$spec{pushed_changes}); print "Completing $$spec{orig}/pushed ...\n" if ($debug); my (%result, %seen, $fails); foreach my $change (reverse @{$$spec{range}}) { my ($sha1, $changeid) = ($$change{pushed}, $$change{id}); if (!defined($sha1)) { # This would happen when completing a local series containing # Changes which were never pushed from this repo. print "Initial $changeid not pushed.\n" if ($debug); next; } my $ginfo = $gerrit_info_by_sha1{$sha1}; if (!$ginfo) { werr("Initial Change $changeid ($sha1) disappeared from '$remote';" ." breaking off pushed series.\n"); next; } my $rev = $$ginfo{rev_by_id}{$sha1}; if (!$rev) { # This can happen if the ginfo was queried by another SHA1 from traversal. werr("PatchSet $sha1 disappeared from Change $changeid on '$remote';" ." breaking off pushed series.\n"); next; } my $stamp = $$rev{ts}; my $base = $$change{base}; my ($series, $anchor) = assemble_series( $changeid, $sha1, $base, \%seen, \&advance_pushed_series, \$stamp, $$spec{pmap}, $$spec{bmap}, $missing, \$fails); attach_series($series, $anchor, \%result) if (!$fails); } $$spec{pushed_changes} = \%result if (!$fails); } sub advance_remote_series($$@) { my ($commit, $base, $stamp, $bmap, $missing, $fails) = @_; my ($sha1, $changeid) = ($$commit{id}, $$commit{changeid}); my $ginfo = $gerrit_info_by_sha1{$sha1}; if (!$ginfo || !$$ginfo{pick_commit}) { if (!$ginfo && $$commit{did_query}) { werr("Deduced Change $changeid ($sha1) disappeared from '$remote';" ." breaking off series.\n"); return (undef, 1); } # We did not query/fetch the corresponding Change yet. print "Missing $changeid ($sha1).\n" if ($debug); $$missing{$sha1} = $stamp; # We'll retry this later. Continue speculatively, to save roundtrips. $$fails = 1; return undef; } my $rev = $$ginfo{revs}[$$ginfo{pick_idx}]; my $nbase = $$rev{base} // $$bmap{$changeid}; $$base = $nbase if (defined($nbase)); return $$rev{id}; } # Assemble a remote series from its tip commit. On the way record required # Changes that we didn't query yet, and fail the operation if there are any. # Additionally, determine the remote group id for the series, which is # needed for deduce_series(); this is the reason why we always traverse down # to the bottom of the series even if we need only a part of it. sub assemble_remote_series($$$$$$$) { my ($ginfo, $base, $stamp, $bmap, $seen, $missing, $fails) = @_; my $rev = $$ginfo{revs}[$$ginfo{pick_idx}]; my $sha1 = $$rev{id}; my $rbase = $$rev{base} // $base; my ($series, $anchor) = assemble_series( $$ginfo{id}, $sha1, $rbase, $seen, \&advance_remote_series, $stamp, $bmap, $missing, $fails); if (!$$fails) { # Only an optimization. # The first Change in the series identifies it. If we got # cut off, get the ID from the already seen parents. my $pgrp = $anchor ? $$anchor{pgrp} : $$series[0]{changeid}; #print "Assigning remote group $pgrp.\n" if ($debug); $$_{pgrp} = $pgrp foreach (@$series); } return ($series, $anchor); } # To a given set of Changes, add any missing Changes which were part of # the latest push of the Changes in the set. sub complete_remote_series($$) { my ($spec, $missing) = @_; return if ($$spec{changes}); print "Completing $$spec{orig}/remote ...\n" if ($debug); my (%result, %seen, $fails); foreach my $change (reverse @{$$spec{range}}) { my $ginfo = $$change{gerrit}; if (!$ginfo) { # This would happen when completing a local series containing # Changes which were never pushed from anywhere. print "Initial $$change{id} not on Gerrit.\n" if ($debug); next; } my ($series, $anchor) = assemble_remote_series( $ginfo, $$change{base}, $$spec{stamp}, $$spec{bmap}, \%seen, $missing, \$fails); attach_series($series, $anchor, \%result) if (!$fails); } $$spec{changes} = \%result if (!$fails); } # Resolve a remote spec into a series. sub resolve_insertion_spec($$) { my ($spec, $missing) = @_; return if ($$spec{changes}); my ($tip, $count, $base, $stamp) = ($$spec{tip_result}, $$spec{count}, $$spec{base}, $$spec{stamp}); # First assemble the entire series. my $fails = 0; my ($series, undef) = assemble_remote_series( $tip, undef, $stamp, {}, undef, $missing, \$fails); wfail("Range $$spec{orig} is empty.\n") if (!@$series); return if ($fails); # Then take the part we actually need. if (defined($count)) { wfail("Range $$spec{orig} extends beyond remote branch.\n") if ($count > @$series); splice @$series, 0, -$count; } elsif (defined($base)) { my $idx = first { $$series[$_]{changeid} eq $base } 0..$#$series; wfail("$base is no ancestor of $$tip{id}.\n") if (!defined($idx)); splice @$series, 0, $idx + 1; } my %result; attach_series($series, undef, \%result); $$spec{changes} = \%result; } sub changify_remote_series($$) { my ($series, $reference) = @_; my $fails; my @result; foreach my $commit (@$series) { my $change = source_map_assign($commit, $reference); $fails = 1 if (!$change); next if ($fails); $$change{gerrit} = $gerrit_info_by_sha1{$$commit{id}}; push @result, $change; } return $fails ? undef : \@result; } sub do_sort_series($$$$$); sub do_sort_series($$$$$) { my ($name, $commits, $parentid, $exclude, $order) = @_; my @ranges; foreach my $commit (@{$$commits{$parentid} // []}) { my $rng = do_sort_series($name, $commits, $$commit{id}, $exclude, $order); my $changeid = $$commit{changeid}; if (!defined($$exclude{$changeid})) { unshift @$rng, $commit; } else { print "Dropping $changeid ($$exclude{$changeid})\n" if ($debug); } push @ranges, $rng if (@$rng); } return [] if (!@ranges); return $ranges[0] if (@ranges == 1); # Whoops - we got a tree instead of a single branch. Starting with a # supposed series 1,2,3,4, this could happen for example like this: # - Re-order it to 2,1,3,4. If just 2,1 is re-pushed, 2 would now have # both 1 and 3 as children. # - Split it by rebasing 3 to base, re-push it. Now base has both 1 and # 3 as children. # - Drop 3 and re-push. Now 2 has both 3 and 4 as children. # We could linearize these arbitrarily, as the order of these sub-series # is obviously irrelevant. However, re-ordering local Changes for no good # reason would be confusing. The solution is hinting linearization with # the local order. my @fails = map { $$_[0] } @ranges; printf("Got multiple children at $parentid: %s\n", join(" ", map { $$_{changeid} } @fails)) if ($debug); my @result; while (1) { # First we sort the ranges according to the local order of their first # elements. Changes which are locally absent get the index -1, which # sorts before others. my $best = (sort { ($$order{$$a[0]{changeid}} // -1) <=> ($$order{$$b[0]{changeid}} // -1) } @ranges)[0]; if (!defined($$order{$$best[0]{changeid}})) { print "$$best[0]{changeid} is not locally present; giving up sorting.\n" if ($debug); last; } # Then move the chosen Change to our result series. push @result, shift @$best; if (!@$best) { # The range ended up empty, so drop it. @ranges = grep { $_ != $best } @ranges; if (@ranges == 1) { # We're left with just one range, so take it in as-is. push @result, @{$ranges[0]}; return \@result; } } } fail(sprintf("Series %s is inconsistent - %s has multiple children:\n%s", $name, substr($parentid, 0, 10), format_commits(\@fails))); } # Turn a set of Changes into a series via topological sorting. # The series may be also filtered during sorting. Doing this # earlier would distort the parent-child relationships. sub sort_series($$$$$) { my ($name, $range, $commits, $current, $drop_closed) = @_; # For the given set of commits, collect the IDs of Changes which # should not be included into the final series after all. my %exclude; my $report = $current && !$quiet; my %selected = map { $$_{id} => 1 } @$range; my @merged; my @levels; foreach my $commit (map { @$_ } values %$commits) { my $changeid = $$commit{changeid}; if (%selected && !defined($selected{$changeid})) { if (defined($changeid2local{$changeid})) { # Drop locally present Changes which are not part of the # (possibly artificially limited) series. This causes partial # updates being matched with only the corresponding fragments # of the remote series. $exclude{$changeid} = 'excluded'; next; } } my $sha1 = $$commit{id}; my $ginfo = $gerrit_info_by_sha1{$sha1}; # Drop Changes that are already in our upstream, unless forced. if (!$force_merged && !$$ginfo{pending}) { # If such a Change is already in the local branch, don't # drop it - it must have been forced previously. if (!defined($selected{$changeid})) { $exclude{$changeid} = 'MERGED'; push @merged, $commit if ($report); next; } } # Drop Changes "more closed" than the "most open" ones in the # series, unless forced. if ($force_closed < 2) { my $status = $$ginfo{status}; my $lvl = ($status eq 'ABANDONED') ? 2 : ($status eq 'DEFERRED') ? 1 : 0; if ($current) { # For remote series, prepare for execution below. push @{$levels[$lvl]}, $commit; } else { # For pushed series, execute immediately, using to the # threshold previously established for the remote series, # so the divergences are minimized. $exclude{$changeid} = $status if ($lvl >= $drop_closed); } } } print "NOT adding already pulled MERGED Change(s) from series $name:\n" .format_commits(\@merged) ."Use --force-merged to add them nonetheless.\n" if (@merged); # The execution part of dropping closed Changes (in remote series). my $have_levels = 0; for my $lvl (0..2) { if (defined($levels[$lvl])) { if ($have_levels > $force_closed) { my $status = ($lvl == 1) ? 'DEFERRED' : 'ABANDONED'; my @closed; foreach my $commit (@{$levels[$lvl]}) { my $changeid = $$commit{changeid}; # If such a Change is already in the local branch, it might be # so because it was forced, or because it was still open at the # time of the previous push/pick. We prefer to follow recent # abandonments, and the user may need to use force again. # So don't examine %$selected here. $exclude{$changeid} = $status; push @closed, $commit if ($report); } if (@closed) { my $twice = ($have_levels > 1) ? " (twice)" : ""; print "NOT adding $status Change(s) from series $name:\n" .format_commits(\@closed) ."Use --force-closed$twice to add them nonetheless.\n"; } $drop_closed //= $lvl; } $have_levels++; } } $drop_closed //= 3; my %order = map { $$_{id} => $$_{index} } @$range; print "Ordering $name ...\n" if ($debug); my $result = do_sort_series($name, $commits, "base", \%exclude, \%order); printf("Result:\n %s\n", join("\n ", map { $$_{changeid} } @$result)) if ($debug); return ($result, $drop_closed); } sub finalize_pushed_series($$$$) { my ($name, $range, $commits, $drop_closed) = @_; # Note that unlike for remote series, it is just fine if the series # turns out to be empty. The most likely reason for that is that # %$commits was empty to start with (i.e., the series was never # gpush'd from this repo). my ($sorted, undef) = sort_series($name, $range, $commits, 0, $drop_closed); return $sorted; } sub finalize_remote_series($$$) { my ($name, $range, $commits) = @_; my ($sorted, $drop_closed) = sort_series($name, $range, $commits, 1, undef); fail("Series $name is empty after filtering.\n") if (!@$sorted && !$check && !$pick_all); return ($sorted, $drop_closed); } # Download the metadata and PatchSets referenced by the specs. sub complete_spec_heads($) { my ($specs) = @_; print "Completing commit specs ...\n" if ($debug); my (%picks, @queried, %visits); foreach my $spec (@$specs) { my $action = $$spec{action}; if ($action == INSERT) { my ($tip, $base) = ($$spec{tip}, $$spec{base}); $picks{$tip} = 1; $picks{$base} = 1 if (defined($base)); } else { foreach my $change (@{$$spec{range}}) { my $changeid = $$change{id}; # We do this for deletions as well, to know which Changes # can be dropped safely. $picks{$changeid} = 1; my $pushed = $$change{pushed}; if (defined($pushed)) { $visits{$pushed} = 1; # To be able to set {did_query} once we have the commits. push @queried, $pushed; } my $orig = $$change{orig}; $visits{$orig} = 1 if (defined($orig)); } } } # When an unspecified local series is matched up, at least one # of its Changes was already queried. Skip these. # Note that this will skip possible alternatives if a Change # was already queried by SHA1, which is fine ... kinda. my @queries = grep { !$gerrit_infos_by_id{$_} } keys %picks; query_gerrit([ map { "change:".$_ } @queries ]) if (@queries); my (@fetches, $any_errors); foreach my $spec (@$specs) { my $action = $$spec{action}; if ($action == INSERT) { map_insertion_spec($spec, \@fetches, \$any_errors); } else { map_update_spec($spec, \@fetches, \$any_errors); } } exit(1) if ($any_errors); fetch_patchsets(\@fetches, \%visits) if (@fetches); print "Visiting fetched and pushed PatchSets ...\n" if ($debug); visit_local_commits([ keys %visits ]); # We queried the pushed commits implicitly (by Change-Id), # so mark them as such. $$_{did_query} = 1 foreach (grep { $_ } map { $commit_by_id{$_} } @queried); analyze_patchsets(\@fetches) if (@fetches); } sub complete_spec_tails($) { my ($specs) = @_; foreach my $spec (@$specs) { next if ($$spec{action} != UPDATE); my (%bmap, %pmap); foreach my $change (@{$$spec{range}}) { my $changeid = $$change{id}; my $base = $$change{base}; $bmap{$changeid} = $base if (defined($base)); next if ($ignore_struct || $force_struct); my $pushed = $$change{pushed}; $pmap{$changeid} = $pushed if (defined($pushed)); } $$spec{bmap} = \%bmap; next if ($ignore_struct || $force_struct); $$spec{pmap} = \%pmap; } # Now that we have the commits for the tips of all remote specifications, # we can resolve them into series as well. # This is iterative, because we need to use the latest PatchSet for each # Change in the series, which is not guaranteed by ancestor traversal: # Suppose series 1,2,3,4. Re-order to 2,1,3,4. Re-push 2,1,3 - 4 can be # omitted, as it has the same diff and the same parent Change. Traversing # the ancestor commit chain of 4 would now a) get outdated PatchSets and # b) produce a different order than the current PatchSets of the series. while (1) { print "Resolving remote commit specs ...\n" if ($debug); # This loop tries to resolve spec objects, and records which # additional Changes need to be fetched from Gerrit. # It also collects remote series corresponding with local ones. my (%picks, %pushes); foreach my $spec (@$specs) { my $action = $$spec{action}; if ($action == INSERT) { resolve_insertion_spec($spec, \%picks); } elsif ($action == UPDATE) { # This is a local spec. We need to obtain the corresponding remote # series. It might have a different structure including a different # tip, so just try to collect all Changes belonging to it and sort # them later. We will still miss Changes which were added at the # end - the only fix for that would be making use of Gerrit's reverse # dependencies. The workaround is specifying these Changes manually. complete_remote_series($spec, \%picks); # Note that we cannot skip the above even in --ignore-struct mode, # as it is necessary for deduce_series(). next if ($ignore_struct || $force_struct); # Prepare reconstruction of the series' last push from this repo. complete_pushed_series($spec, \%pushes); } } # Termination always happens at some point, as cycles are # impossible between the latest PatchSets. last if (!%picks && !%pushes); # Deduced Changes are queried by SHA1, as that is unambiguous. # Don't re-query Changes which were previously queried (but not fetched). my %queries = map { $_ => 1 } grep { !defined($gerrit_info_by_sha1{$_}) } (keys %picks, keys %pushes); my @sha1s = keys %queries; query_gerrit([ map { "commit:".$_ } @sha1s ]) if (@sha1s); my %good_picks; map_queries(\%picks, 'pick_idx', \%good_picks); map_queries(\%pushes, 'push_idx', \%good_picks); my @fetches = values %good_picks; my %visits; fetch_patchsets(\@fetches, \%visits, "deduced ") if (@fetches); print "Visiting deduced fetched and pushed PatchSets ...\n" if ($debug); visit_local_commits([ keys %visits ]); $$_{did_query} = 1 foreach (grep { $_ } map { $commit_by_id{$_} } @sha1s); analyze_patchsets(\@fetches) if (@fetches); } } sub changify_remote_specs($) { my ($specs) = @_; while (1) { foreach my $spec (@$specs) { my $commits = $$spec{new_range_commits}; next if (!$commits); my $changes = changify_remote_series($commits, $$spec{new_range_reference}); if ($changes) { $$spec{new_range} = $changes; delete $$spec{new_range_commits}; } } last if (!source_map_traverse()); } source_map_finish(); } sub finalize_insertion_specs($) { my ($specs) = @_; # We collected the remote Changes belonging to a series. Now sort # them according to the dependencies of their current PatchSets. foreach my $spec (@$specs) { # We don't finalize remote series from local specs yet, because # the step may be obsoleted by the spec being matched up. next if ($$spec{action} != INSERT); ($$spec{new_range_commits}, undef) = finalize_remote_series($$spec{orig}, [], $$spec{changes}); $$spec{new_range_reference} = $$spec{tip}; } changify_remote_specs($specs); } sub check_specs($) { my ($specs) = @_; print "Checking remote commit specs for collisions ...\n" if ($debug); my @new_specs; foreach my $spec (@$specs) { my $action = $$spec{action}; if ($action == INSERT) { my $new_range = $$spec{new_range}; # Try to match up the remote spec with an explicitly specified local one. my $lspec; # The matched up local spec, if any. my %lchanges; # The Changes in the matched up local series, if any. foreach my $change (@$new_range) { my $ospec = $$change{rspec}; # Intersecting another remote spec is an error. wfail("Specs $$spec{orig} and $$ospec{orig} intersect (change $$change{id}).\n") if ($ospec); $$change{rspec} = $spec; # Check that the Change already has a local spec assigned. $ospec = $$change{lspec}; next if (!$ospec); # We don't match up deletions - re-adding Changes which are being deleted # allows bypassing various checks, which can be useful when the situation # is too complex for the script. next if ($$ospec{action} == DELETE); if ($lspec) { # We are already matched up with a local spec. # Verify that the Change belongs to it. If it doesn't, # the remote spec matches multiple local ones. wfail("Spec $$spec{orig} matches both $$lspec{orig} and $$ospec{orig}.\n" ."Try picking only parts of the series.\n") if ($ospec != $lspec); } else { my $mspec = $$ospec{match}; # Complain if the matched up local spec matches multiple remote ones. wfail("Spec $$ospec{orig} matches both $$mspec{orig} and $$spec{orig}.\n" ."Try picking only parts of the series.\n") if ($mspec); printf("Spec %s matched up with %s (change %s).\n", $$spec{orig}, $$ospec{orig}, $$change{id}) if ($debug); # Merge the relevant parts of the local spec into this one. my $parent = $$ospec{parent}; if (defined($parent)) { wfail("Spec $$spec{orig} and $$ospec{orig} match," ." but have different parents.\n") if (defined($$spec{parent}) && ($parent ne $$spec{parent})); $$spec{parent} = $parent; } my $range = $$ospec{range}; $$spec{range} = $range; $$ospec{match} = $spec; # This remote spec is now an update. $$spec{action} = UPDATE; # Mark the local spec as dead. $$ospec{action} = HOLD; # Remember that we were successful. $lspec = $ospec; %lchanges = map { $$_{id} => 1 } @$range; } # The Change doesn't belong to a local spec any more. delete $$change{lspec}; } # Try to match up the remote spec with an unspecified local series. my $llabel; # The label of the matched up unspecified local series, if any. # Walk backwards, to make tip matches more likely, for less noisy output. foreach my $change (reverse @$new_range) { # Check that the Change has no local spec assigned - updates would # have been matched above, and deletions are not matched here, either. next if ($$change{lspec}); # Check that the Change actually exists locally to start with. next if (!defined($$change{local})); # Check that we don't own it yet. my $changeid = $$change{id}; next if (defined($lchanges{$changeid})); # Determine the local series it belongs to. # FIXME: Even a partial remote series will match the entire local # series, so local Changes will be incorrectly (attempted to be) # dropped. We could fix this for ancestor Changes, but not for # descendants, as long as we don't extract reverse dependencies # from Gerrit. # For the time being, explicitly specify also the partial local # series when specifying a partial remote series, so the previous # loop catches the case. my ($nrange, $gid, undef, undef) = do_determine_series($change); # We deduced the series for all local Changes that have remote # counterparts, so this Cannot Fail (TM). die("inconsistent operation") if ($debug && !defined($gid)); # Calculate a descriptive label: :, and the matched # Change if it is not the tip. my $nchangeid = $$nrange[-1]{id}; my $nlabel = $nchangeid.":".int(@$nrange); $nlabel .= " (change $changeid)" if ($changeid ne $nchangeid); if (%lchanges) { # We already matched up a local series (specified or not). # This is an error, as this Change does not belong to it. wfail("Spec $$spec{orig} matches both $$lspec{orig} and $nlabel.\n" ."Try picking only parts of the series.\n") if ($lspec); wfail("Spec $$spec{orig} matches both $llabel and $nlabel.\n" ."Try picking only parts of the series.\n"); } print "Spec $$spec{orig} matched up with $nlabel.\n" if ($debug); # Assign the local series to the remote spec. $$spec{range} = $nrange; # This remote spec is now an update ... $$spec{action} = UPDATE; # ... and needs re-visiting. push @new_specs, $spec; # Remember that we were successful. $llabel = $nlabel; %lchanges = map { $$_{id} => 1 } @$nrange; } } elsif ($action == DELETE) { foreach my $change (@{$$spec{range}}) { # Deletions may not intersect preceding remote specs, # as that would make quite a mess. my $ospec = $$change{rspec}; wfail("Spec $$spec{orig} intersects preceding $$ospec{orig}" ." (change $$change{id}).\n") if ($ospec); } } } # Matched up unspecified local series need to have their remaining # Changes mapped and their previous push reconstructed. if (@new_specs) { complete_spec_heads(\@new_specs); complete_spec_tails(\@new_specs); } } sub finalize_specs($) { my ($specs) = @_; print "Finalizing commit update specs ...\n" if ($debug); return if ($ignore_struct); # Remote series from unmatched local specs were not finalized yet, so do it now. foreach my $spec (@$specs) { next if ($$spec{action} != UPDATE); next if ($$spec{new_range}); ($$spec{new_range_commits}, $$spec{drop_closed}) = finalize_remote_series("$$spec{orig}/remote", $$spec{range}, $$spec{changes}); $$spec{new_range_reference} = $$spec{range}[-1]; } changify_remote_specs($specs); return if ($force_struct); my $errors; foreach my $spec (@$specs) { # This is applicable only to updates. next if ($$spec{action} != UPDATE); my $range = $$spec{range}; my $new_range = $$spec{new_range}; # Note that these are commits, unlike the other two, which are Changes. my $pushed = finalize_pushed_series("$$spec{orig}/pushed", $range, $$spec{pushed_changes}, $$spec{drop_closed}); # Find the first divergences between the structures of the previous push # and the current local and remote series. my ($ldiff, $rdiff) = (-1, -1); for (my $i = 0; ; $i++) { my $lc = $i < @$range ? $$range[$i]{id} : ""; my $rc = $i < @$new_range ? $$new_range[$i]{id} : ""; last if (!length($lc) && !length($rc)); if ($lc eq $rc) { # If both sides are equal, it does not matter if they changed. } else { my $pc = $i < @$pushed ? $$pushed[$i]{changeid} : ""; $ldiff = $i if ($ldiff < 0 && ($lc ne $pc)); $rdiff = $i if ($rdiff < 0 && ($rc ne $pc)); last if ($ldiff >= 0 && $rdiff >= 0); } } if ($ldiff < 0) { if ($rdiff < 0) { print "Keeping series structure (is identical).\n" if ($debug); } else { # Local structure didn't change, so just use whatever we get from the remote. print "Using remote series structure.\n" if ($debug); } } elsif ($rdiff < 0) { # Remote structure didn't change, but local did. This is a weird situation, # as we are pulling Changes from the remote by definition. Use the local # structure and hope for the best. print "Using local series structure.\n" if ($debug); $$spec{new_range} = $range; } else { # If the local and remote series diverge from the pushed series # at the same point, we need to print it only once. my $locpush = ($ldiff == $rdiff) ? "" : sprintf( " Pushed [%d]: %s\n", $ldiff, $ldiff < @$pushed ? format_commit($$pushed[$ldiff], -15) : "none"); $errors .= sprintf( "Structure of series %s diverged:\n" ." Local [%d]: %s\n%s Remote [%d]: %s\n Pushed [%d]: %s\n", $$spec{orig}, $ldiff, $ldiff < @$range ? format_commit($$range[$ldiff]{local}, -15) : "none", $locpush, $rdiff, $rdiff < @$new_range ? format_commit($$new_range[$rdiff]{gerrit}{pick_commit}, -15) : "none", $rdiff, $rdiff < @$pushed ? format_commit($$pushed[$rdiff], -15) : "none"); } } if ($errors) { $errors .= "Use --force-struct/--ignore-struct, or try picking only parts of the series.\n"; fail($errors) if (!$check); print STDERR $errors."\n"; } } sub prepare_specs($) { my ($raw_changes) = @_; my $specs = parse_specs(\@commit_specs); resolve_specs($specs); complete_spec_heads($specs); complete_spec_tails($specs); finalize_insertion_specs($specs); # Before we can match up remote specs with unspecified local # series, we need to make sure that the series are actually # assigned - we cannot simply sweep up all loose Changes around # the matching ones, because we may catch unrelated ones, which # structural change tracking would make a mess of. # This does not apply to --check mode, as it does not handle remote # specs to start with. However, as it skips the post-adjust series # deduction, the adjusted Change list is not committed. So instead, # do it here before any adjustments are made. deduce_series($raw_changes); check_specs($specs); finalize_specs($specs); return $specs; } sub generate_specs($) { my ($change) = @_; print "Generating specs ...\n" if ($debug); my @specs; while (1) { my $range; ($range, undef, $change, undef) = do_determine_series($change, 0, 0, 1); my $cs = $$range[-1]{id}.":".int(@$range); print "Got $cs.\n" if ($debug); push @specs, { orig => $cs, action => UPDATE, range => $range, new_range => $ignore_struct ? $range : undef }; last if (!$change); } return \@specs; } sub prepare_specs_all($) { my ($raw_changes) = @_; if (!@$raw_changes) { print "No local Changes.\n" if (!$quiet); exit(0); } # Before we can iterate the local series, we need to make sure that # they are actually assigned. For this to work we first resolve a # fake spec which causes the corresponding remote series for all # local ones to be assembled. my $fake_specs = parse_specs([ '..' ]); resolve_specs($fake_specs); complete_spec_heads($fake_specs); complete_spec_tails($fake_specs); deduce_series($raw_changes, 1); my $specs = generate_specs($$raw_changes[-1]); # The fake spec already did the "heads" for us. complete_spec_tails($specs); finalize_specs($specs); return $specs; } # Return a list of actions that construct the second range of Changes # with elements from the first one as naturally as possible. sub transmogrify_series($$$) { my ($range, $new_range, $pairs) = @_; # The layout of the new range is given, so placing UPDATEs and # INSERTs is trivial. The challenge is putting each DELETE in a # place that makes sense, rather than dumping them all at the end. # This only becomes "interesting" when the range is permutated in # addition to gaining and losing Changes. $$_{not_new} = 1 foreach (@$range); $$new_range[$_]{new_idx} = $_ for (0 .. $#$new_range); # First score common sequences - the longer, the higher the score. my @seq; foreach my $chg (@$range) { my $ni = $$chg{new_idx}; next if (!defined($ni)); # Skip over deleted. if (@seq) { my $pi = $seq[-1]{new_idx}; # Skip over added. while (++$pi < $ni && !defined($$new_range[$pi]{not_new})) {} if ($pi != $ni) { $$_{score} = int(@seq) foreach (@seq); @seq = (); } } push @seq, $chg; } $$_{score} = int(@seq) foreach (@seq); my $prev; my $del = []; foreach my $chg (@$range) { my $ni = $$chg{new_idx}; if (!defined($ni)) { push @$del, [ DELETE, $chg ]; } else { if (@$del) { # "Attach" the sequence of deletions to the Change in front of or # after it, depending on which one has the higher score. In the # trivial case both are part of the same sequence (no permutation), # so the scores are equal and the choice is irrelevant. if ($prev && $$prev{score} < $$chg{score}) { # Instead of actually prepending to the next Change, append to the # first non-addition in front of it in the new range. This ensures # that deletions always come before additions. while (1) { $prev = undef; last if (--$ni < 0); $prev = $$new_range[$ni]; last if (defined($$prev{not_new})); } } if (!$prev) { push @$pairs, @$del; } else { $$prev{del_tail} = $del; } $del = []; } $prev = $chg; } } if (@$del) { if (!$prev) { push @$pairs, @$del; } else { $$prev{del_tail} = $del; } } push @$pairs, map { [ defined($$_{not_new}) ? UPDATE : INSERT, $_ ], @{$$_{del_tail} // []} } @$new_range; } # Modify the list of local Changes as requested, and mark the Changes # that should be updated. sub apply_specs($$) { my ($specs, $raw_changes) = @_; print "Applying specs ...\n" if ($debug); # We do not store the action inside the Change itself, as a Change # may be DELETEd and INSERTed in the same run, thus requiring # positional assignment of multiple actions. my @pairs = map { [ HOLD, $_ ] } @$raw_changes; my $idx; foreach my $spec (@$specs) { my $action = $$spec{action}; next if ($action == HOLD); # Was absorbed by remote spec my $parent = $$spec{parent}; if (defined($parent)) { if ($parent eq '*') { $idx = 0; } elsif ($parent eq '@') { # Use the index right after the previously manipulated Changes. } else { $idx = first { $pairs[$_][1]{id} eq $parent } 0..$#pairs; die("Lost track of specified parent $parent.\n") if (!defined($idx)); # Insisting on unmanipulated Changes avoids ambiguity in case # of DELETE+INSERT of the same Change. Also, it enforces the # more expressive and less verbose suffix notation. wfail("Specified parent $parent is being manipulated." ." Use suffix-\@ notation instead.\n") if ($pairs[$idx][0] != HOLD); $idx++; } } else { $idx = undef; } my $new_range = $$spec{new_range}; my @new_pairs; if ($action == INSERT) { $idx = @pairs if (!defined($idx)); printf("Adding +%s:%d at %d.\n", $$new_range[-1]{id}, int(@$new_range), $idx) if ($debug); @new_pairs = map { [ INSERT, $_ ] } @$new_range; } else { my $range = $$spec{range}; my $rmidx = first { $pairs[$_][1] == $$range[0] } 0..$#pairs; for my $tidx (1..$#$range) { # An insertion does not overlap existing Changes, so it's not caught earlier. fail("Inconsistent operation - insertion within updated or deleted range.\n") if ($pairs[$rmidx + $tidx][1] != $$range[$tidx]); } if ($action == DELETE) { printf("Deleting %s:%d at %d.\n", $$range[-1]{id}, int(@$range), $rmidx) if ($debug); $idx = $rmidx; $pairs[$idx++][0] = DELETE for (0..$#$range); } else { if (defined($idx) && ($idx ne $rmidx)) { printf("Moving %s:%d from %d to %d.\n", $$range[-1]{id}, int(@$range), $rmidx, $idx) if ($debug); if ($idx >= $rmidx) { wfail("Parent ".$pairs[$idx][1]{id}." lies within updated range" ." ($$spec{orig}).\n") if ($idx < $rmidx + @$range); $idx -= @$range; } } else { printf("Updating %s:%d at %d.\n", $$range[-1]{id}, int(@$range), $rmidx) if ($debug); $idx = $rmidx; } splice @pairs, $rmidx, @$range; transmogrify_series($range, $new_range, \@new_pairs); } } splice @pairs, $idx, 0, @new_pairs; $idx += @new_pairs; } return \@pairs; } sub can_recycle($$$) { my ($commit, $orig_id, $pushed_commit) = @_; return 0 if (!$pushed_commit || !defined($orig_id)); return 1 if ($$commit{id} eq $orig_id); # If we have no direct hit, we may still get lucky with # the commit's contents. First compare the trees themselves ... my $orig_commit = $commit_by_id{$orig_id}; return 0 if (!$orig_commit || ($$commit{tree} ne $$orig_commit{tree})); # ... and then also the base trees, as otherwise we would miss # it if commits were split or joined. my $parent_tree = get_1st_parent_tree($commit); return 0 if ($parent_tree ne get_1st_parent_tree($orig_commit)); # Finally compare the meta data. return 1 if (commit_metas_equal($commit, $orig_commit) && (get_more_parents($commit) eq get_more_parents($orig_commit))); # While the commit is not the same, the diff is, so pre-populate # the tree cache for later comparisons. my $pushed_parent_tree = get_1st_parent_tree($pushed_commit); my $ck = $pushed_parent_tree.$parent_tree.$$commit{tree}; $commit2diff{$ck} = $$pushed_commit{tree}; return 0; } sub format_parts($;$) { my ($parts, $masked) = @_; return "-" if (!$parts); my @out; for my $i (0..3) { my $b = 1 << $i; next if (!($parts & $b)); my $p = ('auth', 'msg', 'diff', 'merge')[$i]; $p = '('.$p.')' if (($masked // 0) & $b); push @out, $p; } return join(",", @out); } # Determine which parts of a commit differ from a reference commit. sub verify_commit($$) { my ($commit, $ref_commit) = @_; return 0 if ($commit == $ref_commit); my ($tree, $ref_tree) = ($$commit{tree}, $$ref_commit{tree}); my ($base_tree, $ref_base_tree) = (get_1st_parent_tree($commit), get_1st_parent_tree($ref_commit)); if ($base_tree eq $ref_base_tree) { # We got lucky, no need to rebase. print "No rebase.\n" if ($debug); } else { # First see if we have a cached reverse rebase. my $rck = $base_tree.$ref_base_tree.$ref_tree; my $rct = $commit2diff{$rck}; if (defined($rct)) { print "Have cached reverse rebase.\n" if ($debug); $ref_tree = $rct; } else { # Then see if we have a cached forward rebase. my $ck = $ref_base_tree.$base_tree.$tree; my $ct = $commit2diff{$ck}; if (defined($ct)) { print "Have cached rebase.\n" if ($debug); $tree = $ct; } else { if (defined($commit2diff{$$ref_commit{id}})) { print "Have cached reverse diff.\n" if ($debug); # If we have a cached reverse diff, it's more efficient to do a # reverse cherry-pick on top of the target commit's parent. ($ref_tree, undef) = apply_diff( $ref_commit, get_1st_parent($commit), NUL_STDERR); $ref_tree = 'CONFLICT-'.$rck if (!defined($ref_tree)); $commit2diff{$rck} = $ref_tree; } else { # Else do a cherry-pick on top of the reference commit's parent. ($tree, undef) = apply_diff( $commit, get_1st_parent($ref_commit), NUL_STDERR); $tree = 'CONFLICT-'.$ck if (!defined($tree)); # ... and cache it for later. # Yes, we re-use the diff cache for that. $commit2diff{$ck} = $tree; } } } } my $parts = 0; $parts |= AUTHOR_PART if ("@{$$commit{author}}" ne "@{$$ref_commit{author}}"); $parts |= MESSAGE_PART if ($$commit{message} ne $$ref_commit{message}); $parts |= DIFF_PART if ($tree ne $ref_tree); $parts |= MERGE_PART if (get_more_parents($commit) ne get_more_parents($ref_commit)); return $parts; } sub verify_local_commit($$$;$) { my ($lcl_commit, $rmt_commit, $rmt_ps, $mask) = @_; my $parts = verify_commit($lcl_commit, $rmt_commit); printf("Verified local %s against remote %s (PS%s) on top of %s: %s%s\n", format_id($$lcl_commit{id}), format_id($$rmt_commit{id}), $rmt_ps, format_id(get_1st_parent($rmt_commit)), format_parts($parts), defined($mask) ? " (eff ".format_parts($parts & $mask).")" : "") if ($debug); return $parts & ($mask // ALL_PARTS); } sub verify_remote_commit($$$$$) { my ($rmt_commit, $rmt_ps, $pushed_commit, $pushed_ps, $mask) = @_; my $parts = verify_commit($rmt_commit, $pushed_commit); printf("Verified remote %s (PS%d) against remote %s (PS%s) on top of %s: %s (eff %s)\n", format_id($$rmt_commit{id}), $rmt_ps, format_id($$pushed_commit{id}), $pushed_ps, format_id(get_1st_parent($pushed_commit)), format_parts($parts), format_parts($parts & $mask)) if ($debug); return $parts & $mask; } use constant { UPD_NOOP => 0, UPD_PUSHED => 1, UPD_META => 2, UPD_INFO => 4, UPD_DROP => 8 }; sub format_upd_cmd($) { my ($cmd) = @_; return "NOOP" if (!$cmd); return join(",", map { ('PUSHED', 'META', 'INFO', 'DROP')[$_] } grep { $cmd & (1 << $_) } 0..3); } sub verify_delete($$$$$$$) { my ($reports, $lcl_commit, $rmt_commit, $rmt_ps, $pushed_commit, $pushed_ps, $need_force_del) = @_; my ($pfx, $annot, $forcing) = (undef, "", 0); # The current local commit can be safely discarded # if it matches a PatchSet on Gerrit. # TODO: Compare against all PatchSets of the Change, # not just the current and the previously pushed one. my $parts = verify_local_commit($lcl_commit, $rmt_commit, $rmt_ps); if ($parts) { if ($pushed_commit && ($pushed_ps ne "?")) { # The PatchSet must be actually still present. $parts = verify_local_commit($lcl_commit, $pushed_commit, $pushed_ps); if ($parts) { $annot = format_parts($parts)." PS$pushed_ps"; $forcing = 1; } } else { # We have no previously pushed commit to compare against. $annot = "PSx:".format_parts($parts); $forcing = 1; } } if ($forcing && !$force) { $$need_force_del = 1; $pfx = "(DROP) "; return (UPD_INFO, $pfx, $annot); } $pfx = $forcing ? "DROP " : "Drop " if (!$quiet); if ($check) { # Can happen only as a result of a structural change. # Cannot be a forced drop in --check mode, so update state. return (UPD_PUSHED | UPD_INFO, $pfx, $annot); } return (UPD_DROP, $pfx, $annot); } sub verify_update($$$$$$$$) { my ($reports, $lcl_commit, $rmt_commit, $rmt_ps, $pushed_commit, $pushed_ps, $need_force_repl, $can_merge) = @_; my ($pfx, $annot, $forcing) = (undef, "", 0); # First make a direct comparison. This should be the common case (the user # is expected to push out updates timely), and also covers bootstrapping. my $parts = verify_local_commit($lcl_commit, $rmt_commit, $rmt_ps); if (!$parts) { # The current local and remote commits are identical, so picking # would be a no-op. # It is possible that the commit was pushed/picked without gpush/gpick, # and therefore the record of the previously pushed commit is outdated # or missing. Therefore, we record the remote SHA1 to reflect the fact # that we are actually up-to-date, but use the local commit to avoid # pointless rewriting. $pfx = "Keep " if (!$quiet); return (UPD_PUSHED | UPD_META | UPD_INFO, NO_PARTS, $pfx, $annot); } my $eff_parts = $parts & $pick_parts; if ($pushed_commit) { # Do a primitive 3-way diff with the previously pushed commit. # Note that we mask away modifications which are identical on # both sides. This might be mildly confusing, but otoh it is # consistent with the equality shortcut above, and knowing # about non-differences is not too useful anyway. my $rmt_parts = verify_remote_commit($rmt_commit, $rmt_ps, $pushed_commit, $pushed_ps, $parts); if (!$rmt_parts) { # As the remote commit equals the previously pushed one, but the local # and remote ones differ, we know that the local one was modified. ($pfx, $annot) = ("Keep ", format_parts($parts)." PS$pushed_ps") if (!$quiet); return (UPD_META | UPD_INFO, NO_PARTS, $pfx, $annot); } # The current remote commit differs from the previously pushed one. if ($ignore) { ($pfx, $annot) = ("Keep ", "PS$pushed_ps ".format_parts($rmt_parts)." (IGNORED)") if (!$quiet); # Do not update the base when ignoring. The presumption # is that the push was wholly bogus and should be reverted # including the base, so inter-diffs can skip over it. And # if we are actually seeing a bona fide conflict, the base # was probably not modified anyway. # INFO, as the source is "applied" even if the update is ignored. return (UPD_PUSHED | UPD_INFO, NO_PARTS, $pfx, $annot); } my $eff_rmt_parts = $rmt_parts & $pick_parts; my $lcl_parts = verify_local_commit($lcl_commit, $pushed_commit, $pushed_ps, $parts); my $fl = $lcl_parts & ~$eff_rmt_parts; my $fc = $lcl_parts & $eff_rmt_parts; my $fr = ~($lcl_parts & $pick_parts) & $rmt_parts; $annot = ($fl ? format_parts($fl)." " : "") ."PS".$pushed_ps .($fc ? ":".uc(format_parts($fc)) : "") .($fr ? " ".format_parts($fr, ~$pick_parts) : ""); if (!$eff_rmt_parts) { # The remote modifications are masked out. $pfx = "Keep " if (!$quiet); return (UPD_INFO, NO_PARTS, $pfx, $annot) if ($check); # Forget about the modifications, as per documentation. The # alternative of requiring an explicit --ignore (possibly in # a separate run) is conceivable, but does not appear to add # value - what else would one want to do after a partial pick? my $upd = ($pick_parts & DIFF_PART) ? UPD_META : 0; return (UPD_PUSHED | UPD_INFO | $upd, NO_PARTS, $pfx, $annot); } if ($lcl_parts) { # The current local commit differs from the previously pushed one, # and we already know that the current remote commit also differs. # However, individual parts may be unmodified on either side. if (!$fc && $merge) { # No textual conflict, as each side modified different parts. # We do not merge without being asked to, as the modifications # might still conflict logically. $pfx = "Merge " if (!$quiet); return (UPD_INFO, NO_PARTS, $pfx, $annot) if ($check); return (UPD_PUSHED | UPD_META | UPD_INFO, $eff_rmt_parts, $pfx, $annot); } if (!$force) { my $pfx; if (!$fc) { $$can_merge = 1; $pfx = "(Merge) "; } elsif ($fl) { $$can_merge = 1 if (!$merge); $$need_force_repl = 1; $pfx = "(MERGE) "; } else { # TODO: We could still attempt 3-way merges here. $$need_force_repl = 1; $pfx = "(REPLACE) "; } return (UPD_INFO, NO_PARTS, $pfx, $annot); } if ($fl && $merge) { # Retain the non-conflicting local modifications. $pfx = "MERGE " if (!$quiet); return (UPD_PUSHED | UPD_META | UPD_INFO, $eff_rmt_parts, $pfx, $annot); } $forcing = 1; } elsif ($pushed_ps eq "?") { # The previously pushed commit disappeared from Gerrit. # Treat it the same as if it had never been pushed. $forcing = 1; } } else { # We have no previously pushed commit to compare against. if ($ignore) { ($pfx, $annot) = ("Keep ", "PSx:".format_parts($parts)." (IGNORED)") if (!$quiet); return (UPD_PUSHED | UPD_INFO, NO_PARTS, $pfx, $annot); } $annot = "PSx:".format_parts($parts, ~$pick_parts); if (!$eff_parts) { # The differing parts are masked out. $pfx = "Keep " if (!$quiet); return (UPD_INFO, NO_PARTS, $pfx, $annot) if ($check); my $upd = ($pick_parts & DIFF_PART) ? UPD_META : 0; return (UPD_PUSHED | UPD_INFO | $upd, NO_PARTS, $pfx, $annot); } $forcing = 1; } if ($forcing && !$force) { $$need_force_repl = 1; $pfx = "(REPLACE) "; return (UPD_INFO, NO_PARTS, $pfx, $annot); } if ($parts & ~$pick_parts) { # The commits differ also in parts which are not being picked. $pfx = $forcing ? "P. REPL. " : "P. Repl. " if (!$quiet); return (UPD_INFO, NO_PARTS, $pfx, $annot) if ($check); return (UPD_PUSHED | UPD_META | UPD_INFO, $eff_parts, $pfx, $annot); } $pfx = $forcing ? "REPLACE " : "Replace " if (!$quiet); return (UPD_INFO, NO_PARTS, $pfx, $annot) if ($check); return (UPD_PUSHED | UPD_META | UPD_INFO, ALL_PARTS, $pfx, $annot); } # Create a new commit by replacing named parts of the # source commit with these from the reference commit. sub merge_commits($$$) { my ($commit, $ref_commit, $parts) = @_; my ($parents, $tree, $message, $author, $committer) = ($$commit{parents}, $$commit{tree}, $$commit{message}, $$commit{author}, $$commit{committer}); if ($parts & AUTHOR_PART) { $author = $$ref_commit{author}; } if ($parts & MESSAGE_PART) { # This is probably the most common case: commit message edited # directly on Gerrit, while amending the diff locally. $message = $$ref_commit{message}; } if ($parts & DIFF_PART) { # To get the correct diff, we need to replace both the parent(s) # and the tree. The implicit rebasing does not matter, as the # result is going to be cherry-picked anyway. $parents = $$ref_commit{parents}; $tree = $$ref_commit{tree}; } return create_commit($parents, $tree, $message, $author, $committer); } # Turn the list of local Changes into the final list of commits the # local branch should be reconstructed from. sub do_adjust_changes($) { my ($pairs) = @_; print "Adjusting Changes ...\n" if ($debug); my ($need_force_del, $need_force_repl, $can_merge); my ($any_missing, $any_crossed, $any_merged); my (@commits, @changes, @reports); foreach my $pair (@$pairs) { my ($action, $change) = @$pair; my $ginfo = $$change{gerrit}; my $rmt_commit = $ginfo && $$ginfo{pick_commit}; my $pushed_id = $$change{pushed}; my $lcl_commit = $$change{local}; my ($upd, $parts) = (UPD_NOOP, NO_PARTS); if ($action == HOLD) { report_local(\@reports, "Hold ", "", $lcl_commit) if ($debug); } elsif ($action == INSERT) { report_remote(\@reports, "Add ", "", $ginfo) if (!$quiet); if ($check) { # Can happen only as a result of a structural change. $upd = UPD_DROP | UPD_INFO; } else { ($upd, $parts) = (UPD_PUSHED | UPD_META | UPD_INFO, ALL_PARTS); } } elsif (!$ginfo) { my $sts = "NEW"; if (defined($pushed_id)) { $sts = "ORPHANED"; $any_missing = 1; } if ($action == DELETE) { if ($force) { report_local(\@reports, "DROP ", $sts, $lcl_commit) if (!$quiet); $upd = UPD_DROP; } else { report_local(\@reports, "(DROP) ", $sts, $lcl_commit); $need_force_del = 1; } } else { report_local(\@reports, "Keep ", $sts, $lcl_commit) if (!$quiet); } } else { my $rmt_ps = $$ginfo{rev_by_id}{$$rmt_commit{id}}{ps}; my ($pushed_commit, $pushed_ps); if (defined($pushed_id)) { $pushed_commit = $commit_by_id{$pushed_id}; my $pushed_rev = $$ginfo{rev_by_id}{$pushed_id}; $pushed_ps = $pushed_rev ? $$pushed_rev{ps} : "?"; } my $commit = $lcl_commit; if (can_recycle($commit, $$change{orig}, $pushed_commit)) { # If we are testing a commit which did not change since the last time # it was pushed, we can just use the rebased pushed commit instead. # This makes some of the verify_commit() calls very cheap. printf("Recycling for verification (local %s => pushed %s).\n", format_id($$commit{id}), format_id($pushed_id)) if ($debug); $commit = $pushed_commit; } my ($pfx, $annot); if ($action == DELETE) { ($upd, $pfx, $annot) = verify_delete(\@reports, $commit, $rmt_commit, $rmt_ps, $pushed_commit, $pushed_ps, \$need_force_del); } else { ($upd, $parts, $pfx, $annot) = verify_update(\@reports, $commit, $rmt_commit, $rmt_ps, $pushed_commit, $pushed_ps, \$need_force_repl, \$can_merge); } report_update(\@reports, $pfx, $annot, $lcl_commit, $ginfo) if (defined($pfx)); # Note: we don't clear the diff cache here, so entries filled # by verify_cherrypick() remain available. } printf("%s %s, %s <= %s of %s\n", format_id($$change{id}), format_upd_cmd($upd), $lcl_commit ? format_id($$lcl_commit{id}) : "", format_parts($parts), $rmt_commit ? format_id($$rmt_commit{id}) : "") if ($debug); if (!($upd & UPD_DROP)) { my $commit; if ($parts == NO_PARTS) { $commit = $lcl_commit; } elsif ($parts != ALL_PARTS) { $commit = merge_commits($lcl_commit, $rmt_commit, $parts); } else { $commit = $rmt_commit; } push @commits, $commit; push @changes, $change; if ($upd & UPD_PUSHED) { $$change{pushed} = $$rmt_commit{id}; # Make the downloaded commit recyclable. This won't help in many # cases, as any added/dropped/tree-modified commits will invalidate # it anyway, but tracking that would add complexity for little benefit. $$change{orig} = $$commit{id} if ($parts == NO_PARTS); } # Generally, the base should match the latest PatchSet even if # we push over it in the end, to optimize inter-diffs. $$change{base} = $$rmt_commit{base} if (($upd & UPD_META) || ($ginfo && !defined($$change{base}))); # If {pushed} was already set, branch tracking already made this a no-op. $$change{tgt} = $$ginfo{branch} if (($upd & UPD_META) || ($ginfo && !defined($$change{tgt}))); # Note: We don't clear the n* properties; they are meant to # be applied regardless of what happens on Gerrit meanwhile. } if ($upd & UPD_INFO) { $any_crossed = 1 if ($$ginfo{cross_branch}); $any_merged = 1 if ($$ginfo{status} eq "MERGED"); } } print format_reports(\@reports); my $any_msg; if ($any_missing && !$quiet) { wout("\nWarning: Some Changes went missing on Gerrit.\n"); $any_msg = 1; } if ($any_merged && !$quiet) { # Note: This is intentionally vague, because MERGED Changes which # were targeting a different branch would be rebased out only if that # branch is forward-merged into the current one. We would have to # inspect the upstream branch to know, and that seems like overkill. wout("\nNotice: Applying MERGED Changes. Using git gpull might rebase them out.\n"); $any_msg = 1; } if ($any_crossed && !$quiet) { nwout("\nNotice: Applying Changes which are targeting a different branch" ." than the current branch's upstream ($upstream_branch).\n"); $any_msg = 1; } if ($can_merge || $need_force_repl || $need_force_del) { my $err; if ($need_force_del) { $err .= "\nLocal modifications found in Change(s) scheduled for deletion;" ." add --force to proceed nonetheless."; } if ($can_merge || $need_force_repl) { my $lmt = " local modifications found in Change(s) scheduled for update;"; my $ipt = " or --ignore to disregard any new PatchSet(s)."; if ($can_merge) { my $mmt = " add --merge after verifying compatibility,"; if ($need_force_repl) { $err .= "\nMergeable and conflicting".$lmt.$mmt ." and/or --force to overwrite unmerged modifications,".$ipt; } else { $err .= "\nMergeable".$lmt.$mmt." --force to overwrite them,".$ipt; } } else { $err .= "\nConflicting".$lmt." use --force to overwrite them,".$ipt; } } nwfail($err."\n") if (!$check); nwout($err."\n"); $any_msg = 1; } print "\n" if ($any_msg); # All Changes which are grouped as an effect of this gpick run get # a new gid assigned. A side effect of this is that series which # got fragmented due to moving around Changes locally may be # permanently split, which seems quite reasonable. deduce_series(\@changes) if (!$check); return \@commits; } sub adjust_changes($) { my ($pairs) = @_; return with_local_git_index(\&do_adjust_changes, $pairs); } # Find the base for the rewrite, skipping over unmodified commits. # Git would do that for us, but writing the todo relies on skipping # over possibly present merge commits. sub determine_base($$) { my ($raw_commits, $commits) = @_; my $base = $local_base; while (@$raw_commits && @$commits) { my $commit = $$raw_commits[0]; last if ($commit != $$commits[0]); $base = $$commit{id}; print "Skipping over unmodified $base\n" if ($debug); shift @$raw_commits; shift @$commits; } return $base; } # Reconstruct the local branch from the specified commits. sub write_rebase_todo($$) { my ($raw_commits, $commits) = @_; my $todo = $gitdir.'/gpick-todo'; print "Writing $todo:\n" if ($debug); open(TODO, "> $todo") or wfail("Cannot write $todo: $!\n"); my %drop = map { $$_{id} => 1 } @$raw_commits; foreach (@$commits) { my $sha1 = $$_{id}; delete $drop{$sha1}; print "> pick $sha1\n" if ($debug); print TODO "pick $sha1\n"; fail("Manipulating merges is not supported (Change $$_{changeid}).\n") if (@{$$_{parents}} > 1); } if (git_config("rebase.missingcommitscheck", "ignore") ne "ignore") { foreach (keys %drop) { print "> drop $_\n" if ($debug); print TODO "drop $_\n"; } } elsif (!@$commits) { # An empty rebase-todo would be a signal to abort. print "> noop\n" if ($debug); print TODO "noop\n"; } close(TODO) or wfail("Failed to write $todo: $!\n"); return $todo; } sub install_hook($$) { my ($hook, $code) = @_; my $file = $gitcommondir."/hooks/".$hook; my $magic = "### git-gpick $hook callback - DO NOT MODIFY\n"; my $content = "#!/bin/sh\n".$magic.$code."exec \"$script\" --x--$hook \"\$\@\"\n"; if (open(HOOK, $file)) { my @lines = ; close(HOOK); fail("There is already another $hook hook installed. Cannot proceed.\n") if (@lines < 3 || ($lines[1] ne $magic)); if (join("", @lines) eq $content) { print "$hook hook is up-to-date.\n" if ($debug); return; } print "$hook hook is outdated; reinstalling.\n" if ($debug); } open(HOOK, "> $file") or wfail("Cannot write $file: $!\n"); print HOOK $content; close(HOOK) or wfail("Failed to write $file: $!\n"); chmod(0755, $file) or wfail("Failed to make $file executable: $!\n"); } sub invoke_rebase($$) { my ($base, $todo) = @_; install_hook("post-rewrite", "[ \"x\$1\" != xrebase ] && exit\n"); print "Applying modifications ...\n" if (!$quiet); $ENV{GPICK_TODO} = $todo; $ENV{GPICK_DEBUG} = $debug; $ENV{GIT_EDITOR} = $script; # Make sure that every commit is picked, otherwise the state todo might # not match if we constructed a first commit that needs no rebasing. my @gitcmd = ('git', 'rebase', '-i', '--no-ff', $base); push @gitcmd, '-q' if ($quiet); if (!$dry_run) { print "+ @gitcmd\n" if ($debug); exec(@gitcmd) or wfail("Cannot run git: $!\n"); # notreached } print "+ @gitcmd [DRY]\n" if ($debug); } sub rewrite_changes($$) { my ($raw_commits, $commits) = @_; my $base = determine_base($raw_commits, $commits); if (!@$raw_commits && !@$commits) { print "Everything already up-to-date.\n" if (!$quiet); return; } my $verify = sha1_hex(join(" ", map { $$_{id} } @$commits)); save_state($dry_run, $verify); # Abuse git-rebase -i to do the commit replacement. # We do that instead of resetting and cherry-picking ourselves, so that # the user is faced with a familiar environment when something goes wrong. my $todo = write_rebase_todo($raw_commits, $commits); invoke_rebase($base, $todo); # We get here only in dry-run mode. } ################### main() #################### # Note: this must come before the GIT_EDITOR clause, because the # variable is still set when the hook is invoked after a clean rebase. if (@ARGV && ($ARGV[0] eq "--x--post-rewrite")) { # The script was launched as a git rebase post-rewrite hook. my @lines = ; $debug = $ENV{GPICK_DEBUG}; # Note that this may make it undefined. print "Called as post-rewrite hook.\n" if ($debug); goto_gitdir(); # The loading of the state refs is delayed until the new state was # verified, so we don't accidentally delete them as orphans. print "Loading new state ...\n" if ($debug); my $verify = load_state_file(1); if (defined($verify)) { my $got_verify = sha1_hex(join(" ", map { substr($_, 0, 40 ) } @lines)); if ($got_verify eq $verify) { # The new state is good, commit it. # We still need to load the state refs to know which ones # to update; they don't overwrite values from the new state. load_refs("refs/gpush/i*"); save_state(); } else { # We were called from an unrelated rebase after ours was aborted. print "New state is outdated.\n" if ($debug); } update_refs(0, [ "delete refs/gpush/state-new\n" ]); } else { # We were called from an unrelated rebase after successfully # completing the previous run. print "No new state present.\n" if ($debug); } exit; } my $editor = $ENV{GIT_EDITOR}; if (defined($editor)) { if ($^O eq "msys") { $editor =~ s,\\,/,g; $editor =~ s,^(.):/,/$1/,g; } if ($editor eq $script) { # The script was launched as the $GIT_EDITOR from git rebase -i. # Replace the todo file and exit. rename($ENV{GPICK_TODO}, $ARGV[0]) or wfail("Cannot move rebase-todo into place: $!\n"); exit; } } parse_arguments(@ARGV); load_config(); goto_gitdir(); load_state(1); determine_local_branch(); my $raw_commits = get_changes(); my $raw_changes = changes_from_commits($raw_commits); my $specs = $pick_all ? prepare_specs_all($raw_changes) : prepare_specs($raw_changes); my $pairs = apply_specs($specs, $raw_changes); my $commits = adjust_changes($pairs); rewrite_changes($raw_commits, $commits) if (!$check); save_state($dry_run);