repobrowse: make git diff output use qspawn

This is a potentially expensive operation, so we may want to
give it it's own limiter channel.

2 files changed, 56 insertions, 109 deletions

diff --git a/lib/PublicInbox/RepobrowseGitDiff.pm b/lib/PublicInbox/RepobrowseGitDiff.pm
index 4ed4d02e..eb64d1fd 100644
--- a/lib/PublicInbox/RepobrowseGitDiff.pm
+++ b/lib/PublicInbox/RepobrowseGitDiff.pm
@@ -4,128 +4,72 @@
 # shows the /diff endpoint for git repositories for cgit compatibility
 # usage: /repo.git/diff?id=COMMIT_ID&id2=COMMIT_ID2
 #
-# FIXME: much duplicated code between this and RepobrowseGitCommit.pm
-#
 # We probably will not link to this outright because it's expensive,
-# but exists to preserve URL compatibility.
+# but exists to preserve URL compatibility with cgit.
 package PublicInbox::RepobrowseGitDiff;
 use strict;
 use warnings;
 use base qw(PublicInbox::RepobrowseBase);
-use PublicInbox::Hval qw(utf8_html to_attr);
-use PublicInbox::RepobrowseGit qw(git_unquote git_commit_title);
-use PublicInbox::RepobrowseGitDiffCommon qw/git_diffstat_emit
-        git_diff_ab_index git_diff_ab_hdr git_diff_ab_hunk/;
+use PublicInbox::Hval qw(utf8_html);
+use PublicInbox::RepobrowseGitDiffCommon;
+use PublicInbox::Qspawn;
+
+sub git_diff_sed ($$) {
+        my ($self, $req) = @_;
+        git_diff_sed_init($req);
+        $req->{dstate} = DSTATE_STAT;
+        # this filters for $fh->write or $body->getline (see Qspawn)
+        sub {
+                my $dst = delete $req->{dhtml} || '';
+                if (defined $_[0]) { # $_[0] == scalar buffer
+                        $req->{dbuf} .= $_[0];
+                        git_diff_sed_run(\$dst, $req);
+                } else { # undef means EOF from "git show", flush the last bit
+                        git_diff_sed_close(\$dst, $req);
+                        $dst .= '</pre></body></html>';
+                }
+                $dst;
+        }
+}
 
 sub call_git_diff {
         my ($self, $req) = @_;
-        my $git = $req->{repo_info}->{git};
-        my $q = PublicInbox::RepobrowseGitQuery->new($req->{env});
+        my $env = $req->{env};
+        my $q = PublicInbox::RepobrowseGitQuery->new($env);
         my $id = $q->{id};
         my $id2 = $q->{id2};
 
-        my @cmd = (qw(diff-tree -z --numstat -p --encoding=UTF-8
-                        --no-notes --no-color -M -B -D -r),
-                        $id2, $id, '--');
+        my $git = $req->{repo_info}->{git};
+        my $cmd = [ 'git', "--git-dir=$git->{git_dir}", qw(diff-tree
+                        -z --numstat -p --encoding=UTF-8
+                        --no-color -M -B -D -r),
+                        $id2, $id, '--' ];
         my $expath = $req->{expath};
-        push @cmd, $expath if $expath ne '';
-        $req->{rpipe} = $git->popen(\@cmd, undef, { 2 => $git->err_begin });
-        my $env = $req->{env};
-        my $err = $env->{'psgi.errors'};
-        my ($vin);
-        $req->{dbuf} = '';
+        push @$cmd, $expath if $expath ne '';
+        my $o = { nofollow => 1, noindex => 1 };
+        my $ex = $expath eq '' ? '' : " $expath";
+        $req->{dhtml} = $self->html_start($req, 'diff', $o). "\n\n".
+                                utf8_html("git diff-tree -r -M -B -D ".
+                                "$id2 $id --$ex"). "\n\n";
         $req->{p} = [ $id2 ];
         $req->{h} = $id;
-        my $end = sub {
-                if (my $fh = delete $req->{fh}) {
-                        # write out the last bit that was buffered
-                        my @buf = split(/\n/, delete $req->{dbuf}, -1);
-                        my $s = '';
-                        $s .= git_diff_line_i($req, $_) foreach @buf;
-                        $s .= '</pre></body></html>';
-                        $fh->write($s);
-
-                        $fh->close;
-                } elsif (my $res = delete $req->{res}) {
-                        $res->($self->r(500));
-                }
-                if (my $rpipe = delete $req->{rpipe}) {
-                        $rpipe->close; # _may_ be Danga::Socket::close
+        my $rdr = { 2 => $git->err_begin };
+        my $qsp = PublicInbox::Qspawn->new($cmd, undef, $rdr);
+        # $env->{'qspawn.quiet'} = 1;
+        $qsp->psgi_return($env, undef, sub { # parse header
+                my ($r) = @_;
+                if (!defined $r) {
+                        [ 500, [ 'Content-Type', 'text/html' ], [ $git->err ]];
+                } elsif ($r == 0) {
+                        [ 200, [ 'Content-Type', 'text/html' ], [
+                                delete($req->{dhtml}).
+                                'No differences</pre></body></html>' ]
+                        ]
+                } else {
+                        $env->{'qspawn.filter'} = git_diff_sed($self, $req);
+                        [ 200, [ 'Content-Type', 'text/html' ] ];
                 }
-        };
-        my $fail = sub {
-                if ($!{EAGAIN} || $!{EINTR}) {
-                        select($vin, undef, undef, undef) if defined $vin;
-                        # $vin is undef on async, so this is a noop on EAGAIN
-                        return;
-                }
-                my $e = $!;
-                $end->();
-                $err->print("git diff ($git->{git_dir}): $e\n");
-        };
-        my $cb = sub {
-                my $off = length($req->{dbuf});
-                my $n = $req->{rpipe}->sysread($req->{dbuf}, 8192, $off);
-                return $fail->() unless defined $n;
-                return $end->() if $n == 0;
-                if (my $res = delete $req->{res}) {
-                        my $h = ['Content-Type', 'text/html; charset=UTF-8'];
-                        my $fh = $req->{fh} = $res->([200, $h]);
-                        my $o = { nofollow => 1, noindex => 1 };
-                        my $ex = $expath eq '' ? '' : " $expath";
-                        $fh->write($self->html_start($req, 'diff', $o).
-                                        "\n\n".
-                                        utf8_html("git diff-tree -r -M -B -D ".
-                                                "$id2 $id --$ex"). "\n\n");
-                }
-                git_diff_to_html($req);
-        };
-        if (my $async = $env->{'pi-httpd.async'}) {
-                $req->{rpipe} = $async->($req->{rpipe}, $cb);
-                sub { $req->{res} = $_[0] } # let Danga::Socket handle the rest.
-        } else { # synchronous loop for other PSGI servers
-                $vin = '';
-                vec($vin, fileno($req->{rpipe}), 1) = 1;
-                sub {
-                        $req->{res} = $_[0];
-                        while ($req->{rpipe}) { $cb->() }
-                }
-        }
-}
-
-sub git_diff_line_i {
-        my ($req, $l) = @_;
-        my $cmt = '[a-f0-9]+';
-
-        if ($l =~ m{^diff --git ("?a/.+) ("?b/.+)$}) { # regular
-                $l = git_diff_ab_hdr($req, $1, $2);
-        } elsif ($l =~ /^index ($cmt)\.\.($cmt)(.*)$/o) { # regular
-                $l = git_diff_ab_index($1, $2, $3);
-        } elsif ($l =~ /^@@ (\S+) (\S+) @@(.*)$/) { # regular
-                $l = git_diff_ab_hunk($req, $1, $2, $3);
-        } else {
-                $l = utf8_html($l);
-        }
-        $l .= "\n";
-}
-
-sub git_diff_to_html {
-        my ($req) = @_;
-        my $fh = $req->{fh};
-        if (!$req->{diff_state}) {
-                my ($stat, $buf) = split(/\0\0/, $req->{dbuf}, 2);
-                return unless defined $buf;
-                $req->{dbuf} = $buf;
-                git_diffstat_emit($req, $fh, $stat);
-                $req->{diff_state} = 1;
-        }
-        my @buf = split(/\n/, $req->{dbuf}, -1);
-        $req->{dbuf} = pop @buf; # last line, careful...
-        if (@buf) {
-                my $s = '';
-                $s .= git_diff_line_i($req, $_) foreach @buf;
-                $fh->write($s) if $s ne '';
-        }
+        });
 }
 
 1;
diff --git a/lib/PublicInbox/RepobrowseGitDiffCommon.pm b/lib/PublicInbox/RepobrowseGitDiffCommon.pm
index ae0e6821..bca50e15 100644
--- a/lib/PublicInbox/RepobrowseGitDiffCommon.pm
+++ b/lib/PublicInbox/RepobrowseGitDiffCommon.pm
@@ -216,6 +216,9 @@ sub git_diff_sed_stat ($$) {
         my $ndel = \($req->{ndel});
         if (!$req->{dstat_started}) {
                 $req->{dstat_started} = 1;
+
+                # merges start with an extra '\0' before the diffstat
+                # non-merge commits start with an extra '\n', instead
                 if ($req->{mhelp}) {
                         if ($stat[0] eq '') {
                                 shift @stat;
@@ -224,8 +227,8 @@ sub git_diff_sed_stat ($$) {
 'initial merge diffstat line was not empty';
                         }
                 } else {
-                        $stat[0] =~ s/\A\n//s or warn
-'failed to remove initial newline from diffstat';
+                        # for commits, only (not diff-tree)
+                        $stat[0] =~ s/\A\n//s;
                 }
         }
         while (defined(my $l = shift @stat)) {