podman/hack/markdown-preprocess-review

#!/usr/bin/perl

(our $ME = $0) =~ s|^.*/||;

use v5.20;

our $DSM = 'docs/source/markdown';

my ($oldname, $newname);
my %oldname;
my %changed;
open my $git_diff, '-|', 'git', 'log', '-1', '-p'
    or die "$ME: Cannot fork: $!\n";
while (my $line = <$git_diff>) {
    chomp $line;

    if ($line =~ m!^\-\-\-\s+a/$DSM/(podman-\S+\.md(\.in)?)!) {
        $oldname = $1;
        $newname = undef;
    }
    elsif ($line =~ m!^\+\+\+\s+b/$DSM/(podman-\S+\.md(\.in)?)!) {
        $newname = $1;
        $oldname{$newname} = $oldname;
    }
    elsif ($newname) {
        if ($line =~ s/^-####\s+//) {
            $line =~ /^\*\*--(\S+?)\*\*/
                or die "$ME: in $newname: weird '$line'";
            $changed{$newname}{$1}{name} //= $1;
        }
        # Usually the same, but not for host.container and host.pod.md
        elsif ($line =~ /^\+\@\@option\s+(\S+)/) {
            my $optfile = $1;
            if ($optfile =~ /^(.*)\.\S+$/) {
                $changed{$newname}{$1}{name} = $optfile;
            }
        }
    }
}
close $git_diff;

# Pass 2: read each oldfile, parse changed options
for my $f (sort keys %changed) {
    my $oldfile = $oldname{$f};
    open my $git_fh, '-|', 'git', 'show', "HEAD^:$DSM/$oldfile"
        or die "$ME: Cannot fork: $!\n";
    my $opt;
    while (my $line = <$git_fh>) {
        if ($line =~ /^####\s+\*\*--(\S+?)\*\*/) {
            $opt = $1;
            if ($changed{$f}{$opt}) {
                $changed{$f}{$opt}{text} = $line;
            }
            else {
                undef $opt;
            }
        }
        elsif ($line =~ /^#/ || $line =~ /^\@\@option\s/) {
            undef $opt;
        }
        elsif ($opt) {
            $changed{$f}{$opt}{text} .= $line;
        }
    }
    close $git_fh
        or die "$ME: Error running git on $oldfile\n";
}

# Pass 3: write out files
my $tempdir = "/tmp/$ME.diffs";
system('rm', '-rf', $tempdir);
mkdir $tempdir, 0755;

for my $md_file (sort keys %changed) {
    for my $opt (sort keys %{$changed{$md_file}}) {
        my $d = "$tempdir/$changed{$md_file}{$opt}{name}";
        mkdir $d, 0755;

        my $outfile = "$d/$md_file";
        open my $fh, '>', $outfile
            or die "$ME: Cannot create $outfile: $!\n";
        # strip all trailing newlines
        (my $text = $changed{$md_file}{$opt}{text}) =~ s/\n+$/\n/s;
        print { $fh } $text;
        close $fh
            or die "$ME: Error writing $outfile: $!\n";

        my $new_text = "$DSM/options/$changed{$md_file}{$opt}{name}.md";
        die "$ME: $md_file: File does not exist: $new_text\n" if ! -e $new_text;
        system('cp', $new_text, "$d/zzz-chosen.md");
    }
}

# Now run diffuse
chdir $tempdir or die;
my @all_opts = glob("*");
for my $i (0..$#all_opts) {
    my $opt = $all_opts[$i];
    chdir "$tempdir/$opt"
        or die "??? Internal error, cannot cd $tempdir/$opt: $!";

    $| = 1; printf "--%s (%d/%d) ", $opt, $i+1, scalar(@all_opts);

    my @all_files = glob("*");
    if (all_files_identical(@all_files)) {
        pop @all_files;
        print "[identical between @all_files]\n";
        next;
    }

    # Prompt
    print "[Y/n/q] ";
    my $ans = <STDIN>;
    next if $ans =~ /^n/i;
    exit 0 if $ans =~ /^q/i;

    # Try to cull the files (remove identical ones)
    my @files = glob("*");
    my $winner = pop @files;

    for my $f (@files) {
        system('cmp', '-s', $f, $winner);
        if ($? == 0) {
            print "[ $f is the one we went with; removing from list ]\n";
            unlink $f;
            next;
        }

        system('wdiff', '-1', '-2', '-3', $f, $winner);
        if ($? == 0) {
            print "[ $f is whitespace-identical with what we went with ]\n";
            unlink $f;
            next;
        }
    }

    # Recompute @files, in case some were deleted above
    @files = glob("*"); pop @files;

    for (my $i=0; $i < $#files; $i++) {
        my $f1 = $files[$i];
        next unless -e $f1;

        for (my $j=$i+1; $j <= $#files; $j++) {
            my $f2 = $files[$j];
            next unless -e $f2;

            system('wdiff', '-1', '-2', '-3', $f1, $f2);
            if ($? == 0) {
                print "[ $f2 : removing, it =~ $f1 ]\n";
                unlink $f2;
            }
        }
    }

    # Recompute @files, in case some were deleted above
    @files = glob("*");

    # diffuse works great for 3-4 files, passable for 5, not at all for >5
    if (@files <= 5) {
        system("diffuse", "-w", @files) == 0
            or die "Diffuse failed\n";
    }
    else {
        # Too many files. Go by threes.
        my $winner = pop @files;
        for (my $i=0; $i < @files; $i += 3) {
            system("diffuse", "-w", @files[$i..$i+2], $winner);
        }
    }
}


sub all_files_identical {
    my %sha;
    for my $f (@_) {
        my $result = qx{sha256sum $f};
        $result =~ /^([0-9a-f]+)\s/
            or die "Internal error: unexpected result from sha256sum $f: $result";
        $sha{$1}++;
    }

    return (keys(%sha) == 1);
}