Skip to content
This repository has been archived by the owner on Apr 29, 2024. It is now read-only.

Commit

Permalink
major rewrite in progress
Browse files Browse the repository at this point in the history
  • Loading branch information
harleypig committed Feb 4, 2013
1 parent c98f475 commit e46067c
Showing 1 changed file with 190 additions and 132 deletions.
322 changes: 190 additions & 132 deletions handle_metadata
View file
Original file line number Diff line number Diff line change
Expand Up @@ -8,156 +8,214 @@

use strict;
use warnings;
#no warnings 'File::Find';

use Data::Dumper ();
use Fcntl ':mode';
use File::Basename ();
use File::Find ();
use File::Spec ();
use File::stat ':FIELDS'; # overrides perl's stat and lstat
use IPC::Run3::Simple;

my $metafile = shift
or die "Expecting file to save metadata to.\n";

open my $METAFILE, '>', $metafile
or die "Unable to open $metafile for writing: $!\n";

my $base_dir = get_toplevel();

chdir $base_dir
#use Data::Dumper ();
#use Fcntl ':mode';
#use File::Basename ();
#use File::Find ();
#use File::Spec ();
#use File::stat ':FIELDS'; # overrides perl's stat and lstat
#use IPC::Run3::Simple;
use Getopt::Long;

# Are we in a git repository?
# If not, die
# Are we running as post-commit or post-{checkout,merge}?
# If post-commit
# get list of files being committed
# for each file being committed
# gather metadata
# save metadata in note
# Else if post-{checkout,merge}
# get list of files that were changed
# for each file that changed
# get metadata from note
# modify metadata for file

chdir get_toplevel();
or die "Unable to change to git top level directory: $!\n";

my $git_file = get_gitfiles();
GetOptions(
save => \my $save,
set => \my $set,
) or die "Problem parsing options, exiting.\n";

my ( @metadata, %hardlink );
die "either --save or --set is required\n"
if ! defined $save && ! defined $set;

File::Find::find({
die "--save and --set are mutually exclusive\n"
if defined $save && defined $set;

'wanted' => \&wanted,
'preprocess' => \&directories,
'no_chdir' => 1,
if ( defined $save ) {

}, $base_dir );
my @files = run3([qw( git diff-tree HEAD )]);

for my $hardlink ( keys %hardlink ) {
} elsif ( defined $set ) {
} else {

my $meta = $hardlink{ $hardlink }{ 'meta' };
my $names = join "\0", @{ $hardlink{ $hardlink }{ 'names' } };
printf $METAFILE "HL: %s\0%s\n", $meta, $names;
die "Unknown and unhandled error! How'd you get here?\n";

}

exit 0;

###############################################################################################
#my %format = (
# RF => "%d\0%d\0%d\0%s\n", # Regular file
# CF => "CF: %d\0%d\0%d\0%d\0%s\n", # Character file
# HL => "HL: %s\0%s\n", # Hardlink
#);
#
##for my $file (
#my ( @metadata, %hardlink );
#
##File::Find::find({
##
## 'wanted' => \&wanted,
## 'preprocess' => \&directories,
## 'no_chdir' => 1,
##
##}, $base_dir );
#
##for my $hardlink ( keys %hardlink ) {
##
## my $meta = $hardlink{ $hardlink }{ 'meta' };
## my $names = join "\0", @{ $hardlink{ $hardlink }{ 'names' } };
## push @metadata, sprintf $format{ HL }, $meta, $names;
##
##}
#
#exit 0;
#
################################################################################################

sub get_toplevel {

run3({ 'cmd' => [qw( git rev-parse --show-toplevel )], 'stdout' => \my $toplevel });
$toplevel = File::Spec->canonpath( $toplevel );
return $toplevel;

}

sub get_gitfiles {

run3({ 'cmd' => [qw( git ls-files )], 'stdout' => \my @gitfiless });
my %gitfile; @gitfile{ @gitfiless } = undef;

for my $file ( @gitfiles ) {

$gitfile{ $file } = undef;

my $dirname = File::Basename::dirname( $file );
next if $dirname eq '.';

# Is it faster to just reassaign the same value or to check for existence?
# next if exists $gitfile{ $dirname };

$gitfile{ $dirname } = undef;

}
}

sub directories {

# Skip this repositories .git directory
if ( $File::Find::dir eq File::Spec->catfile( $base_dir, '.git' ) ) {

$File::Find::prune = 1;
return;

}

if ( $File::Find::dir ne $base_dir ) {

if ( lstat( $File::Find::dir ) ) {

my $dir = File::Spec->abs2rel( $File::Find::dir, $base_dir );
my ( $toplevel, $err, $syserr ) = run3([qw( git rev-parse --show-toplevel )]);

# If this directory isn't in the list of git files then skip it.
if ( ! exists $git_file->{ $dir } ) {
die "System error: $syserr\n"
if $syserr;

$File::Find::prune = 1;
return;
die "Unable to get top level directory: $err\n"
if $err;

}

printf $METAFILE "%d\0%d\0%d\0%s\n", $st_uid, $st_gid, $st_mode, $dir;

} else {

warn "Problem getting stat on $File::Find::dir ($!), skipping.\n";
$File::Find::prune = 1;
return;

}
}

File::Spec->no_upwards( @_ );
$toplevel = File::Spec->canonpath( $toplevel );
return $toplevel;

}

sub wanted {

# We've already handled directories
return if -d $File::Find::name;

lstat( $File::Find::name ) or do {
warn "Problem getting stat on $File::Find::name ($!), skipping.\n";
return;
};

# return if ! $st_dev || $st_dev != $File::Find::topdev;

my $name = File::Spec->abs2rel( $File::Find::name, $base_dir );

my $ftype = ( $st_mode & 0170000 ) >> 12;

if ( $st_rdev || $ftype == 1 || $ftype == 12 ) {

printf $METAFILE "CF: %d\0%d\0%d\0%d\0%s\n", $st_uid, $st_gid, $st_mode, $st_rdev, $name;
return;

}

return unless exists $git_file->{ $name };

if ( $st_nlink > 1 ) {

my $meta = sprintf "%d\0%d\0%d", $st_uid, $st_gid, $st_mode;

( die sprintf "This shouldn't happen! $name is a hardlink and has different meta data ($meta) than %s\n", Data::Dumper::Dumper $hardlink{ $st_ino } )
if exists $hardlink{ $st_ino } && $meta ne $hardlink{ $st_ino }{ 'meta' };

$hardlink{ $st_ino }{ 'meta' } = $meta;
push @{ $hardlink{ $st_ino }{ 'names' } }, $name;
return;

}

printf $METAFILE "%d\0%d\0%d\0%s\n", $st_uid, $st_gid, $st_mode, $name;

}
#sub get_gitfiles {
#
# run3({ cmd => [qw( git ls-files --full-name )], stdout => \my @gitfiles });
# run3({ cmd => [qw( git config -f /.gitmodules --get-regexp ^submodule\..*\.path$ )], stdout => \my @submodules });
#
# my %submodule = map { ( split /\s/ )[1], undef } @submodules;
#
# # Directories are not explicitly handled in git, so we need to find the
# # directories ourselves.
#
# my %gitfile;
#
# for my $file ( @gitfiles ) {
#
# # submodules need to be handled differently, and is outside the scope of
# # this script.
#
# next if exists $submodule{ $file };
#
# $gitfile{ $file } = undef;
#
# my $dirname = File::Basename::dirname( $file );
# next if $dirname eq '.';
#
# # Is it faster to just reassaign the same value or to check for existence?
# # next if exists $gitfile{ $dirname };
#
# $gitfile{ $dirname } = undef;
#
# }
#
# return \%gitfile;
#
#}
#
#sub directories {
#
# # Skip this repositories .git directory
# if ( $File::Find::dir eq File::Spec->catfile( $base_dir, '.git' ) ) {
#
# $File::Find::prune = 1;
# return;
#
# }
#
# if ( $File::Find::dir ne $base_dir ) {
#
# if ( lstat( $File::Find::dir ) ) {
#
# my $dir = File::Spec->abs2rel( $File::Find::dir, $base_dir );
#
# # If this directory isn't in the list of git files then skip it.
# if ( ! exists $git_file->{ $dir } ) {
#
# $File::Find::prune = 1;
# return;
#
# }
#
# push @metadata, sprintf $format{ RF }, $st_uid, $st_gid, $st_mode, $dir;
#
# } else {
#
# warn "Problem getting stat on $File::Find::dir ($!), skipping.\n";
# $File::Find::prune = 1;
# return;
#
# }
# }
#
# File::Spec->no_upwards( @_ );
#
#}
#
#sub wanted {
#
# # We've already handled directories
# return if -d $File::Find::name;
#
# lstat( $File::Find::name ) or do {
# warn "Problem getting stat on $File::Find::name ($!), skipping.\n";
# return;
# };
#
# my $name = File::Spec->abs2rel( $File::Find::name, $base_dir );
#
# # If it's not in git we don't care about it
# return unless exists $git_file->{ $name };
#
# my $ftype = ( $st_mode & 0170000 ) >> 12;
#
# if ( $st_rdev || $ftype == 1 || $ftype == 12 ) {
#
# # this is a character file
#
# push @metadata, sprintf $format{ CF }, $st_uid, $st_gid, $st_mode, $st_rdev, $name;
# return;
#
# }
#
# if ( $st_nlink > 1 ) {
#
# # this is a hardlinked file
#
# my $meta = sprintf "%d\0%d\0%d", $st_uid, $st_gid, $st_mode;
#
# ( die sprintf "This shouldn't happen! $name is a hardlink and has different meta data ($meta) than %s\n", Data::Dumper::Dumper $hardlink{ $st_ino } )
# if exists $hardlink{ $st_ino } && $meta ne $hardlink{ $st_ino }{ 'meta' };
#
# $hardlink{ $st_ino }{ 'meta' } = $meta;
# push @{ $hardlink{ $st_ino }{ 'names' } }, $name;
# return;
#
# }
#
# push @metadata, sprintf $format{ RF }, $st_uid, $st_gid, $st_mode, $name;
#
#}

0 comments on commit e46067c

Please sign in to comment.