如果您想查看所有blob的大小，Vi的答案中的git-fatfiles脚本很不错，但是它是如此之慢以至于无法使用。我取消了40行的输出限制，它试图使用计算机的所有RAM而不是完成内存。因此，我重写了它：速度提高了数千倍，增加了功能（可选），并且删除了一些奇怪的错误-如果您对输出求和以查看文件使用的总空间，则旧版本的计数将不准确。
#!/usr/bin/perl
use warnings;
use strict;
use IPC::Open2;
use v5.14;
# Try to get the "format_bytes" function:
my $canFormat = eval {
require Number::Bytes::Human;
Number::Bytes::Human->import('format_bytes');
1;
};
my $format_bytes;
if ($canFormat) {
$format_bytes = \&format_bytes;
}
else {
$format_bytes = sub { return shift; };
}
# parse arguments:
my ($directories, $sum);
{
my $arg = $ARGV[0] // "";
if ($arg eq "--sum" || $arg eq "-s") {
$sum = 1;
}
elsif ($arg eq "--directories" || $arg eq "-d") {
$directories = 1;
$sum = 1;
}
elsif ($arg) {
print "Usage: $0 [ --sum, -s | --directories, -d ]\n";
exit 1;
}
}
# the format is [hash, file]
my %revList = map { (split(' ', $_))[0 => 1]; } qx(git rev-list --all --objects);
my $pid = open2(my $childOut, my $childIn, "git cat-file --batch-check");
# The format is (hash => size)
my %hashSizes = map {
print $childIn $_ . "\n";
my @blobData = split(' ', <$childOut>);
if ($blobData[1] eq 'blob') {
# [hash, size]
$blobData[0] => $blobData[2];
}
else {
();
}
} keys %revList;
close($childIn);
waitpid($pid, 0);
# Need to filter because some aren't files--there are useless directories in this list.
# Format is name => size.
my %fileSizes =
map { exists($hashSizes{$_}) ? ($revList{$_} => $hashSizes{$_}) : () } keys %revList;
my @sortedSizes;
if ($sum) {
my %fileSizeSums;
if ($directories) {
while (my ($name, $size) = each %fileSizes) {
# strip off the trailing part of the filename:
$fileSizeSums{$name =~ s|/[^/]*$||r} += $size;
}
}
else {
while (my ($name, $size) = each %fileSizes) {
$fileSizeSums{$name} += $size;
}
}
@sortedSizes = map { [$_, $fileSizeSums{$_}] }
sort { $fileSizeSums{$a} <=> $fileSizeSums{$b} } keys %fileSizeSums;
}
else {
# Print the space taken by each file/blob, sorted by size
@sortedSizes = map { [$_, $fileSizes{$_}] }
sort { $fileSizes{$a} <=> $fileSizes{$b} } keys %fileSizes;
}
for my $fileSize (@sortedSizes) {
printf "%s\t%s\n", $format_bytes->($fileSize->[1]), $fileSize->[0];
}
将其命名为git-fatfiles.pl并运行它。要查看文件的所有修订版使用的磁盘空间，请使用该--sum选项。要查看相同的内容，但对于每个目录中的文件，请使用--directories选项。如果您安装Number :: Bytes :: Human cpan模块（运行“ cpan Number :: Bytes :: Human”），则大小将被格式化为：“ 21M /path/to/file.mp4”。