git-filehistory (first cut at detecting renames)

Hi all,

This is my stupid perl script to detect renames all the way to it's
origin.  This is my first attempt at writing a perl script so it looks
a bit awful.  It's not perfect, if a file is renamed, the entire
commit is dumped out.  Maybe if git-diff-tree -M <pathname> does post
processing to output changes relevant to <pathname> then it will work
quite well.

David

[davidho@penguin git-tutorial]$ cat git-filehistory.perl
#!/usr/bin/perl

use warnings;
use strict;

sub usage($);

# Sanity checks:
my $GIT_DIR = $ENV{'GIT_DIR'} || ".git";

unless ( -d $GIT_DIR && -d $GIT_DIR . "/objects" &&
        -d $GIT_DIR . "/objects/" && -d $GIT_DIR . "/refs") {
        usage("Git repository not found.");
}

usage("") if scalar @ARGV != 1;

my ($file) = @ARGV;

unless (-f $file) {
        usage("git filehistory: '$file' does not exist");
}

open(F,"-|","git-rev-parse",,"--default","HEAD","--revs-only",$file);
my $git_rev_args = <F>;
chomp ($git_rev_args);
close (F);

open(F,"-|","git-rev-list", $git_rev_args);
my @commits = <F>;
close (F);

my $nextname = $file;
for (my $i=0; $i < scalar @commits; $i++)
{
        my $commit = $commits[$i];
        chomp ($commit);

        open(F,"-|","git-diff-tree","-r","-M", $commit)
                or die "Failed to open pipe from git-diff-tree: " . $!;
        my @lines = grep /R\d{3}.+$nextname/, <F>;
        close(F);
        if (scalar @lines >=1) {
                #The file is renamed!
                my $rc = system
("git-diff-tree","-r","-M","-p","--pretty=medium",$commit);
                die "git-diff-tree failed" if $rc;
                print "\n";
                if ($lines[0] =~ /R\d{3}\t(.+)\t(.+)/) {
                        $nextname = $1;
                }
        }
        else {
                #Not renamed
                my $rc = system
("git-diff-tree","-r","-m","-p","--pretty=medium",$commit, $nextname);
                die "git-diff-tree failed" if $rc;
                print "\n";
        }
}

sub usage($) {
        my $s = shift;
        print $s, "\n" if (length $s != 0);
        print <<EOT;
$0 <file>
This script traces renames to find the origin of the file
EOT
        exit(1);
}

Actually, just tried it on the git repository.  It is very slow. 
Maybe because I'm doing a complete diff-tree on each commit.  Mmm...
maybe there is a better way to detect renames  when walking up the
file history.

David

On 10/20/05, David Ho [off-list ref] wrote:

Hi all,

This is my stupid perl script to detect renames all the way to it's
origin.  This is my first attempt at writing a perl script so it looks
a bit awful.  It's not perfect, if a file is renamed, the entire
commit is dumped out.  Maybe if git-diff-tree -M <pathname> does post
processing to output changes relevant to <pathname> then it will work
quite well.

David

[davidho@penguin git-tutorial]$ cat git-filehistory.perl
#!/usr/bin/perl

use warnings;
use strict;

sub usage($);

# Sanity checks:
my $GIT_DIR = $ENV{'GIT_DIR'} || ".git";

unless ( -d $GIT_DIR && -d $GIT_DIR . "/objects" &&
        -d $GIT_DIR . "/objects/" && -d $GIT_DIR . "/refs") {
        usage("Git repository not found.");
}

usage("") if scalar @ARGV != 1;

my ($file) = @ARGV;

unless (-f $file) {
        usage("git filehistory: '$file' does not exist");
}

open(F,"-|","git-rev-parse",,"--default","HEAD","--revs-only",$file);
my $git_rev_args = <F>;
chomp ($git_rev_args);
close (F);

open(F,"-|","git-rev-list", $git_rev_args);
my @commits = <F>;
close (F);

my $nextname = $file;
for (my $i=0; $i < scalar @commits; $i++)
{
        my $commit = $commits[$i];
        chomp ($commit);

        open(F,"-|","git-diff-tree","-r","-M", $commit)
                or die "Failed to open pipe from git-diff-tree: " . $!;
        my @lines = grep /R\d{3}.+$nextname/, <F>;
        close(F);
        if (scalar @lines >=1) {
                #The file is renamed!
                my $rc = system
("git-diff-tree","-r","-M","-p","--pretty=medium",$commit);
                die "git-diff-tree failed" if $rc;
                print "\n";
                if ($lines[0] =~ /R\d{3}\t(.+)\t(.+)/) {
                        $nextname = $1;
                }
        }
        else {
                #Not renamed
                my $rc = system
("git-diff-tree","-r","-m","-p","--pretty=medium",$commit, $nextname);
                die "git-diff-tree failed" if $rc;
                print "\n";
        }
}

sub usage($) {
        my $s = shift;
        print $s, "\n" if (length $s != 0);
        print <<EOT;
$0 <file>
This script traces renames to find the origin of the file
EOT
        exit(1);
}