NAME

Git::Raw::Repository - Git repository class

VERSION

version 0.90

SYNOPSIS

use Git::Raw;

# clone a Git repository
my $url  = 'git://github.com/jacquesg/p5-Git-Raw.git';
my $callbacks = {
  'transfer_progress' => sub {
    my ($total_objects, $received_objects, $local_objects, $total_deltas,
      $indexed_deltas, $received_bytes) = @_;

    print "Objects: $received_objects/$total_objects", "\n";
    print "Received: ", int($received_bytes/1024), "KB", "\n";
  }
};
my $repo = Git::Raw::Repository -> clone($url, 'p5-Git-Raw', {}, {
  'callbacks' => $callbacks
});

# print all the tags of the repository
foreach my $tag ($repo -> tags) {
  say $tag -> name;
}

# merge a reference
my $ref = Git::Raw::Reference -> lookup('HEAD', $repo);
$repo -> merge ($ref);

my @conflicts = $repo -> index -> conflicts;
print "Got ", scalar (@conflicts), " conflicts!\n";

DESCRIPTION

A Git::Raw::Repository represents a Git repository.

WARNING: The API of this module is unstable and may change without warning (any change will be appropriately documented in the changelog).

METHODS

init( $path, $is_bare )

Initialize a new repository at $path.

clone( $url, $path, \%opts, [\%fetch_opts, \%checkout_opts])

Clone the repository at $url to $path. Valid fields for the %opts hash are:

  • "bare"

    If true (default is false) create a bare repository.

  • "checkout_branch"

    The name of the branch to checkout (default is to use the remote's HEAD).

  • "disable_checkout"

    If true (default is false) files will not be checked out after the clone completes.

  • "callbacks"

    • "remote_create"

      Remote customization callback. If a non-default remote is required, i.e. a remote with a remote name other than 'origin', this callback should be used. The callback receives a Git::Raw::Repository object, a string containing the default name for the remote, typically 'origin', and a string containing the URL of the remote. This callback should return a Git::Raw::Remote object. The returned object and the repository object passed to this callback are ephemeral. Note: Do not take any references to it, as it may be freed internally.

See Git::Raw::Remote->fetch() for valid %fetch_opts values and Git::Raw::Repository->checkout() for valid %checkout_opts values.

open( $path )

Open the repository at $path.

discover( $path )

Discover the path to the repository directory given a subdirectory.

new( )

Create a new repository with neither backends nor config object.

config( )

Retrieve the default Git::Raw::Config of the repository.

commondir( )

Get the path of the shared common directory for this repository.

is_worktree( )

Check if the repository is a linked work tree.

index( [$new_index] )

Retrieve the index of the repository. If $new_index is passed, it will be used as the index of the repository. If $new_index is undef the index associated with the repository will be disassociated. Returns a Git::Raw::Index object or undef if index has been disassociated as part of the call.

odb( [$new_odb] )

Retrieve the object database of the repository. If $odb is passed, it will be used as the object database. Returns a Git::Raw::Odb object.

head( [$new_head] )

Retrieve the Git::Raw::Reference pointed by the HEAD of the repository. If the Git::Raw::Reference $new_head is passed, the HEAD of the repository will be changed to point to it.

head_for_worktree( $worktree )

Retrieve the Git::Raw::Reference pointed by the HEAD of $worktree.

detach_head( $commitish )

Make the repository HEAD point directly to a commit. $commitish should be peelable to a Git::Raw::Commit object, that is, it should be a Git::Raw::Commit or Git::Raw::Reference object, or alternatively a commit id or commit id prefix.

lookup( $id )

Retrieve the object corresponding to $id.

checkout( $object, \%opts )

Updates the files in the index and working tree to match the content of $object. Valid fields for the %opts hash are:

  • "checkout_strategy"

    Hash representing the desired checkout strategy. Valid fields are:

    • "none"

      Dry-run checkout strategy. It doesn't make any changes, but checks for conflicts.

    • "force"

      Take any action to make the working directory match the target (pretty much the opposite of "none").

    • "safe_create"

      Recreate missing files.

    • "safe"

      Make only modifications that will not lose changes (to be used in order to simulate git checkout).

    • "allow_conflicts"

      Apply safe updates even if there are conflicts.

    • "remove_untracked"

      Remove untracked files from the working directory.

    • "remove_ignored"

      Remove ignored files from the working directory.

    • "update_only"

      Only update files that already exist (files won't be created or deleted).

    • "dont_update_index"

      Do not write the updated files' info to the index.

    • "dont_remove_existing"

      Don not overwrite existing files or folders.

    • "dont_write_index"

      Prevent writing of the index upon completion.

    • "no_refresh"

      Do not reload the index and git attrs from disk before operations.

    • "skip_unmerged"

      Skip files with unmerged index entries, instead of treating them as conflicts.

  • "notify"

    Notification flags for the notify callback. A list of the following options:

    • "conflict"

      Notifies about conflicting paths.

    • "dirty"

      Notifies about files that don't need an update but no longer match the baseline. Core git displays these files when checkout runs, but won't stop the checkout.

    • "updated"

      Notification on any file changed.

    • "untracked"

      Notification about untracked files.

    • "ignored"

      Notifies about ignored files.

    • "all"

      All of the above.

  • "callbacks"

    Hash containing progress and notification callbacks. Valid fields are:

    • "notify"

      This callback is called for each file matching one of the notify options selected. It runs before modifying any files on disk. This callback should return a non-zero value should the checkout be cancelled. The callback receives a string containing the path of the file $path and an array reference containing the reason $why.

    • "progress"

      The callback to be invoked as a file is checked out. The callback receives a string containing the path of the file $path, an integer $completed_steps and an integer $total_steps.

  • "paths"

    An optional array representing the list of files that should be checked out. If "paths" is not specified, all files will be checked out (default).

  • "our_label"

    The name of the "our" side of conflicts.

  • "their_label"

    The name of the "their" side of conflicts.

  • "ancestor_label"

    The name of the common ancestor side of conflicts.

  • "target_directory"

    Alternative checkout path to the working directory.

Example:

$repo -> checkout($repo -> head -> target, {
  'checkout_strategy' => { 'safe'  => 1 },
  'notify'    => [ 'all' ],
  'callbacks' => {
     'notify' => sub {
       my ($path, $why) = @_;

       print "File: $path: ", join(' ', @$why), "\n";
     },
     'progress' => sub {
        my ($path, $completed_steps, $total_steps) = @_;

        print "File: $path", "\n" if defined ($path);
        print "Progress: $completed_steps/$total_steps", "\n";
     }
  },
  'paths' => [ 'myscript.pl' ]
});

reset( $target, \%opts )

Reset the current HEAD to the given commit. Valid fields for the %opts hash are:

  • "type"

    Set the type of the reset to be performed. Valid values are: "soft" (the head will be moved to the commit), "mixed" (trigger a soft reset and replace the index with the content of the commit tree) or "hard" (trigger a "mixed" reset and the working directory will be replaced with the content of the index).

  • "paths"

    List of entries in the index to be updated from the target commit tree. This is particularly useful to implement "git reset HEAD -- file file" behaviour. Note, if this parameter is specified, a value of "mixed" will be used for "type" (setting "type" to "soft" or "hard" has no effect).

status( \%opts, [$file, $file, ...] )

Retrieve the status of files in the index and/or working directory. This function returns a hash reference with an entry for each $file, or all files if no file parameters are provided. Each $file entry has a list of "flags", which may include: "index_new", "index_modified", "index_deleted", "index_renamed", "worktree_new", "worktree_modified", "worktree_deleted", "worktree_renamed", "worktree_unreadable", "conflicted" and "ignored".

If $file has been renamed in either the index or worktree or both, $file will also have a corresponding entry "index" and/or "worktree", containing the previous filename "old_file".

Valid fields for the %opts hash are:

  • "flags"

    Flags for the status. Valid values include:

    • "include_untracked"

      Callbacks should be made on untracked files. These will only be made if the workdir files are included in the $show option.

    • "include_ignored"

      Callbacks should be made on ignored files. These will only be made if the ignored files get callbacks.

    • "include_unmodified"

      Include even unmodified files.

    • "exclude_submodules"

      Submodules should be skipped. This only applies if there are no pending typechanges to the submodule (either from or to another type).

    • "recurse_untracked_dirs"

      All files in untracked directories should be included. Normally if an entire directory is new, then just the top-level directory is included (with a trailing slash on the entry name). This flag includes all of the individual files in the directory instead.

    • "disable_pathspec_match"

      Each $file specified should be treated as a literal path, and not as a pathspec pattern.

    • "recurse_ignored_dirs"

      The contents of ignored directories should be included in the status. This is like doing git ls-files -o -i --exclude-standard with core git.

    • "renames_head_to_index"

      Rename detection should be processed between the head and the index.

    • "renames_index_to_workdir"

      Rename detection should be run between the index and the working directory.

    • "sort_case_sensitively"

      Override the native case sensitivity for the file system and force the output to be in case-sensitive order.

    • "sort_case_insensitively"

      Override the native case sensitivity for the file system and force the output to be in case-insensitive order.

    • "renames_from_rewrites"

      Rename detection should include rewritten files.

    • "no_refresh"

      Bypass the default status behavior of doing a "soft" index reload (i.e. reloading the index data if the file on disk has been modified outside Git::Raw).

    • "update_index"

      Refresh the stat cache in the index for files that are unchanged but have out-of-date stat information in the index. It will result in less work being done on subsequent calls to status. This is mutually exclusive with the "no_refresh" option.

    • "include_unreadable"

      Include unreadable files.

    • "include_unreadable_as_untracked"

      Include unreadable files as untracked files.

  • "show"

    One of the following values (Defaults to index_and_worktree):

    • "index_and_worktree"

    • "index"

    • "worktree"

Example:

my $opts = {
  'flags' => {
    'include_untracked'        => 1,
    'renames_head_to_index'    => 1,
    'renames_index_to_workdir' => 1,
  },
  'show' => 'index_and_worktree'
};
my $file_statuses = $repo -> status($opts);
while (my ($file, $status) = each %$file_statuses) {
  my $flags = $status -> {'flags'};
  print "File: $file: Status: ", join (' ', @$flags), "\n";

  if (grep { $_ eq 'index_renamed' } @$flags) {
    print "Index previous filename: ",
    $status -> {'index'} -> {'old_file'}, "\n";
  }

  if (grep { $_ eq 'worktree_renamed' } @$flags) {
    print "Worktree previous filename: ",
    $status -> {'worktree'} -> {'old_file'}, "\n";
  }
}

merge_base( @objects )

Find the merge base between @objects. Each element in @objects should be peelable to a Git::Raw::Commit object, that is, it should be a Git::Raw::Commit or Git::Raw::Reference object, or alternatively a commit id or commit id prefix.

merge_analysis( $reference )

Analyzes the given $reference and determines the opportunities for merging them into the HEAD of the repository. This function returns an array reference with optional members "normal", "up_to_date", "fast_forward" and/or "unborn".

  • "normal"

    A "normal" merge. Both HEAD and the given merge input have diverged from their common ancestor. The divergent commits must be merged.

  • "up_to_date"

    All given merge inputs are reachable from HEAD, meaning the repository is up-to-date and no merge needs to be performed.

  • "fast_forward"

    The given merge input is a fast-forward from HEAD and no merge needs to be performed. Instead, the given merge input may be checked out.

  • "unborn"

    The HEAD of the current repository is "unborn" and does not point to a valid commit. No merge can be performed, but the caller may wish to simply set HEAD to the target commit(s).

merge( $ref, [\%merge_opts, \%checkout_opts])

Merge the given $ref into HEAD, writing the results into the working directory. Any changes are staged for commit and any conflicts are written to the index. The index should be inspected for conflicts after this method completes and any conflicts should be resolved. At this stage a commit may be created to finalise the merge.

  • "flags"

    Merge flags. Valid values include:

    • "find_renames"

      Detect renames.

  • "file_flags"

    See Git::Raw::Index->merge() for options.

  • "favor"

    Specify content automerging behaviour. Valid values are "ours", "theirs", and "union".

  • "rename_threshold"

    Similarity metric for considering a file renamed (default is 50).

  • "target_limit"

    Maximum similarity sources to examine (overrides the "merge.renameLimit" configuration entry) (default is 200).

Example:

my $branch = Git::Raw::Branch -> lookup($repo, 'branch', 1);
my $analysis = $repo -> merge_analysis($branch);
my $merge_opts = {
  'favor' => 'theirs'
};
my $checkout_opts = {
  'checkout_strategy' => {
    'force' => 1
  }
};
$repo -> merge($branch1, $merge_opts, $checkout_opts);

ignore( $rules )

Add an ignore rules to the repository. The format of the rules is the same one of the .gitignore file (see the gitignore(5) manpage). Example:

$repo -> ignore("*.o\n");

path_is_ignored( $path )

Checks the ignore rules to see if they would apply to the given file. This indicates if the file would be ignored regardless of whether the file is already in the index or committed to the repository.

diff( [\%diff_opts] )

Compute the Git::Raw::Diff between the repo's default index and another tree. Valid fields for the %diff_opts hash are:

  • "tree"

    If provided, the diff is computed between "tree" and the repo's default index. The default is the repo's working directory.

  • "flags"

    Flags for generating the diff. Valid values include:

    • "reverse"

      Reverse the sides of the diff.

    • "include_ignored"

      Include ignored files in the diff.

    • "include_typechange"

      Enable the generation of typechange delta records.

    • "recurse_ignored_dirs"

      Even if "include_ignored" is specified, an entire ignored directory will be marked with only a single entry in the diff. This flag adds all files under the directory as ignored entries, too.

    • "include_untracked"

      Include untracked files in the diff.

    • "recurse_untracked_dirs"

      Even if "include_untracked" is specified, an entire untracked directory will be marked with only a single entry in the diff (core git behaviour). This flag adds all files under untracked directories as untracked entries, too.

    • "ignore_filemode"

      Ignore file mode changes.

    • "ignore_case"

      Use case insensitive filename comparisons.

    • "ignore_submodules"

      Treat all submodules as unmodified.

    • "ignore_whitespace"

      Ignore all whitespace.

    • "ignore_whitespace_change"

      Ignore changes in amount of whitespace.

    • "ignore_whitespace_eol"

      Ignore whitespace at end of line.

    • "skip_binary_check"

      Disable updating of the binary flag in delta records. This is useful when iterating over a diff if you don't need hunk and data callbacks and want to avoid having to load file completely.

    • "enable_fast_untracked_dirs"

      When diff finds an untracked directory, to match the behavior of core git, it scans the contents for ignored and untracked files. If all contents are ignore, then the directory is ignored. If any contents are not ignored, then the directory is untracked. This is extra work that may not matter in many cases. This flag turns off that scan and immediately labels an untracked directory as untracked (changing the behavior to not match core git).

    • "show_untracked_content"

      Include the content of untracked files. This implies "include_untracked".

    • "show_unmodified"

      Include the names of unmodified files.

    • "patience"

      Use the "patience diff" algorithm.

    • "minimal"

      Take extra time to find minimal diff.

    • "show_binary"

      Include the necessary deflate / delta information so that git apply can apply given diff information to binary files.

    • "force_text"

      Treat all files as text, disabling binary attributes and detection.

    • "force_binary"

      Treat all files as binary, disabling text diffs.

  • "prefix"

    • "a"

      The virtual "directory" to prefix to old file names in hunk headers. (Default is "a".)

    • "b"

      The virtual "directory" to prefix to new file names in hunk headers. (Default is "b".)

  • "context_lines"

    The number of unchanged lines that define the boundary of a hunk (and to display before and after)

  • "interhunk_lines"

    The maximum number of unchanged lines between hunk boundaries before the hunks will be merged into a one.

  • "paths"

    A list of paths to constrain diff.

blob( $buffer )

Create a new Git::Raw::Blob. Shortcut for Git::Raw::Blob->create().

branch( $name, $target )

Create a new Git::Raw::Branch. Shortcut for Git::Raw::Branch->create().

branches( [$type] )

Retrieve a list of Git::Raw::Branch objects. Possible values for $type include "local", "remote" or "all".

commit( $msg, $author, $committer, \@parents, $tree [, $update_ref ] )

Create a new Git::Raw::Commit. Shortcut for Git::Raw::Commit->create().

tag( $name, $msg, $tagger, $target )

Create a new Git::Raw::Tag. Shortcut for Git::Raw::Tag->create().

tags( [$type] )

Retrieve the list of annotated and/or lightweight tag objects. Possible values for $type include "all", "annotated" or "lightweight".

stash( $repo, $msg )

Save the local modifications to a new stash. Shortcut for Git::Raw::Stash->save().

remotes( )

Retrieve the list of Git::Raw::Remote objects.

refs( )

Retrieve the list of Git::Raw::Reference objects.

walker( )

Create a new Git::Raw::Walker. Shortcut for Git::Raw::Walker->create().

path( )

Retrieve the complete path of the repository.

workdir( [$new_dir] )

Retrieve the working directory of the repository. If $new_dir is passed, the working directory of the repository will be set to the directory.

blame( $path )

Retrieve blame information for $path. Returns a Git::Raw::Blame object.

cherry_pick( $commit, [\%merge_opts, \%checkout_opts, $mainline] )

Cherry-pick the given $commit, producing changes in the index and working directory. See Git::Raw::Repository->merge() for valid %merge_opts and %checkout_opts values. For merge commits $mainline specifies the parent.

revert( $commit, [\%merge_opts, \%checkout_opts, $mainline] )

Revert the given $commit, producing changes in the index and working directory. See Git::Raw::Repository->merge() for valid %merge_opts and %checkout_opts values. For merge commits $mainline specifies the parent.

revparse( $spec )

Parse the revision string $spec to for from, to and intent. Returns a list of objects in list context and the number of objects parsed from $spec in scalar context.

my ($from, $to) = $repo -> revparse('HEAD~..HEAD');
print "Range is $from -> $to", "\n";

state( )

Determine the state of the repository. One of the following values is returned:

  • "none"

    Normal state

  • "merge"

    Repository is in a merge.

  • "revert"

    Repository is in a revert.

  • "cherry_pick"

    Repository is in a cherry-pick.

  • "bisect"

    Repository is bisecting.

  • "rebase"

    Repository is rebasing.

  • "rebase_interactive"

    Repository is in an interactive rebase.

  • "rebase_merge"

    Repository is in an rebase merge.

  • "apply_mailbox"

    Repository is applying patches.

  • "mailbox_or_rebase"

    Repository is applying patches or rebasing.

state_cleanup( )

Remove all the metadata associated with an ongoing command like merge, revert, cherry-pick, etc.

message( )

Retrieve the content of git's prepared message i.e. ".git/MERGE_MSG".

is_empty( )

Check if the repository is empty.

is_bare( )

Check if the repository is bare.

is_shallow( )

Check if the repository is a shallow clone.

is_head_detached( )

Check if the repository's HEAD is detached, that is, it points directly to a commit.

AUTHOR

Alessandro Ghedini <alexbio@cpan.org>

Jacques Germishuys <jacquesg@cpan.org>

LICENSE AND COPYRIGHT

Copyright 2012 Alessandro Ghedini.

This program is free software; you can redistribute it and/or modify it under the terms of either: the GNU General Public License as published by the Free Software Foundation; or the Artistic License.

See http://dev.perl.org/licenses/ for more information.