class Rugged::Rebase

Public Class Methods

new(repo, branch, upstream[, onto][, options]) → new_rebase

Source

static VALUE rb_git_rebase_new(int argc, VALUE* argv, VALUE klass)
{
        int error = 0, exception = 0;
        git_rebase *rebase = NULL;
        git_repository *repo;
        git_annotated_commit *branch = NULL, *upstream = NULL, *onto = NULL;
        VALUE rb_repo, rb_branch, rb_upstream, rb_onto, rb_options;
        git_rebase_options options = GIT_REBASE_OPTIONS_INIT;

        rb_scan_args(argc, argv, "31:", &rb_repo, &rb_branch, &rb_upstream, &rb_onto, &rb_options);
        Data_Get_Struct(rb_repo, git_repository, repo);

        if ((exception = rugged_get_annotated_commit(&branch, rb_repo, rb_branch)))
                goto cleanup;

        if ((exception = rugged_get_annotated_commit(&upstream, rb_repo, rb_upstream)))
                goto cleanup;

        if (!NIL_P(rb_onto)) {
                if ((exception = rugged_get_annotated_commit(&onto, rb_repo, rb_onto)))
                        goto cleanup;
        }

        parse_rebase_options(&options, rb_options);

        error = git_rebase_init(&rebase, repo, branch, upstream, onto, &options);

cleanup:
        git_annotated_commit_free(branch);
        git_annotated_commit_free(upstream);
        git_annotated_commit_free(onto);

        if (exception)
                rb_jump_tag(exception);

        rugged_exception_check(error);

        return rugged_rebase_new(klass, rb_repo, rebase);
}

Initialize a new rebase operation. This will put repo in a rebase state.

branch is the branch to be rebased, and upstream is the branch which is to be the new base. If onto is specified, this will be the one base, and upstream is used to determine which commits from branch to use for the rebase.

You can pass merge and checkout options for the options hash, plus a few rebase-specific ones:

:quiet: If true, a flag will be set to ask tools to activate quiet mode. This does not do anything for libgit2/rugged but can be used to interact with other implementations.
:inmemory: Do not put the repository in a rebase state but perform all the operations in-memory. In case of conflicts, the rebase operation Hash returned by next will contain the index which can be used to resolve conflicts.
:rewrite_notes_ref: Name of the notes reference used to rewrite notes for rebased commits when finishing the rebase. If nil, it will be taken from the configuration.

Public Instance Methods

abort → nil

Source

static VALUE rb_git_rebase_abort(VALUE self)
{
        git_rebase *rebase;

        Data_Get_Struct(self, git_rebase, rebase);
        rugged_exception_check(git_rebase_abort(rebase));

        return Qnil;
}

Abort the rebase currently in process, resetting the repository and working directory to their state before the rebase began.

commit(author: nil, committer: committer, message: nil) → oid or nil

Source

static VALUE rb_git_rebase_commit(int argc, VALUE *argv, VALUE self)
{
        int error;
        git_oid id;
        git_rebase *rebase;
        git_signature *author = NULL, *committer;
        const char *message = NULL;
        VALUE rb_options, rb_author, rb_committer, rb_message;

        Data_Get_Struct(self, git_rebase, rebase);
        rb_scan_args(argc, argv, ":", &rb_options);

        Check_Type(rb_options, T_HASH);

        rb_author = rb_hash_aref(rb_options, CSTR2SYM("author"));
        rb_committer = rb_hash_aref(rb_options, CSTR2SYM("committer"));
        rb_message = rb_hash_aref(rb_options, CSTR2SYM("message"));

        if (!NIL_P(rb_message)) {
                Check_Type(rb_message, T_STRING);
                message = StringValueCStr(rb_message);
        }

        if (NIL_P(rb_committer))
                rb_raise(rb_eArgError, "Expected non-nil committer");
        else
                committer = rugged_signature_get(rb_committer, NULL);

        if (!NIL_P(rb_author))
                author = rugged_signature_get(rb_author, NULL);

        error = git_rebase_commit(&id, rebase, author, committer, NULL, message);
        git_signature_free(author);
        git_signature_free(committer);

        if (error == GIT_EAPPLIED) {
                giterr_clear();
                return Qnil;
        }

        rugged_exception_check(error);

        return rugged_create_oid(&id);
}

Commit the current patch. Any conflicts must have been resolved.

If author is nil, the existing author for the commit will be used. If message is nil, the existing message will be used.

Returns a string containing the oid of the newly created commit, or nil if there are no changes to be committed.

finish → nil

Source

static VALUE rb_git_rebase_finish(VALUE self, VALUE rb_sig)
{
        git_rebase *rebase;
        git_signature *sig;
        int error;

        Data_Get_Struct(self, git_rebase, rebase);
        sig = rugged_signature_get(rb_sig, NULL);
        error = git_rebase_finish(rebase, sig);
        git_signature_free(sig);

        rugged_exception_check(error);

        return Qnil;
}

Finish the rebase currently in progress once all patches have been applied.

inmemory_index → index

Source

static VALUE rb_git_rebase_inmemory_index(VALUE self)
{
        git_rebase *rebase;
        git_index *index;

        Data_Get_Struct(self, git_rebase, rebase);
        rugged_exception_check(git_rebase_inmemory_index(&index, rebase));

        return rugged_index_new(rb_cRuggedIndex, self, index);
}

Gets the index produced by the last operation, which is the result of next and which will be committed by the next invocation of commit. This is useful for resolving conflicts in an in-memory rebase before committing them.

This is only applicable for in-memory rebases; for rebases within a working directory, the changes were applied to the repository’s index.

next() → operation or nil

Source

static VALUE rb_git_rebase_next(VALUE self)
{
        int error;
        git_rebase *rebase;
        git_rebase_operation *operation;
        VALUE hash, val;

        Data_Get_Struct(self, git_rebase, rebase);
        error = git_rebase_next(&operation, rebase);
        if (error == GIT_ITEROVER)
                return Qnil;

        rugged_exception_check(error);

        /* Create the operation hash out of the relevant details */
        hash = rb_hash_new();

        val = rebase_operation_type(operation);
        rb_hash_aset(hash, CSTR2SYM("type"), val);

        if (operation->type != GIT_REBASE_OPERATION_EXEC) {
                val = rugged_create_oid(&operation->id);
                rb_hash_aset(hash, CSTR2SYM("id"), val);
        }

        if (operation->exec) {
                val = rb_str_new_utf8(operation->exec);
                rb_hash_aset(hash, CSTR2SYM("exec"), val);
        }

        return hash;
}

Perform the next step in the rebase. The returned operation is a Hash with its details or nil if there are no more operations to perform. The Hash contains some of the following entries:

:type: The type of operation being done. Can be one of :pick, :reword, :edit, :squash, :fixup or :exec.
:id: The id of the commit being cherry-picked. Exists for all but :exec operations.
:exec: If the operatin is :exec this is what the user asked to be executed.