RF: Move get_git_dir() into GitRepo

datalad/distribution/dataset.py

@@ -10,7 +10,6 @@
 """
 
 import logging
-from os.path import abspath
 from os.path import curdir
 from os.path import exists
 from os.path import join as opj
@@ -40,7 +39,6 @@
 from datalad.utils import optional_args, expandpath, is_explicit_path
 from datalad.utils import get_dataset_root
 from datalad.utils import dlabspath
-from datalad.distribution.utils import get_git_dir
 
 
 lgr = logging.getLogger('datalad.dataset')

datalad/distribution/tests/test_utils.py

@@ -13,7 +13,8 @@
 from os.path import join as opj
 
 from datalad.distribution.utils import _get_flexible_source_candidates
-from datalad.distribution.utils import get_git_dir
+
+from datalad.support.gitrepo import GitRepo
 
 from datalad.tests.utils import with_tempfile
 from datalad.tests.utils import eq_
@@ -64,7 +65,7 @@ def test_get_flexible_source_candidates():
 @with_tempfile
 def test_get_git_dir(path):
     # minimal, only missing coverage
-    assert_raises(RuntimeError, get_git_dir, path)
+    assert_raises(RuntimeError, GitRepo.get_git_dir, path)
 
     srcpath = opj(path, 'src')
     targetpath = opj(path, 'target')
@@ -74,9 +75,9 @@ def test_get_git_dir(path):
     if not on_windows:
         # with PY3 would also work with Windows 6+
         os.symlink(srcpath, targetgitpath)
-        eq_(srcpath, get_git_dir(targetpath))
+        eq_(srcpath, GitRepo.get_git_dir(targetpath))
         # cleanup for following test
         unlink(targetgitpath)
     with open(targetgitpath, 'w') as f:
         f.write('gitdir: {}'.format(srcpath))
-    eq_(srcpath, get_git_dir(targetpath))
+    eq_(srcpath, GitRepo.get_git_dir(targetpath))

datalad/distribution/utils.py

@@ -11,18 +11,15 @@
 """
 
 import logging
-import re
 
-from os.path import exists
-from os.path import isdir
 from os.path import join as opj
-from os.path import islink
 from os.path import isabs
 from os.path import normpath
 import posixpath
 
 from six.moves.urllib.parse import unquote as urlunquote
 
+from datalad.support.annexrepo import GitRepo
 from datalad.support.annexrepo import AnnexRepo
 from datalad.support.network import DataLadRI
 from datalad.support.network import URL
@@ -43,50 +40,13 @@ def _fixup_submodule_dotgit_setup(ds, relativepath):
     # move .git to superrepo's .git/modules, remove .git, create
     # .git-file
     path = opj(ds.path, relativepath)
-    src_dotgit = get_git_dir(path)
+    src_dotgit = GitRepo.get_git_dir(path)
 
     # at this point install always yields the desired result
     # just make sure
     assert(src_dotgit == '.git')
 
 
-def get_git_dir(path):
-    """figure out a repo's gitdir
-
-    '.git' might be a  directory, a symlink or a file
-
-    Parameter
-    ---------
-    path: str
-      currently expected to be the repos base dir
-
-    Returns
-    -------
-    str
-      relative path to the repo's git dir; So, default would be ".git"
-    """
-
-    from os.path import isfile
-
-    dot_git = opj(path, ".git")
-    if not exists(dot_git):
-        raise RuntimeError("Missing .git in %s." % path)
-    elif islink(dot_git):
-        # readlink cannot be imported on windows, but there should also
-        # be no symlinks
-        from os import readlink
-        git_dir = readlink(dot_git)
-    elif isdir(dot_git):
-        git_dir = ".git"
-    elif isfile(dot_git):
-        with open(dot_git) as f:
-            git_dir = f.readline()
-            if git_dir.startswith("gitdir:"):
-                git_dir = git_dir[7:]
-            git_dir = git_dir.strip()
-
-    return git_dir
-
 
 def _get_git_url_from_source(source):
     """Return URL for cloning associated with a source specification

datalad/interface/clean.py

@@ -19,11 +19,11 @@
 from ..consts import ANNEX_TEMP_DIR
 from ..consts import SEARCH_INDEX_DOTGITDIR
 
+from datalad.support.gitrepo import GitRepo
 from datalad.support.constraints import EnsureNone
 from datalad.distribution.dataset import EnsureDataset
 from datalad.distribution.dataset import require_dataset
 from datalad.distribution.dataset import datasetmethod
-from datalad.distribution.utils import get_git_dir
 from datalad.interface.annotate_paths import AnnotatePaths
 from datalad.interface.common_opts import recursion_flag
 from datalad.interface.common_opts import recursion_limit
@@ -90,7 +90,7 @@ def __call__(dataset=None, what=None, recursive=False, recursion_limit=None):
                 yield ap
                 continue
             d = ap['path']
-            gitdir = get_git_dir(d)
+            gitdir = GitRepo.get_git_dir(d)
             for dirpath, flag, msg, sing_pl in [
                 (ARCHIVES_TEMP_DIR, "cached-archives",
                  "temporary archive", ("directory", "directories")),

datalad/metadata/search.py

@@ -34,7 +34,7 @@
 from datalad.distribution.dataset import Dataset
 from datalad.distribution.dataset import datasetmethod, EnsureDataset, \
     require_dataset
-from datalad.distribution.utils import get_git_dir
+from datalad.support.gitrepo import GitRepo
 from datalad.support.param import Parameter
 from datalad.support.constraints import EnsureNone
 from datalad.support.constraints import EnsureInt
@@ -263,7 +263,7 @@ def __init__(self, ds, force_reindex=False, **kwargs):
 
         self.idx_obj = None
         # where does the bunny have the eggs?
-        self.index_dir = opj(self.ds.path, get_git_dir(self.ds.path), SEARCH_INDEX_DOTGITDIR)
+        self.index_dir = opj(self.ds.path, GitRepo.get_git_dir(ds), SEARCH_INDEX_DOTGITDIR)
         self._mk_search_index(force_reindex)
 
     def show_keys(self, mode):

datalad/support/gitrepo.py

@@ -802,6 +802,44 @@ def is_valid_repo(cls, path):
         """Returns if a given path points to a git repository"""
         return exists(opj(path, '.git'))
 
+    @staticmethod
+    def get_git_dir(repo):
+        """figure out a repo's gitdir
+
+        '.git' might be a  directory, a symlink or a file
+
+        Parameter
+        ---------
+        repo: path or Repo instance
+          currently expected to be the repos base dir
+
+        Returns
+        -------
+        str
+          relative path to the repo's git dir; So, default would be ".git"
+        """
+        if hasattr(repo, 'path'):
+            # repo instance like given
+            repo = repo.path
+        dot_git = op.join(repo, ".git")
+        if not op.exists(dot_git):
+            raise RuntimeError("Missing .git in %s." % repo)
+        elif op.islink(dot_git):
+            # readlink cannot be imported on windows, but there should also
+            # be no symlinks
+            from os import readlink
+            git_dir = readlink(dot_git)
+        elif op.isdir(dot_git):
+            git_dir = ".git"
+        elif op.isfile(dot_git):
+            with open(dot_git) as f:
+                git_dir = f.readline()
+                if git_dir.startswith("gitdir:"):
+                    git_dir = git_dir[7:]
+                git_dir = git_dir.strip()
+
+        return git_dir
+
     @property
     def config(self):
         """Get an instance of the parser for the persistent repository