Refs #78. Fix handling of export-ignore for directories

Files contained by ignored directories should be ignored as well unless export-ignore is explicitly unset either for the file itself or for any preceding directory.
Kentzo · Feb 10, 2020 · 43f0728 · 43f0728
1 parent ed50459
commit 43f0728
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 4 deletions.
diff --git a/git_archive_all.py b/git_archive_all.py
@@ -165,6 +165,7 @@ def __init__(self, prefix='', exclude=True, force_sub=False, extra=None, main_re
             If None, tries to resolve via Git's CLI.
         """
         self._check_attr_gens = {}
+        self._ignored_paths_cache = {}
 
         if git_version is None:
             git_version = self.get_git_version()
@@ -271,9 +272,26 @@ def is_file_excluded(self, repo_abspath, repo_file_path):
 
         @return: True if file should be excluded. Otherwise False.
         """
-        next(self._check_attr_gens[repo_abspath])
-        attrs = self._check_attr_gens[repo_abspath].send(repo_file_path)
-        return attrs['export-ignore'] == b'set'
+        cache = self._ignored_paths_cache.setdefault(repo_abspath, {})
+
+        if repo_file_path not in cache:
+            next(self._check_attr_gens[repo_abspath])
+            attrs = self._check_attr_gens[repo_abspath].send(repo_file_path)
+            export_ignore_attr = attrs['export-ignore']
+
+            if export_ignore_attr == b'set':
+                cache[repo_file_path] = True
+            elif export_ignore_attr == b'unset':
+                cache[repo_file_path] = False
+            else:
+                repo_file_dir_path = path.dirname(repo_file_path)
+
+                if repo_file_dir_path:
+                    cache[repo_file_path] = self.is_file_excluded(repo_abspath, repo_file_dir_path)
+                else:
+                    cache[repo_file_path] = False
+
+        return cache[repo_file_path]
 
     def archive_all_files(self, archiver):
         """

diff --git a/git_archive_all.pyi b/git_archive_all.pyi
@@ -24,6 +24,8 @@ class GitArchiver(object):
     LOG: ClassVar[logging.Logger]
 
     _check_attr_gens: Dict[str, CheckGitAttrGen]
+    _ignored_paths_cache: Dict[PathStr, Dict[PathStr, bool]]
+
     git_version: Optional[Tuple[int]]
     main_repo_abspath: PathStr
     prefix: PathStr

diff --git a/test_git_archive_all.py b/test_git_archive_all.py
@@ -323,6 +323,16 @@ def make_actual_tree(tar_file):
     b'\"\\\xc2.dat\"': FileRecord('Although practicality beats purity.')
 })
 
+ignore_dir = {
+    '.gitattributes': FileRecord('.gitattributes export-ignore\n**/src export-ignore\ndata/src/__main__.py -export-ignore', excluded=True),
+    '__init__.py': FileRecord('#Beautiful is better than ugly.'),
+    'data': DirRecord({
+        'src': DirRecord({
+            '__init__.py': FileRecord('#Explicit is better than implicit.', excluded=True),
+            '__main__.py': FileRecord('#Simple is better than complex.')
+        })
+    })
+}
 
 skipif_file_darwin = pytest.mark.skipif(sys.platform.startswith('darwin'), reason='Invalid macOS filename.')
 skipif_file_win32 = pytest.mark.skipif(sys.platform.startswith('win32'), reason="Invalid Windows filename.")
@@ -349,7 +359,8 @@ def make_actual_tree(tar_file):
     pytest.param(backslash_base, id='Backslash', marks=skipif_file_win32),
     pytest.param(backslash_quoted, id='Backslash (Quoted)', marks=skipif_file_win32),
     pytest.param(non_unicode_backslash_base, id='Non-Unicode Backslash', marks=[skipif_file_win32, skipif_file_darwin]),
-    pytest.param(non_unicode_backslash_quoted, id='Non-Unicode Backslash (Quoted)', marks=[skipif_file_win32, skipif_file_darwin])
+    pytest.param(non_unicode_backslash_quoted, id='Non-Unicode Backslash (Quoted)', marks=[skipif_file_win32, skipif_file_darwin]),
+    pytest.param(ignore_dir, id='Ignore Directory')
 ])
 def test_ignore(contents, tmpdir, git_env, monkeypatch):
     """