From 28df74b7c9ce9d104fa803032322d4bb10bda7dc Mon Sep 17 00:00:00 2001
From: Ilya Kulakov <kulakov.ilya@gmail.com>
Date: Fri, 4 Dec 2020 18:33:51 +0600
Subject: [PATCH] Use git to get list of submodules

Parsing .gitmodules may yield paths that are ignored by git.
Refs #85
---
 git_archive_all.py  | 55 ++++++++++++++++++++++++++++-----------------
 git_archive_all.pyi |  5 ++++-
 2 files changed, 39 insertions(+), 21 deletions(-)

diff --git a/git_archive_all.py b/git_archive_all.py
index 5ace362..8e58744 100755
--- a/git_archive_all.py
+++ b/git_archive_all.py
@@ -345,30 +345,19 @@ def walk_git_files(self, repo_path=fspath('')):
                 self.run_git_shell('git submodule init', repo_abspath)
                 self.run_git_shell('git submodule update', repo_abspath)
 
-            try:
-                repo_gitmodules_abspath = path.join(repo_abspath, fspath(".gitmodules"))
-
-                with open(repo_gitmodules_abspath) as f:
-                    lines = f.readlines()
-
-                for l in lines:
-                    m = re.match("^\\s*path\\s*=\\s*(.*)\\s*$", l)
+            for repo_submodule_path in self.list_repo_submodules(repo_abspath):  # relative to repo_path
+                if self.is_file_excluded(repo_abspath, repo_submodule_path):
+                    continue
 
-                    if m:
-                        repo_submodule_path = fspath(m.group(1))  # relative to repo_path
-                        main_repo_submodule_path = path.join(repo_path, repo_submodule_path)  # relative to main_repo_abspath
+                main_repo_submodule_path = path.join(repo_path, repo_submodule_path)  # relative to main_repo_abspath
 
-                        if self.is_file_excluded(repo_abspath, repo_submodule_path):
-                            continue
+                for main_repo_submodule_file_path in self.walk_git_files(main_repo_submodule_path):
+                    repo_submodule_file_path = path.relpath(main_repo_submodule_file_path, repo_path)  # relative to repo_path
 
-                        for main_repo_submodule_file_path in self.walk_git_files(main_repo_submodule_path):
-                            repo_submodule_file_path = path.relpath(main_repo_submodule_file_path, repo_path)  # relative to repo_path
-                            if self.is_file_excluded(repo_abspath, repo_submodule_file_path):
-                                continue
+                    if self.is_file_excluded(repo_abspath, repo_submodule_file_path):
+                        continue
 
-                            yield main_repo_submodule_file_path
-            except IOError:
-                pass
+                    yield main_repo_submodule_file_path
         finally:
             self._check_attr_gens[repo_abspath].close()
             del self._check_attr_gens[repo_abspath]
@@ -538,6 +527,9 @@ def get_git_version(cls):
 
     @classmethod
     def list_repo_files(cls, repo_abspath):
+        """
+        Return a list of all files as seen by git in a given repo.
+        """
         repo_file_paths = cls.run_git_shell(
             'git ls-files -z --cached --full-name --no-empty-directory',
             cwd=repo_abspath
@@ -551,6 +543,29 @@ def list_repo_files(cls, repo_abspath):
 
         return repo_file_paths
 
+    @classmethod
+    def list_repo_submodules(cls, repo_abspath):
+        """
+        Return a list of all direct submodules as seen by git in a given repo.
+        """
+        if sys.platform.startswith('win32'):
+            shell_command = 'git submodule foreach --quiet "\\"{0}\\" -c \\"from __future__ import print_function; print(\'"$sm_path"\', end=chr(0))\\""'
+        else:
+            shell_command = 'git submodule foreach --quiet \'"{0}" -c "from __future__ import print_function; print(\\"$sm_path\\", end=chr(0))"\''
+
+        python_exe = sys.executable or 'python'
+        shell_command = shell_command.format(python_exe)
+
+        repo_submodule_paths = cls.run_git_shell(shell_command, cwd=repo_abspath)
+        repo_submodule_paths = repo_submodule_paths.split(b'\0')[:-1]
+
+        if sys.platform.startswith('win32'):
+            repo_submodule_paths = (git_fspath(p.replace(b'/', b'\\')) for p in repo_submodule_paths)
+        else:
+            repo_submodule_paths = map(git_fspath, repo_submodule_paths)
+
+        return repo_submodule_paths
+
 
 def main(argv=None):
     if argv is None:
diff --git a/git_archive_all.pyi b/git_archive_all.pyi
index 28733b9..710c1ea 100644
--- a/git_archive_all.pyi
+++ b/git_archive_all.pyi
@@ -64,4 +64,7 @@ class GitArchiver(object):
     def get_git_version(cls) -> Optional[Tuple[int]]: ...
 
     @classmethod
-    def list_repo_files(cls, repo_abspath: PathStr) -> Generator[PathStr, None, None]: ...
\ No newline at end of file
+    def list_repo_files(cls, repo_abspath: PathStr) -> Generator[PathStr, None, None]: ...
+
+    @classmethod
+    def list_repo_submodules(cls, repo_abspath: PathStr) -> Generator[PathStr, None, None]: ...