Replace GitPython with pygit2

conda_smithy/feedstock_io.py

@@ -7,22 +7,28 @@
 def get_repo(path, search_parent_directories=True):
     repo = None
     try:
-        import git
+        import pygit2
 
-        repo = git.Repo(
-            path, search_parent_directories=search_parent_directories
-        )
+        if search_parent_directories:
+            path = pygit2.discover_repository(path)
+        if path is not None:
+            try:
+                no_search = pygit2.enums.RepositoryOpenFlag.NO_SEARCH
+            except AttributeError:  # pygit2 < 1.14
+                no_search = pygit2.GIT_REPOSITORY_OPEN_NO_SEARCH
+
+            repo = pygit2.Repository(path, no_search)
     except ImportError:
         pass
-    except git.InvalidGitRepositoryError:
+    except pygit2.GitError:
         pass
 
     return repo
 
 
 def get_repo_root(path):
     try:
-        return get_repo(path).working_tree_dir
+        return get_repo(path).workdir.rstrip(os.path.sep)
     except AttributeError:
         return None
 
@@ -32,8 +38,13 @@ def set_exe_file(filename, set_exe=True):
 
     repo = get_repo(filename)
     if repo:
-        mode = "+x" if set_exe else "-x"
-        repo.git.execute(["git", "update-index", f"--chmod={mode}", filename])
+        index_entry = repo.index[os.path.relpath(filename, repo.workdir)]
+        if set_exe:
+            index_entry.mode |= all_execute_permissions
+        else:
+            index_entry.mode &= ~all_execute_permissions
+        repo.index.add(index_entry)
+        repo.index.write()
 
     mode = os.stat(filename).st_mode
     if set_exe:
@@ -54,7 +65,8 @@ def write_file(filename):
 
     repo = get_repo(filename)
     if repo:
-        repo.index.add([filename])
+        repo.index.add(os.path.relpath(filename, repo.workdir))
+        repo.index.write()
 
 
 def touch_file(filename):
@@ -68,7 +80,8 @@ def remove_file_or_dir(filename):
 
     repo = get_repo(filename)
     if repo:
-        repo.index.remove([filename], r=True)
+        repo.index.remove_all(["filename/**"])
+        repo.index.write()
     shutil.rmtree(filename)
 
 
@@ -77,7 +90,11 @@ def remove_file(filename):
 
     repo = get_repo(filename)
     if repo:
-        repo.index.remove([filename])
+        try:
+            repo.index.remove(os.path.relpath(filename, repo.workdir))
+            repo.index.write()
+        except OSError:  # this is specifically "file not in index"
+            pass
 
     os.remove(filename)
 
@@ -106,4 +123,5 @@ def copy_file(src, dst):
 
     repo = get_repo(dst)
     if repo:
-        repo.index.add([dst])
+        repo.index.add(os.path.relpath(dst, repo.workdir))
+        repo.index.write()

conda_smithy/feedstock_tokens.py

@@ -27,11 +27,12 @@
 import json
 import os
 import secrets
+import subprocess
 import tempfile
 import time
 from contextlib import contextmanager, redirect_stderr, redirect_stdout
 
-import git
+import pygit2
 import requests
 import scrypt
 
@@ -175,7 +176,10 @@ def feedstock_token_exists(user, project, token_repo, provider=None):
                 .replace("$GH_TOKEN", github_token)
                 .replace("${GH_TOKEN}", github_token)
             )
-            git.Repo.clone_from(_token_repo, tmpdir, depth=1)
+            subprocess.run(
+                ["git", "clone", "-q", "--depth", "1", _token_repo, tmpdir],
+                check=True,
+            )
             token_file = os.path.join(
                 tmpdir,
                 "tokens",
@@ -250,7 +254,10 @@ def is_valid_feedstock_token(
                 .replace("$GH_TOKEN", github_token)
                 .replace("${GH_TOKEN}", github_token)
             )
-            git.Repo.clone_from(_token_repo, tmpdir, depth=1)
+            subprocess.run(
+                ["git", "clone", "-q", "--depth", "1", _token_repo, tmpdir],
+                check=True,
+            )
             token_file = os.path.join(
                 tmpdir,
                 "tokens",
@@ -351,7 +358,12 @@ def register_feedstock_token(
                 .replace("$GH_TOKEN", github_token)
                 .replace("${GH_TOKEN}", github_token)
             )
-            repo = git.Repo.clone_from(_token_repo, tmpdir, depth=1)
+            # cloning via subprocesses to take advantage of git SSH support
+            subprocess.run(
+                ["git", "clone", "-q", "--depth", "1", _token_repo, tmpdir],
+                check=True,
+            )
+            repo = pygit2.Repository(tmpdir)
             token_file = os.path.join(
                 tmpdir,
                 "tokens",
@@ -413,17 +425,25 @@ def register_feedstock_token(
                 json.dump(token_data, fp)
 
             # push
-            repo.index.add(token_file)
-            repo.index.commit(
-                "[ci skip] [skip ci] [cf admin skip] ***NO_CI*** "
-                "added token for {}/{} on provider{}".format(
-                    user,
-                    project,
-                    "" if provider is None else " " + provider,
-                )
+            repo.index.add(os.path.relpath(token_file, tmpdir))
+            repo.index.write()
+            subprocess.run(
+                [
+                    "git",
+                    "commit",
+                    "-q",
+                    "-m",
+                    "[ci skip] [skip ci] [cf admin skip] ***NO_CI*** "
+                    f"added token for {user}/{project} on provider{'' if provider is None else ' ' + provider}",
+                ],
+                check=True,
+                cwd=tmpdir,
+            )
+
+            subprocess.run(
+                ["git", "pull", "-q", "--rebase"], check=True, cwd=tmpdir
             )
-            repo.remote().pull(rebase=True)
-            repo.remote().push()
+            subprocess.run(["git", "push", "-q"], check=True, cwd=tmpdir)
         except Exception as e:
             if "DEBUG_FEEDSTOCK_TOKENS" in os.environ:
                 raise e

conda_smithy/feedstocks.py

@@ -3,8 +3,7 @@
 import multiprocessing
 import os
 
-import git
-from git import GitCommandError, Repo
+import pygit2
 from github import Github
 
 from conda_smithy import github as smithy_github
@@ -50,11 +49,11 @@ def cloned_feedstocks(feedstocks_directory):
 
 def fetch_feedstock(repo_dir):
     """Git fetch --all a single git repository."""
-    repo = Repo(repo_dir)
+    repo = pygit2.Repository(repo_dir)
     for remote in repo.remotes:
         try:
             remote.fetch()
-        except GitCommandError:
+        except pygit2.GitError:
             print(f"Failed to fetch {remote.name} from {remote.url}.")
 
 
@@ -87,12 +86,14 @@ def clone_feedstock(feedstock_gh_repo, feedstocks_dir):
     clone_directory = os.path.join(feedstocks_dir, repo.name)
     if not os.path.exists(clone_directory):
         print(f"Cloning {repo.name}")
-        clone = Repo.clone_from(repo.clone_url, clone_directory)
-        clone.delete_remote("origin")
-    clone = Repo(clone_directory)
-    if "upstream" in [remote.name for remote in clone.remotes]:
-        clone.delete_remote("upstream")
-    clone.create_remote("upstream", url=repo.clone_url)
+        clone = pygit2.clone_repository(repo.clone_url, clone_directory)
+        clone.remotes.delete("origin")
+    else:
+        clone = pygit2.Repository(clone_directory)
+
+    if "upstream" in clone.remotes.names():
+        clone.remotes.delete("upstream")
+    clone.remotes.create("upstream", repo.clone_url)
 
 
 def clone_all(gh_org, feedstocks_dir):
@@ -193,8 +194,8 @@ def feedstocks_repos(
         random.shuffle(feedstocks)
 
     for feedstock in feedstocks:
-        repo = git.Repo(feedstock.directory)
-        upstream = repo.remotes.upstream
+        repo = pygit2.Repository(feedstock.directory)
+        upstream = repo.remotes["upstream"]
 
         if pull_up_to_date:
             print("Fetching ", feedstock.package)
@@ -231,23 +232,23 @@ def feedstocks_yaml(
     for repo, feedstock in feedstocks_repos(
         organization, feedstocks_directory, **feedstocks_repo_kwargs
     ):
-        upstream = repo.remotes.upstream
-        try:
-            refs = upstream.refs
-        except AssertionError:
-            # In early versions of gitpython and empty list of refs resulted in an
-            # assertion error (https://github.com/gitpython-developers/GitPython/pull/499).
-            refs = []
+        upstream = repo.remotes["upstream"]
+        refs = [
+            ref
+            for ref in repo.references.iterator()
+            if ref.name.startswith("refs/remotes/upstream/")
+        ]
 
         if not refs:
             upstream.fetch()
-            refs = upstream.refs
+            refs = [
+                ref
+                for ref in repo.references.iterator()
+                if ref.name.startswith("refs/remotes/upstream/")
+            ]
 
         for ref in refs:
-            remote_branch = (
-                ref.remote_head
-            )  # .replace('{}/'.format(gh_me.login), '')
-            if remote_branch.endswith("HEAD"):
+            if ref.name == "refs/remotes/upstream/HEAD":
                 continue
 
             try:
@@ -259,9 +260,8 @@ def feedstocks_yaml(
                     ) as fh:
                         content = "".join(fh.readlines())
                 else:
-                    blob = ref.commit.tree["recipe"]["meta.yaml"]
-                    stream = blob.data_stream
-                    content = stream.read().decode("utf-8")
+                    blob = repo[ref.resolve().target].tree["recipe/meta.yaml"]
+                    content = blob.data.decode("utf-8")
                 yaml = yaml_meta(content)
             except:
                 # Add a helpful comment so we know what we are working with and reraise.

conda_smithy/github.py

@@ -2,7 +2,7 @@
 from random import choice
 
 import github
-from git import Repo
+import pygit2
 from github import Github
 from github.GithubException import GithubException
 from github.Organization import Organization
@@ -191,18 +191,18 @@ def create_github_repo(args):
         print("Github repository already exists.")
 
     # Now add this new repo as a remote on the local clone.
-    repo = Repo(args.feedstock_directory)
+    repo = pygit2.Repository(args.feedstock_directory)
     remote_name = args.remote_name.strip()
     if remote_name:
-        if remote_name in [remote.name for remote in repo.remotes]:
+        if remote_name in repo.remotes.names():
             existing_remote = repo.remotes[remote_name]
             if existing_remote.url != gh_repo.ssh_url:
                 print(
                     f"Remote {remote_name} already exists, and doesn't point to {gh_repo.ssh_url} "
                     f"(it points to {existing_remote.url})."
                 )
         else:
-            repo.create_remote(remote_name, gh_repo.ssh_url)
+            repo.remotes.create(remote_name, gh_repo.ssh_url)
 
     if args.extra_admin_users is not None:
         for user in args.extra_admin_users:

environment.yml

@@ -21,7 +21,7 @@ dependencies:
   - jinja2
   - requests
   - pycryptodome
-  - gitpython
+  - pygit2
   - pygithub >=2,<3
   - ruamel.yaml
   - conda-forge-pinning

news/2120-replace-gitpython-with-pygit2.rst

@@ -0,0 +1,23 @@
+**Added:**
+
+* <news item>
+
+**Changed:**
+
+* Replaced the ``GitPython`` dependency with ``pygit2`` (#2120).
+
+**Deprecated:**
+
+* <news item>
+
+**Removed:**
+
+* <news item>
+
+**Fixed:**
+
+* Fixed compatibility with git index versions 3 and 4 (#2120).
+
+**Security:**
+
+* <news item>