def get_dirty_paths_by_status(self) -> Dict[str, List[Path]]:
"""
Returns all paths that have a git status, grouped by change type.
These can be staged, unstaged, or untracked.
"""
if self._dirty_paths_by_status is not None:
return self._dirty_paths_by_status
debug_echo("Initializing dirty paths")
sub_out = subprocess.run(
["git", "status", "--porcelain", "-z", "':!.semgrep_logs/'"],
timeout=GIT_SH_TIMEOUT,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
)
git_status_output = sub_out.stdout.decode("utf-8", errors="replace")
debug_echo(f"Git status output: {git_status_output}")
output = zsplit(git_status_output)
debug_echo("finished getting dirty paths")
dirty_paths = bucketize(
output,
key=lambda line: line[0],
value_transform=lambda line: Path(line[3:]),
)
debug_echo(str(dirty_paths))
# Cache dirty paths
self._dirty_paths_by_status = dirty_paths
return dirty_paths
def get_dirty_paths_by_status(self) -> Dict[str, List[Path]]:
"""
Returns all paths that have a git status, grouped by change type.
These can be staged, unstaged, or untracked.
"""
output = zsplit(git.status("--porcelain", "-z").stdout.decode())
return bucketize(
output,
key=lambda line: line[0],
value_transform=lambda line: Path(line[3:]),
)
def get_git_status(self) -> GitStatus:
"""
Returns Absolute Paths to all files that are staged
Ignores files that are symlinks to directories
"""
import gitdb.exc # type: ignore
repo = get_git_repo()
if not repo or self._base_commit is None:
return GitStatus([], [], [], [])
try:
repo.rev_parse(self._base_commit)
except gitdb.exc.BadName:
raise ActionFailure(f"Unknown git ref '{self._base_commit}'")
# Output of git command will be relative to git project root
status_output = zsplit(
git.diff(
"--cached",
"--name-status",
"--no-ext-diff",
"-z",
"--diff-filter=ACDMRTUXB",
"--ignore-submodules",
self._base_commit,
).stdout.decode())
added = []
modified = []
removed = []
unmerged = []
while status_output:
code = status_output[0]
fname = status_output[1]
trim_size = 2
if not code.strip():
continue
if code == StatusCode.Untracked or code == StatusCode.Ignored:
continue
resolved_name = self._fname_to_path(repo, fname)
# If file is symlink to directory, skip
absolute_name = Path(repo.working_tree_dir) / fname
if absolute_name.is_symlink() and resolved_name.is_dir():
click.echo(
f"| Skipping {absolute_name} since it is a symlink to a directory: {resolved_name}",
err=True,
)
else:
# The following detection for unmerged codes comes from `man git-status`
if code == StatusCode.Unmerged:
unmerged.append(resolved_name)
if (code[0] == StatusCode.Renamed
): # code is RXXX, where XXX is percent similarity
removed.append(resolved_name)
fname = status_output[2]
trim_size += 1
added.append(resolved_name)
if code == StatusCode.Added:
added.append(resolved_name)
if code == StatusCode.Modified:
modified.append(resolved_name)
if code == StatusCode.Deleted:
removed.append(resolved_name)
status_output = status_output[trim_size:]
debug_echo(
f"Git status:\nadded: {added}\nmodified: {modified}\nremoved: {removed}\nunmerged: {unmerged}"
)
return GitStatus(added, modified, removed, unmerged)
