filter-repo: move sanity_check to put analyze functions before filtering ones

Signed-off-by: Elijah Newren <newren@gmail.com>
This commit is contained in:
Elijah Newren 2018-12-25 07:44:49 -08:00
parent fc90cf8ca9
commit 9887dd5cbe

View File

@ -1891,67 +1891,6 @@ class FilteringOptions(object):
FilteringOptions.sanity_check_args(args)
return args
def sanity_check(refs, is_bare):
def abort(reason):
raise SystemExit(
"Aborting: Refusing to overwrite repo history since this does not\n"
"look like a fresh clone.\n"
" ("+reason+")\n"
"To override, use --force.")
# Make sure repo is fully packed, just like a fresh clone would be
output = subprocess.check_output('git count-objects -v'.split())
stats = dict(x.split(': ') for x in output.splitlines())
if stats['count'] != '0' or stats['packs'] != '1':
abort("expected freshly packed repo")
# Make sure there is precisely one remote, named "origin"
output = subprocess.check_output('git remote'.split()).strip()
if output != "origin":
abort("expected one remote, origin")
# Avoid letting people running with weird setups and overwriting GIT_DIR
# elsewhere
git_dir = GitUtils.determine_git_dir()
if is_bare and git_dir != '.':
abort("GIT_DIR must be .")
elif not is_bare and git_dir != '.git':
abort("GIT_DIR must be .git")
# Make sure that all reflogs have precisely one entry
reflog_dir=os.path.join(git_dir, 'logs')
for root, dirs, files in os.walk(reflog_dir):
for filename in files:
pathname = os.path.join(root, filename)
with open(pathname) as f:
if len(f.read().splitlines()) > 1:
shortpath = pathname[len(reflog_dir)+1:]
abort("expected at most one entry in the reflog for " + shortpath)
# Make sure there are no stashed changes
if 'refs/stash' in refs:
abort("has stashed changes")
# Do extra checks in non-bare repos
if not is_bare:
# Avoid uncommitted, unstaged, or untracked changes
if subprocess.call('git diff --staged'.split()):
abort("you have uncommitted changes")
if subprocess.call('git diff --quiet'.split()):
abort("you have unstaged changes")
if len(subprocess.check_output('git ls-files -o'.split())) > 0:
abort("you have untracked changes")
# Avoid unpushed changes
for refname, rev in refs.iteritems():
if not refname.startswith('refs/heads/'):
continue
origin_ref = refname.replace('refs/heads/', 'refs/remotes/origin/')
if origin_ref not in refs:
abort('{} exists, but {} not found'.format(refname, origin_ref))
if rev != refs[origin_ref]:
abort('{} does not match {}'.format(refname, origin_ref))
def analyze_commit(stats, graph, commit, parents, date, file_changes):
def equiv_class(filename):
return stats['equivalence'].get(filename, (filename,))
@ -2393,6 +2332,67 @@ def do_analysis(args, git_dir):
write_report(reportdir, stats)
sys.stdout.write("done.\n")
def sanity_check(refs, is_bare):
def abort(reason):
raise SystemExit(
"Aborting: Refusing to overwrite repo history since this does not\n"
"look like a fresh clone.\n"
" ("+reason+")\n"
"To override, use --force.")
# Make sure repo is fully packed, just like a fresh clone would be
output = subprocess.check_output('git count-objects -v'.split())
stats = dict(x.split(': ') for x in output.splitlines())
if stats['count'] != '0' or stats['packs'] != '1':
abort("expected freshly packed repo")
# Make sure there is precisely one remote, named "origin"
output = subprocess.check_output('git remote'.split()).strip()
if output != "origin":
abort("expected one remote, origin")
# Avoid letting people running with weird setups and overwriting GIT_DIR
# elsewhere
git_dir = GitUtils.determine_git_dir()
if is_bare and git_dir != '.':
abort("GIT_DIR must be .")
elif not is_bare and git_dir != '.git':
abort("GIT_DIR must be .git")
# Make sure that all reflogs have precisely one entry
reflog_dir=os.path.join(git_dir, 'logs')
for root, dirs, files in os.walk(reflog_dir):
for filename in files:
pathname = os.path.join(root, filename)
with open(pathname) as f:
if len(f.read().splitlines()) > 1:
shortpath = pathname[len(reflog_dir)+1:]
abort("expected at most one entry in the reflog for " + shortpath)
# Make sure there are no stashed changes
if 'refs/stash' in refs:
abort("has stashed changes")
# Do extra checks in non-bare repos
if not is_bare:
# Avoid uncommitted, unstaged, or untracked changes
if subprocess.call('git diff --staged'.split()):
abort("you have uncommitted changes")
if subprocess.call('git diff --quiet'.split()):
abort("you have unstaged changes")
if len(subprocess.check_output('git ls-files -o'.split())) > 0:
abort("you have untracked changes")
# Avoid unpushed changes
for refname, rev in refs.iteritems():
if not refname.startswith('refs/heads/'):
continue
origin_ref = refname.replace('refs/heads/', 'refs/remotes/origin/')
if origin_ref not in refs:
abort('{} exists, but {} not found'.format(refname, origin_ref))
if rev != refs[origin_ref]:
abort('{} does not match {}'.format(refname, origin_ref))
def tweak_commit(args, commit):
def filename_matches(path_expression, pathname):
if path_expression == '':