def update_database(repo, fs_dir, do_checksum):
lists_of_files = status.compare_fs_db(repo, fs_dir, do_checksum, updating=True)
new = lists_of_files[config.NEW_FILES]
missing = lists_of_files[config.MISSING_FILES]
good = lists_of_files[config.GOOD_FILES]
different = lists_of_files[config.DIFFERENT_FILES]
moved = lists_of_files[config.MOVED_FILES]
# only skip MISSING
to_save = list(good)
to_update = list(different)
if not do_checksum:
# force checksum for database entry
to_update += new
else:
to_save += new
for fname, old_info in to_update:
fs_fullpath = os.path.join(fs_dir, fname)
info = common.get_file_info(fs_fullpath, do_checksum=True)
to_save.append((fname, info))
for fname, info in moved:
del info["moved_from"]
to_save.append((fname, info))
# sort to match DB order
to_save = sorted(to_save, key=lambda entry: entry[0])
tmp_db_file = "{}.tmp".format(repo.db_file)
with open(tmp_db_file, "w+") as tmp_db_f:
for fname, info in to_save:
common.print_a_file(fname, info, tmp_db_f)
try: os.remove(repo.db_file)
except OSError: pass
os.rename(tmp_db_file, repo.db_file)
log.warn("Database updated.")
log.info("{} entries untouched".format(len(good)))
log.info("{} entries added".format(len(missing)))
log.info("{} entries updated".format(len(different)))
log.info("{} entries removed".format(len(new)))
log.info("{} entries moved".format(len(moved)))
status.do_clean(repo)
def compare_fs_db(repo, fs_dir, do_checksum, updating=False):
progress = progress_on_fs_and_db(repo, fs_dir, do_checksum)
good, missing, new, different, moved = [], [], [], [], []
try:
while True:
state, diff, db_entry, fs_entry = next(progress)
try:
fs_fullpath, fs_relpath, fs_info = fs_entry
except Exception: pass
try:
db_relpath, db_info = db_entry
except Exception: pass
if state is FileState.OK:
good.append((db_relpath, db_info))
elif state is FileState.DIFFERENT:
assert diff
different.append((db_relpath, diff))
elif state is FileState.MISSING_IN_FS:
if not updating:
assert not os.path.exists("{}/{}".format(fs_dir, db_relpath))
missing.append((db_relpath, db_info))
elif state is FileState.MISSING_ON_DB:
if not updating:
command = '/usr/bin/grep "{}" "{}" --quiet'.format(fs_relpath, repo.db_file)
assert os.system(command) # assert !0 (text not found)
new.append((fs_relpath, fs_info))
else: # MOVED should not happend here
log.critical("Incorrect state: {}".format(state))
assert False # should not come here
except StopIteration:
pass
# compute moved files
for new_file_info in list(new):
new_file, new_info = new_file_info
if not do_checksum: # if checksum not computed before
fs_fullpath = os.path.join(fs_dir, new_file)
new_info = common.get_file_info(fs_fullpath, do_checksum=True)
missing_file_info = [(i, missing_file_info[0]) for i, missing_file_info in enumerate(missing)
if missing_file_info[1]["md5sum"] == new_info["md5sum"]]
if not missing_file_info:
continue
assert len(missing_file_info) == 1
idx, missing_file = missing_file_info[0]
info = new_info.copy()
info["moved_from"] = missing_file
moved.append((new_file, info))
log.warning("{} # actually moved from {}".format(new_file, missing_file))
missing.pop(idx)
new.remove(new_file_info)
return OrderedDict((
(config.GOOD_FILES, good),
(config.NEW_FILES, new),
(config.MISSING_FILES, missing),
(config.DIFFERENT_FILES, different),
(config.MOVED_FILES, moved),
))