robertvokac/bitrot
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Activity

Optimization: don't SELECT the path twice if it's not there

Browse Source
This commit is contained in:
Łukasz Langa 2020-05-17 21:58:30 +02:00
parent 45ab4501ee
commit 52677d2b5d
No known key found for this signature in database
GPG Key ID: B26995E310250568
1 changed files with 18 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
src/bitrot.py
View File

@ -221,7 +221,6 @@ class Bitrot(object):
if self.verbosity: if self.verbosity:
self.report_progress(current_size, total_size) self.report_progress(current_size, total_size)
missing_paths.discard(p_uni)
try: try:
new_sha1 = sha1(p, self.chunk_size) new_sha1 = sha1(p, self.chunk_size)
except (IOError, OSError) as e: except (IOError, OSError) as e:
@ -231,24 +230,37 @@ class Bitrot(object):
), ),
file=sys.stderr, file=sys.stderr,
) )
missing_paths.discard(p_uni)
continue continue
cur.execute('SELECT mtime, hash, timestamp FROM bitrot WHERE path=?', if p_uni not in missing_paths:
(p_uni,)) # We are not expecting this path, it wasn't in the database yet.
row = cur.fetchone() # It's either new or a rename. Let's handle that.
if not row:
stored_path = self.handle_unknown_path( stored_path = self.handle_unknown_path(
cur, p_uni, new_mtime, new_sha1, paths_uni, hashes cur, p_uni, new_mtime, new_sha1, paths_uni, hashes
) )
self.maybe_commit(conn) self.maybe_commit(conn)
if p_uni == stored_path: if p_uni == stored_path:
new_paths.append(p_uni) new_paths.append(p_uni)
missing_paths.discard(p_uni)
else: else:
renamed_paths.append((stored_path, p_uni)) renamed_paths.append((stored_path, p_uni))
missing_paths.discard(stored_path) missing_paths.discard(stored_path)
continue continue
# At this point we know we're seeing an expected file.
missing_paths.discard(p_uni)
cur.execute('SELECT mtime, hash, timestamp FROM bitrot WHERE path=?',
(p_uni,))
row = cur.fetchone()
if not row:
print(
'\rwarning: path disappeared from the database while running:',
p_uni,
file=sys.stderr,
)
continue
stored_mtime, stored_sha1, stored_ts = row stored_mtime, stored_sha1, stored_ts = row
if int(stored_mtime) != new_mtime: if int(stored_mtime) != new_mtime:
updated_paths.append(p_uni) updated_paths.append(p_uni)