Add support for filtering orphaned files with --owner orphan #1841
Conversation
Extends the --owner filter to support finding files with no valid user/group by using the 'orphan' keyword. This provides equivalent functionality to find's -nouser and -nogroup flags. Examples: fd --owner orphan # equivalent to find -nouser fd --owner :orphan # equivalent to find -nogroup fd --owner orphan:orphan # both -nouser and -nogroup Implementation: - Added Check::Orphaned variant to owner filter - Extended OwnerFilter::from_string to recognize 'orphan' keyword - Updated OwnerFilter::matches to check if uid/gid maps to valid user/group - Added unit tests for orphan parsing
src/filter/owner.rs
Outdated
|
|
||
| self.uid.check(md.uid()) && self.gid.check(md.gid()) | ||
| let uid_match = match self.uid { | ||
| Check::Orphaned => User::from_uid(md.uid().into()).ok().flatten().is_none(), |
There was a problem hiding this comment.
It might be worth using a cache, so we don't have to do a lookup for every single file. We could probably even use a Set of uids known not to be orphaned.
src/filter/owner.rs
Outdated
| Check::Equal(x) => v == *x, | ||
| Check::NotEq(x) => v != *x, | ||
| Check::Ignore => true, | ||
| Check::Orphaned => unreachable!("Orphaned check handled in OwnerFilter::matches"), |
There was a problem hiding this comment.
I don't really like how the caller is responsible for handling this separately. It would be nice if we could just include a Set of allowed ids in the Orphaned vairant, but I think on some systems, the user ids could be dynamic, and reading from /etc/passwd isn't sufficient.
Another option could be to have the Orphaned variant include a function to check if an id exists, or even just a value indicating if it is for user ids or group ids.
src/filter/owner.rs
Outdated
| both_negate:"!4:!3" => Ok(OwnerFilter { uid: NotEq(4), gid: NotEq(3) }), | ||
| uid_not_gid:"6:!8" => Ok(OwnerFilter { uid: Equal(6), gid: NotEq(8) }), | ||
|
|
||
| orphan_uid: "orphan" => Ok(OwnerFilter { uid: Orphaned, gid: Ignore }), |
There was a problem hiding this comment.
It is possible that a user or group is actually named "orphan". I think it would probably be better to use a a different option, or maybe use something that isn't a valid username like "-".
src/cli.rs
Outdated
| /// Filter files by their user and/or group. | ||
| /// Format: [(user|uid)][:(group|gid)]. Either side is optional. | ||
| /// Precede either side with a '!' to exclude files instead. | ||
| /// Use 'orphan' to match files with no valid user/group. |
There was a problem hiding this comment.
If we move forward with this, you'll also need to update the man page and CHANGELOG
Changes based on maintainer feedback: - Changed keyword from 'orphan' to '-' to avoid conflicts with actual usernames - Refactored Check::Orphan to include validator function (fn(u32) -> bool) - Simplified matches() to use check() uniformly - no special case handling - Added caching with LazyLock<Mutex<HashSet>> to avoid repeated uid/gid lookups - Updated documentation (man page and CHANGELOG) The new design stores the validation logic in the Orphan variant itself, making it self-contained and consistent with other Check variants. All checks now go through the generic check() method.
|
Addressed all feedback: changed keyword to |
|
bump @tmccombs |
|
@tmccombs I should have fixed it |
|
bump @tmccombs is there sth I should fix? |
src/filter/owner.rs
Outdated
| use std::sync::{LazyLock, Mutex}; | ||
|
|
||
| #[derive(Clone, Copy, Debug, PartialEq, Eq)] | ||
| static VALID_UIDS: LazyLock<Mutex<HashSet<u32>>> = LazyLock::new(|| Mutex::new(HashSet::new())); |
There was a problem hiding this comment.
This will not be performant. I'd use a thread-local instead. You should also use something like a HashMap<u32, bool> so that you can cache both negative and positive results.
tmccombs
left a comment
There was a problem hiding this comment.
Mostly looks good. Just address @tavianator's comment.
- Replace LazyLock<Mutex<HashSet>> with thread_local RefCell<HashMap> - Cache both positive and negative lookup results - Avoids mutex contention in parallel traversal
|
@tavianator I should have addressed ur review in the last commit |
@tmccombs |
|
@tavianator bump. ty in advance |
Extends the --owner filter to support finding files with no valid
user/group by using the 'orphan' keyword. This provides equivalent
functionality to find's -nouser and -nogroup flags.
Examples:
fd --owner orphan # equivalent to find -nouser
fd --owner :orphan # equivalent to find -nogroup
fd --owner orphan:orphan # both -nouser and -nogroup
Implementation: