The tool could only act on duplicates through the default listing or the
manual per-group interactive prompt, with no way to resolve groups in bulk
by a rule. This blocked any scripted or automated use, and bulk resolution
is the top item in the README's proposed operations.
Add a resolver module that reduces each duplicate group to a single kept
file. --keep <newest|oldest|first|last|shortest|shallowest> selects the
keeper: newest/oldest by mtime, first/last and shortest/shallowest by
path, each resolved to a unique keeper through total-order path tiebreaks
so the outcome is deterministic and order-independent. Selection is a pure,
independently tested function; deletion is a separate step.
Deletion is guarded: --keep alone previews (KEEP/DELETE lines and a
would-free summary) and removes nothing, so a mistaken invocation is
harmless. --force performs the deletion, continues past individual
failures, reports freed space, and exits non-zero if any file could not be
removed. clap enforces that --keep excludes --interactive and that --force
requires --keep, so misuse fails before any file is touched.
Also fixes the interactive table's column width, which sized on path
component count instead of path string length so the columns never aligned.
The deduplication core was a three-stage streaming producer/consumer
(scan -> group-by-size -> group-by-hash) orchestrated by a Server struct
that ran the stages concurrently on a threadpool. Coordination relied on
AtomicBool flags polled in busy-wait loops, an Arc<Mutex<Vec>> hand-off
queue, DashMap stores, and a per-file Arc<Mutex<FileState>>. Every stage
had to receive its own hand-cloned Arcs with ad-hoc names, and the
busy-wait branches burned a core spinning while waiting for the producer.
Replace it with a staged batch pipeline over owned collections:
pipeline::run(&Params) drives scan -> group_by_size -> group_by_hash in
sequence, with rayon supplying parallelism. With no shared mutable state
between stages, all the Arc plumbing, the AtomicBool coordination, the
Mutex queue, the per-file lock, server.rs, and the threadpool/dashmap
dependencies are gone. FileInfo becomes plain data and each stage is a
pure, independently testable function.
Behavior is preserved except for three authorized deviations: progress
spinners render sequentially rather than concurrently under -p; the
interactive empty-result message prints once instead of twice; and the
interactive "Duplicate Set X of Y" total now counts the confirmed
duplicate groups shown instead of the internal candidate-hash store size.
This lands as the foundation for the cache, CLI, and TUI work that follows.
Sreedev Kodichath