Struct DuplicateDetector 

pub struct DuplicateDetector;

Duplicate detection service

Implementations

impl DuplicateDetector

pub fn find_duplicates( conn: &Connection, drive_root: &Path, ) -> Result<Vec<DuplicateGroup>>

Find all exact duplicate groups in the database

Returns groups where 2+ photos share the same SHA256 hash.

pub fn find_perceptual_duplicates( conn: &Connection, drive_root: &Path, exclude_ids: &HashSet<i64>, ) -> Result<Vec<DuplicateGroup>>

Find perceptually-similar duplicate groups (re-edits, re-saves, quality-adjusted exports). Computes phash on demand for any photo where it’s NULL and persists into photos.phash, then groups by Hamming distance ≤ 10 across the 64-bit hashes.

Excludes photos already grouped as exact duplicates so the UI doesn’t show the same pair twice.

pub fn find_perceptual_duplicates_with_progress( conn: &Connection, drive_root: &Path, exclude_ids: &HashSet<i64>, cancel: Option<&AtomicBool>, progress: impl FnMut(DuplicateProgress), ) -> Result<Vec<DuplicateGroup>>

pub fn calculate_wasted_space(conn: &Connection) -> Result<u64>

Wasted bytes across the duplicate listing.

Counts every detected duplicate group — both byte-exact dupes (same file_hash) AND perceptual ones (pHash match). The earlier version of this function only summed exact matches, which on a perceptual-only library reported “0 MB potentially wasted” even when the listing had dozens of groups. Now: walk the duplicate_groups table directly and sum (total_size - largest) per group, since the user’s win is keeping one copy and trashing the rest.