ncakanghan/Archive
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

중복 탐지 범위 선택 추가 — 현재 폴더/소스 또는 전체

Browse Source 
기존엔 항상 라이브러리 전체를 대상으로 했다. 이제 중복 검토 패널에서
현재 보고 있는 뷰(폴더=하위 포함 / 소스)로 범위를 좁힐 수 있고,
폴더/소스 뷰에서는 기본값이 '현재 위치'다. 전체/검색/태그/스마트 뷰는
전체 라이브러리로 고정.

- dedupe::Scope { All | Source | Folder{recursive} } 도입, load_files/run에 적용
  (folder_snapshot과 동일한 경로 프리픽스 스코핑)
- start_dedupe에 folder_id/recursive 파라미터 추가
- DedupePanel: view() 기반 범위 세그먼트 컨트롤
- 테스트: folder_scope_restricts_detection (folderA=1그룹, 전체=2그룹, 누출 없음)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
강 한
2026-07-19 19:57:40 +09:00
co-authored by Claude Fable 5
parent a734ba21b9
commit 8ba0c24ce8
5 changed files with 165 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src-tauri/crates/archive-indexer/src/dedupe.rs
+91 -11
View File
@@ -59,7 +59,18 @@ struct FileMeta {
duration_ms: Option<i64>,
}
fn load_files(db: &Db, kind: u8, scope: Option<i64>) -> Result<Vec<FileMeta>> {
/// 중복 탐지 범위. `folder_snapshot`과 동일한 스코핑 규칙을 따른다.
#[derive(Clone, Copy, Debug)]
pub enum Scope {
/// 전체 라이브러리(모든 소스)
All,
/// 특정 소스
Source(i64),
/// 특정 폴더 (recursive=true면 하위 폴더 전체 포함)
Folder { id: i64, recursive: bool },
}
fn load_files(db: &Db, kind: u8, scope: Scope) -> Result<Vec<FileMeta>> {
Ok(db.with_read(move |conn| {
let base = "SELECT f.id, fo.path, f.name, f.size, f.width, f.height, f.mtime_ms, f.duration_ms
FROM files f JOIN folders fo ON fo.id = f.folder_id
@@ -78,14 +89,33 @@ fn load_files(db: &Db, kind: u8, scope: Option<i64>) -> Result<Vec<FileMeta>> {
})
};
match scope {
Some(sid) => {
Scope::All => {
let mut stmt = conn.prepare(base)?;
let rows = stmt.query_map([kind], map)?;
rows.collect()
}
Scope::Source(sid) => {
let mut stmt = conn.prepare(&format!("{base} AND f.source_id = ?2"))?;
let rows = stmt.query_map(rusqlite::params![kind, sid], map)?;
rows.collect()
}
None => {
let mut stmt = conn.prepare(base)?;
let rows = stmt.query_map([kind], map)?;
Scope::Folder { id, recursive: false } => {
let mut stmt = conn.prepare(&format!("{base} AND f.folder_id = ?2"))?;
let rows = stmt.query_map(rusqlite::params![kind, id], map)?;
rows.collect()
}
Scope::Folder { id, recursive: true } => {
// 폴더 경로 프리픽스로 하위 전체 (folder_snapshot과 동일)
let (src, path): (i64, String) = conn.query_row(
"SELECT source_id, path FROM folders WHERE id = ?1",
[id],
|r| Ok((r.get(0)?, r.get(1)?)),
)?;
let prefix = format!("{}{}%", path, std::path::MAIN_SEPARATOR);
let mut stmt = conn.prepare(&format!(
"{base} AND fo.source_id = ?2 AND (fo.path = ?3 OR fo.path LIKE ?4)"
))?;
let rows = stmt.query_map(rusqlite::params![kind, src, path, prefix], map)?;
rows.collect()
}
}
@@ -385,7 +415,7 @@ fn detect_similar_videos(
pub fn run(
db: &Arc<Db>,
tools: Option<&FfTools>,
scope_source: Option<i64>,
scope: Scope,
image_threshold: u32,
include_video: bool,
cancel: &AtomicBool,
@@ -397,14 +427,14 @@ pub fn run(
let mut grouped: std::collections::HashSet<i64> = std::collections::HashSet::new();
// 완전 동일 (이미지+영상 모두)
let mut all_files = load_files(db, 0, scope_source)?;
all_files.extend(load_files(db, 1, scope_source)?);
let mut all_files = load_files(db, 0, scope)?;
all_files.extend(load_files(db, 1, scope)?);
stats.exact_groups = detect_exact(db, &all_files, &mut grouped, cancel, |c, t| {
on_progress(DedupeProgress { phase: "exact".into(), current: c, total: t });
})?;
// 유사 이미지 (완전 동일 그룹에 속한 파일은 제외)
let images = load_files(db, 0, scope_source)?;
let images = load_files(db, 0, scope)?;
stats.image_groups = detect_similar_images(db, &images, image_threshold, &dismissed, &grouped, cancel, |c, t| {
on_progress(DedupeProgress { phase: "image".into(), current: c, total: t });
})?;
@@ -412,7 +442,7 @@ pub fn run(
// 유사 영상 (opt-in + ffmpeg 필요)
if include_video {
if let Some(tools) = tools {
let videos = load_files(db, 1, scope_source)?;
let videos = load_files(db, 1, scope)?;
stats.video_groups = detect_similar_videos(db, &videos, tools, 8, &dismissed, cancel, |c, t| {
on_progress(DedupeProgress { phase: "video".into(), current: c, total: t });
})?;
@@ -471,7 +501,7 @@ mod tests {
crate::pipeline::scan_source(&db, sid, root, &cancel, |_| {}).unwrap();
crate::meta::extract_image_meta(&db, sid, &cancel).unwrap();
let stats = run(&db, None, Some(sid), 5, false, &cancel, |_| {}).unwrap();
let stats = run(&db, None, Scope::Source(sid), 5, false, &cancel, |_| {}).unwrap();
// 완전 동일 그룹 1개 (original + exact_copy)
assert_eq!(stats.exact_groups, 1, "완전 동일 그룹");
@@ -493,6 +523,56 @@ mod tests {
assert_eq!(keepers, 1);
}
#[test]
fn folder_scope_restricts_detection() {
let (_d, media, db, sid) = setup();
let root = media.path();
std::fs::create_dir_all(root.join("folderA")).unwrap();
std::fs::create_dir_all(root.join("folderB")).unwrap();
// folderA: 완전 동일 쌍
let a = image::RgbImage::from_fn(200, 200, |x, y| {
image::Rgb([(x % 256) as u8, (y % 256) as u8, 10])
});
a.save(root.join("folderA/a1.png")).unwrap();
std::fs::copy(root.join("folderA/a1.png"), root.join("folderA/a2.png")).unwrap();
// folderB: 다른 완전 동일 쌍
let b = image::RgbImage::from_fn(200, 200, |x, y| {
image::Rgb([10, (x % 256) as u8, (y % 256) as u8])
});
b.save(root.join("folderB/b1.png")).unwrap();
std::fs::copy(root.join("folderB/b1.png"), root.join("folderB/b2.png")).unwrap();
let cancel = AtomicBool::new(false);
crate::pipeline::scan_source(&db, sid, root, &cancel, |_| {}).unwrap();
crate::meta::extract_image_meta(&db, sid, &cancel).unwrap();
let fa: i64 = db
.with_read(|c| c.query_row("SELECT id FROM folders WHERE name='folderA'", [], |r| r.get(0)))
.unwrap();
// folderA 범위: 완전 동일 그룹 1개 + folderA 외 파일 없음
let sa = run(&db, None, Scope::Folder { id: fa, recursive: true }, 5, false, &cancel, |_| {}).unwrap();
assert_eq!(sa.exact_groups, 1, "folderA 범위는 그룹 1개");
let outside: i64 = db
.with_read(|c| {
c.query_row(
"SELECT count(*) FROM dupe_members dm
JOIN files f ON f.id = dm.file_id
JOIN folders fo ON fo.id = f.folder_id
WHERE fo.name != 'folderA'",
[],
|r| r.get(0),
)
})
.unwrap();
assert_eq!(outside, 0, "folderA 범위인데 다른 폴더 파일이 포함됨");
// 전체 범위: 그룹 2개 (folderA + folderB)
let sall = run(&db, None, Scope::All, 5, false, &cancel, |_| {}).unwrap();
assert_eq!(sall.exact_groups, 2, "전체 범위는 그룹 2개");
}
#[test]
fn signature_distance_median() {
let a = vec![0u64, 0, 0, 0, 0];
src-tauri/src/commands.rs
+10 -1
View File
@@ -1039,6 +1039,8 @@ pub fn start_dedupe(
app: tauri::AppHandle,
state: tauri::State<'_, AppState>,
source_id: Option<i64>,
folder_id: Option<i64>,
recursive: Option<bool>,
image_threshold: u32,
include_video: bool,
on_progress: Channel<DedupeProgress>,
@@ -1048,6 +1050,13 @@ pub fn start_dedupe(
}
state.dedupe_cancel.store(false, Ordering::SeqCst);
// 범위 결정: 폴더 > 소스 > 전체 (folder_snapshot과 동일 우선순위)
let scope = match (folder_id, source_id) {
(Some(fid), _) => dedupe::Scope::Folder { id: fid, recursive: recursive.unwrap_or(true) },
(None, Some(sid)) => dedupe::Scope::Source(sid),
(None, None) => dedupe::Scope::All,
};
let db = state.db.clone();
let tools = state.tools.clone();
let cancel = state.dedupe_cancel.clone();
@@ -1059,7 +1068,7 @@ pub fn start_dedupe(
let result = dedupe::run(
&db,
tools.as_ref(),
source_id,
scope,
image_threshold,
include_video,
&cancel,