videre_core/
library_stats.rs

1//! Aggregate library statistics for dashboard-style callers.
2//! Plain queries over an open `rusqlite::Connection`, shared source of truth
3//! for `videre report`'s stats tile, `videre stats`, and any other embedder.
4//! See docs/superpowers/specs/2026-07-31-dashboard-stats-backend-design.md
5//! (Pass A) for what is and isn't in scope.
6
7use rusqlite::{Connection, Result};
8use serde::Serialize;
9
10#[derive(Debug, Clone, PartialEq, Default, Serialize)]
11pub struct LibraryStats {
12    pub total_files: i64,
13    pub total_size_bytes: i64,
14    pub total_photos: i64,
15    pub total_videos: i64,
16    pub duplicate_group_count: i64,
17    pub duplicate_file_count: i64,
18    pub wasted_bytes: i64,
19    pub faces_detected: i64,
20    pub people_named: i64,
21    /// One entry per model with an embedding database for this library.
22    /// Empty when nothing has been embedded, which is a normal state.
23    #[serde(default)]
24    pub embeddings: Vec<crate::embeddings_db::ModelEmbeddingCount>,
25}
26
27use crate::db::table_exists;
28
29const PHOTO_MIME_LIST: &str =
30    "'image/jpeg','image/png','image/gif','image/webp','image/bmp','image/tiff','image/heic'";
31const VIDEO_MIME_LIST: &str = "'video/quicktime','video/mp4'";
32
33const PHOTO_EXTS: &str = "'jpg','jpeg','png','gif','webp','bmp','tiff','heic','dng'";
34const VIDEO_EXTS: &str = "'mov','mp4'";
35
36// `VIDEO_EXTS`'s values must stay in sync with
37// `crate::embeddings::is_video_ext` (the shared "is this a video" check).
38// The SQL below uses `lower(ext)` so this list stays case-insensitive,
39// matching that helper.
40
41pub fn compute(conn: &Connection) -> Result<LibraryStats> {
42    let total_files: i64 = conn.query_row("SELECT COUNT(*) FROM file_hashes", [], |r| r.get(0))?;
43    let total_size_bytes: i64 =
44        conn.query_row("SELECT COALESCE(SUM(size_bytes), 0) FROM file_hashes", [], |r| r.get(0))?;
45    let total_photos: i64 = conn.query_row(
46        &format!(
47            "SELECT COUNT(*) FROM file_hashes
48             WHERE mime IN ({PHOTO_MIME_LIST})
49                OR (mime IS NULL AND lower(ext) IN ({PHOTO_EXTS}))"
50        ),
51        [],
52        |r| r.get(0),
53    )?;
54    let total_videos: i64 = conn.query_row(
55        &format!(
56            "SELECT COUNT(*) FROM file_hashes
57             WHERE mime IN ({VIDEO_MIME_LIST})
58                OR (mime IS NULL AND lower(ext) IN ({VIDEO_EXTS}))"
59        ),
60        [],
61        |r| r.get(0),
62    )?;
63
64    let duplicate_group_count: i64 = conn.query_row(
65        "SELECT COUNT(*) FROM \
66         (SELECT hash FROM file_hashes GROUP BY hash HAVING COUNT(*) > 1)",
67        [],
68        |r| r.get(0),
69    )?;
70    let duplicate_file_count: i64 = conn.query_row(
71        "SELECT COUNT(*) FROM file_hashes \
72         WHERE hash IN (SELECT hash FROM file_hashes GROUP BY hash HAVING COUNT(*) > 1)",
73        [],
74        |r| r.get(0),
75    )?;
76    let wasted_bytes: i64 = conn.query_row(
77        "SELECT COALESCE(SUM(size_bytes * (cnt - 1)), 0) FROM \
78         (SELECT hash, size_bytes, COUNT(*) as cnt \
79          FROM file_hashes GROUP BY hash HAVING cnt > 1)",
80        [],
81        |r| r.get(0),
82    )?;
83
84    let (faces_detected, people_named) = if table_exists(conn, "faces")? {
85        let faces_detected: i64 = conn.query_row("SELECT COUNT(*) FROM faces", [], |r| r.get(0))?;
86        let people_named: i64 = conn.query_row(
87            "SELECT COUNT(DISTINCT person_label) FROM faces \
88             WHERE confirmed = 1 AND person_label IS NOT NULL",
89            [],
90            |r| r.get(0),
91        )?;
92        (faces_detected, people_named)
93    } else {
94        (0, 0)
95    };
96
97    Ok(LibraryStats {
98        total_files,
99        total_size_bytes,
100        total_photos,
101        total_videos,
102        duplicate_group_count,
103        duplicate_file_count,
104        wasted_bytes,
105        faces_detected,
106        people_named,
107        embeddings: Vec::new(),
108    })
109}
110
111/// `compute`, plus the per-model embedding inventory.
112///
113/// Separate from `compute` because embeddings live outside the connection, in
114/// files addressed by the library's own path. `videre report`'s stats tile
115/// keeps calling `compute` and is unaffected.
116pub fn compute_full(conn: &Connection, db_path: &std::path::Path) -> anyhow::Result<LibraryStats> {
117    let mut stats = compute(conn)?;
118    stats.embeddings = crate::embeddings_db::counts_by_model(db_path)?;
119    Ok(stats)
120}
121
122#[cfg(test)]
123mod tests {
124    use super::*;
125
126    fn test_db() -> Connection {
127        let conn = Connection::open_in_memory().unwrap();
128        conn.execute_batch(
129            "CREATE TABLE file_hashes (
130                path        TEXT PRIMARY KEY,
131                hash        TEXT NOT NULL,
132                size_bytes  INTEGER,
133                ext         TEXT
134            );",
135        )
136        .unwrap();
137        crate::db::ensure_file_hashes_columns(&conn);
138        conn
139    }
140
141    fn insert_file(conn: &Connection, path: &str, hash: &str, size_bytes: i64, ext: &str) {
142        conn.execute(
143            "INSERT INTO file_hashes (path, hash, size_bytes, ext) VALUES (?1, ?2, ?3, ?4)",
144            rusqlite::params![path, hash, size_bytes, ext],
145        )
146        .unwrap();
147    }
148
149    #[test]
150    fn compute_counts_total_files_and_size() {
151        let conn = test_db();
152        insert_file(&conn, "/a/1.jpg", "h1", 1000, "jpg");
153        insert_file(&conn, "/a/2.png", "h2", 2500, "png");
154
155        let stats = compute(&conn).unwrap();
156        assert_eq!(stats.total_files, 2);
157        assert_eq!(stats.total_size_bytes, 3500);
158    }
159
160    #[test]
161    fn compute_on_empty_db_returns_zeros() {
162        let conn = test_db();
163        let stats = compute(&conn).unwrap();
164        assert_eq!(stats.total_files, 0);
165        assert_eq!(stats.total_size_bytes, 0);
166    }
167
168    #[test]
169    fn compute_splits_photos_and_videos_by_extension() {
170        let conn = test_db();
171        insert_file(&conn, "/a/1.jpg", "h1", 100, "jpg");
172        insert_file(&conn, "/a/2.heic", "h2", 100, "heic");
173        insert_file(&conn, "/a/3.mov", "h3", 100, "mov");
174        insert_file(&conn, "/a/4.mp4", "h4", 100, "mp4");
175        insert_file(&conn, "/a/5.unknown", "h5", 100, "xyz");
176
177        let stats = compute(&conn).unwrap();
178        assert_eq!(stats.total_photos, 2);
179        assert_eq!(stats.total_videos, 2);
180        assert_eq!(stats.total_files, 5); // unrecognized ext still counts toward total_files
181    }
182
183    #[test]
184    fn compute_counts_video_exts_case_insensitively() {
185        let conn = test_db();
186        insert_file(&conn, "/a/1.MOV", "h1", 100, "MOV");
187        insert_file(&conn, "/a/2.Mp4", "h2", 100, "Mp4");
188        insert_file(&conn, "/a/3.mov", "h3", 100, "mov");
189
190        let stats = compute(&conn).unwrap();
191        assert_eq!(stats.total_videos, 3); // uppercase/mixed-case exts still count as video
192    }
193
194    #[test]
195    fn compute_counts_duplicate_groups_and_wasted_bytes() {
196        let conn = test_db();
197        insert_file(&conn, "/a/1.jpg", "dup-hash", 1000, "jpg");
198        insert_file(&conn, "/b/1-copy.jpg", "dup-hash", 1000, "jpg");
199        insert_file(&conn, "/a/2.jpg", "dup-hash", 1000, "jpg");
200        insert_file(&conn, "/a/3.jpg", "unique-hash", 500, "jpg");
201
202        let stats = compute(&conn).unwrap();
203        assert_eq!(stats.duplicate_group_count, 1);
204        assert_eq!(stats.duplicate_file_count, 3); // all 3 members of the dup group
205        assert_eq!(stats.wasted_bytes, 2000); // (3 - 1) * 1000
206    }
207
208    #[test]
209    fn compute_with_no_duplicates_reports_zero() {
210        let conn = test_db();
211        insert_file(&conn, "/a/1.jpg", "h1", 500, "jpg");
212        insert_file(&conn, "/a/2.jpg", "h2", 500, "jpg");
213
214        let stats = compute(&conn).unwrap();
215        assert_eq!(stats.duplicate_group_count, 0);
216        assert_eq!(stats.duplicate_file_count, 0);
217        assert_eq!(stats.wasted_bytes, 0);
218    }
219
220    #[test]
221    fn compute_counts_faces_and_named_people() {
222        let conn = test_db();
223        conn.execute_batch(
224            "CREATE TABLE faces (
225                id            INTEGER PRIMARY KEY,
226                hash          TEXT NOT NULL,
227                bbox          TEXT NOT NULL,
228                landmark      TEXT,
229                embedding     BLOB NOT NULL,
230                cluster_id    INTEGER,
231                person_label  TEXT,
232                confirmed     INTEGER DEFAULT 0,
233                is_primary    INTEGER DEFAULT 0
234            );",
235        )
236        .unwrap();
237        conn.execute(
238            "INSERT INTO faces (id, hash, bbox, embedding, person_label, confirmed) \
239             VALUES (1, 'h1', '[]', X'00', 'Alice', 1)",
240            [],
241        )
242        .unwrap();
243        conn.execute(
244            "INSERT INTO faces (id, hash, bbox, embedding, person_label, confirmed) \
245             VALUES (2, 'h1', '[]', X'00', 'Alice', 1)",
246            [],
247        )
248        .unwrap();
249        conn.execute(
250            "INSERT INTO faces (id, hash, bbox, embedding, person_label, confirmed) \
251             VALUES (3, 'h2', '[]', X'00', NULL, 0)",
252            [],
253        )
254        .unwrap();
255
256        let stats = compute(&conn).unwrap();
257        assert_eq!(stats.faces_detected, 3);
258        assert_eq!(stats.people_named, 1); // distinct confirmed person_label
259    }
260
261    #[test]
262    fn compute_without_faces_table_returns_zero_not_error() {
263        let conn = test_db(); // no faces table created
264        let stats = compute(&conn).unwrap();
265        assert_eq!(stats.faces_detected, 0);
266        assert_eq!(stats.people_named, 0);
267    }
268}
videre_core/library_stats.rs

videre_core/
library_stats.rs
