videre-core 0.49.1

//! Aggregate library statistics for dashboard-style callers.
//! Plain queries over an open `rusqlite::Connection`, shared source of truth
//! for the gallery's stats tile, `videre stats`, and any other embedder.

use rusqlite::{Connection, Result};
use serde::Serialize;

#[derive(Debug, Clone, PartialEq, Default, Serialize)]
pub struct LibraryStats {
    pub total_files: i64,
    pub total_size_bytes: i64,
    pub total_photos: i64,
    pub total_videos: i64,
    pub duplicate_group_count: i64,
    pub duplicate_file_count: i64,
    pub wasted_bytes: i64,
    pub faces_detected: i64,
    pub people_named: i64,
    /// Counts of rated/picked/labelled/liked photos.
    #[serde(default)]
    pub marks: crate::marks::MarksSummary,
    /// One entry per model with an embedding database for this library.
    /// Empty when nothing has been embedded, which is a normal state.
    #[serde(default)]
    pub embeddings: Vec<crate::embeddings_db::ModelEmbeddingCount>,
}

use crate::db::table_exists;

const PHOTO_MIME_LIST: &str =
    "'image/jpeg','image/png','image/gif','image/webp','image/bmp','image/tiff','image/heic'";
const VIDEO_MIME_LIST: &str = "'video/quicktime','video/mp4'";

const PHOTO_EXTS: &str = "'jpg','jpeg','png','gif','webp','bmp','tiff','heic','dng'";
const VIDEO_EXTS: &str = "'mov','mp4'";

// `VIDEO_EXTS`'s values must stay in sync with
// `crate::embeddings::is_video_ext` (the shared "is this a video" check).
// The SQL below uses `lower(ext)` so this list stays case-insensitive,
// matching that helper.

pub fn compute(conn: &Connection) -> Result<LibraryStats> {
    let total_files: i64 = conn.query_row("SELECT COUNT(*) FROM file_hashes", [], |r| r.get(0))?;
    let total_size_bytes: i64 = conn.query_row(
        "SELECT COALESCE(SUM(size_bytes), 0) FROM file_hashes",
        [],
        |r| r.get(0),
    )?;
    let total_photos: i64 = conn.query_row(
        &format!(
            "SELECT COUNT(*) FROM file_hashes
             WHERE mime IN ({PHOTO_MIME_LIST})
                OR (mime IS NULL AND lower(ext) IN ({PHOTO_EXTS}))"
        ),
        [],
        |r| r.get(0),
    )?;
    let total_videos: i64 = conn.query_row(
        &format!(
            "SELECT COUNT(*) FROM file_hashes
             WHERE mime IN ({VIDEO_MIME_LIST})
                OR (mime IS NULL AND lower(ext) IN ({VIDEO_EXTS}))"
        ),
        [],
        |r| r.get(0),
    )?;

    let duplicate_group_count: i64 = conn.query_row(
        "SELECT COUNT(*) FROM \
         (SELECT hash FROM file_hashes GROUP BY hash HAVING COUNT(*) > 1)",
        [],
        |r| r.get(0),
    )?;
    let duplicate_file_count: i64 = conn.query_row(
        "SELECT COUNT(*) FROM file_hashes \
         WHERE hash IN (SELECT hash FROM file_hashes GROUP BY hash HAVING COUNT(*) > 1)",
        [],
        |r| r.get(0),
    )?;
    let wasted_bytes: i64 = conn.query_row(
        "SELECT COALESCE(SUM(size_bytes * (cnt - 1)), 0) FROM \
         (SELECT hash, size_bytes, COUNT(*) as cnt \
          FROM file_hashes GROUP BY hash HAVING cnt > 1)",
        [],
        |r| r.get(0),
    )?;

    let (faces_detected, people_named) = if table_exists(conn, "faces")? {
        let faces_detected: i64 = conn.query_row("SELECT COUNT(*) FROM faces", [], |r| r.get(0))?;
        let people_named: i64 = conn.query_row(
            "SELECT COUNT(DISTINCT person_label) FROM faces \
             WHERE confirmed = 1 AND person_label IS NOT NULL",
            [],
            |r| r.get(0),
        )?;
        (faces_detected, people_named)
    } else {
        (0, 0)
    };

    Ok(LibraryStats {
        total_files,
        total_size_bytes,
        total_photos,
        total_videos,
        duplicate_group_count,
        duplicate_file_count,
        wasted_bytes,
        faces_detected,
        people_named,
        marks: crate::marks::summary(conn)?,
        embeddings: Vec::new(),
    })
}

/// Compute database and model-store statistics for one selected library.
pub fn compute_full_in(
    conn: &Connection,
    ctx: &crate::library::LibraryContext,
) -> anyhow::Result<LibraryStats> {
    ctx.ensure_root_identity()?;
    let mut stats = compute(conn)?;
    stats.embeddings = crate::embeddings_db::counts_by_model_in(ctx)?;
    Ok(stats)
}

/// What the library is made of, by file type.
///
/// `stats` could say how many files there were and how big they were in total,
/// but not what they *are* - and "70,601 files, 480GB" answers a different
/// question from "12,000 of those are HEIC and they are 22GB of it".
///
/// Grouped by extension rather than mime, with the mime shown alongside,
/// because extension is what a user recognises and types into `--ext`.
#[derive(Debug, Clone, Serialize)]
pub struct TypeBreakdown {
    pub ext: String,
    pub mime: String,
    pub files: i64,
    pub bytes: i64,
}

pub fn by_type(conn: &rusqlite::Connection, limit: usize) -> rusqlite::Result<Vec<TypeBreakdown>> {
    let mut stmt = conn.prepare(
        "SELECT LOWER(COALESCE(NULLIF(ext,''),'(none)')),
                COALESCE(NULLIF(mime,''),'(unknown)'),
                COUNT(*), COALESCE(SUM(size_bytes),0)
           FROM file_hashes
          GROUP BY 1, 2
          ORDER BY 4 DESC",
    )?;
    let rows = stmt.query_map([], |r| {
        Ok(TypeBreakdown {
            ext: r.get(0)?,
            mime: r.get(1)?,
            files: r.get(2)?,
            bytes: r.get(3)?,
        })
    })?;
    let mut out: Vec<TypeBreakdown> = rows.collect::<rusqlite::Result<_>>()?;
    out.truncate(limit);
    Ok(out)
}

/// A scanned path whose filename extension conflicts with detected content.
#[derive(Debug, Clone, PartialEq, Serialize)]
pub struct MismatchFile {
    pub path: String,
    pub ext: String,
    pub mime: String,
}

/// Exact mismatch count with a bounded or complete path list.
#[derive(Debug, Clone, PartialEq, Serialize)]
pub struct MismatchReport {
    pub count: usize,
    pub files: Vec<MismatchFile>,
    pub truncated: bool,
}

/// Report disagreements from the last scan without opening source files.
///
/// `Some(limit)` retains at most that many paths; `None` retains all paths.
/// Every row is still examined so `count` is exact even for a bounded report.
pub fn mismatched_files(conn: &Connection, limit: Option<usize>) -> Result<MismatchReport> {
    let mut stmt = conn.prepare("SELECT path, ext, mime FROM file_hashes ORDER BY path")?;
    let rows = stmt.query_map([], |r| {
        Ok((
            r.get::<_, String>(0)?,
            r.get::<_, Option<String>>(1)?,
            r.get::<_, Option<String>>(2)?,
        ))
    })?;
    let mut count = 0;
    let mut files = Vec::new();
    for row in rows {
        let (path, ext, mime) = row?;
        let (Some(ext), Some(mime)) = (ext, mime) else {
            continue;
        };
        if crate::mime_probe::is_content_mismatch(&ext, Some(&mime)) {
            count += 1;
            if limit.is_none_or(|bound| files.len() < bound) {
                files.push(MismatchFile { path, ext, mime });
            }
        }
    }
    Ok(MismatchReport {
        count,
        truncated: count > files.len(),
        files,
    })
}

#[cfg(test)]
mod tests {
    use super::*;

    fn test_db() -> Connection {
        let conn = Connection::open_in_memory().unwrap();
        conn.execute_batch(
            "CREATE TABLE file_hashes (
                path        TEXT PRIMARY KEY,
                hash        TEXT NOT NULL,
                size_bytes  INTEGER,
                ext         TEXT
            );",
        )
        .unwrap();
        crate::library_db::ensure_scan_schema(&conn).unwrap();
        conn
    }

    fn insert_file(conn: &Connection, path: &str, hash: &str, size_bytes: i64, ext: &str) {
        conn.execute(
            "INSERT INTO file_hashes (path, hash, size_bytes, ext) VALUES (?1, ?2, ?3, ?4)",
            rusqlite::params![path, hash, size_bytes, ext],
        )
        .unwrap();
    }

    fn insert_scanned_file(conn: &Connection, path: &str, hash: &str, ext: &str, mime: &str) {
        conn.execute(
            "INSERT INTO file_hashes (path, hash, ext, mime) VALUES (?1, ?2, ?3, ?4)",
            rusqlite::params![path, hash, ext, mime],
        )
        .unwrap();
    }

    #[test]
    fn mismatched_files_counts_paths_not_hashes() {
        let conn = test_db();
        insert_scanned_file(&conn, "/a/1.png", "same", "png", "image/jpeg");
        insert_scanned_file(&conn, "/a/2.png", "same", "png", "image/jpeg");
        insert_scanned_file(&conn, "/a/3.jpg", "same", "jpg", "image/jpeg");

        let report = mismatched_files(&conn, Some(10)).unwrap();
        assert_eq!(report.count, 2);
        assert_eq!(report.files.len(), 2);
        assert_eq!(report.files[0].path, "/a/1.png");
        assert_eq!(report.files[0].ext, "png");
        assert_eq!(report.files[0].mime, "image/jpeg");
        assert_eq!(report.files[1].path, "/a/2.png");
        assert!(!report.truncated);
    }

    #[test]
    fn mismatched_files_samples_in_path_order() {
        let conn = test_db();
        for i in (0..12).rev() {
            insert_scanned_file(
                &conn,
                &format!("/a/{i:02}.png"),
                &format!("hash-{i}"),
                "png",
                "image/jpeg",
            );
        }
        insert_scanned_file(&conn, "/a/movie.mov", "movie", "mov", "video/mp4");
        insert_scanned_file(
            &conn,
            "/a/unknown.jpg",
            "unknown",
            "jpg",
            crate::mime_probe::UNKNOWN_MIME,
        );

        let bounded = mismatched_files(&conn, Some(10)).unwrap();
        assert_eq!(bounded.count, 12);
        assert_eq!(bounded.files.len(), 10);
        assert_eq!(bounded.files.first().unwrap().path, "/a/00.png");
        assert_eq!(bounded.files.last().unwrap().path, "/a/09.png");
        assert!(bounded.truncated);

        let full = mismatched_files(&conn, None).unwrap();
        assert_eq!(full.count, 12);
        assert_eq!(full.files.len(), 12);
        assert_eq!(full.files.first().unwrap().path, "/a/00.png");
        assert_eq!(full.files.last().unwrap().path, "/a/11.png");
        assert!(!full.truncated);
    }

    #[test]
    fn mismatched_files_propagates_sql_errors() {
        let conn = test_db();
        conn.execute_batch("DROP TABLE file_hashes").unwrap();
        assert!(mismatched_files(&conn, Some(10)).is_err());
    }

    #[test]
    fn compute_counts_total_files_and_size() {
        let conn = test_db();
        insert_file(&conn, "/a/1.jpg", "h1", 1000, "jpg");
        insert_file(&conn, "/a/2.png", "h2", 2500, "png");

        let stats = compute(&conn).unwrap();
        assert_eq!(stats.total_files, 2);
        assert_eq!(stats.total_size_bytes, 3500);
    }

    #[test]
    fn compute_on_empty_db_returns_zeros() {
        let conn = test_db();
        let stats = compute(&conn).unwrap();
        assert_eq!(stats.total_files, 0);
        assert_eq!(stats.total_size_bytes, 0);
    }

    #[test]
    fn compute_splits_photos_and_videos_by_extension() {
        let conn = test_db();
        insert_file(&conn, "/a/1.jpg", "h1", 100, "jpg");
        insert_file(&conn, "/a/2.heic", "h2", 100, "heic");
        insert_file(&conn, "/a/3.mov", "h3", 100, "mov");
        insert_file(&conn, "/a/4.mp4", "h4", 100, "mp4");
        insert_file(&conn, "/a/5.unknown", "h5", 100, "xyz");

        let stats = compute(&conn).unwrap();
        assert_eq!(stats.total_photos, 2);
        assert_eq!(stats.total_videos, 2);
        assert_eq!(stats.total_files, 5); // unrecognized ext still counts toward total_files
    }

    #[test]
    fn compute_counts_video_exts_case_insensitively() {
        let conn = test_db();
        insert_file(&conn, "/a/1.MOV", "h1", 100, "MOV");
        insert_file(&conn, "/a/2.Mp4", "h2", 100, "Mp4");
        insert_file(&conn, "/a/3.mov", "h3", 100, "mov");

        let stats = compute(&conn).unwrap();
        assert_eq!(stats.total_videos, 3); // uppercase/mixed-case exts still count as video
    }

    #[test]
    fn compute_counts_duplicate_groups_and_wasted_bytes() {
        let conn = test_db();
        insert_file(&conn, "/a/1.jpg", "dup-hash", 1000, "jpg");
        insert_file(&conn, "/b/1-copy.jpg", "dup-hash", 1000, "jpg");
        insert_file(&conn, "/a/2.jpg", "dup-hash", 1000, "jpg");
        insert_file(&conn, "/a/3.jpg", "unique-hash", 500, "jpg");

        let stats = compute(&conn).unwrap();
        assert_eq!(stats.duplicate_group_count, 1);
        assert_eq!(stats.duplicate_file_count, 3); // all 3 members of the dup group
        assert_eq!(stats.wasted_bytes, 2000); // (3 - 1) * 1000
    }

    #[test]
    fn compute_with_no_duplicates_reports_zero() {
        let conn = test_db();
        insert_file(&conn, "/a/1.jpg", "h1", 500, "jpg");
        insert_file(&conn, "/a/2.jpg", "h2", 500, "jpg");

        let stats = compute(&conn).unwrap();
        assert_eq!(stats.duplicate_group_count, 0);
        assert_eq!(stats.duplicate_file_count, 0);
        assert_eq!(stats.wasted_bytes, 0);
    }

    #[test]
    fn compute_counts_faces_and_named_people() {
        let conn = test_db();
        conn.execute_batch(
            "CREATE TABLE faces (
                id            INTEGER PRIMARY KEY,
                hash          TEXT NOT NULL,
                bbox          TEXT NOT NULL,
                landmark      TEXT,
                embedding     BLOB NOT NULL,
                cluster_id    INTEGER,
                person_label  TEXT,
                confirmed     INTEGER DEFAULT 0,
                is_primary    INTEGER DEFAULT 0
            );",
        )
        .unwrap();
        conn.execute(
            "INSERT INTO faces (id, hash, bbox, embedding, person_label, confirmed) \
             VALUES (1, 'h1', '[]', X'00', 'Alice', 1)",
            [],
        )
        .unwrap();
        conn.execute(
            "INSERT INTO faces (id, hash, bbox, embedding, person_label, confirmed) \
             VALUES (2, 'h1', '[]', X'00', 'Alice', 1)",
            [],
        )
        .unwrap();
        conn.execute(
            "INSERT INTO faces (id, hash, bbox, embedding, person_label, confirmed) \
             VALUES (3, 'h2', '[]', X'00', NULL, 0)",
            [],
        )
        .unwrap();

        let stats = compute(&conn).unwrap();
        assert_eq!(stats.faces_detected, 3);
        assert_eq!(stats.people_named, 1); // distinct confirmed person_label
    }

    #[test]
    fn compute_without_faces_table_returns_zero_not_error() {
        let conn = test_db(); // no faces table created
        let stats = compute(&conn).unwrap();
        assert_eq!(stats.faces_detected, 0);
        assert_eq!(stats.people_named, 0);
    }

    #[test]
    fn compute_full_in_reads_models_without_creating_missing_stores() {
        let temp = tempfile::tempdir().unwrap();
        let root = temp.path().join("library");
        let cache = temp.path().join("cache");
        std::fs::create_dir(&root).unwrap();
        let ctx = crate::library::LibraryContext::new(&root, &cache).unwrap();
        let conn = crate::library_db::initialize(&ctx).unwrap();
        let stats = compute_full_in(&conn, &ctx).unwrap();
        assert!(stats.embeddings.is_empty());
        assert!(!ctx.paths.embeddings.exists());
    }
}