velesdb-core 3.8.1

use super::*;
use crate::point::Point;
use crate::velesql::Parser;
use crate::DistanceMetric;
use tempfile::tempdir;

// =========================================================================
// execute_query end-to-end
// =========================================================================

#[test]
fn test_execute_query_select_all_returns_inserted_points() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    db.create_collection("docs", 4, DistanceMetric::Cosine)
        .unwrap();

    let coll = db.get_vector_collection("docs").unwrap();
    coll.upsert(vec![
        Point::new(
            1,
            vec![1.0, 0.0, 0.0, 0.0],
            Some(serde_json::json!({"title": "alpha"})),
        ),
        Point::new(
            2,
            vec![0.0, 1.0, 0.0, 0.0],
            Some(serde_json::json!({"title": "beta"})),
        ),
    ])
    .unwrap();

    let query = Parser::parse("SELECT * FROM docs").unwrap();
    let results = db
        .execute_query(&query, &std::collections::HashMap::new())
        .unwrap();

    assert_eq!(results.len(), 2);
}

#[test]
fn test_execute_query_with_limit() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    db.create_collection("items", 4, DistanceMetric::Cosine)
        .unwrap();

    let coll = db.get_vector_collection("items").unwrap();
    let points: Vec<Point> = (1..=5)
        .map(|i| {
            #[allow(clippy::cast_precision_loss)]
            let v = vec![i as f32, 0.0, 0.0, 0.0];
            Point::new(i, v, Some(serde_json::json!({})))
        })
        .collect();
    coll.upsert(points).unwrap();

    let query = Parser::parse("SELECT * FROM items LIMIT 2").unwrap();
    let results = db
        .execute_query(&query, &std::collections::HashMap::new())
        .unwrap();

    assert_eq!(
        results.len(),
        2,
        "LIMIT 2 over 5 inserted points must return exactly 2 rows"
    );
}

#[test]
fn test_execute_query_nonexistent_collection_returns_error() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();

    let query = Parser::parse("SELECT * FROM ghost").unwrap();
    let err = db
        .execute_query(&query, &std::collections::HashMap::new())
        .unwrap_err();

    assert!(matches!(err, crate::Error::CollectionNotFound(_)));
}

#[test]
fn test_execute_query_validation_error_returns_query_error() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    db.create_collection_typed("t", &crate::CollectionType::MetadataOnly)
        .unwrap();
    // Parses fine, but similarity() without a score context (no NEAR / similarity in WHERE)
    // is rejected by QueryValidator -> execute_query maps it to Error::Query.
    let query = Parser::parse("SELECT similarity() FROM t WHERE name = 'x'").unwrap();
    let err = db
        .execute_query(&query, &std::collections::HashMap::new())
        .unwrap_err();
    assert!(matches!(err, crate::Error::Query(_)));
}

// =========================================================================
// explain_query
// =========================================================================

#[test]
fn test_explain_query_returns_valid_plan() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    db.create_collection("plans", 4, DistanceMetric::Cosine)
        .unwrap();

    let query = Parser::parse("SELECT * FROM plans").unwrap();
    let plan = db.explain_query(&query).unwrap();

    // First call is a cache miss.
    assert_eq!(plan.cache_hit, Some(false));
    assert_eq!(plan.plan_reuse_count, Some(0));
}

#[test]
fn test_explain_query_cache_hit_after_execute() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    db.create_collection("cached", 4, DistanceMetric::Cosine)
        .unwrap();

    let query = Parser::parse("SELECT * FROM cached").unwrap();

    // execute_query populates the cache on miss.
    db.execute_query(&query, &std::collections::HashMap::new())
        .unwrap();

    // explain_query should now report a cache hit.
    let plan = db.explain_query(&query).unwrap();
    assert_eq!(plan.cache_hit, Some(true));
}

// =========================================================================
// DML: INSERT / UPDATE via execute_query
// =========================================================================

#[test]
fn test_execute_query_insert_into_metadata_collection() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    db.create_collection_typed("items", &crate::CollectionType::MetadataOnly)
        .unwrap();

    let query =
        Parser::parse("INSERT INTO items (id, tag, score) VALUES (1, 'hello', 42.0)").unwrap();
    let results = db
        .execute_query(&query, &std::collections::HashMap::new())
        .unwrap();

    assert_eq!(results.len(), 1);
    assert_eq!(results[0].point.id, 1);
    let payload = results[0].point.payload.as_ref().unwrap();
    assert_eq!(payload["tag"], serde_json::json!("hello"));
}

#[test]
fn test_execute_query_update_modifies_payload() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    db.create_collection_typed("items", &crate::CollectionType::MetadataOnly)
        .unwrap();

    let coll = db.get_metadata_collection("items").unwrap();
    coll.upsert_metadata(vec![Point::metadata_only(
        1,
        serde_json::json!({"status": "draft", "count": 0}),
    )])
    .unwrap();

    let query = Parser::parse("UPDATE items SET status = 'published' WHERE id = 1").unwrap();
    let results = db
        .execute_query(&query, &std::collections::HashMap::new())
        .unwrap();
    assert_eq!(results.len(), 1);

    let updated = coll.get(&[1]).into_iter().flatten().next().unwrap();
    let payload = updated.payload.unwrap();
    assert_eq!(payload["status"], serde_json::json!("published"));
    // Unmodified fields are preserved.
    assert_eq!(payload["count"], serde_json::json!(0));
}

// =========================================================================
// Schema version interaction with plan cache
// =========================================================================

#[test]
fn test_schema_version_increments_on_create_and_delete() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    let v0 = db.schema_version();

    db.create_collection("a", 4, DistanceMetric::Cosine)
        .unwrap();
    let v1 = db.schema_version();
    assert!(v1 > v0, "schema_version should increment after create");

    db.delete_collection("a").unwrap();
    let v2 = db.schema_version();
    assert!(v2 > v1, "schema_version should increment after delete");
}

// =========================================================================
// join_row_budget: LIMIT must NOT bound joined INPUT rows for shapes where
// downstream stages aggregate/dedup/reorder (GROUP BY / HAVING / DISTINCT /
// ORDER BY). SQL LIMIT bounds output groups/rows, not input rows.
// =========================================================================

use crate::collection::search::query::pushdown::PushdownAnalysis;
use crate::collection::search::query::JOIN_ROW_CEILING;

fn select_of(sql: &str) -> crate::velesql::SelectStatement {
    Parser::parse(sql).unwrap().select
}

#[test]
fn test_join_row_budget_plain_limit_uses_limit() {
    // No GROUP BY / DISTINCT / HAVING / ORDER BY: LIMIT bounds input rows.
    let select = select_of("SELECT d.id FROM docs AS d JOIN meta AS m ON d.id = m.id LIMIT 5");
    let budget = Database::join_row_budget(&select, &PushdownAnalysis::default());
    assert_eq!(budget, 5);
}

#[test]
fn test_join_row_budget_group_by_uses_ceiling() {
    // GROUP BY ... LIMIT n bounds GROUPS, not input rows: must NOT truncate to n.
    let select = select_of(
        "SELECT m.tag FROM docs AS d JOIN meta AS m ON d.id = m.id GROUP BY m.tag LIMIT 2",
    );
    let budget = Database::join_row_budget(&select, &PushdownAnalysis::default());
    assert_eq!(budget, JOIN_ROW_CEILING);
}

#[test]
fn test_join_row_budget_distinct_uses_ceiling() {
    // DISTINCT dedups output rows: LIMIT must not truncate the join input.
    let select =
        select_of("SELECT DISTINCT m.tag FROM docs AS d JOIN meta AS m ON d.id = m.id LIMIT 2");
    let budget = Database::join_row_budget(&select, &PushdownAnalysis::default());
    assert_eq!(budget, JOIN_ROW_CEILING);
}

#[test]
fn test_join_row_budget_order_by_uses_ceiling() {
    // ORDER BY can reorder past the window: still falls back to the ceiling.
    let select =
        select_of("SELECT d.id FROM docs AS d JOIN meta AS m ON d.id = m.id ORDER BY d.id LIMIT 3");
    let budget = Database::join_row_budget(&select, &PushdownAnalysis::default());
    assert_eq!(budget, JOIN_ROW_CEILING);
}

// =========================================================================
// Database::execute_aggregate (single-source aggregation entry)
// =========================================================================

/// `Database::execute_aggregate` resolves the target collection from the
/// `_collection` param when the query has no explicit `FROM` (the convention the
/// CLI REPL and SDKs use to inject the active collection), and runs GROUP BY.
#[test]
fn test_execute_aggregate_resolves_collection_via_param() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    db.create_collection("orders", 2, DistanceMetric::Cosine)
        .unwrap();
    let coll = db.get_vector_collection("orders").unwrap();
    let points: Vec<Point> = [(10, "x"), (11, "x"), (12, "y"), (13, "y")]
        .into_iter()
        .map(|(id, cat)| {
            Point::new(
                id,
                vec![1.0, 0.0],
                Some(serde_json::json!({ "category": cat })),
            )
        })
        .collect();
    coll.upsert(points).unwrap();

    // Parse with FROM (the grammar requires it), then clear it to simulate the
    // programmatic / REPL convention of supplying the target via `_collection`.
    let mut query =
        Parser::parse("SELECT category, COUNT(*) AS n FROM orders GROUP BY category").unwrap();
    query.select.from = String::new();
    let mut params = std::collections::HashMap::new();
    params.insert(
        "_collection".to_string(),
        serde_json::Value::String("orders".to_string()),
    );

    let value = db.execute_aggregate(&query, &params).unwrap();
    let groups = value
        .as_array()
        .expect("GROUP BY returns an array of groups");
    assert_eq!(groups.len(), 2, "two category groups (x, y); got {value:?}");
}

/// `Database::execute_aggregate` surfaces a `CollectionNotFound` error for an
/// unresolved target rather than silently returning an empty result.
#[test]
fn test_execute_aggregate_unknown_collection_errors() {
    let dir = tempdir().unwrap();
    let db = Database::open(dir.path()).unwrap();
    let query = Parser::parse("SELECT COUNT(*) AS n FROM ghost").unwrap();
    let params = std::collections::HashMap::new();
    let err = db.execute_aggregate(&query, &params).unwrap_err();
    assert!(
        matches!(err, crate::Error::CollectionNotFound(_)),
        "expected CollectionNotFound, got {err:?}"
    );
}