-- Matching spans and the earliest/latest spans deliberately live in different regions.
CREATE TABLE having_input_bounds (
    trace_id STRING,
    service_name STRING,
    span_name STRING,
    product STRING,
    duration BIGINT,
    ts TIMESTAMP(9) TIME INDEX,
    PRIMARY KEY (service_name)
)
PARTITION ON COLUMNS (service_name) (
    service_name < 'm',
    service_name >= 'm'
)
WITH ('append_mode' = 'true');

INSERT INTO having_input_bounds VALUES
    ('a', 'other', 'parent', NULL, 30, 5),
    ('a', 'cart', 'GetCart', NULL, 6, 10),
    ('a', 'frontend', 'GetProduct', 'product-7', 8, 15),
    ('a', 'cart', 'GetCart', NULL, 6, 20),
    ('a', 'other', 'parent', NULL, 30, 25),
    ('b', 'other', 'parent', NULL, 1, 2),
    ('b', 'frontend', 'GetProduct', 'product-x', 18, 12),
    ('b', 'other', 'parent', NULL, 30, 22),
    ('c', 'cart', 'GetCart', NULL, 5, 30),
    ('d', 'frontend', NULL, 'product-7', 8, 15),
    (NULL, 'other', 'parent', NULL, 2, 3),
    (NULL, 'cart', 'GetCart', NULL, 6, 11),
    (NULL, 'other', 'parent', NULL, 30, 25);

-- T02 shape: only the upper bound may be inferred; preserve earlier spans and NULL keys.
SELECT trace_id, MIN(CAST(ts AS BIGINT)) AS trace_start
FROM having_input_bounds
GROUP BY trace_id
HAVING SUM(CASE WHEN service_name = 'cart' AND span_name = 'GetCart'
    AND CAST(ts AS BIGINT) >= 10 AND CAST(ts AS BIGINT) < 20 THEN 1 ELSE 0 END) > 0
ORDER BY trace_start, trace_id
LIMIT 20;

-- Pin the optimization itself: the remote scan has only the inferred time upper
-- bound, while HAVING stays above aggregation. Result equality alone cannot prove this.
-- SQLNESS REPLACE (-+) -
-- SQLNESS REPLACE (\s\s+) _
-- SQLNESS REPLACE (RepartitionExec:.*) RepartitionExec: REDACTED
-- SQLNESS REPLACE (MergeScanExec:.*) MergeScanExec: REDACTED
EXPLAIN SELECT trace_id, MIN(CAST(ts AS BIGINT)) AS trace_start
FROM having_input_bounds
GROUP BY trace_id
HAVING SUM(CASE WHEN service_name = 'cart' AND span_name = 'GetCart'
    AND CAST(ts AS BIGINT) >= 10 AND CAST(ts AS BIGINT) < 20 THEN 1 ELSE 0 END) > 0
ORDER BY trace_start, trace_id
LIMIT 20;

-- T03 shape: a span attribute is part of the existence condition, not an input filter.
SELECT trace_id, MIN(CAST(ts AS BIGINT)) AS trace_start
FROM having_input_bounds
GROUP BY trace_id
HAVING SUM(CASE WHEN service_name = 'frontend' AND span_name = 'GetProduct'
    AND product = 'product-7' AND CAST(ts AS BIGINT) >= 10
    AND CAST(ts AS BIGINT) < 20 THEN 1 ELSE 0 END) > 0
ORDER BY trace_start, trace_id;

-- T04 shape: bounds on a different expression must not become aggregate input bounds.
SELECT trace_id, MIN(CAST(ts AS BIGINT)) AS trace_start
FROM having_input_bounds
GROUP BY trace_id
HAVING SUM(CASE WHEN service_name = 'frontend' AND span_name = 'GetProduct'
    AND duration >= 5 AND duration < 20 AND CAST(ts AS BIGINT) >= 10
    AND CAST(ts AS BIGINT) < 20 THEN 1 ELSE 0 END) > 0
ORDER BY trace_start, trace_id;

-- MAX is symmetric: preserve later spans by inferring only the lower bound.
-- Also check that the conditional SUM itself remains unchanged.
SELECT trace_id, MAX(CAST(ts AS BIGINT)) AS trace_end,
    SUM(CASE WHEN service_name = 'cart' AND CAST(ts AS BIGINT) >= 10
        AND CAST(ts AS BIGINT) < 20 THEN 1 ELSE 0 END) AS matches
FROM having_input_bounds
GROUP BY trace_id
HAVING matches >= 1
ORDER BY trace_end, trace_id;

-- Full-group aggregates forbid pruning, even when the MIN would be preserved.
SELECT trace_id, MIN(CAST(ts AS BIGINT)) AS trace_start,
    MAX(CAST(ts AS BIGINT)) AS trace_end, COUNT(*) AS spans
FROM having_input_bounds
GROUP BY trace_id
HAVING SUM(CASE WHEN service_name = 'cart' AND CAST(ts AS BIGINT) >= 10
    AND CAST(ts AS BIGINT) < 20 THEN 1 ELSE 0 END) > 0
ORDER BY trace_start, trace_id;

-- OR does not guarantee a matching span in every surviving group.
SELECT trace_id, MIN(CAST(ts AS BIGINT)) AS trace_start
FROM having_input_bounds
GROUP BY trace_id
HAVING SUM(CASE WHEN service_name = 'cart' AND CAST(ts AS BIGINT) < 20 THEN 1 ELSE 0 END) > 0
    OR MIN(CAST(ts AS BIGINT)) > 20
ORDER BY trace_start, trace_id;

DROP TABLE having_input_bounds;

-- A derived column must not move fallible arithmetic/casts ahead of its WHERE guard.
CREATE TABLE having_guarded(k STRING, v BIGINT, d BIGINT, text_value STRING, ts TIMESTAMP TIME INDEX);
INSERT INTO having_guarded VALUES ('a', 12, 0, 'bad', 1), ('a', 5, 1, '5', 2), ('b', 25, 1, '25', 3);

SELECT k, MIN(r) FROM (
    SELECT k, v / d AS r FROM having_guarded WHERE abs(d) > 0
) s GROUP BY k
HAVING SUM(CASE WHEN r < 20 THEN 1 ELSE 0 END) > 0
ORDER BY k;

SELECT k, MIN(r) FROM (
    SELECT k, CAST(text_value AS BIGINT) AS r FROM having_guarded WHERE abs(d) > 0
) s GROUP BY k
HAVING SUM(CASE WHEN r < 20 THEN 1 ELSE 0 END) > 0
ORDER BY k;

CREATE VIEW having_guarded_view AS
SELECT k, v / d AS r FROM having_guarded WHERE abs(d) > 0;

SELECT k, MIN(r) FROM having_guarded_view GROUP BY k
HAVING SUM(CASE WHEN r < 20 THEN 1 ELSE 0 END) > 0
ORDER BY k;

DROP VIEW having_guarded_view;
DROP TABLE having_guarded;