[Need Migration] Improve query performance of get_cells in rich-indexer

[EthanYuan]

What problem does this PR solve?

Problem Summary:

When querying Rich Indexer get_cells or get_cells_capacity, if there are too many DEAD cells under the specified address, it will pull down the performance of the query.

What is changed and how it works?

What's Changed:

• Added a field is_spent to the output table to know if the cell is live or dead without having to join the input table.
• Auto migrations

EXPLAIN ANALYZE
SELECT
    output.id,
    output.output_index,
    output.capacity,
    query_script.code_hash AS lock_code_hash,
    query_script.hash_type AS lock_hash_type,
    query_script.args AS lock_args,
    type_script.code_hash AS type_code_hash,
    type_script.hash_type AS type_hash_type,
    type_script.args AS type_args,
    ckb_transaction.tx_index,
    ckb_transaction.tx_hash,
    block.block_number,
	output.data as output_data
FROM
    output
    JOIN (
        SELECT
            script.id,
            script.code_hash,
            script.hash_type,
            script.args
        FROM
            script
        WHERE
            (code_hash = '\xd00c84f0ec8fd441c38bc3f87a371f547190f2fcff88e642bc5bf54b9e318323')
            AND (hash_type = 1)
            AND (args LIKE '\x25ee39ecfae122adbdcc557207119fae07c6ae788925')
    ) AS query_script ON output.lock_script_id = query_script.id
    JOIN ckb_transaction ON output.tx_id = ckb_transaction.id
    JOIN block ON ckb_transaction.block_id = block.id
    LEFT JOIN script AS type_script ON output.type_script_id = type_script.id
WHERE
	output.is_spent = 0
ORDER BY
    output.id DESC
LIMIT
    1;

 QUERY PLAN

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 Limit  (cost=362735.97..362735.98 rows=1 width=293) (actual time=36.386..43.978 rows=1 loops=1)
   ->  Sort  (cost=362735.97..362735.98 rows=2 width=293) (actual time=36.384..43.975 rows=1 loops=1)
         Sort Key: output.id DESC
         Sort Method: top-N heapsort  Memory: 25kB
         ->  Gather  (cost=16570.86..362735.96 rows=2 width=293) (actual time=36.181..43.936 rows=120 loops=1)
               Workers Planned: 2
               Workers Launched: 2
               ->  Nested Loop Left Join  (cost=15570.86..361735.76 rows=1 width=293) (actual time=27.775..31.739 rows=40 loops=3)
                     ->  Nested Loop  (cost=15570.43..361728.23 rows=1 width=194) (actual time=27.766..31.670 rows=40 loops=3)
                           ->  Nested Loop  (cost=15569.99..361720.70 rows=1 width=194) (actual time=27.755..31.558 rows=40 loops=3)
                                 ->  Nested Loop  (cost=15569.56..361713.05 rows=1 width=157) (actual time=27.744..31.443 rows=40 loops=3)
                                       ->  Parallel Bitmap Heap Scan on script  (cost=15349.14..90708.40 rows=6 width=115) (actual time=27.420..30.277 rows=0 loops=3)
                                             Recheck Cond: ((code_hash = '\xd00c84f0ec8fd441c38bc3f87a371f547190f2fcff88e642bc5bf54b9e318323'::bytea) AND (hash_type = 1))
                                             Filter: (args ~~ '\x25ee39ecfae122adbdcc557207119fae07c6ae788925'::bytea)
                                             Rows Removed by Filter: 43845
                                             Heap Blocks: exact=6426
                                             ->  Bitmap Index Scan on script_code_hash_hash_type_args_key  (cost=0.00..15349.14 rows=149246 width=0) (actual time=16.896..16.897 rows=131537 loops=1)
                                                   Index Cond: ((code_hash = '\xd00c84f0ec8fd441c38bc3f87a371f547190f2fcff88e642bc5bf54b9e318323'::bytea) AND (hash_type = 1))
                                       ->  Bitmap Heap Scan on output  (cost=220.41..45166.83 rows=61 width=58) (actual time=0.953..3.456 rows=120 loops=1)
                                             Recheck Cond: (lock_script_id = script.id)
                                             Filter: (is_spent = 0)
                                             Rows Removed by Filter: 572
                                             ->  Bitmap Index Scan on idx_output_table_lock_script_id  (cost=0.00..220.40 rows=12244 width=0) (actual time=0.246..0.246 rows=1953 loops=1)
                                                   Index Cond: (lock_script_id = script.id)
                                 ->  Index Scan using ckb_transaction_pkey on ckb_transaction  (cost=0.44..7.64 rows=1 width=53) (actual time=0.002..0.002 rows=1 loops=120)
                                       Index Cond: (id = output.tx_id)
                           ->  Index Scan using block_pkey on block  (cost=0.43..7.54 rows=1 width=16) (actual time=0.002..0.002 rows=1 loops=120)
                                 Index Cond: (id = ckb_transaction.block_id)
                     ->  Index Scan using script_pkey on script type_script  (cost=0.43..7.53 rows=1 width=115) (actual time=0.001..0.001 rows=1 loops=120)
                           Index Cond: (id = output.type_script_id)
 Planning Time: 1.040 ms
 Execution Time: 44.067 ms
(32 rows)

Bitmap Heap Scan on output (cost=220.41..45166.83 rows=61 width=58) (actual time=0.953..3.456 rows=120 loops=1) shows
that the number of rows scanned has become less.

The previous scan was:

Bitmap Heap Scan on output (cost=223.46..45139.27 rows=12244 width=58) (actual time=2.271..245.854 rows=692 loops=1)

Check List

Tests

• Unit test
• Integration test
• Manual test
• No code ci-runs-only: [ quick_checks,linters ]

Release note

Title Only: Include only the PR title in the release note.