Add ArrayRows hint

Makefile

@@ -12,7 +12,7 @@ OBJS = \
 
 HINTPLANVER = 1.9.0
 
-REGRESS = init base_plan pg_hint_plan ut-init ut-A ut-S ut-J ut-L ut-G ut-R \
+REGRESS = init base_plan pg_hint_plan ut-init ut-A ut-AS ut-S ut-J ut-L ut-G ut-R \
 	ut-fdw ut-W ut-T ut-fini plpgsql hint_table disable_index \
 	query_parser oldextversions
 REGRESS_OPTS = --encoding=UTF8

docs/hint_list.md

@@ -28,5 +28,6 @@ The available hints are listed below.
 | Behavior control on Join | `Memoize(table table[ table...])` | Allows the topmost join of a join among the specified tables to Memoize the inner result. Not enforced. |
 | | `NoMemoize(table table[ table...])` | Inhibits the topmost join of a join among the specified tables from Memoizing the inner result. |
 | Row number correction | `Rows(table table[ table...] correction)` | Corrects row number of a result of the joins on the tables specified.  The available correction methods are absolute (#<n>), addition (+<n>), subtract (-<n>) and multiplication (*<n>).  <n> should be a string that strtod() can understand. |
+| | `ArrayRows(table[ qualifier] correction)` | Adjusts row estimation for array predicates on the tables specified.  Targets array operators (`&&`, `@>`, `<@`) and scalar-array comparisons (`<cmp> ANY/ALL (array)`).  The correction is absolute (#<n>), addition (+<n>), subtraction (-<n>) or multiplication (*<n>).  `qualifier` is optional and can be an operator (`&&`, `@>`, `<@`), a quantifier (`ANY`, `ALL`) or a comparison+quantifier (e.g. `= ANY`, `<> ALL`). |
 | Parallel query configuration | `Parallel(table <# of workers> [soft\|hard])` | Enforces or inhibits parallel execution of the specified table.  <# of workers> is the desired number of parallel workers, where zero means inhibiting parallel execution.  If the third parameter is soft (default), it just changes max\_parallel\_workers\_per\_gather and leaves everything else to the planner.  Hard enforces the specified number of workers. |
 | GUC | `Set(GUC-param value)` | Sets GUC parameter to the value defined while planner is running. |

docs/hint_table.md

@@ -105,6 +105,45 @@ from restrictions in the planner.  For example:
 =# /*+ Rows(a b *10) */ SELECT... ; Makes the number 10 times larger.
 ```
 
+### Hints for ArrayRows (array predicate row corrections)
+
+This hint, named `ArrayRows`, adjusts row estimation for array-related
+predicates.  It can be useful when PostgreSQL's default estimate is too crude
+(especially when the array expression is not a constant).
+
+`ArrayRows` targets:
+
+- Array operators: `&&`, `@>`, `<@`
+- Scalar-array comparisons: `<cmp> ANY (array)` and `<cmp> ALL (array)`
+
+`<cmp>` can be `=`, `<>` (or `!=`), `<`, `<=`, `>` and `>=`.
+
+The last argument is the correction term:
+
+- `#<n>` sets the estimated number of rows produced by the matching predicate(s)
+- `+<n>` increases the estimated rows by the given amount
+- `-<n>` decreases the estimated rows by the given amount
+- `*<n>` scales the matching predicate(s) by the given factor
+
+The optional `qualifier` can narrow the hint down to only a specific operator
+or quantifier, for example `&&`, `ANY`, or `= ANY`.  If multiple matching
+predicates exist for the same relation set, the correction is distributed
+evenly across them.
+
+If multiple `ArrayRows` hints match the same predicate, the most specific
+qualifier wins (comparison+quantifier like `= ANY` is preferred over `ANY`/`ALL`
+or an array operator like `&&`, which are preferred over no qualifier).
+
+For inheritance/partitioned tables, it is usually best to target the parent
+relation name/alias; child rels may appear in `EXPLAIN` with planner-generated
+aliases like `parent_1`, `parent_2`, but the parent hint still applies.
+
+```sql
+=# /*+ ArrayRows(t && #10) */ EXPLAIN SELECT * FROM t WHERE t.tags && some_array();
+=# /*+ ArrayRows(t = ANY #100) */ EXPLAIN SELECT * FROM t WHERE t.id = ANY(some_array());
+=# /*+ ArrayRows(t ANY *0.1) */ EXPLAIN SELECT * FROM t WHERE t.id < ANY(some_array());
+```
+
 ### Hints for parallel plans
 
 This hint, named `Parallel`, enforces parallel execution configuration