update bq2cst & add functions

expected_output/core.sql

@@ -217,5 +217,34 @@ WINDOW
   named_window2 AS (PARTITION BY str ORDER BY int)
 ;
 
+----- chained function call -----
+SELECT (col).UPPER().LOWER();
+
+SELECT
+  TIMESTAMP '2024-01-01'.TIMESTAMP_TRUNC(month),
+  CASE WHEN TRUE THEN TIMESTAMP '2024-01-01' END.TIMESTAMP_TRUNC(month),
+;
+
+SELECT (col).(safe.left)(3);
+
+SELECT STRUCT('a' AS b).TO_JSON().b.JSON_VALUE().CONCAT('c');
+
+SELECT
+  "foo"
+  .longfunctionname()
+  .longfunctionname()
+  .longfunctionname()
+  .longfunctionname()
+  -- comment
+  .longfunctionname()
+  .longfunctionname()
+  .longfunctionname()
+  .longfunctionname()
+  .longfunctionname(),
+;
+
+----- with expression -----
+SELECT WITH(a AS 'a', UPPER(a));
+
 ----- template -----
 SELECT {{variable}}, {variable};

expected_output/ddl.sql

@@ -105,6 +105,11 @@ WITH CONNECTION ident
 OPTIONS (dummy = 'dummy')
 ;
 
+CREATE EXTERNAL TABLE tablename
+WITH CONNECTION DEFAULT
+OPTIONS (dummy = 'dummy')
+;
+
 ----- CREATE VIEW statement -----
 CREATE VIEW view_name AS SELECT * FROM t;
 
@@ -180,6 +185,10 @@ AS
   SELECT 1 AS one
 ;
 
+CREATE TABLE FUNCTION foo.bar(tablename TABLE<col1 INT64, col2 FLOAT64>)
+AS (SELECT 1)
+;
+
 -- remote function
 CREATE FUNCTION dataset.abc()
 RETURNS INT64
@@ -247,6 +256,12 @@ STORING (a, b, c)
 OPTIONS (dummy = 'dummy')
 ;
 
+CREATE VECTOR INDEX new_index
+ON tablename (col)
+PARTITION BY foo
+OPTIONS (dummy = 'dummy')
+;
+
 ----- ALTER BI_CAPACITY statement -----
 ALTER BI_CAPACITY `project.region-us.default`
 SET OPTIONS (preferred_tables = ['table1', 'table2'])
@@ -375,6 +390,9 @@ ALTER VIEW viewname ALTER COLUMN colname SET OPTIONS (dummy = 'dummy');
 -- MATERIALIZED
 ALTER MATERIALIZED VIEW IF EXISTS example SET OPTIONS (dummy = 'dummy');
 
+----- ALTER VECTOR INDEX statement -----
+ALTER VECTOR INDEX index_name ON table_name REBUILD;
+
 ----- DROP statement -----
 -- general
 DROP TABLE example;

expected_output/ml.sql

@@ -8,6 +8,8 @@ OPTIONS (endpoint = '')
 AS (SELECT 1)
 ;
 
+CREATE MODEL ident REMOTE WITH CONNECTION DEFAULT OPTIONS (endpoint = '');
+
 CREATE MODEL ident
 AS (
   training_data AS (SELECT 1),

expected_output/pipe.sql

@@ -17,6 +17,17 @@ FROM table
   SELECT col
 ;
 
+WITH t AS (SELECT 1)
+FROM t
+;
+
+SELECT *
+FROM (
+  FROM t
+  |> SELECT *
+)
+;
+
 ----- select statement -----
 SELECT 1
 |> SELECT *
@@ -36,6 +47,13 @@ SELECT 2
 |> SELECT *
 ;
 
+FROM t
+|>
+  SELECT col
+  -- break parent
+  WINDOW a AS (PARTITION BY b)
+;
+
 ----- from statement -----
 FROM tabe AS t1
 INNER JOIN table AS p2 USING(col)
@@ -107,6 +125,11 @@ FROM t
     col2
 ;
 
+----- distinct pipe operator -----
+FROM t
+|> DISTINCT
+;
+
 ----- union pipe operator -----
 FROM t
 |> UNION ALL (SELECT 1), (SELECT 2)
@@ -171,3 +194,27 @@ FROM t
 FROM t
 |> unpivot INCLUDE NULLS (sales FOR quarter IN (q1, q2)) AS q
 ;
+
+----- match recognize pipe operator -----
+FROM t
+|>
+  MATCH_RECOGNIZE (
+    ORDER BY col1
+    PATTERN (symbol)
+  )
+;
+
+----- with pipe operator -----
+FROM t
+|> WITH u AS (SELECT 1 AS key)
+|> INNER JOIN u USING(key)
+;
+
+FROM t
+|>
+  WITH
+    -- CTE
+    u AS (SELECT 1 AS key),
+    v AS (SELECT 2 AS key),
+|> INNER JOIN u USING(key)
+;

expected_output/select.sql

@@ -276,6 +276,59 @@ FROM
   ) AS unpivot
 ;
 
+-- MATCH_RECOGNIZE
+SELECT *
+FROM
+  t MATCH_RECOGNIZE (
+    PARTITION BY col1, col2
+    ORDER BY col3, col4 DESC NULLS FIRST
+    MEASURES ANY_VALUE(col4) AS col4
+    AFTER MATCH SKIP TO NEXT ROW
+    PATTERN (symbol1)
+    DEFINE
+      symbol1 AS col3 = 'foo',
+      symbol2 AS col4 = 'bar'
+    OPTIONS (use_longest_match = TRUE)
+  ) AS u
+;
+
+SELECT *
+FROM
+  t MATCH_RECOGNIZE (
+    ORDER BY col1
+    PATTERN (^ ()*? symbol1{0}{1,2}{,3} $)
+  )
+;
+
+SELECT *
+FROM
+  t MATCH_RECOGNIZE (
+    ORDER BY col1
+    PATTERN (symbol1 | symbol2 symbol3)
+  )
+;
+
+SELECT *
+FROM
+  t MATCH_RECOGNIZE (
+    ORDER BY col1
+    PATTERN (
+      symbol1 symbol2
+      | symbol3 symbol4
+      -- comment
+      | symbol5 symbol6
+    )
+  )
+;
+
+SELECT *
+FROM
+  t MATCH_RECOGNIZE (
+    ORDER BY col1
+    PATTERN ((symbol1 | ) ( | symbol2))
+  )
+;
+
 -- TABLESAMPLE
 SELECT * FROM t TABLESAMPLE SYSTEM (20 PERCENT);
 

input/core.sql

@@ -228,5 +228,25 @@ window
   named_window2 as (partition by str order by int)
 ;
 
+----- chained function call -----
+select (col).upper().lower();
+
+select
+  timestamp '2024-01-01'.timestamp_trunc(month),
+  case when true then timestamp '2024-01-01' end.timestamp_trunc(month),
+;
+
+select (col).(safe.left)(3);
+
+select struct('a' as b).to_json().b.json_value().concat('c');
+
+select "foo"
+    .longfunctionname().longfunctionname().longfunctionname().longfunctionname()
+    -- comment
+    .longfunctionname().longfunctionname().longfunctionname().longfunctionname().longfunctionname();
+
+----- with expression -----
+select with(a as 'a', upper(a));
+
 ----- template -----
 select {{variable}}, {variable};

input/ddl.sql

@@ -103,6 +103,10 @@ create external table dataset.new_table
 with connection ident
 options (dummy = 'dummy');
 
+create external table tablename
+with connection default
+options (dummy = 'dummy');
+
 ----- CREATE VIEW statement -----
 create view view_name
 as
@@ -182,6 +186,12 @@ as
   select 1 as one
 ;
 
+create table function foo.bar (
+  tablename table<col1 int64, col2 float64>
+) as (
+  select 1
+);
+
 -- remote function
 create function dataset.abc()
 returns int64
@@ -245,6 +255,12 @@ storing (a, b, c)
 options (dummy = 'dummy')
 ;
 
+create vector index new_index
+on tablename(col)
+partition by foo
+options(dummy='dummy')
+;
+
 ----- ALTER BI_CAPACITY statement -----
 alter bi_capacity `project.region-us.default`
 set options(preferred_tables = ['table1', 'table2']);
@@ -372,6 +388,9 @@ alter column colname set options(dummy='dummy')
 -- MATERIALIZED
 alter materialized view if exists example set options (dummy = 'dummy');
 
+----- ALTER VECTOR INDEX statement -----
+alter vector index index_name on table_name rebuild;
+
 ----- DROP statement -----
 -- general
 drop table example;

input/ml.sql

@@ -12,6 +12,11 @@ options (endpoint = '')
 as (select 1)
 ;
 
+create model ident
+remote with connection default
+options (endpoint = '')
+;
+
 create model ident
 as (
   training_data as (select 1),

input/pipe.sql

@@ -11,6 +11,10 @@ from table
 -- comment before select
 select col;
 
+with t as (select 1) from t;
+
+select * from (from t |> select *);
+
 ----- select statement -----
 select 1 |> select *;
 
@@ -20,6 +24,12 @@ select 1 union all select 2 |> select *;
 
 (select 1) order by 1 limit 1 |> SELECT *;
 
+from t |>
+  select col
+  -- break parent
+  window a as (partition by b)
+;
+
 ----- from statement -----
 from tabe as t1
 join table as p2 using (col);
@@ -64,6 +74,9 @@ from t
   aggregate count(*)
   group by col1 as col_a desc nulls last, col2;
 
+----- distinct pipe operator -----
+from t |> distinct;
+
 ----- union pipe operator -----
 from t |> union all (select 1), (select 2);
 
@@ -108,3 +121,29 @@ sum(sales) for quarter in ('Q1', 'Q2')) as q;
 from t |> unpivot (sales for quarter in (q1, q2)) as q;
 
 from t |> unpivot include nulls (sales for quarter in (q1, q2)) as q;
+
+----- match recognize pipe operator -----
+from t |> match_recognize (
+  order by col1
+  pattern (symbol)
+);
+
+----- with pipe operator -----
+from t
+|> with u as (
+    select 1 as key
+  )
+|> inner join u using (key)
+;
+
+from t
+|> with
+  -- CTE
+  u as (
+    select 1 as key
+  ),
+  v as (
+    select 2 as key
+  ),
+|> inner join u using (key)
+;