Merge pull request taozhi8833998#2610 from taozhi8833998/fix-struct-expr-bigquery

fix: struct expr with orderby and limit in bigquery

Author: taozhi8833998

pegjs/bigquery.pegjs

@@ -2201,13 +2201,15 @@ json_expr
   }
 
 struct_expr
-  = s:(struct_type / KW_STRUCT) __ LPAREN __ c:column_clause __ RPAREN {
+  = s:(struct_type / KW_STRUCT) __ LPAREN __ c:column_clause __ RPAREN __ o:order_by_clause?  __ l:limit_clause? {
     return {
       definition: s,
       expr_list: c,
       type: 'struct',
       keyword: s && 'struct',
-      parentheses: true
+      parentheses: true,
+      order_by: o,
+      limit: l
     }
   }
 
@@ -2598,6 +2600,7 @@ aggr_func_list
 aggr_func
   = aggr_fun_count
   / aggr_fun_smma
+  / aggr_array_agg
 
 aggr_fun_smma
   = name:KW_SUM_MAX_MIN_AVG  __ LPAREN __ e:additive_expr __ RPAREN __ bc:over_partition? {
@@ -2675,6 +2678,16 @@ count_arg
   }
   / d:KW_DISTINCT? __ c:or_and_expr __ or:order_by_clause?  { return { distinct: d, expr: c, orderby: or, ...getLocationObject() }; }
 
+aggr_array_agg
+  = pre:(ident __ DOT)? __ name:(KW_ARRAY_AGG / KW_STRING_AGG) __ LPAREN __ arg:expr __ RPAREN {
+    // => { type: 'aggr_func'; args:count_arg; name: 'ARRAY_AGG' | 'STRING_AGG';  }
+      return {
+        type: 'aggr_func',
+        name: pre ? `${pre[0]}.${name}` : name,
+        args: arg,
+      };
+    }
+
 star_expr
   = "*" { return { type: 'star', value: '*' }; }
 
@@ -3089,6 +3102,8 @@ KW_CAST     = "CAST"i       !ident_start { return 'CAST' }
 KW_SAFE_CAST     = "SAFE_CAST"i   !ident_start { return 'SAFE_CAST' }
 
 KW_ARRAY     = "ARRAY"i     !ident_start { return 'ARRAY'; }
+KW_ARRAY_AGG = "ARRAY_AGG"i !ident_start { return 'ARRAY_AGG'; }
+KW_STRING_AGG = "STRING_AGG"i !ident_start { return 'STRING_AGG'; }
 KW_BYTES     = "BYTES"i     !ident_start { return 'BYTES'; }
 KW_BOOL     = "BOOL"i     !ident_start { return 'BOOL'; }
 KW_CHAR     = "CHAR"i     !ident_start { return 'CHAR'; }

src/array-struct.js

@@ -1,5 +1,6 @@
 import { columnsToSQL } from './column'
-import { exprToSQL } from './expr'
+import { exprToSQL, orderOrPartitionByToSQL } from './expr'
+import { limitToSQL } from './limit'
 import { arrayStructTypeToSQL, hasVal, toUpper } from './util'
 
 function arrayExprListToSQL(expr) {
@@ -32,13 +33,13 @@ function arrayStructValueToSQL(expr) {
 }
 
 function arrayStructExprToSQL(expr) {
-  const { definition, keyword } = expr
+  const { definition, keyword, orderby, limit } = expr
   const result = [toUpper(keyword)]
   if (definition && typeof definition === 'object') {
     result.length = 0
     result.push(arrayStructTypeToSQL(definition))
   }
-  result.push(arrayStructValueToSQL(expr))
+  result.push(arrayStructValueToSQL(expr), orderOrPartitionByToSQL(orderby, 'order by'), limitToSQL(limit))
   return result.filter(hasVal).join('')
 }
 

test/bigquery.spec.js

@@ -897,7 +897,52 @@ describe('BigQuery', () => {
         'SELECT COUNT(*) FROM `bigquery-public-data.blah.events_*`',
         'SELECT COUNT(*) FROM `bigquery-public-data.blah.events_*`'
       ]
-    }
+    },
+    {
+      title: 'struct expr',
+      sql: [
+        `WITH thingie AS (
+          SELECT
+            thingie2,
+            thingie3,
+            COUNT(*) as thingie4
+          FROM thingie5.thingie6
+          WHERE thingie7 BETWEEN '2025-10-24' AND '2025-10-30'
+            AND thingie8 = TRUE
+            AND thingie2 IS NOT NULL
+          GROUP BY thingie2, thingie3
+        ),
+        path_totals AS (
+          SELECT
+            thingie2,
+            SUM(thingie4) as thingie9
+          FROM thingie
+          GROUP BY thingie2
+        ),
+        thingie13 AS (
+          SELECT SUM(thingie9) as overall_total
+          FROM path_totals
+        ),
+        thingie10 AS (
+          SELECT
+            thingie2,
+            ARRAY_AGG(STRUCT(thingie3, thingie4) ORDER BY thingie4 DESC LIMIT 3) as thingie11
+          FROM thingie
+          GROUP BY thingie2
+        )
+        SELECT
+          pt.thingie2,
+          pt.thingie9 as thingie4,
+          ROUND(100.0 * pt.thingie9 / tar.overall_total, 2) as thingie12,
+          tc.thingie11
+        FROM path_totals pt
+        CROSS JOIN thingie13 tar
+        JOIN thingie10 tc ON pt.thingie2 = tc.thingie2
+        ORDER BY pt.thingie9 DESC
+        LIMIT 20`,
+        "WITH thingie AS (SELECT thingie2, thingie3, COUNT(*) AS thingie4 FROM thingie5.thingie6 WHERE thingie7 BETWEEN '2025-10-24' AND '2025-10-30' AND thingie8 = TRUE AND thingie2 IS NOT NULL GROUP BY thingie2, thingie3), path_totals AS (SELECT thingie2, SUM(thingie4) AS thingie9 FROM thingie GROUP BY thingie2), thingie13 AS (SELECT SUM(thingie9) AS overall_total FROM path_totals), thingie10 AS (SELECT thingie2, ARRAY_AGG(undefined) AS thingie11 FROM thingie GROUP BY thingie2) SELECT pt.thingie2, pt.thingie9 AS thingie4, ROUND(100.0 * pt.thingie9 / tar.overall_total, 2) AS thingie12, tc.thingie11 FROM path_totals AS pt CROSS JOIN thingie13 AS tar JOIN thingie10 AS tc ON pt.thingie2 = tc.thingie2 ORDER BY pt.thingie9 DESC LIMIT 20"
+      ]
+    },
   ]
 
   SQL_LIST.forEach(sqlInfo => {