HIVE-29786: Support SELECT * EXCLUDE syntax for column omission

parser/src/java/org/apache/hadoop/hive/ql/parse/FromClauseParser.g

@@ -48,10 +48,10 @@ catch (RecognitionException e) {
 //-----------------------------------------------------------------------------------
 
 tableAllColumns
-    : STAR
-        -> ^(TOK_ALLCOLREF)
-    | tableName DOT STAR
-        -> ^(TOK_ALLCOLREF tableName)
+    : STAR (KW_EXCLUDE LPAREN columnNameList RPAREN)?
+        -> ^(TOK_ALLCOLREF columnNameList?)
+    | tableName DOT STAR (KW_EXCLUDE LPAREN columnNameList RPAREN)?
+        -> ^(TOK_ALLCOLREF tableName columnNameList?)
     ;
 
 // (table|column)

parser/src/java/org/apache/hadoop/hive/ql/parse/HiveLexerParent.g

@@ -399,6 +399,7 @@ KW_SYSTEM_TIME: 'SYSTEM_TIME';
 KW_SYSTEM_VERSION: 'SYSTEM_VERSION';
 KW_EXPIRE_SNAPSHOTS: 'EXPIRE_SNAPSHOTS';
 KW_REWRITE_MANIFESTS: 'REWRITE_MANIFESTS';
+KW_EXCLUDE: 'EXCLUDE';
 KW_SET_CURRENT_SNAPSHOT: 'SET_CURRENT_SNAPSHOT';
 KW_BRANCH: 'BRANCH';
 KW_SNAPSHOTS: 'SNAPSHOTS';

parser/src/java/org/apache/hadoop/hive/ql/parse/IdentifiersParser.g

@@ -1026,6 +1026,7 @@ nonReserved
     | KW_TRIM
     | KW_SPEC
     | KW_SYSTEM_TIME | KW_SYSTEM_VERSION
+    | KW_EXCLUDE
     | KW_EXPIRE_SNAPSHOTS
     | KW_REWRITE_MANIFESTS
     | KW_SET_CURRENT_SNAPSHOT

ql/src/java/org/apache/hadoop/hive/ql/parse/CalcitePlanner.java

@@ -4366,8 +4366,11 @@ private Pair<RelNode, RowResolver> internalGenSelectLogicalPlan(QB qb, RelNode s
 
           // 6.4 Build ExprNode corresponding to colums
           if (expr.getType() == HiveParser.TOK_ALLCOLREF) {
-            pos = genRexNodeRegex(".*",
-                expr.getChildCount() == 0 ? null : getUnescapedName((ASTNode) expr.getChild(0)).toLowerCase(),
+            // Parse SELECT * EXCLUDE columns and pass them to the Calcite engine for exclusion
+            ExcludeResult excludeResult = processAllColRefAndExclude(expr, inputRR);
+            String starTabAlias = excludeResult.tableAlias();
+            excludedColumns.addAll(excludeResult.excludedColumns());
+            pos = genRexNodeRegex(".*", starTabAlias,
                 expr, columnList, excludedColumns, inputRR, starRR, pos, outputRR, qb.getAliases(), true);
           } else if (expr.getType() == HiveParser.TOK_TABLE_OR_COL
                   && !hasAsClause

ql/src/java/org/apache/hadoop/hive/ql/parse/SemanticAnalyzer.java

@@ -4737,6 +4737,40 @@ static boolean isRegex(String pattern, HiveConf conf) {
     return false;
   }
 
+  protected record ExcludeResult(String tableAlias, Set<ColumnInfo> excludedColumns) {}
+
+  protected ExcludeResult processAllColRefAndExclude(ASTNode expr, RowResolver inputRR)
+      throws SemanticException {
+    // Check if the query uses SELECT * EXCLUDE. If it does, grab the table
+    // alias (like t.*) and build a list of the columns the user wants to exclude.
+    String starTabAlias = null;
+    ASTNode excludeNode = null;
+    Set<ColumnInfo> excludedColumns = new HashSet<>();
+
+    if (expr.getChildren() != null) {
+      for (Node childNode : expr.getChildren()) {
+        ASTNode child = (ASTNode) childNode;
+        switch (child.getType()) {
+          case HiveParser.TOK_TABNAME -> starTabAlias = getUnescapedName(child).toLowerCase();
+          case HiveParser.TOK_TABCOLNAME -> excludeNode = child;
+          default ->
+              throw new SemanticException(
+                  "Unexpected node type in TOK_ALLCOLREF: " + child.getType());
+        }
+      }
+    }
+
+    if (excludeNode != null) {
+      for (int e = 0; e < excludeNode.getChildCount(); e++) {
+        String excludeColName = unescapeIdentifier(excludeNode.getChild(e).getText()).toLowerCase();
+        ColumnInfo colInfo = inputRR.get(starTabAlias, excludeColName);
+        if (colInfo != null) {
+          excludedColumns.add(colInfo);
+        }
+      }
+    }
+    return new ExcludeResult(starTabAlias, excludedColumns);
+  }
 
   private Operator<?> genSelectPlan(String dest, QB qb, Operator<?> input,
                                     Operator<?> inputForSelectStar) throws SemanticException {
@@ -4914,9 +4948,16 @@ private Operator<?> genSelectPlan(String dest, ASTNode selExprList, QB qb, Opera
       // The real expression
       if (expr.getType() == HiveParser.TOK_ALLCOLREF) {
         int initPos = pos;
-        pos = genExprNodeDescRegex(".*", expr.getChildCount() == 0 ? null
-                : getUnescapedName((ASTNode) expr.getChild(0)).toLowerCase(),
-            expr, colList, null, inputRR, starRR, pos, out_rwsch, qb.getAliases(), false);
+
+        ExcludeResult excludeResult = processAllColRefAndExclude(expr, inputRR);
+        String starTabAlias = excludeResult.tableAlias();
+        Set<ColumnInfo> excludeCols = excludeResult.excludedColumns();
+        if (excludeCols.isEmpty()) {
+          excludeCols = null;
+        }
+
+        pos = genExprNodeDescRegex(".*", starTabAlias,
+            expr, colList, excludeCols, inputRR, starRR, pos, out_rwsch, qb.getAliases(), false);
         if (unparseTranslator.isEnabled()) {
           offset += pos - initPos - 1;
         }

ql/src/test/queries/clientpositive/select_exclude.q

@@ -0,0 +1,39 @@
+set hive.cli.print.header=true;
+
+CREATE TABLE test_exclude (
+  id INT,
+  name STRING,
+  email STRING,
+  address STRING,
+  phone STRING
+);
+
+INSERT INTO test_exclude VALUES
+(1, 'Alice', 'alice@test.com', '123 Apple St', '555-0100'),
+(2, 'Bob', 'bob@test.com', '456 Banana Ave', '555-0200');
+
+-- Exclude a single column
+EXPLAIN SELECT * EXCLUDE (email) FROM test_exclude;
+SELECT * EXCLUDE (email) FROM test_exclude;
+
+-- Exclude multiple columns
+EXPLAIN SELECT * EXCLUDE (email, address, phone) FROM test_exclude;
+SELECT * EXCLUDE (email, address, phone) FROM test_exclude;
+
+-- Exclude with table alias
+EXPLAIN SELECT t.* EXCLUDE (id, phone) FROM test_exclude t;
+SELECT t.* EXCLUDE (id, phone) FROM test_exclude t;
+
+-- Exclude with JOIN
+CREATE TABLE test_exclude_join (
+  id INT,
+  department STRING
+);
+INSERT INTO test_exclude_join VALUES (1, 'Engineering'), (2, 'Sales');
+
+EXPLAIN
+SELECT a.* EXCLUDE (address, phone), b.* EXCLUDE (id)
+FROM test_exclude a JOIN test_exclude_join b ON a.id = b.id;
+
+SELECT a.* EXCLUDE (address, phone), b.* EXCLUDE (id)
+FROM test_exclude a JOIN test_exclude_join b ON a.id = b.id;