Allow more expressions for selected statements (#242)

dey4ss · web-flow · commit 1969a3c131e0 · 2024-06-12T12:52:06.000+02:00
Carefully extend the range of expressions allowed in the `INSERT` statements, but also in `EXECUTE` statements and in hints.
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -27,19 +27,19 @@ jobs:
             os: ubuntu-latest
             container: ubuntu:18.04
 
-          - name: gcc-12
-            cc: gcc-12
-            cxx: g++-12
+          - name: gcc-13
+            cc: gcc-13
+            cxx: g++-13
             os: ubuntu-latest
-            container: ubuntu:22.04
+            container: ubuntu:24.04
             # We need relaxed builds for debug mode with current GCC versions, see #218.
             build_options: "relaxed_build=on"
 
-          - name: clang-15
-            cc: clang-15
-            cxx: clang++-15
+          - name: clang-18
+            cc: clang-18
+            cxx: clang++-18
             os: ubuntu-latest
-            container: ubuntu:22.04
+            container: ubuntu:24.04
 
           - name: clang-macOS
             cc: clang
diff --git a/src/parser/bison_parser.cpp b/src/parser/bison_parser.cpp
diff --git a/src/parser/bison_parser.h b/src/parser/bison_parser.h
diff --git a/src/parser/bison_parser.y b/src/parser/bison_parser.y
@@ -257,7 +257,7 @@
 %type <expr>                   function_expr between_expr expr_alias param_expr
 %type <expr>                   column_name literal int_literal num_literal string_literal bool_literal date_literal interval_literal
 %type <expr>                   comp_expr opt_where join_condition opt_having case_expr case_list in_expr hint
-%type <expr>                   array_expr array_index null_literal
+%type <expr>                   array_expr array_index null_literal extended_literal casted_extended_literal
 %type <limit>                  opt_limit opt_top
 %type <order>                  order_desc
 %type <order_type>             opt_order_type
@@ -284,7 +284,7 @@
 %type <import_type_t>          opt_file_type file_type
 
 %type <str_vec>                ident_commalist opt_column_list
-%type <expr_vec>               expr_list select_list opt_literal_list literal_list hint_list opt_hints opt_partition
+%type <expr_vec>               expr_list select_list opt_extended_literal_list extended_literal_list hint_list opt_hints opt_partition
 %type <table_vec>              table_ref_commalist
 %type <order_vec>              opt_order order_list
 %type <with_description_vec>   opt_with_clause with_clause with_description_list
@@ -397,7 +397,7 @@ hint : IDENTIFIER {
   $$ = Expr::make(kExprHint);
   $$->name = $1;
 }
-| IDENTIFIER '(' literal_list ')' {
+| IDENTIFIER '(' extended_literal_list ')' {
   $$ = Expr::make(kExprHint);
   $$->name = $1;
   $$->exprList = $3;
@@ -423,13 +423,13 @@ prepare_statement : PREPARE IDENTIFIER FROM prepare_target_query {
   $$->query = $4;
 };
 
-prepare_target_query : STRING
+prepare_target_query : STRING;
 
-                           execute_statement : EXECUTE IDENTIFIER {
+execute_statement : EXECUTE IDENTIFIER {
   $$ = new ExecuteStatement();
   $$->name = $2;
 }
-| EXECUTE IDENTIFIER '(' opt_literal_list ')' {
+| EXECUTE IDENTIFIER '(' opt_extended_literal_list ')' {
   $$ = new ExecuteStatement();
   $$->name = $2;
   $$->parameters = $4;
@@ -708,7 +708,7 @@ truncate_statement : TRUNCATE table_name {
  * INSERT INTO students VALUES ('Max', 1112233, 'Musterhausen', 2.3)
  * INSERT INTO employees SELECT * FROM stundents
  ******************************/
-insert_statement : INSERT INTO table_name opt_column_list VALUES '(' literal_list ')' {
+insert_statement : INSERT INTO table_name opt_column_list VALUES '(' extended_literal_list ')' {
   $$ = new InsertStatement(kInsertValues);
   $$->schema = $3.schema;
   $$->tableName = $3.name;
@@ -914,18 +914,34 @@ expr_list : expr_alias {
   $$ = $1;
 };
 
-opt_literal_list : literal_list { $$ = $1; }
+// Literals, casted literals, and negative numbers/intervals are allowed for INSERT and EXECUTE statements or hints.
+opt_extended_literal_list : extended_literal_list { $$ = $1; }
 | /* empty */ { $$ = nullptr; };
 
-literal_list : literal {
+extended_literal_list : casted_extended_literal {
   $$ = new std::vector<Expr*>();
   $$->push_back($1);
 }
-| literal_list ',' literal {
+| extended_literal_list ',' casted_extended_literal {
   $1->push_back($3);
   $$ = $1;
 };
 
+casted_extended_literal : extended_literal | CAST '(' extended_literal AS column_type ')' {
+  $$ = Expr::makeCast($3, $5);
+};
+
+extended_literal : literal {
+  if ($1->type == ExprType::kExprParameter) {
+    delete $1;
+    yyerror(&yyloc, result, scanner, "Parameter ? is not a valid literal.");
+    YYERROR;
+  }
+  $$ = $1;
+}
+| '-' num_literal { $$ = Expr::makeOpUnary(kOpUnaryMinus, $2); };
+| '-' interval_literal { $$ = Expr::makeOpUnary(kOpUnaryMinus, $2); };
+
 expr_alias : expr opt_alias {
   $$ = $1;
   if ($2) {
@@ -1078,10 +1094,7 @@ date_literal : DATE STRING {
   $$ = Expr::makeDateLiteral($2);
 };
 
-interval_literal : int_literal duration_field {
-  $$ = Expr::makeIntervalLiteral($1->ival, $2);
-  delete $1;
-}
+interval_literal : INTVAL duration_field { $$ = Expr::makeIntervalLiteral($1, $2); }
 | INTERVAL STRING datetime_field {
   int duration{0}, chars_parsed{0};
   // If the whole string is parsed, chars_parsed points to the terminating null byte after the last character
@@ -1314,9 +1327,11 @@ ident_commalist : IDENTIFIER {
 
 // clang-format off
 %%
-    // clang-format on
-    /*********************************
+
+/*********************************
  ** Section 4: Additional C code
  *********************************/
 
-    /* empty */
+/* empty */
+
+    // clang-format on
diff --git a/src/sql/Table.h b/src/sql/Table.h
@@ -60,9 +60,9 @@ struct JoinDefinition {
   TableRef* left;
   TableRef* right;
   Expr* condition;
+  std::vector<char*>* namedColumns;
 
   JoinType type;
-  std::vector<char*>* namedColumns;
 };
 
 }  // namespace hsql
diff --git a/src/sql/statements.cpp b/src/sql/statements.cpp
@@ -368,12 +368,20 @@ const char* TableRef::getName() const {
 }
 
 // JoinDefinition
-JoinDefinition::JoinDefinition() : left(nullptr), right(nullptr), condition(nullptr), type(kJoinInner) {}
+JoinDefinition::JoinDefinition()
+    : left(nullptr), right(nullptr), condition(nullptr), namedColumns(nullptr), type(kJoinInner) {}
 
 JoinDefinition::~JoinDefinition() {
   delete left;
   delete right;
   delete condition;
+
+  if (namedColumns) {
+    for (auto* column : *namedColumns) {
+      free(column);
+    }
+    delete namedColumns;
+  }
 }
 
 SetOperation::SetOperation() : nestedSelectStatement(nullptr), resultOrder(nullptr), resultLimit(nullptr) {}
diff --git a/test/queries/queries-bad.sql b/test/queries/queries-bad.sql
@@ -68,3 +68,13 @@
 !SELECT * FROM foo INNER JOIN bar USING (a b);
 !SELECT * FROM foo INNER JOIN bar USING (a AS b);
 !SELECT * FROM foo INNER JOIN bar USING (1);
+# INSERT, EXECUTE, and HINTS only allow specific expressions.
+!INSERT INTO foo VALUES (?);
+!INSERT INTO foo VALUES (CAST(column_a AS INT));
+!INSERT INTO foo VALUES (AVG(another_column));
+!EXECUTE statement_a(?);
+!EXECUTE statement_a(CAST(column_a AS INT));
+!EXECUTE statement_a(AVG(another_column));
+!SELECT * FROM foo WITH HINT (?);
+!SELECT * FROM foo WITH HINT (CAST(column_a AS INT));
+!SELECT * FROM foo WITH HINT (AVG(another_column));
diff --git a/test/queries/queries-good.sql b/test/queries/queries-good.sql
@@ -61,7 +61,7 @@ DROP INDEX IF EXISTS myindex;
 # PREPARE
 PREPARE prep_inst FROM 'INSERT INTO test VALUES (?, ?, ?)';
 PREPARE prep2 FROM 'INSERT INTO test VALUES (?, 0, 0); INSERT INTO test VALUES (0, ?, 0); INSERT INTO test VALUES (0, 0, ?);';
-EXECUTE prep_inst(1, 2, 3);
+EXECUTE prep_another_inst(1, 2, 3, CAST(1 AS LONG), -2.5, INTERVAL '3 HOURS', -2 SECONDS, TRUE, NULL, DATE '2000-01-01');
 EXECUTE prep;
 DEALLOCATE PREPARE prep;
 # COPY
@@ -80,7 +80,8 @@ COPY (SELECT firstname, COUNT(*) FROM students GROUP BY firstname) TO 'student_n
 # HINTS
 SELECT * FROM test WITH HINT(NO_CACHE);
 SELECT * FROM test WITH HINT(NO_CACHE, NO_SAMPLING);
-SELECT * FROM test WITH HINT(NO_CACHE, SAMPLE_RATE(0.1), OMW(1.0, 'test'));
+SELECT * FROM test WITH HINT(NO_CACHE, SAMPLE_RATE(0.1), OMW(1.0, 'test'), START_DATE(CAST('2000-01-01' AS DATE)));
+SELECT * FROM test WITH HINT(TIME_DIFFERENCE(-3 HOURS), ALLOW_RESULT_CACHE(FALSE), DEFAULT_VALUE(NULL), TIMETRAVEL_TO(DATE '2000-01-01'));
 SHOW TABLES;
 SHOW COLUMNS students;
 DESCRIBE students;
diff --git a/test/select_tests.cpp b/test/select_tests.cpp
@@ -803,7 +803,7 @@ TEST(WithClauseSingle) {
       "WITH "
       "a AS (SELECT name FROM peopleA)"
       "SELECT name FROM a;",
-      kStmtSelect, SelectStatement, result, stmt)
+      kStmtSelect, SelectStatement, result, stmt);
 
   // with_description_list – count
   ASSERT_EQ(stmt->withDescriptions->size(), 1);
@@ -828,7 +828,7 @@ TEST(WithClauseDouble) {
       "a AS (SELECT nameA FROM peopleA), "
       "b AS (SELECT nameB, cityB FROM peopleB) "
       "SELECT nameA FROM a;",
-      kStmtSelect, SelectStatement, result, stmt)
+      kStmtSelect, SelectStatement, result, stmt);
 
   // with_description_list – count
   ASSERT_EQ(stmt->withDescriptions->size(), 2);
diff --git a/test/sql_asserts.h b/test/sql_asserts.h
@@ -5,15 +5,15 @@
   hsql::SQLParserResult result;                            \
   hsql::SQLParser::parse(query, &result);                  \
   ASSERT(result.isValid());                                \
-  ASSERT_EQ(result.size(), numStatements);
+  ASSERT_EQ(result.size(), numStatements)
 
 #define TEST_PARSE_SINGLE_SQL(query, stmtType, stmtClass, result, outputVar) \
   TEST_PARSE_SQL_QUERY(query, result, 1);                                    \
   ASSERT_EQ(result.getStatement(0)->type(), stmtType);                       \
-  const stmtClass* outputVar = (const stmtClass*)result.getStatement(0);
+  const stmtClass* outputVar = (const stmtClass*)result.getStatement(0)
 
 #define TEST_CAST_STMT(result, stmt_index, stmtType, stmtClass, outputVar) \
   ASSERT_EQ(result.getStatement(stmt_index)->type(), stmtType);            \
-  const stmtClass* outputVar = (const stmtClass*)result.getStatement(stmt_index);
+  const stmtClass* outputVar = (const stmtClass*)result.getStatement(stmt_index)
 
 #endif
diff --git a/test/sql_tests.cpp b/test/sql_tests.cpp
@@ -224,11 +224,45 @@ TEST(UpdateStatementTest) {
 }
 
 TEST(InsertStatementTest) {
-  TEST_PARSE_SINGLE_SQL("INSERT INTO students VALUES ('Max Mustermann', 12345, 'Musterhausen', 2.0)", kStmtInsert,
-                        InsertStatement, result, stmt);
-
-  ASSERT_EQ(stmt->values->size(), 4);
-  // TODO
+  TEST_PARSE_SINGLE_SQL(
+      "INSERT INTO students VALUES ('Max Mustermann', 12345, 'Musterhausen', 2.0, -1, 1 month, "
+      " CAST('2000-02-02' AS DATE), - INTERVAL '3 seconds', DATE '2000-02-02', FALSE, NULL)",
+      kStmtInsert, InsertStatement, result, stmt);
+
+  ASSERT_EQ(stmt->values->size(), 11);
+  ASSERT_EQ(stmt->values->at(0)->type, kExprLiteralString);
+  ASSERT_STREQ(stmt->values->at(0)->name, "Max Mustermann");
+  ASSERT_EQ(stmt->values->at(1)->type, kExprLiteralInt);
+  ASSERT_EQ(stmt->values->at(1)->ival, 12345);
+  ASSERT_EQ(stmt->values->at(2)->type, kExprLiteralString);
+  ASSERT_STREQ(stmt->values->at(2)->name, "Musterhausen");
+  ASSERT_EQ(stmt->values->at(3)->type, kExprLiteralFloat);
+  ASSERT_EQ(stmt->values->at(3)->fval, 2.0);
+  ASSERT_EQ(stmt->values->at(4)->type, kExprOperator);
+  ASSERT_EQ(stmt->values->at(4)->opType, kOpUnaryMinus);
+  ASSERT(stmt->values->at(4)->expr);
+  ASSERT_EQ(stmt->values->at(4)->expr->type, kExprLiteralInt);
+  ASSERT_EQ(stmt->values->at(4)->expr->ival, 1);
+  ASSERT_EQ(stmt->values->at(5)->type, kExprLiteralInterval);
+  ASSERT_EQ(stmt->values->at(5)->ival, 1);
+  ASSERT_EQ(stmt->values->at(5)->datetimeField, kDatetimeMonth);
+  ASSERT_EQ(stmt->values->at(6)->type, kExprCast);
+  ASSERT_EQ(stmt->values->at(6)->columnType, ColumnType{DataType::DATE});
+  ASSERT(stmt->values->at(6)->expr);
+  ASSERT_EQ(stmt->values->at(6)->expr->type, kExprLiteralString);
+  ASSERT_STREQ(stmt->values->at(6)->expr->name, "2000-02-02");
+  ASSERT_EQ(stmt->values->at(7)->type, kExprOperator);
+  ASSERT_EQ(stmt->values->at(7)->opType, kOpUnaryMinus);
+  ASSERT(stmt->values->at(7)->expr);
+  ASSERT_EQ(stmt->values->at(7)->expr->type, kExprLiteralInterval);
+  ASSERT_EQ(stmt->values->at(7)->expr->ival, 3);
+  ASSERT_EQ(stmt->values->at(7)->expr->datetimeField, kDatetimeSecond);
+  ASSERT_EQ(stmt->values->at(8)->type, kExprLiteralDate);
+  ASSERT_STREQ(stmt->values->at(8)->name, "2000-02-02");
+  ASSERT_EQ(stmt->values->at(9)->type, kExprLiteralInt);
+  ASSERT_EQ(stmt->values->at(9)->ival, 0);
+  ASSERT_TRUE(stmt->values->at(9)->isBoolLiteral);
+  ASSERT_EQ(stmt->values->at(10)->type, kExprLiteralNull);
 }
 
 TEST(AlterStatementDropActionTest) {
diff --git a/test/test.sh b/test/test.sh
@@ -11,10 +11,10 @@ NC='\033[0m'
 BOLD='\033[1;39m'
 
 RET=0
-SQL_TEST_RET=0
-MEM_LEAK_EXECUTED=$?
-MEM_LEAK_RET=0
-CONFLICT_RET=0
+SQL_TEST_RET=1
+MEM_LEAK_EXECUTED=false
+MEM_LEAK_RET=1
+CONFLICT_RET=1
 
 #################################################
 # Running SQL parser tests.
@@ -25,7 +25,6 @@ SQL_TEST_RET=$?
 if [ $SQL_TEST_RET -eq 0 ]; then
   printf "${GREEN}SQL parser tests succeeded!${NC}\n"
 else
-  RET=1
   printf "${RED}SQL parser tests failed!${NC}\n"
 fi
 
@@ -38,14 +37,11 @@ if [[ "$unamestr" == 'Linux' ]]; then
     bin/tests -f "test/queries/queries-good.sql" -f "test/queries/queries-bad.sql" \
     3>&1>/dev/null;
 
-  MEM_LEAK_EXECUTED=true
   MEM_LEAK_RET=$?
-  RET=1
+  MEM_LEAK_EXECUTED=true
 
   if [ $MEM_LEAK_RET -eq 0 ]; then
     printf "${GREEN}Memory leak check succeeded!${NC}\n"
-    MEM_LEAK_RET=0
-    RET=0
   elif [ $MEM_LEAK_RET -eq 200 ]; then
     printf "${RED}Memory leak check failed!${NC}\n"
   elif [ $MEM_LEAK_RET -eq 127 ]; then
@@ -55,7 +51,6 @@ if [[ "$unamestr" == 'Linux' ]]; then
   fi
 else
   printf "\n${YELLOW}Skipping memory leak checks (can only be executed on Linux)!${NC}\n"
-  MEM_LEAK_EXECUTED=false
 fi
 
 #################################################
@@ -68,7 +63,6 @@ CONFLICT_RET=$?
 if [ $CONFLICT_RET -eq 0 ]; then
   printf "${GREEN}Conflict check succeeded!${NC}\n"
 else
-  RET=1
   printf "${RED}Conflict check failed!${NC}\n"
 fi
 
@@ -86,9 +80,22 @@ fi
 if [ $CONFLICT_RET -eq 0 ]; then printf "Grammar Conflict Check: ${GREEN}Success${BOLD}\n";
 else                             printf "Grammar Conflict Check: ${RED}Failure${BOLD}\n"; fi
 
-if [ $RET -ne 0 ]; then                     printf "${RED}Some tests failed!${NC}\n"
-elif [ "$MEM_LEAK_EXECUTED" = false ]; then printf "${YELLOW}Some tests were skipped!${NC}\n"
-else                                        printf "${GREEN}All tests passed!${NC}\n"
+if [[ $SQL_TEST_RET -ne 0 || $CONFLICT_RET -ne 0 ]]; then
+  RET=1
+fi
+
+if [ $MEM_LEAK_RET -ne 0 ]; then
+  if [ "$MEM_LEAK_EXECUTED" = true ]; then
+    RET=1
+  fi
+fi
+
+
+if [ $RET -eq 0 ]; then
+  if [ "$MEM_LEAK_EXECUTED" = true ]; then printf "${GREEN}All tests passed!${NC}\n"
+  else                                     printf "${YELLOW}Some tests were skipped!${NC}\n"
+  fi
+else                                       printf "${RED}Some tests failed!${NC}\n"
 fi
 printf "${NC}----------------------------------\n"
 
