Implemented UNION and OFFSET

run_tests.sh

@@ -10,8 +10,9 @@ make grammar_test -C src/
 echo "\n\n"
 
 
-./bin/analysis "SELECT t1.a AS id, t1.b, t2.c FROM \"tbl\" AS t1 JOIN (SELECT * FROM foo JOIN bar ON foo.id = bar.id) t2 ON t1.a = t2.b WHERE (t1.b OR NOT t1.a) AND t2.c = 12.5"
+# ./bin/analysis "SELECT t1.a AS id, t1.b, t2.c FROM \"tbl\" AS t1 JOIN (SELECT * FROM foo JOIN bar ON foo.id = bar.id) t2 ON t1.a = t2.b WHERE (t1.b OR NOT t1.a) AND t2.c = 12.5"
 # ./bin/analysis "CREATE TABLE \"table\" FROM TBL FILE 'students.tbl'"
+./bin/analysis "SELECT * FROM t1 UNION (SELECT abc AS t FROM t2 ORDER BY col3 LIMIT 10) ORDER BY col1;"
 
 echo "\n\n"
 

src/lib/CreateStatement.h

@@ -19,7 +19,11 @@ typedef enum {
  * @struct CreateStatement
  */
 struct CreateStatement : Statement {
-	CreateStatement() : Statement(kStmtCreate) {};
+	CreateStatement() :
+		Statement(kStmtCreate),
+		file_path(NULL),
+		table_name(NULL) {};
+		
 	virtual ~CreateStatement(); // defined in destructors.cpp
 
 	CreateType create_type;

src/lib/Expr.h

@@ -50,7 +50,13 @@ struct Expr {
 
 
 
-	Expr(ExprType type) : type(type), expr(NULL), expr2(NULL), name(NULL), table(NULL), alias(NULL) {};
+	Expr(ExprType type) :
+		type(type),
+		expr(NULL),
+		expr2(NULL),
+		name(NULL),
+		table(NULL),
+		alias(NULL) {};
 
 	// Interesting side-effect:
 	// Making the destructor virtual causes segmentation faults

src/lib/ImportStatement.h

@@ -19,7 +19,11 @@ typedef enum {
  * @struct ImportStatement
  */
 struct ImportStatement : Statement {
-	ImportStatement() : Statement(kStmtImport) {};
+	ImportStatement() :
+		Statement(kStmtImport),
+		file_path(NULL),
+		table_name(NULL) {};
+		
 	virtual ~ImportStatement(); // defined in destructors.cpp
 
 	ImportFileType file_type;

src/lib/SelectStatement.h

@@ -20,7 +20,10 @@ typedef enum {
 } OrderType;
 
 struct OrderDescription {
-	OrderDescription(OrderType type, Expr* expr) : type(type), expr(expr) {}
+	OrderDescription(OrderType type, Expr* expr) :
+		type(type),
+		expr(expr) {}
+		
 	virtual ~OrderDescription(); // defined in destructors.cpp
 
 	OrderType type;
@@ -34,7 +37,10 @@ struct OrderDescription {
 const int64_t kNoLimit = -1;
 const int64_t kNoOffset = -1;
 struct LimitDescription {
-	LimitDescription(int64_t limit, int64_t offset) : limit(limit), offset(offset) {}
+	LimitDescription(int64_t limit, int64_t offset) :
+		limit(limit),
+		offset(offset) {}
+
 	int64_t limit;
 	int64_t offset;	
 };
@@ -44,15 +50,24 @@ struct LimitDescription {
  * Representation of a full select statement.
  */
 struct SelectStatement : Statement {
-	SelectStatement() : Statement(kStmtSelect) {};
+	SelectStatement() : 
+		Statement(kStmtSelect),
+		from_table(NULL),
+		select_list(NULL),
+		where_clause(NULL),
+		group_by(NULL),
+		order(NULL),
+		limit(NULL),
+		union_select(NULL) {};
+
 	virtual ~SelectStatement(); // defined in destructors.cpp
 
 	TableRef* from_table;
 	List<Expr*>* select_list;
 	Expr* where_clause;	
-	
 	List<Expr*>* group_by;
 
+	SelectStatement* union_select;
 	OrderDescription* order;
 	LimitDescription* limit;
 };

src/lib/Statement.h

@@ -11,7 +11,7 @@ namespace hsql {
 
 
 typedef enum {
-	kStmtError,
+	kStmtError, // Unused
 	kStmtSelect,
 	kStmtImport,
 	// Following types are planned but not supported yet
@@ -26,7 +26,9 @@ typedef enum {
 
 
 struct Statement {
-	Statement(StatementType type) : type(type) {};
+	Statement(StatementType type) :
+		type(type) {};
+
 	virtual ~Statement(); // defined in destructors.cpp
 
 	StatementType type;
@@ -35,8 +37,16 @@ struct Statement {
 
 class StatementList : public List<Statement*> {
 public:
-	StatementList() : List<Statement*>(), isValid(true) {};
+	StatementList() :
-	StatementList(Statement* stmt) : List<Statement*>(stmt), isValid(true) {};
+		List<Statement*>(),
+		isValid(true),
+		parser_msg(NULL) {};
+
+	StatementList(Statement* stmt) :
+		List<Statement*>(stmt),
+		isValid(true),
+		parser_msg(NULL) {};
+		
 	virtual ~StatementList(); // defined in destructors.cpp
 
 	bool isValid;

src/lib/Table.h

@@ -29,7 +29,16 @@ typedef enum {
 typedef struct TableRef TableRef;
 
 struct TableRef {
-	TableRef(TableRefType type) : type(type) {}
+	TableRef(TableRefType type) :
+		type(type),
+		name(NULL),
+		alias(NULL),
+		select(NULL),
+		list(NULL),
+		left(NULL),
+		right(NULL),
+		join_condition(NULL) {}
+		
 	virtual ~TableRef(); // defined in destructors.cpp
 
 	TableRefType type;

src/lib/sqlhelper.cpp

@@ -70,7 +70,7 @@ void printExpression(Expr* expr, uint num_indent) {
     default: fprintf(stderr, "Unrecognized expression type %d\n", expr->type); return;
   }
   if (expr->alias != NULL) {
-    inprint("Alias", num_indent); inprint(expr->alias, num_indent+1);
+    inprint("Alias", num_indent+1); inprint(expr->alias, num_indent+2);
   }
 }
 
@@ -87,6 +87,12 @@ void printSelectStatementInfo(SelectStatement* stmt, uint num_indent) {
     printExpression(stmt->where_clause, num_indent+2);
   }
 
+
+  if (stmt->union_select != NULL) {
+    inprint("Union:", num_indent+1);
+    printSelectStatementInfo(stmt->union_select, num_indent+2);
+  }
+
   if (stmt->order != NULL) {
     inprint("OrderBy:", num_indent+1);
     printExpression(stmt->order->expr, num_indent+2);

src/parser/bison_parser.y

@@ -105,12 +105,12 @@ typedef void* yyscan_t;
 %token <uval> NOTEQUALS LESSEQ GREATEREQ
 
 /* SQL Keywords */
-%token DISTINCT DATABASE NATURAL CONTROL BETWEEN SELECT
+%token DATABASE DISTINCT BETWEEN CONTROL NATURAL COLUMN
-%token HAVING OFFSET CREATE IMPORT RENAME DELETE INSERT
+%token CREATE DELETE HAVING IMPORT INSERT ISNULL OFFSET
-%token UPDATE UNLOAD COLUMN ISNULL WHERE GROUP ORDER LIMIT
+%token RENAME SELECT UNLOAD UPDATE ALTER CROSS GROUP INDEX
-%token INNER OUTER RIGHT CROSS USING TABLE INDEX ALTER FROM
+%token INNER LIMIT ORDER OUTER RIGHT TABLE UNION USING WHERE
-%token DESC JOIN LEFT FILE DROP LOAD INTO NULL LIKE TOP ASC
+%token DESC DROP FILE FROM INTO JOIN LEFT LIKE LOAD NULL ALL
-%token CSV TBL NOT AND BY ON AS OR IN IS
+%token AND ASC CSV NOT TBL TOP AS BY IN IS ON OR
 
 
 /*********************************
@@ -118,7 +118,7 @@ typedef void* yyscan_t;
  *********************************/
 %type <stmt_list>	statement_list
 %type <statement> 	statement
-%type <select_stmt> select_statement
+%type <select_stmt> select_statement select_ref select_with_paren select_no_paren select_clause
 %type <import_stmt> import_statement
 %type <create_stmt> create_statement
 %type <sval> 		table_name opt_alias alias file_path
@@ -126,12 +126,12 @@ typedef void* yyscan_t;
 %type <table>		join_clause join_table
 %type <expr> 		expr scalar_expr unary_expr binary_expr function_expr star_expr expr_alias
 %type <expr> 		column_name literal int_literal num_literal string_literal
-%type <expr> 		comp_expr where_clause join_condition
+%type <expr> 		comp_expr opt_where join_condition
-%type <expr_list> 	expr_list group_clause select_list
+%type <expr_list> 	expr_list opt_group select_list
 %type <table_list> 	table_ref_commalist
-%type <order>		order_by_clause
+%type <order>		opt_order
-%type <limit>		limit_clause
+%type <limit>		opt_limit
-%type <order_type>	order_type
+%type <order_type>	opt_order_type
 %type <uval>		import_file_type
 
 
@@ -226,18 +226,42 @@ create_statement:
  ******************************/
 
 select_statement:
-		SELECT select_list from_clause where_clause group_clause order_by_clause limit_clause
+		select_with_paren
-		{
+	|	select_no_paren
-			SelectStatement* s = new SelectStatement();
+	;
-			s->select_list = $2;
+
-			s->from_table = $3;
+select_with_paren:
-			s->where_clause = $4;
+		'(' select_no_paren ')' { $$ = $2; }
-			s->group_by = $5;
+	|	'(' select_with_paren ')' { $$ = $2; }
-			s->order = $6;
+	;
-			s->limit = $7;
+
-			$$ = s;
+select_no_paren:
+		select_clause opt_order opt_limit {
+			$$ = $1;
+			$$->order = $2;
+			$$->limit = $3;
+		}
+	|	select_ref UNION select_ref opt_order opt_limit {
+			$$ = $1;
+			$$->union_select = $3;
+			$$->order = $4;
+			$$->limit = $5;
+		}
+	;
+
+select_ref:
+		select_clause
+	|	select_with_paren
+	;
+
+select_clause:
+		SELECT select_list from_clause opt_where opt_group {
+			$$ = new SelectStatement();
+			$$->select_list = $2;
+			$$->from_table = $3;
+			$$->where_clause = $4;
+			$$->group_by = $5;
 		}
-	|	'(' select_statement ')' { $$ = $2; }
 	;
 
 
@@ -251,31 +275,33 @@ from_clause:
 	;
 
 
-where_clause:
+opt_where:
 		WHERE expr { $$ = $2; }
 	|	/* empty */ { $$ = NULL; }
 	;
 
 // TODO: having
-group_clause:
+opt_group:
 		GROUP BY expr_list { $$ = $3; }
 	|	/* empty */ { $$ = NULL; }
 	;
 
 
-order_by_clause:
+opt_order:
-		ORDER BY expr order_type { $$ = new OrderDescription($4, $3); }
+		ORDER BY expr opt_order_type { $$ = new OrderDescription($4, $3); }
 	|	/* empty */ { $$ = NULL; }
 	;
 
-order_type:
+opt_order_type:
 		ASC { $$ = kOrderAsc; }
 	|	DESC { $$ = kOrderDesc; }
 	|	/* empty */ { $$ = kOrderAsc; }
 	;
 
-limit_clause:
+
+opt_limit:
 		LIMIT int_literal { $$ = new LimitDescription($2->ival, kNoOffset); delete $2; }
+	|	LIMIT int_literal OFFSET int_literal { $$ = new LimitDescription($2->ival, $4->ival); delete $2; delete $4; }
 	|	/* empty */ { $$ = NULL; }
 	;
 

src/parser/flex_lexer.l

@@ -56,57 +56,59 @@
 [ \t\n]+	/* skip whitespace */;
 
 
-DISTINCT	TOKEN(DISTINCT)
 DATABASE	TOKEN(DATABASE)
-NATURAL		TOKEN(NATURAL)
+DISTINCT	TOKEN(DISTINCT)
-CONTROL		TOKEN(CONTROL)
 BETWEEN		TOKEN(BETWEEN)
-SELECT		TOKEN(SELECT)
+CONTROL		TOKEN(CONTROL)
-HAVING		TOKEN(HAVING)
+NATURAL		TOKEN(NATURAL)
-OFFSET		TOKEN(OFFSET)
-CREATE		TOKEN(CREATE)
-IMPORT		TOKEN(IMPORT)
-RENAME		TOKEN(RENAME)
-DELETE		TOKEN(DELETE)
-INSERT		TOKEN(INSERT)
-UPDATE		TOKEN(UPDATE)
-UNLOAD		TOKEN(UNLOAD)
 COLUMN		TOKEN(COLUMN)
+CREATE		TOKEN(CREATE)
+DELETE		TOKEN(DELETE)
+HAVING		TOKEN(HAVING)
+IMPORT		TOKEN(IMPORT)
+INSERT		TOKEN(INSERT)
 ISNULL		TOKEN(ISNULL)
-WHERE		TOKEN(WHERE)
+OFFSET		TOKEN(OFFSET)
+RENAME		TOKEN(RENAME)
+SELECT		TOKEN(SELECT)
+UNLOAD		TOKEN(UNLOAD)
+UPDATE		TOKEN(UPDATE)
+ALTER		TOKEN(ALTER)
+CROSS		TOKEN(CROSS)
 GROUP		TOKEN(GROUP)
-ORDER		TOKEN(ORDER)
+INDEX		TOKEN(INDEX)
-LIMIT		TOKEN(LIMIT)
 INNER		TOKEN(INNER)
+LIMIT		TOKEN(LIMIT)
+ORDER		TOKEN(ORDER)
 OUTER		TOKEN(OUTER)
 RIGHT		TOKEN(RIGHT)
-CROSS		TOKEN(CROSS)
-USING		TOKEN(USING)
 TABLE		TOKEN(TABLE)
-INDEX		TOKEN(INDEX)
+UNION		TOKEN(UNION)
-ALTER		TOKEN(ALTER)
+USING		TOKEN(USING)
-FROM		TOKEN(FROM)
+WHERE		TOKEN(WHERE)
 DESC		TOKEN(DESC)
+DROP		TOKEN(DROP)
+FILE		TOKEN(FILE)
+FROM		TOKEN(FROM)
+INTO		TOKEN(INTO)
 JOIN		TOKEN(JOIN)
 LEFT		TOKEN(LEFT)
-FILE		TOKEN(FILE)
-DROP		TOKEN(DROP)
-LOAD		TOKEN(LOAD)
-INTO		TOKEN(INTO)
-NULL		TOKEN(NULL)
 LIKE		TOKEN(LIKE)
-TOP			TOKEN(TOP)
+LOAD		TOKEN(LOAD)
+NULL		TOKEN(NULL)
+ALL			TOKEN(ALL)
+AND			TOKEN(AND)
 ASC			TOKEN(ASC)
 CSV			TOKEN(CSV)
-TBL			TOKEN(TBL)
 NOT			TOKEN(NOT)
-AND			TOKEN(AND)
+TBL			TOKEN(TBL)
-BY			TOKEN(BY)
+TOP			TOKEN(TOP)
-ON			TOKEN(ON)
 AS			TOKEN(AS)
-OR			TOKEN(OR)
+BY			TOKEN(BY)
 IN			TOKEN(IN)
 IS			TOKEN(IS)
+ON			TOKEN(ON)
+OR			TOKEN(OR)
 
 
 "<>" 		TOKEN(NOTEQUALS)

src/parser/keywordlist_generator.py

@@ -5,7 +5,8 @@ import math
 with open("sql_keywords.txt", 'r') as fh:
 	keywords = [line.strip() for line in fh.readlines() if not line.strip().startswith("//") and len(line.strip()) > 0]
 
-	keywords = sorted(keywords, key=lambda x: len(x), reverse=True)
+	keywords = sorted(keywords) # Sort by name
+	keywords = sorted(keywords, key=lambda x: len(x), reverse=True) # Sort by length
 
 	#################
 	# Flex

src/parser/sql_keywords.txt

@@ -14,6 +14,8 @@ DESC
 LIMIT
 DISTINCT
 OFFSET
+UNION
+ALL
 
 // Join clause
 JOIN

test/valid_queries.sql

@@ -5,6 +5,10 @@ SELECT col1 AS myname, col2, 'test' FROM "table", foo AS t WHERE age > 12 AND zi
 SELECT * from "table" JOIN table2 ON a = b WHERE (b OR NOT a) AND a = 12.5
 (SELECT a FROM foo WHERE a > 12 OR b > 3 AND c NOT LIKE 's%' LIMIT 10);
 SELECT t1.a, t1.b, t2.c FROM "table" AS t1 JOIN (SELECT * FROM foo JOIN bar ON foo.id = bar.id) t2 ON t1.a = t2.b WHERE (t1.b OR NOT t1.a) AND t2.c = 12.5
+SELECT * FROM "table" LIMIT 10 OFFSET 10;
+SELECT * FROM t1 UNION SELECT * FROM t2 ORDER BY col1; 
+SELECT * FROM t1 UNION (SELECT * FROM t2) ORDER BY col1; 
+SELECT * FROM t1 UNION (SELECT * FROM t2 UNION SELECT * FROM t3) ORDER BY col1; 
 # CREATE statement
 CREATE TABLE "table" FROM TBL FILE 'students.tbl'
 # Multiple statements