able to parse most of column attributes; need more work to support more datatypes

Author: xwb1989

ast.go

@@ -200,50 +200,44 @@ type DDL struct {
 	NewName []byte
 }
 
+type ColumnAtts []string
+
+func (node ColumnAtts) Format(buf *TrackedBuffer) {
+	prefix := " "
+	for _, v := range node {
+		if v != "" {
+			buf.Myprintf("%s%s", prefix, v)
+		}
+	}
+}
+
 type ColumnDefinition struct {
-	ColName      string
-	ColType      string
-	IsPrimaryKey string
+	ColName    string
+	ColType    string
+	ColumnAtts ColumnAtts
 }
 
 func (node ColumnDefinition) Format(buf *TrackedBuffer) {
-	s := ""
-	s += node.ColName
-	s += " " + node.ColType
-	if node.IsPrimaryKey != "" {
-		s += " " + node.IsPrimaryKey
-	}
-	buf.Myprintf(s)
+	buf.Myprintf("%s %s%v", node.ColName, node.ColType, node.ColumnAtts)
 }
 
 type ColumnDefinitions []ColumnDefinition
 
 func (node ColumnDefinitions) Format(buf *TrackedBuffer) {
-	sep := ",\n"
+	prefix := ""
 	buf.Myprintf("(\n")
 	for i := 0; i < len(node); i++ {
-		if i == len(node)-1 {
-			sep = "\n"
-		}
-		buf.Myprintf("\t%v%s", node[i], sep)
+		buf.Myprintf("%s\t%v", prefix, node[i])
+		prefix = ",\n"
 	}
-	buf.Myprintf(")")
+	buf.Myprintf("\n)")
 }
 
 type CreateTable struct {
 	Name              []byte
 	ColumnDefinitions ColumnDefinitions
 }
 
-func (node *CreateTable) FindPrimaryKey() string {
-	for _, col := range node.ColumnDefinitions {
-		if col.IsPrimaryKey != "" {
-			return col.ColName
-		}
-	}
-	return ""
-}
-
 func (node *CreateTable) Format(buf *TrackedBuffer) {
 	buf.Myprintf("create table %s %v", node.Name, node.ColumnDefinitions)
 }
@@ -254,10 +248,6 @@ const (
 	AST_VIEW  = "view"
 )
 
-const (
-	AST_PRIMARY_KEY = "primary key"
-)
-
 const (
 	AST_CREATE = "create"
 	AST_ALTER  = "alter"
@@ -1040,4 +1030,12 @@ const (
 	AST_CHAR    = "char"
 	AST_VARCHAR = "varchar"
 	AST_TEXT    = "text"
+
+	AST_PRIMARY_KEY = "primary key"
+
+	AST_UNIQUE_KEY     = "unique key"
+	AST_AUTO_INCREMENT = "auto_increment"
+	AST_NOT_NULL       = "not null"
+	AST_DEFAULT        = "default"
+	AST_KEY            = "key"
 )

parse_test.go

@@ -24,7 +24,7 @@ func TestParse(t *testing.T) {
 	}
 }
 
-func TestCreatTable(t *testing.T) {
+func TestCreatTable1(t *testing.T) {
 	sql := `create table t1 (
 	ID int primary key,
 	LastName varchar(255),
@@ -39,22 +39,34 @@ func TestCreatTable(t *testing.T) {
 	assert.Equal(t, sql, s)
 }
 
-func TestPrimaryKey(t *testing.T) {
+func TestCreatTable2(t *testing.T) {
 	sql := `create table t1 (
+	ID int primary key not null auto_increment,
 	LastName varchar(255),
-	FirstName varchar(255),
-	ID int primary key
+	FirstName varchar(255)
 )`
 	tree, err := Parse(sql)
 	if err != nil {
 		t.Fatal(err)
 	}
-	create, ok := tree.(*CreateTable)
-	if !ok {
-		t.Fatal("not CreateTable")
+	s := String(tree)
+
+	assert.Equal(t, sql, s)
+}
+
+func TestCreatTable3(t *testing.T) {
+	sql := `create table t1 (
+	ID int unique key not null auto_increment,
+	LastName varchar(255),
+	FirstName varchar(255)
+)`
+	tree, err := Parse(sql)
+	if err != nil {
+		t.Fatal(err)
 	}
-	primary_key := create.FindPrimaryKey()
-	assert.Equal(t, "ID", primary_key)
+	s := String(tree)
+
+	assert.Equal(t, sql, s)
 }
 
 func BenchmarkParse1(b *testing.B) {

sql.y

@@ -63,18 +63,24 @@ var (
   updateExprs UpdateExprs
   updateExpr  *UpdateExpr
 
+/*
+for CreateTable
+*/
   createTableStmt CreateTable
   columnDefinition ColumnDefinition
   columnDefinitions []ColumnDefinition
+  columnAtts ColumnAtts
 }
 
 %token LEX_ERROR
-%token <empty> SELECT INSERT UPDATE DELETE FROM WHERE GROUP HAVING ORDER BY LIMIT FOR PRIMARY
+%token <empty> SELECT INSERT UPDATE DELETE FROM WHERE GROUP HAVING ORDER BY LIMIT FOR
 %token <empty> ALL DISTINCT AS EXISTS IN IS LIKE BETWEEN NULL ASC DESC VALUES INTO DUPLICATE KEY DEFAULT SET LOCK
 %token <bytes> ID STRING NUMBER VALUE_ARG LIST_ARG COMMENT
 %token <empty> LE GE NE NULL_SAFE_EQUAL
 %token <empty> '(' '=' '<' '>' '~'
 
+%token <empty> PRIMARY
+%token <empty> UNIQUE
 %left <empty> UNION MINUS EXCEPT INTERSECT
 %left <empty> ','
 %left <empty> JOIN STRAIGHT_JOIN LEFT RIGHT INNER OUTER CROSS NATURAL USE FORCE
@@ -85,14 +91,14 @@ var (
 %left <empty> '&' '|' '^'
 %left <empty> '+' '-'
 %left <empty> '*' '/' '%'
-%nonassoc <empty> '.'
-%left <empty> UNARY
+%nonassoc <empty> '.' 
+%left <empty> UNARY 
 %right <empty> CASE WHEN THEN ELSE
 %left <empty> END
 
 // DDL Tokens
 %token <empty> CREATE ALTER DROP RENAME ANALYZE
-%token <empty> TABLE INDEX VIEW TO IGNORE IF UNIQUE USING
+%token <empty> TABLE INDEX VIEW TO IGNORE IF USING
 %token <empty> SHOW DESCRIBE EXPLAIN
 
 %start any_command
@@ -151,17 +157,21 @@ var (
 /*
 Below are modification to extract primary key
 */
-%type <str> data_type
-%type <columnDefinition> column_definition
-%type <columnDefinitions> column_definition_list
-%type <statement> create_table_statement
-%type <str> primary_key_opt
 /*
-Datatypes
+keywords
 */
 %token <empty> BIT TINYINT SMALLINT MEDIUMINT INT INTEGER BIGINT REAL DOUBLE FLOAT UNSIGNED ZEROFILL DECIMAL NUMERIC
 %token <empty> TEXT CHAR VARCHAR
-%type <str> length_opt char_type numeric_type unsigned_opt zero_fill_opt
+
+%token <empty> NULLX AUTO_INCREMENT BOOL APPROXNUM INTNUM
+
+%type <str> data_type
+%type <columnDefinition> column_definition
+%type <columnDefinitions> column_definition_list
+%type <statement> create_table_statement
+%type <str> length_opt char_type numeric_type unsigned_opt zero_fill_opt key_att
+%type <columnAtts> column_atts
+
 
 
 %%
@@ -328,19 +338,58 @@ unsigned_opt:
   {
     $$ = AST_UNSIGNED
   }
-primary_key_opt: 
+
+column_atts:
   {
-    $$ = ""
+    $$ = ColumnAtts{}
   }
-| PRIMARY KEY 
+| column_atts NOT NULL
   {
+    $$ = append($$, AST_NOT_NULL)
+  }
+
+| column_atts NULL
+| column_atts DEFAULT STRING
+  {
+    node := StrVal($3)
+    $$ = append($$, "default " + String(node))
+  }
+| column_atts DEFAULT NUMBER
+  {
+    node := NumVal($3)
+    $$ = append($$, "default " + String(node))
+  }
+| column_atts AUTO_INCREMENT
+  {
+    $$ = append($$, AST_AUTO_INCREMENT)
+  }
+| column_atts key_att
+{
+    $$ = append($$, $2)
+}
+
+key_att:
+  primary_key
+  { 
     $$ = AST_PRIMARY_KEY
   }
+| unique_key
+  {
+    $$ = AST_UNIQUE_KEY
+  }
+
+primary_key:
+  PRIMARY KEY
+| KEY 
+
+unique_key:
+  UNIQUE
+| UNIQUE KEY
 
 column_definition:
-  ID data_type primary_key_opt
+  ID data_type column_atts
   {
-    $$ = ColumnDefinition{ColName: string($1), ColType: $2, IsPrimaryKey: $3  }
+    $$ = ColumnDefinition{ColName: string($1), ColType: $2, ColumnAtts: $3  }
   }
 
 column_definition_list:

token.go

@@ -126,9 +126,10 @@ var keywords = map[string]int{
 	"text":    TEXT,
 
 	//other keywords
-	"unsigned": UNSIGNED,
-	"zerofill": ZEROFILL,
-	"primary":  PRIMARY,
+	"unsigned":       UNSIGNED,
+	"zerofill":       ZEROFILL,
+	"primary":        PRIMARY,
+	"auto_increment": AUTO_INCREMENT,
 }
 
 // Lex returns the next token form the Tokenizer.