DBMS Code Overview

The code files that we have given you for the DBMS programming assignment can be grouped into several collections of related files. Some files appear in more than one collection. Below, we discuss each collection in turn.

More detail can be found in the API documentation for the framework.

1. Files that define and implement the parser

   • sym.java
   • Parser.java
   • Lexer.java

   You do not need to understand the code in these files, and you will not be modifying them.

2. Files related to WHERE clause evaluation

   • ConditionalExpression.java
   • AndExpression.java
   • OrExpression.java
   • NotExpression.java
   • TrueExpression.java
   • Comparison.java
   • CompareTerm.java

   You will not be modifying these files, and you don’t need to understand all of the details of the code that they contain. However, you may want to browse through the files to get some sense of how WHERE clause evaluation is carried out.

   In a parsed SQL command, a WHERE clause is essentially a tree data structure. For example, the WHERE clause

   WHERE age < 30 AND (zip = '02138' OR zip = '02140')
   

   would form a tree that looks like this:

   

   A WHERE clause is represented by a reference to the object that serves as the root of the WHERE clause tree. This object is an instance of one of the subclasses of the abstract class ConditionalExpression. To evaluate the clause, we simply invoke the root object’s isTrue() method, which returns true if the expression is true for the current values of the columns involved, and false otherwise. The leaf nodes of the tree are terms that are being compared. These terms are either columns or literal values. The CompareTerm class is used for literal values. The Column class is used for columns; it is a subclass of CompareTerm.

   We have given you almost all of the code needed for evaluating WHERE clauses. The one exception is the code you will write to retrieve the value of a column from the tuple that is currently being considered for selection. However, this value-retrieval code will be implemented in other files.

3. Files for classes of objects generated by the parser

   • SQLStatement.java (and its subclasses; more on them later)
   • Table.java
   • Column.java
   • ColumnOptions.java
   • Limit.java

   The SQL parser takes the SQL command entered by the user (a string of characters) and breaks it up into a set of tokens. Each token represents one element of the command — a table name, a column name, a column value, etc.

   In our framework, these tokens are represented by objects, and the tokens for a given command are managed by an object that represents the command as a whole. For example, if the user enters a SELECT command, the parser will generate an object of type SelectStatement, and we will use that object to access the objects that represent the individual tokens of that command (e.g., the names of the table or tables on which the SELECT command is operating).

   The objects that represent the command as a whole are instances of one of the subclasses of the abstract class SQLStatement. These subclasses will be described in the next section. The classes used for the tokens include the classes mentioned above (Table, Column, ColumnOptions, Limit), the classes for conditional expressions mentioned in the previous section, and several built-in Java classes (String, Integer, Double). All SQLStatement subclasses have methods that allow you to get information about the tokens of a particular type (e.g., getTable(), getColumn(), numTables(), numColumns(), etc.). These accessor methods are defined in the SQLStatement class itself, and the subclasses inherit them.

4. Files for classes of objects that represent SQL commands

   • BeginStatement.java (for BEGIN TRANSACTION)
   • CommitStatement.java (for COMMIT TRANSACTION)
   • CreateStatement.java
   • DeleteStatement.java
   • DropStatement.java
   • InsertStatement.java
   • RollbackStatement.java (for ROLLBACK TRANSACTION)
   • SelectStatement.java
   • UpdateStatement.java

   These files contain the class definitions for the subclasses of the abstract class SQLStatement that was mentioned above. Each subclass represents one type of SQL command.

   The SQLStatement superclass includes an abstract method called execute(). Each subclass has its own version of that method, that performs the operations needed to carry out that command. We have given you the execute() method for CREATE TABLE and DROP TABLE commands. You will complete the execute() methods of some of the other types of SQL commands.

5. Files for classes of objects that are used to access and manipulate the contents of a table

   • Table.java
   • Column.java
   • InsertRow.java
   • TableIterator.java
   • RowOutput.java
   • RowInput.java

   The Table and Column classes are used for two different purposes. First, they are used to represent tokens generated by the parser, in which case they primarily serve as identifiers that specify the name of the table or column (see section 3 above). Second, they are used to represent the stored representations of actual tables and columns in the relational database. If the parser generates an instance of one of these classes, it may be necessary to fill in some of the fields of the object before it can used to actually represent a stored table or column. You should not need to add to or modify these two classes, but it is important that you understand the methods that they contain. You should at least read the comments that precede each method, and you may also want to look over the bodies of the methods to get some idea of how they work.

   InsertRow objects are used to marshall the values in a row that is being inserted, turning them into a single key/data pair. You will implement the key method in this class. In doing so, you will use objects of type RowOutput to prepare the byte arrays for the key/value pair that you ask Berkeley DB to store.

   TableIterator objects are used to iterate over some or all of the tuples in a table. You will implement some of the methods in this class. In doing so, you will use objects of type RowInput to extract values from the byte arrays that form the basis of the key/value pairs that you obtain from Berkeley DB.

   You should not need to add to or modify the RowOutput or RowInput classes, but it is important that you understand the methods that they contain! In the case of RowOutput, you should also review the methods that it inherits from the DataOutputStream class.

6. Files that implement the DBMS infrastructure

   • DBMS.java
   • Catalog.java
   • Table.java

   DBMS.java contains the main() method for the application. It also has methods/functions for initializing and shutting down the DBMS, including code for configuring and opening the Berkeley DB environment used for the underlying BDB databases.

   Catalog.java contains the code implementing the DBMS catalog, which stores the per-table metadata associated with each table in the DBMS.

   Table.java includes support for an in-memory table cache, which keeps track of all tables whose underlying Berkeley DB databases have already been opened. When a table is opened, the corresponding Table object is stored in this cache for possible future use. The table cache is implemented using an instance of Java’s HashTable class.

   You should not need to modify these files, but it is important that you understand the methods that they contain. You should at least read the comments that precede each method, and you may also want to look over the bodies of the methods to get some idea of how they work.

   Note: To allow access to the DBMS and Catalog methods from all other classes, we make all of these methods static, so that the class name can be used to invoke them (e.g., Catalog.putMetadata(tablename);).

Last updated on September 15, 2024.