C++辅导：CSE340 Type Checker辅导留学生R设计、R编程辅导

，，、。

Goal

Your goal is to finish an incomplete parser and write a type checker for a given language. The input to your project will be a program and the output will be either error messages if there is a type mismatch or lists of equivalent types if there is no type mismatch. Your type checker will enforce semantic checks on the input program, and will be described in the following.

Grammar Description

program → decl body decl → type_decl_section var_decl_section type_decl_section → TYPE type_decl_list type_decl_section → ε type_decl_list → type_decl type_decl_list type_decl_list → type_decl type_decl → id_list COLON type_name SEMICOLON type_name → REAL type_name → INT type_name → BOOLEAN type_name → STRING type_name → LONG type_name → ID var_decl_section → VAR var_decl_list var_decl_section → ε var_decl_list → var_decl var_decl_list var_decl_list → var_decl var_decl → id_list COLON type_name SEMICOLON id_list → ID COMMA id_list id_list → ID body → LBRACE stmt_list RBRACE stmt_list → stmt stmt_list stmt_list → stmt stmt → while_stmt stmt → assign_stmt stmt → do_stmt stmt → switch_stmt while_stmt → WHILE condition body assign_stmt → ID EQUAL expr SEMICOLON do_stmt → DO body WHILE condition SEMICOLON switch_stmt → SWITCH ID LBRACE case_list RBRACE case_list → case case_list case_list → case case → CASE NUM COLON body expr → term PLUS expr expr → term MINUS expr expr → term term → factor MULT term term → factor DIV term term → factor factor → LPAREN expr RPAREN factor → NUM factor → REALNUM factor → ID condition → ID condition → primary relop primary primary → ID primary → NUM primary → REALNUM relop → GREATER relop → GTEQ relop → LESS relop → NOTEQUAL relop → LTEQ

The tokens used in the grammar description are:

TYPE = TYPE COLON = : SEMICOLON = ; REAL = REAL INT = INT BOOLEAN = BOOLEAN STRING = STRING LONG = LONG VAR = VAR COMMA = , LBRACE = { RBRACE = } WHILE = WHILE EQUAL = = DO = DO SWITCH = SWITCH CASE = CASE PLUS = + MINUS = ‐ MULT = * DIV = / LPAREN = ( RPAREN = ) GREATER = gt; GTEQ = gt;= LESS = lt; LTEQ = lt;= NOTEQUAL = lt;gt; ID = letter(letter | digit)* NUM = 0 | (digit digit*) REALNUM = NUM \. digit*

Language Semantics

As can be seen from the grammar, in this language types are first declared, then variables are declared, then the body of the program follows.

Types

The language has five builtin types: INT, REAL, BOOLEAN, STRING, and LONG.

Programmers can declare types either explicitly or implicitly.

Explicit types are names that are not builtin types and that have their first appearance in the program as part of the id_list of a type_decl.
Implicit types are not builtin types and not explicit programmerdeclared types. Implicit types have their first appearance as a type_name in a var_decl or a type_decl.

Example

Consider the following program written in our language:

TYPE
  a : INT;
  b : a;
VAR
  x : b;
  y : c;
#123;
  y = x;
#125;

There are three types declared by the programmer in this example, a, b, and c, where a and b are explicit types and c is an implicit type.

Variables

Programmers can declare variables either explicitly or implicitly.

Explicit variables are declared in an id_list of a var_decl.
A variable is declared implicitly if it is not declared explicitly but it appears in the program body.

Example

Consider the following program written in our language:

TYPE
  a : INT;
  b : a;
VAR
  x : b;
  y : c;
#123;
  y = x;
  z = 10;
  w = z * 5;
#125;

This program has four variables declared: x, y, z, and w, with x and y explicitly declared and z and w implicitly declared. Note that the implicitly declared variables z and w also have an implicitly declared type.

Declaration vs. Use

Any appearance of a name (type or variable) in the program is either a declaration or a use.

The following lists all possible declarations of a name:

Any appearance of a name in the (left of COLON) id_list part of a type_decl is a type declaration
Any appearance of a name in the (left of COLON) id_list part of a var_decl is a variable declaration
The first appearance of a name in the entire program, if the name appears as type_name in a type_decl
The first appearance of a name in the entire program, if the name appears as type_name in a var_decl
The first appearance of a name inside the body of a program is a variable declaration

Any other appearance of a name is considered a use of that name.
Note that the above definitions exclude the builtin type names.
Given the following example (the line numbers are not part of the input):

TYPE
  a : INT;
  b : a;
VAR
  x : b;
  y : c;
#123;
  y = x;
  z = 10;
  w = z * 5;
#125;

We can categorize all appearances of names as declaration or use:

Line 2, the appearance of name a is a declaration
Line 3, the appearance of name b is a declaration
Line 3, the appearance of name a is a use
Line 5, the appearance of name x is a declaration
Line 5, the appearance of name b is a use
Line 6, the appearance of name y is a declaration
Line 6, the appearance of name c is a declaration
Line 8, the appearance of name y is a use
Line 8, the appearance of name x is a use
Line 9, the appearance of name z is a declaration
Line 10, the appearance of name w is a declaration
Line 10, the appearance of name z is a use

Type System

Our language uses structural equivalence for checking type equivalence.

Implicit types (in variable declarations or on implicitly declared variables) will be inferred from the usage (in a simplified form of HindleyMilner type inference).

Here are all the type rules/constraints that your type checker will enforce (constraints are labeled from C1 to C5 for reference):

C1: The left hand side of an assignment should have the same type as the right hand side of that assignment
C2: The operands of an operation (PLUS, MINUS, MULT, and DIV) should have the same type (it can be any type, including STRING and BOOLEAN)
C3: The operands of a relational operator (see relop in grammar) should have the same type (it can be any type, including STRING and BOOLEAN )
C4: condition should be of type BOOLEAN
C5: The variable that follows the SWITCH keyword in switch_stmt should be of type INT
The type of an expr is the same as the type of its operands
The result of p1 relop p2 is of type BOOLEAN (assuming that p1 and p2 have the same type)
NUM constants are of type INT
REALNUM constants are of type REAL
If two types cannot be determined to be the same according to the above rules, the two types are different

Incomplete Parser

The parser given on the submission site is incomplete, as it is missing an implementation for while_stmt, condition, do_stmt, switch_stmt, case_list, and case.

As described in the evaluation section, you must implement parsing for while_stmt, condition, and do_stmt, while switch_stmt, case_list, and case are extra credit (though note that to receive full extra credit, you must implement all of the type checks in addition to the parsing cases.

It is strongly recommended that you finish the incomplete parser before implementing the type checking part. You should make sure that your parser generates a syntax error message if the input program does not follow the proper syntax. We recommend that you check your code on the submission website to make sure it passes all the syntax error test cases before moving on to implementing the type checking part.

Output

Your program will check for the following semantic errors and output the correct message when it encounters that error. Note that there will only be at most one error per test case.

Duplication Errors

Errors involving programmerdefined types:
- Programmerdefined type declared more than once:
  - Explicit type redeclared explicitly (error code 1.1)
    An explicitly declared type can be declared again explicitly by appearing as part of an id_list in a type declaration.
  - Implicit type redeclared explicitly (error code 1.2)
    An implicitly declared type can be declared again explicitly by appearing as part of an id_list in a type declaration.
- Programmerdefined type redeclared as variable (error code 1.3)
  If a previously declared type appears again in an id_list of a variable declaration, the type is redeclared as a variable.
- Programmerdefined type used as variable (error code 1.4)
  If a previously declared type appears in the body of the program, the type is used as a variable.
Errors involving variable declarations:
- Variable declared more than once (error code 2.1)
  An explicitly declared variable can be declared again explicitly by appearing as part of an id_list in a variable declaration.
- Variable used as a type (error code 2.2)
  If an explicitly declared variable is used as type_name in a variable declaration, the variable is used as a type.

Duplication Error Output Format

For these errors, you should output one line in the following format:

ERROR CODE lt;codegt; lt;symbol_namegt;

Type Mismatch

If any of the type constraints (listed in the Type System section above) is violated in the input program, then the output of your program should be:

TYPE MISMATCH lt;line_numbergt; lt;constraintgt;

No Semantic Errors

If there are no semantic errors in the program, then your program should output lists of types and variables that are typeequivalent. The symbols should be listed in the order they appear in the program and builtin types should be listed first in the following order: BOOLEAN, INT, LONG, REAL, STRING. Each list must be on a single line of the output and each symbol in the list should be separated by a single space character. Each list must be terminated by a # character.

联系我们