Implement phase 02 syntax analysis

02_syntax/lexer.py

@@ -0,0 +1,128 @@
+#!/bin/env python3
+
+import argparse
+import datetime
+import ply.lex as lex
+
+reserved = {
+    'var':          'VAR',
+    'is':           'IS',
+    'unless':       'UNLESS',
+    'otherwise':    'OTHERWISE',
+    'until':        'UNTIL',
+    'do':           'DO',
+    'done':         'DONE',
+    'procedure':    'PROCEDURE',
+    'function':     'FUNCTION',
+    'return':       'RETURN',
+    'print':        'PRINT',
+    'end':          'END',
+}
+
+tokens = [
+    'LPAREN',
+    'RPAREN',
+    'LSQUARE',
+    'RSQUARE',
+    'LCURLY',
+    'RCURLY',
+    'APOSTROPHE',
+    'AMPERSAND',
+    'COMMA',
+    'DOT',
+    'EQ',
+    'LT',
+    'PLUS',
+    'MINUS',
+    'MULT',
+    'DIV',
+    'STRING',
+    'DATE_LITERAL',
+    'INT_LITERAL',
+    'IDENT',
+    'FUNC_IDENT',
+    'PROC_IDENT',
+] + list(reserved.values())
+
+def t_whitespace(t):
+    r'[ \t\n]+'
+    t.lexer.lineno += t.value.count('\n')
+
+def t_comment(t):
+    r'\(%(.|\n)*%\)'
+    t.lexer.lineno += t.value.count('\n')
+
+t_LPAREN        = r'\('
+t_RPAREN        = r'\)'
+t_LSQUARE       = r'\['
+t_RSQUARE       = r'\]'
+t_LCURLY        = r'\{'
+t_RCURLY        = r'\}'
+t_APOSTROPHE    = r'\''
+t_AMPERSAND     = r'&'
+t_COMMA         = r','
+t_DOT           = r'\.'
+t_EQ            = r'='
+t_LT            = r'<'
+t_PLUS          = r'\+'
+t_MINUS         = r'-'
+t_MULT          = r'\*'
+t_DIV           = r'/'
+
+def t_STRING(t):
+    r'".*?"'
+    t.value = t.value[1:-1]
+    return t
+
+def t_DATE_LITERAL(t):
+    r'\d{4}-\d{2}-\d{2}'
+    t.value = datetime.date.fromisoformat(t.value)
+    return t
+
+def t_INT_LITERAL(t):
+    r'-?\d{1,3}(\'\d{3})*'
+    t.value = int(t.value.replace('\'', ''))
+    return t
+
+def t_IDENT(t):
+    r'[a-z][a-zA-Z0-9_]+'
+    t.type = reserved.get(t.value, 'IDENT')
+    return t
+
+def t_FUNC_IDENT(t):
+    r'[A-Z][a-z0-9_]+'
+    return t
+
+def t_PROC_IDENT(t):
+    r'[A-Z]{2}[A-Z0-9_]*'
+    return t
+
+def t_error(t):
+    print(f'Illegal character \'{t.value[0]}\' at line {t.lexer.lineno}')
+    raise SystemExit
+
+lexer = lex.lex()
+
+def tokenize_file(file_path: str):
+    with open(file_path, 'r', encoding='utf-8') as file:
+        lexer.input(file.read())
+
+    tok = lexer.token()
+    while tok:
+        print(tok)
+        tok = lexer.token()
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    group = parser.add_mutually_exclusive_group(required=True)
+    group.add_argument('--who', action='store_true', help='print out student IDs and NAMEs of authors')
+    group.add_argument('-f', '--file', help='filename to process')
+
+    args = parser.parse_args()
+    
+    if args.who:
+        print('Author')
+        print('  Student ID: 150189237')
+        print('  Name:       Oskari Alaranta')
+    else:
+        tokenize_file(args.file)

02_syntax/main.py

@@ -0,0 +1,186 @@
+#!/bin/env python3
+
+import argparse
+import ply.lex as lex
+import ply.yacc as yacc
+import lexer
+
+# Simple debuggin function to call from syntax rules
+symbolnum = 0
+def debug_print(p):
+    global symbolnum
+    symbolnum += 1
+    p[0] = symbolnum
+    msg = ''
+    for i, s in enumerate(p.slice):
+        if s is not None:
+            if type(s) is lex.LexToken:
+                msg += str(s.type) + '<' + str(s.value) + '> '
+            else:
+                msg += str(s) + '(' + str(p[i]) + ') '
+        else:
+            msg += '?? '
+        if i == 0:
+            msg += ':: '
+    print(msg)
+
+tokens = lexer.tokens
+
+def p_program(p):
+    '''program : statement_list
+               | definitions program'''
+    debug_print(p)
+
+def p_statement_list(p):
+    '''statement_list : statement
+                      | statement COMMA statement_list'''
+    debug_print(p)
+
+def p_definitions(p):
+    '''definitions : function_definition
+                   | procedure_definition
+                   | variable_definition'''
+    debug_print(p)
+
+def p_variable_definition(p):
+    '''variable_definition : VAR IDENT EQ expression'''
+    debug_print(p)
+
+def p_empty(p):
+    '''empty :'''
+    debug_print(p)
+
+def p_opt_formals(p):
+    '''opt_formals : empty
+                   | formals'''
+    debug_print(p)
+
+def p_opt_variable_defitions(p):
+    '''opt_variable_definitions : empty
+                                | variable_definition opt_variable_definitions'''
+    debug_print(p)
+
+def p_function_definition(p):
+    '''function_definition : FUNCTION FUNC_IDENT LCURLY opt_formals RCURLY RETURN IDENT opt_variable_definitions IS rvalue END FUNCTION'''
+    debug_print(p)
+
+def p_procedure_definition(p):
+    '''procedure_definition : PROCEDURE PROC_IDENT LCURLY opt_formals RCURLY opt_variable_definitions IS statement_list END PROCEDURE
+                            | PROCEDURE PROC_IDENT LCURLY opt_formals RCURLY RETURN IDENT opt_variable_definitions IS statement_list END PROCEDURE'''
+    debug_print(p)
+
+def p_formals(p):
+    '''formals : formal_arg
+               | formal_arg COMMA formals'''
+    debug_print(p)
+
+def p_formal_arg(p):
+    '''formal_arg : IDENT LSQUARE IDENT RSQUARE'''
+    debug_print(p)
+
+def p_procedure_call(p):
+    '''procedure_call : PROC_IDENT LPAREN RPAREN
+                      | PROC_IDENT LPAREN arguments RPAREN'''
+    debug_print(p)
+
+def p_arguments(p):
+    '''arguments : expression
+                 | expression COMMA arguments'''
+    debug_print(p)
+
+def p_assignment(p):
+    '''assignment : lvalue EQ rvalue'''
+    debug_print(p)
+
+def p_lvalue(p):
+    '''lvalue : IDENT
+              | IDENT DOT IDENT'''
+    debug_print(p)
+
+def p_rvalue(p):
+    '''rvalue : expression
+              | unless_expression'''
+    debug_print(p)
+
+def p_print_statement(p):
+    '''print_statement : PRINT print_item
+                       | print_statement AMPERSAND print_item'''
+    debug_print(p)
+
+def p_print_item(p):
+    '''print_item : STRING
+                  | expression'''
+    debug_print(p)
+
+def p_statement(p):
+    '''statement : procedure_call
+                 | assignment
+                 | print_statement
+                 | DO statement_list UNTIL expression
+                 | DO statement_list UNLESS expression DONE
+                 | DO statement_list UNLESS expression OTHERWISE statement_list DONE
+                 | RETURN expression'''
+    debug_print(p)
+
+def p_expression(p):
+    '''expression : simple_expr
+                  | expression EQ simple_expr
+                  | expression LT simple_expr'''
+    debug_print(p)
+
+def p_simple_expr(p):
+    '''simple_expr : term
+                   | simple_expr PLUS term
+                   | simple_expr MINUS term
+       term : factor
+            | term MULT factor
+            | term DIV factor
+       factor : atom
+              | MINUS atom
+              | PLUS atom
+       atom : IDENT
+            | IDENT APOSTROPHE IDENT
+            | INT_LITERAL
+            | DATE_LITERAL
+            | function_call
+            | procedure_call
+            | LPAREN expression RPAREN'''
+    debug_print(p)
+
+def p_function_call(p):
+    '''function_call : FUNC_IDENT LPAREN RPAREN
+                     | FUNC_IDENT LPAREN arguments RPAREN'''
+    debug_print(p)
+
+def p_unless_expression(p):
+    '''unless_expression : DO expression UNLESS expression OTHERWISE expression DONE'''
+    debug_print(p)
+
+def p_error(p):
+    if p is not None:
+        print(f"{{{p.lineno}}}:Syntax Error (token:'{p.value}')")
+    else:
+        print('Syntax Error at the end of file')
+    raise SystemExit
+
+def syntax_check_file(file_path: str, debug: bool):
+    parser = yacc.yacc()
+    with open(file_path, 'r', encoding='utf-8') as file:
+        parser.parse(file.read(), lexer=lexer.lexer, debug=debug)
+    print('syntax OK')
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('-d', '--debug', action='store_true', help='debug?')
+    group = parser.add_mutually_exclusive_group(required=True)
+    group.add_argument('--who', action='store_true', help='print out student IDs and NAMEs of authors')
+    group.add_argument('-f', '--file', help='filename to process')
+
+    args = parser.parse_args()
+
+    if args.who:
+        print('Author')
+        print('  Student ID: 150189237')
+        print('  Name:       Oskari Alaranta')
+    else:
+        syntax_check_file(args.file, args.debug)