Various other WIP attempts

compiler/engine.py

@@ -0,0 +1,10 @@
+from tokenizer import JackTokenizer
+
+"""
+New Compilation Engine
+"""
+class Engine:
+  def __init__(self, input_file):
+    self.i = 0
+    self.jt = JackTokenizer(input_file, False)
+    self.file = open(input_file + ".xml", 'w')

compiler/grammar.py

@@ -1,5 +1,67 @@
 from keywords import Keyword
 
+class Element:
+  def __init__(name, grammar, terminal = False):
+    self.name = name
+    self.grammar = grammar
+    self.terminal = terminal
+
+CLASS = Element('class', [
+  # class className {
+  Keyword.CLASS,
+  Keyword.IDENTIFIER,
+  Keyword.BRACE_OPEN,
+  # class Variable Declarations (one or more) = list
+  CLASSVARDEC,
+  # subroutine declarations (one or more) = list
+  SUBROUTINEDEC,
+  # }
+  Keyword.BRACE_CLOSE
+])
+
+CLASSVARDEC = Element('classVarDec', [
+  # static|field type (, name)* ;
+  Keyword.STATIC | Keyword.FIELD,
+  TYPE,
+  [Keyword.COMMA, Keyword.IDENTIFIER],
+  Keyword.SEMICOLON
+])
+
+# Parameter List =
+#  (
+#    (type varName) (, type varName)*
+#  )?
+# we use tuples for zero OR one of a sequence
+PARAMETER_LIST = Element('parameterList', (
+  TYPE,
+  Keyword.IDENTIFIER,
+  [Keyword.COMMA, TYPE, Keyword.IDENTIFIER]
+))
+
+TYPE = Element('type', Keyword.INT | Keyword.CHAR | Keyword.BOOLEAN | Keyword.IDENTIFIER, True)
+
+SUBROUTINE_BODY = Element('subroutineBody', [
+  # One or more variable declarations
+  # `var type varName (, varName)* ;`
+    [VARDEC],
+    STATEMENTS
+])
+
+SUBROUTINEDEC = Element('subroutineDec', [
+  # (constructor | function | method) (void | type) subRoutineName '(' parameterList ')'
+  # subroutineBody
+  Keyword.CONSTRUCTOR | Keyword.FUNCTION | Keyword.METHOD,
+  Keyword.VOID | TYPE,
+  Keyword.IDENTIFIER,
+  Keyword.PARAN_OPEN,
+  PARAMETER_LIST,
+  Keyword.PARAN_CLOSE,
+  # Subroutine Body {
+  Keyword.BRACE_OPEN,
+  SUBROUTINE_BODY,
+  Keyword.BRACE_CLOSE,
+])
+
 """
 The grammar is defined by the following constructs:
 
@@ -11,25 +73,26 @@ Inside this list, each element can be any of the following:
 - a bitwise mask of the Keyword enum to denote multiple possibilities
 - Another list, to denote zero-or-more of a inner-sequence
 - A tuple, to denote zero-or-one of a inner-sequence
+- A lambda denotes a non-terminal part of the grammar
 
 This is basically an attempt to translate Figure 10.5 from the book into
 a Python structure.
 
 """
 
-TYPE = Keyword.INT | Keyword.CHAR | Keyword.BOOLEAN | Keyword.IDENTIFIER
+UNARY_OP = Element('unaryOp', Keyword.NOT | Keyword.MINUS, True)
 
-UNARY_OP = Keyword.NOT | Keyword.MINUS
+CONSTANT = Element('KeywordConstant', Keyword.TRUE | Keyword.FALSE, Keyword.NULL, Keyword.THIS, True)
 
-TERM = Keyword.INTEGERCONSTANT | Keyword.STRINGCONSTANT | Keyword.TRUE | Keyword.FALSE | Keyword.IDENTIFIER
+TERM = Element('term', Keyword.INTEGERCONSTANT | Keyword.STRINGCONSTANT | Keyword.TRUE | Keyword.FALSE | Keyword.IDENTIFIER)
 
-OP = Keyword.PLUS | Keyword.MINUS | Keyword.MUL | Keyword.DIV | Keyword.AND | Keyword.OR | Keyword.GT | Keyword.LT | Keyword.EQ
+OP = Element('op', Keyword.PLUS | Keyword.MINUS | Keyword.MUL | Keyword.DIV | Keyword.AND | Keyword.OR | Keyword.GT | Keyword.LT | Keyword.EQ, True)
 
-EXPRESSION = [TERM, [OP, TERM]]
+EXPRESSION = Element('expression', [TERM, [OP, TERM]])
 
-EXPRESSIONLIST = (EXPRESSION, [Keyword.COMMA, EXPRESSION])
+EXPRESSIONLIST = Element('expressionList', (EXPRESSION, [Keyword.COMMA, EXPRESSION]))
 
-SUBROUTINE_CALL = {
+SUBROUTINE_CALL = Element('subroutineCall', {
   (Keyword.IDENTIFIER, Keyword.PARAN_OPEN): [
     EXPRESSIONLIST,
     Keyword.PARAN_CLOSE,
@@ -40,81 +103,29 @@ SUBROUTINE_CALL = {
     EXPRESSIONLIST,
     Keyword.PARAN_CLOSE
   ]
-}
+})
 
-STATEMENTS = {
+STATEMENT = Element('statement', {
   (Keyword.LET): [Keyword.IDENTIFIER, (Keyword.SQUARE_OPEN, EXPRESSION, Keyword.SQUARE_CLOSE)],
   (Keyword.IF): [
     Keyword.PARAN_OPEN,
     EXPRESSION,
     Keyword.PARAN_CLOSE,
     Keyword.BRACE_OPEN,
-    lambda:STATEMENTS,
+    lambda: STATEMENTS,
     Keyword.BRACE_CLOSE,
-    ( Keyword.ELSE, Keyword.BRACE_OPEN, lambda:STATEMENTS, Keyword.BRACE_CLOSE)
+    ( Keyword.ELSE, Keyword.BRACE_OPEN, lambda:STATEMENT, Keyword.BRACE_CLOSE)
   ],
   (Keyword.WHILE): [
     Keyword.PARAN_OPEN,
     EXPRESSION,
     Keyword.PARAN_CLOSE,
     Keyword.BRACE_OPEN,
-    lambda:STATEMENTS,
+    lambda: STATEMENTS,
     Keyword.BRACE_CLOSE,
   ],
   (Keyword.DO): SUBROUTINE_CALL,
   (Keyword.RETURN): [(EXPRESSION), Keyword.SEMICOLON]
-}
+})
 
-SUBROUTINEDEC = [
+STATEMENTS = Element('statements', [STATEMENT])
-  # (constructor | function | method) (void | type) subRoutineName '(' parameterList ')'
-  # subroutineBody
-  Keyword.CONSTRUCTOR | Keyword.FUNCTION | Keyword.METHOD,
-  Keyword.VOID | TYPE,
-  Keyword.IDENTIFIER,
-  Keyword.PARAN_OPEN,
-  # Parameter List =
-  #  (
-  #    (type varName) (, type varName)*
-  #  )?
-  # we use tuples for zero OR one of a sequence
-  (
-    TYPE,
-    Keyword.IDENTIFIER,
-    [Keyword.COMMA, TYPE, Keyword.IDENTIFIER]
-  ),
-  Keyword.PARAN_CLOSE,
-  # Subroutine Body {
-  Keyword.BRACE_OPEN,
-  # One or more variable declarations
-  # `var type varName (, varName)* ;`
-  [
-    Keyword.VAR,
-    TYPE,
-    Keyword.IDENTIFIER,
-    [Keyword.COMMA, Keyword.IDENTIFIER],
-    Keyword.SEMICOLON
-  ],
-  STATEMENTS,
-  Keyword.BRACE_CLOSE,
-]
-
-CLASSVARDEC = [
-  # static|field type (, name)* ;
-  Keyword.STATIC | Keyword.FIELD,
-  TYPE,
-  [Keyword.COMMA, Keyword.IDENTIFIER],
-  Keyword.SEMICOLON
-]
-
-GRAMMAR = [
-  # class className {
-  Keyword.CLASS,
-  Keyword.IDENTIFIER,
-  Keyword.BRACE_OPEN,
-  # class Variable Declarations (one or more) = list
-  CLASSVARDEC,
-  # subroutine declarations (one or more) = list
-  SUBROUTINEDEC,
-  # }
-  Keyword.BRACE_CLOSE
-]

compiler/grammar.yml

@@ -0,0 +1,50 @@
+keyword:
+  - 'class'
+  - 'method'
+  - 'function'
+  - 'constructor'
+  - 'int'
+  - 'boolean'
+  - 'char'
+  - 'void'
+  - 'var'
+  - 'static'
+  - 'field'
+  - 'let'
+  - 'do'
+  - 'if'
+  - 'else'
+  - 'while'
+  - 'return'
+  - 'true'
+  - 'false'
+  - 'null'
+  - 'this'
+symbol:
+  - '{'
+  - '}'
+  - '('
+  - ')'
+  - '['
+  - ']'
+  - '.'
+  - ';'
+  - '+'
+  - '-'
+  - '*'
+  - '/'
+  - '&'
+  - '|'
+  - '<'
+  - '>'
+  - '='
+  - '~'
+  - ','
+# Predefined
+integerConstant: null
+StringConstant: null
+identifier: null
+className:
+class:
+  - class
+  - *className