src/soc/decoder/pseudo/parser.py

   1 # Based on GardenSnake - a parser generator demonstration program
   2 # GardenSnake was released into the Public Domain by Andrew Dalke.
   3
   4 # Portions of this work are derived from Python's Grammar definition
   5 # and may be covered under the Python copyright and license
   6 #
   7 #          Andrew Dalke / Dalke Scientific Software, LLC
   8 #             30 August 2006 / Cape Town, South Africa
   9
  10 # Modifications for inclusion in PLY distribution
  11 from pprint import pprint
  12 from ply import lex, yacc
  13 import astor
  14
  15 from soc.decoder.power_decoder import create_pdecode
  16 from soc.decoder.pseudo.lexer import IndentLexer
  17
  18 # I use the Python AST
  19 #from compiler import ast
  20 import ast
  21
  22 # Helper function
  23 def Assign(left, right):
  24     names = []
  25     if isinstance(left, ast.Name):
  26         # Single assignment on left
  27         # XXX when doing IntClass, which will have an "eq" function,
  28         # this is how to access it
  29         #   eq = ast.Attribute(left, "eq")   # get eq fn
  30         #   return ast.Call(eq, [right], []) # now call left.eq(right)
  31         return ast.Assign([ast.Name(left.id, ast.Store())], right)
  32     elif isinstance(left, ast.Tuple):
  33         # List of things - make sure they are Name nodes
  34         names = []
  35         for child in left.getChildren():
  36             if not isinstance(child, ast.Name):
  37                 raise SyntaxError("that assignment not supported")
  38             names.append(child.name)
  39         ass_list = [ast.AssName(name, 'OP_ASSIGN') for name in names]
  40         return ast.Assign([ast.AssTuple(ass_list)], right)
  41     else:
  42         raise SyntaxError("Can't do that yet")
  43
  44
  45 ## I implemented INDENT / DEDENT generation as a post-processing filter
  46
  47 # The original lex token stream contains WS and NEWLINE characters.
  48 # WS will only occur before any other tokens on a line.
  49
  50 # I have three filters.  One tags tokens by adding two attributes.
  51 # "must_indent" is True if the token must be indented from the
  52 # previous code.  The other is "at_line_start" which is True for WS
  53 # and the first non-WS/non-NEWLINE on a line.  It flags the check so
  54 # see if the new line has changed indication level.
  55
  56
  57 ## No using Python's approach because Ply supports precedence
  58
  59 # comparison: expr (comp_op expr)*
  60 # arith_expr: term (('+'|'-') term)*
  61 # term: factor (('*'|'/'|'%'|'//') factor)*
  62 # factor: ('+'|'-'|'~') factor | power
  63 # comp_op: '<'|'>'|'=='|'>='|'<='|'<>'|'!='|'in'|'not' 'in'|'is'|'is' 'not'
  64
  65 def make_lt_compare(arg):
  66     (left, right) = arg
  67     return ast.Compare(left, [ast.Lt()], [right])
  68 def make_gt_compare(arg):
  69     (left, right) = arg
  70     return ast.Compare(left, [ast.Gt()], [right])
  71 def make_eq_compare(arg):
  72     (left, right) = arg
  73     return ast.Compare(left, [ast.Eq()], [right])
  74
  75 binary_ops = {
  76     "+": ast.Add(),
  77     "-": ast.Sub(),
  78     "*": ast.Mult(),
  79     "/": ast.Div(),
  80     "<": make_lt_compare,
  81     ">": make_gt_compare,
  82     "=": make_eq_compare,
  83 }
  84 unary_ops = {
  85     "+": ast.Add,
  86     "-": ast.Sub,
  87     }
  88
  89 def check_concat(node): # checks if the comparison is already a concat
  90     print (node)
  91     if not isinstance(node, ast.Call):
  92         return [node]
  93     print (node.func.id)
  94     if node.func.id != 'concat':
  95         return [node]
  96     return node[1]
  97
  98
  99 ##########   Parser (tokens -> AST) ######
 100
 101 # also part of Ply
 102 #import yacc
 103
 104 class PowerParser:
 105
 106     precedence = (
 107         ("left", "EQ", "GT", "LT"),
 108         ("left", "PLUS", "MINUS"),
 109         ("left", "MULT", "DIV"),
 110         )
 111
 112     def __init__(self):
 113         self.gprs = {}
 114         for rname in ['RA', 'RB', 'RC', 'RT', 'RS']:
 115             self.gprs[rname] = None
 116         self.read_regs = []
 117         self.write_regs = []
 118
 119     # The grammar comments come from Python's Grammar/Grammar file
 120
 121     ## NB: compound_stmt in single_input is followed by extra NEWLINE!
 122     # file_input: (NEWLINE | stmt)* ENDMARKER
 123
 124     def p_file_input_end(self, p):
 125         """file_input_end : file_input ENDMARKER"""
 126         print ("end", p[1])
 127         p[0] = p[1]
 128
 129     def p_file_input(self, p):
 130         """file_input : file_input NEWLINE
 131                       | file_input stmt
 132                       | NEWLINE
 133                       | stmt"""
 134         if isinstance(p[len(p)-1], str):
 135             if len(p) == 3:
 136                 p[0] = p[1]
 137             else:
 138                 p[0] = [] # p == 2 --> only a blank line
 139         else:
 140             if len(p) == 3:
 141                 p[0] = p[1] + p[2]
 142             else:
 143                 p[0] = p[1]
 144
 145
 146     # funcdef: [decorators] 'def' NAME parameters ':' suite
 147     # ignoring decorators
 148     def p_funcdef(self, p):
 149         "funcdef : DEF NAME parameters COLON suite"
 150         p[0] = ast.FunctionDef(p[2], p[3], p[5], ())
 151
 152     # parameters: '(' [varargslist] ')'
 153     def p_parameters(self, p):
 154         """parameters : LPAR RPAR
 155                       | LPAR varargslist RPAR"""
 156         if len(p) == 3:
 157             args=[]
 158         else:
 159             args = p[2]
 160         p[0] = ast.arguments(args=args, vararg=None, kwarg=None, defaults=[])
 161
 162
 163     # varargslist: (fpdef ['=' test] ',')* ('*' NAME [',' '**' NAME] |
 164     # '**' NAME) |
 165     # highly simplified
 166     def p_varargslist(self, p):
 167         """varargslist : varargslist COMMA NAME
 168                        | NAME"""
 169         if len(p) == 4:
 170             p[0] = p[1] + p[3]
 171         else:
 172             p[0] = [p[1]]
 173
 174     # stmt: simple_stmt | compound_stmt
 175     def p_stmt_simple(self, p):
 176         """stmt : simple_stmt"""
 177         # simple_stmt is a list
 178         p[0] = p[1]
 179
 180     def p_stmt_compound(self, p):
 181         """stmt : compound_stmt"""
 182         p[0] = [p[1]]
 183
 184     # simple_stmt: small_stmt (';' small_stmt)* [';'] NEWLINE
 185     def p_simple_stmt(self, p):
 186         """simple_stmt : small_stmts NEWLINE
 187                        | small_stmts SEMICOLON NEWLINE"""
 188         p[0] = p[1]
 189
 190     def p_small_stmts(self, p):
 191         """small_stmts : small_stmts SEMICOLON small_stmt
 192                        | small_stmt"""
 193         if len(p) == 4:
 194             p[0] = p[1] + [p[3]]
 195         else:
 196             p[0] = [p[1]]
 197
 198     # small_stmt: expr_stmt | print_stmt  | del_stmt | pass_stmt | flow_stmt |
 199     #    import_stmt | global_stmt | exec_stmt | assert_stmt
 200     def p_small_stmt(self, p):
 201         """small_stmt : flow_stmt
 202                       | break_stmt
 203                       | expr_stmt"""
 204         if isinstance(p[1], ast.Call):
 205             p[0] = ast.Expr(p[1])
 206         else:
 207             p[0] = p[1]
 208
 209     # expr_stmt: testlist (augassign (yield_expr|testlist) |
 210     #                      ('=' (yield_expr|testlist))*)
 211     # augassign: ('+=' | '-=' | '*=' | '/=' | '%=' | '&=' | '|=' | '^=' |
 212     #             '<<=' | '>>=' | '**=' | '//=')
 213     def p_expr_stmt(self, p):
 214         """expr_stmt : testlist ASSIGN testlist
 215                      | testlist """
 216         if len(p) == 2:
 217             # a list of expressions
 218             #p[0] = ast.Discard(p[1])
 219             p[0] = p[1]
 220         else:
 221             if p[1].id in self.gprs:
 222                 self.write_regs.append(p[1].id) # add to list of regs to write
 223             p[0] = Assign(p[1], p[3])
 224
 225     def p_flow_stmt(self, p):
 226         "flow_stmt : return_stmt"
 227         p[0] = p[1]
 228
 229     # return_stmt: 'return' [testlist]
 230     def p_return_stmt(self, p):
 231         "return_stmt : RETURN testlist"
 232         p[0] = ast.Return(p[2])
 233
 234
 235     def p_compound_stmt(self, p):
 236         """compound_stmt : if_stmt
 237                          | while_stmt
 238                          | for_stmt
 239                          | funcdef
 240         """
 241         p[0] = p[1]
 242
 243     def p_break_stmt(self, p):
 244         """break_stmt : BREAK
 245         """
 246         p[0] = ast.Break()
 247
 248     def p_for_stmt(self, p):
 249         """for_stmt : FOR test EQ test TO test COLON suite
 250         """
 251         p[0] = ast.While(p[2], p[4], [])
 252         # auto-add-one (sigh) due to python range
 253         start = p[4]
 254         end = ast.BinOp(p[6], ast.Add(), ast.Constant(1))
 255         it = ast.Call(ast.Name("range"), [start, end], [])
 256         p[0] = ast.For(p[2], it, p[8], [])
 257
 258     def p_while_stmt(self, p):
 259         """while_stmt : DO WHILE test COLON suite ELSE COLON suite
 260                       | DO WHILE test COLON suite
 261         """
 262         if len(p) == 6:
 263             p[0] = ast.While(p[3], p[5], [])
 264         else:
 265             p[0] = ast.While(p[3], p[5], p[8])
 266
 267     def p_if_stmt(self, p):
 268         """if_stmt : IF test COLON suite ELSE COLON suite
 269                    | IF test COLON suite
 270         """
 271         if len(p) == 5:
 272             p[0] = ast.If(p[2], p[4], [])
 273         else:
 274             p[0] = ast.If(p[2], p[4], p[7])
 275
 276     def p_suite(self, p):
 277         """suite : simple_stmt
 278                  | NEWLINE INDENT stmts DEDENT"""
 279         if len(p) == 2:
 280             p[0] = p[1]
 281         else:
 282             p[0] = p[3]
 283
 284
 285     def p_stmts(self, p):
 286         """stmts : stmts stmt
 287                  | stmt"""
 288         if len(p) == 3:
 289             p[0] = p[1] + p[2]
 290         else:
 291             p[0] = p[1]
 292
 293     def p_comparison(self, p):
 294         """comparison : comparison PLUS comparison
 295                       | comparison MINUS comparison
 296                       | comparison MULT comparison
 297                       | comparison DIV comparison
 298                       | comparison LT comparison
 299                       | comparison EQ comparison
 300                       | comparison GT comparison
 301                       | PLUS comparison
 302                       | MINUS comparison
 303                       | comparison APPEND comparison
 304                       | power"""
 305         if len(p) == 4:
 306             print (list(p))
 307             if p[2] == '||':
 308                 l = check_concat(p[1]) + check_concat(p[3])
 309                 p[0] = ast.Call(ast.Name("concat"), l, [])
 310             elif p[2] in ['<', '>', '=']:
 311                 p[0] = binary_ops[p[2]]((p[1],p[3]))
 312             else:
 313                 p[0] = ast.BinOp(p[1], binary_ops[p[2]], p[3])
 314         elif len(p) == 3:
 315             p[0] = unary_ops[p[1]](p[2])
 316         else:
 317             p[0] = p[1]
 318
 319     # power: atom trailer* ['**' factor]
 320     # trailers enables function calls (and subscripts).
 321     # I only allow one level of calls
 322     # so this is 'trailer'
 323     def p_power(self, p):
 324         """power : atom
 325                  | atom trailer"""
 326         if len(p) == 2:
 327             p[0] = p[1]
 328         else:
 329             if p[2][0] == "CALL":
 330                 #p[0] = ast.Expr(ast.Call(p[1], p[2][1], []))
 331                 p[0] = ast.Call(p[1], p[2][1], [])
 332                 #if p[1].id == 'print':
 333                 #    p[0] = ast.Printnl(ast.Tuple(p[2][1]), None, None)
 334                 #else:
 335                 #    p[0] = ast.CallFunc(p[1], p[2][1], None, None)
 336             else:
 337                 print (p[2][1])
 338                 #raise AssertionError("not implemented %s" % p[2][0])
 339                 subs = p[2][1]
 340                 if len(subs) == 1:
 341                     idx = subs[0]
 342                 else:
 343                     idx = ast.Slice(subs[0], subs[1], None)
 344                 p[0] = ast.Subscript(p[1], idx)
 345
 346     def p_atom_name(self, p):
 347         """atom : NAME"""
 348         p[0] = ast.Name(p[1], ctx=ast.Load())
 349
 350     def p_atom_number(self, p):
 351         """atom : BINARY
 352                 | NUMBER
 353                 | STRING"""
 354         p[0] = ast.Constant(p[1])
 355
 356     #'[' [listmaker] ']' |
 357
 358     def p_atom_listmaker(self, p):
 359         """atom : LBRACK listmaker RBRACK"""
 360         p[0] = p[2]
 361
 362     def p_listmaker(self, p):
 363         """listmaker : test COMMA listmaker
 364                      | test
 365         """
 366         if len(p) == 2:
 367             p[0] = ast.List([p[1]])
 368         else:
 369             p[0] = ast.List([p[1]] + p[3].nodes)
 370
 371     def p_atom_tuple(self, p):
 372         """atom : LPAR testlist RPAR"""
 373         print ("tuple", p[2])
 374         if isinstance(p[2], ast.Name):
 375             print ("tuple name", p[2].id)
 376             if p[2].id in self.gprs:
 377                 self.read_regs.append(p[2].id) # add to list of regs to read
 378                 #p[0] = ast.Subscript(ast.Name("GPR"), ast.Str(p[2].id))
 379                 #return
 380         p[0] = p[2]
 381
 382     # trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
 383     def p_trailer(self, p):
 384         """trailer : trailer_arglist
 385                    | trailer_subscript
 386         """
 387         p[0] = p[1]
 388
 389     def p_trailer_arglist(self, p):
 390         "trailer_arglist : LPAR arglist RPAR"
 391         p[0] = ("CALL", p[2])
 392
 393     def p_trailer_subscript(self, p):
 394         "trailer_subscript : LBRACK subscript RBRACK"
 395         p[0] = ("SUBS", p[2])
 396
 397     #subscript: '.' '.' '.' | test | [test] ':' [test]
 398
 399     def p_subscript(self, p):
 400         """subscript : test COLON test
 401                      | test
 402         """
 403         if len(p) == 4:
 404             p[0] = [p[1], p[3]]
 405         else:
 406             p[0] = [p[1]]
 407
 408
 409     # testlist: test (',' test)* [',']
 410     # Contains shift/reduce error
 411     def p_testlist(self, p):
 412         """testlist : testlist_multi COMMA
 413                     | testlist_multi """
 414         if len(p) == 2:
 415             p[0] = p[1]
 416         else:
 417             # May need to promote singleton to tuple
 418             if isinstance(p[1], list):
 419                 p[0] = p[1]
 420             else:
 421                 p[0] = [p[1]]
 422         # Convert into a tuple?
 423         if isinstance(p[0], list):
 424             p[0] = ast.Tuple(p[0])
 425
 426     def p_testlist_multi(self, p):
 427         """testlist_multi : testlist_multi COMMA test
 428                           | test"""
 429         if len(p) == 2:
 430             # singleton
 431             p[0] = p[1]
 432         else:
 433             if isinstance(p[1], list):
 434                 p[0] = p[1] + [p[3]]
 435             else:
 436                 # singleton -> tuple
 437                 p[0] = [p[1], p[3]]
 438
 439
 440     # test: or_test ['if' or_test 'else' test] | lambdef
 441     #  as I don't support 'and', 'or', and 'not' this works down to 'comparison'
 442     def p_test(self, p):
 443         "test : comparison"
 444         p[0] = p[1]
 445
 446
 447
 448     # arglist: (argument ',')* (argument [',']| '*' test [',' '**' test]
 449     # | '**' test)
 450     # XXX INCOMPLETE: this doesn't allow the trailing comma
 451     def p_arglist(self, p):
 452         """arglist : arglist COMMA argument
 453                    | argument"""
 454         if len(p) == 4:
 455             p[0] = p[1] + [p[3]]
 456         else:
 457             p[0] = [p[1]]
 458
 459     # argument: test [gen_for] | test '=' test  # Really [keyword '='] test
 460     def p_argument(self, p):
 461         "argument : test"
 462         p[0] = p[1]
 463
 464     def p_error(self, p):
 465         #print "Error!", repr(p)
 466         raise SyntaxError(p)
 467
 468
 469 class GardenSnakeParser(PowerParser):
 470     def __init__(self, lexer = None):
 471         PowerParser.__init__(self)
 472         if lexer is None:
 473             lexer = IndentLexer(debug=1)
 474         self.lexer = lexer
 475         self.tokens = lexer.tokens
 476         self.parser = yacc.yacc(module=self, start="file_input_end",
 477                                 debug=False, write_tables=False)
 478
 479         self.sd = create_pdecode()
 480
 481     def parse(self, code):
 482         self.lexer.input(code)
 483         result = self.parser.parse(lexer = self.lexer, debug=False)
 484         return ast.Module(result)
 485
 486
 487 ###### Code generation ######
 488
 489 #from compiler import misc, syntax, pycodegen
 490
 491 class GardenSnakeCompiler(object):
 492     def __init__(self):
 493         self.parser = GardenSnakeParser()
 494     def compile(self, code, mode="exec", filename="<string>"):
 495         tree = self.parser.parse(code)
 496         print ("snake")
 497         pprint(tree)
 498         return tree
 499         #misc.set_filename(filename, tree)
 500         return compile(tree, mode="exec", filename="<string>")
 501         #syntax.check(tree)
 502         gen = pycodegen.ModuleCodeGenerator(tree)
 503         code = gen.getCode()
 504         return code
 505