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
  24
  25 def Assign(left, right):
  26     names = []
  27     print("Assign", left, right)
  28     if isinstance(left, ast.Name):
  29         # Single assignment on left
  30         # XXX when doing IntClass, which will have an "eq" function,
  31         # this is how to access it
  32         #   eq = ast.Attribute(left, "eq")   # get eq fn
  33         #   return ast.Call(eq, [right], []) # now call left.eq(right)
  34         return ast.Assign([ast.Name(left.id, ast.Store())], right)
  35     elif isinstance(left, ast.Tuple):
  36         # List of things - make sure they are Name nodes
  37         names = []
  38         for child in left.getChildren():
  39             if not isinstance(child, ast.Name):
  40                 raise SyntaxError("that assignment not supported")
  41             names.append(child.name)
  42         ass_list = [ast.AssName(name, 'OP_ASSIGN') for name in names]
  43         return ast.Assign([ast.AssTuple(ass_list)], right)
  44     elif isinstance(left, ast.Subscript):
  45         return ast.Assign([left], right)
  46         # XXX HMMM probably not needed...
  47         ls = left.slice
  48         if isinstance(ls, ast.Slice):
  49             lower, upper, step = ls.lower, ls.upper, ls.step
  50             print("slice assign", lower, upper, step)
  51             if step is None:
  52                 ls = (lower, upper, None)
  53             else:
  54                 ls = (lower, upper, step)
  55             ls = ast.Tuple(ls)
  56         return ast.Call(ast.Name("selectassign"),
  57                         [left.value, ls, right], [])
  58     else:
  59         print("Assign fail")
  60         raise SyntaxError("Can't do that yet")
  61
  62
  63 # I implemented INDENT / DEDENT generation as a post-processing filter
  64
  65 # The original lex token stream contains WS and NEWLINE characters.
  66 # WS will only occur before any other tokens on a line.
  67
  68 # I have three filters.  One tags tokens by adding two attributes.
  69 # "must_indent" is True if the token must be indented from the
  70 # previous code.  The other is "at_line_start" which is True for WS
  71 # and the first non-WS/non-NEWLINE on a line.  It flags the check so
  72 # see if the new line has changed indication level.
  73
  74
  75 # No using Python's approach because Ply supports precedence
  76
  77 # comparison: expr (comp_op expr)*
  78 # arith_expr: term (('+'|'-') term)*
  79 # term: factor (('*'|'/'|'%'|'//') factor)*
  80 # factor: ('+'|'-'|'~') factor | power
  81 # comp_op: '<'|'>'|'=='|'>='|'<='|'<>'|'!='|'in'|'not' 'in'|'is'|'is' 'not'
  82
  83 def make_le_compare(arg):
  84     (left, right) = arg
  85     return ast.Compare(left, [ast.LtE()], [right])
  86
  87
  88 def make_ge_compare(arg):
  89     (left, right) = arg
  90     return ast.Compare(left, [ast.GtE()], [right])
  91
  92
  93 def make_lt_compare(arg):
  94     (left, right) = arg
  95     return ast.Compare(left, [ast.Lt()], [right])
  96
  97
  98 def make_gt_compare(arg):
  99     (left, right) = arg
 100     return ast.Compare(left, [ast.Gt()], [right])
 101
 102
 103 def make_eq_compare(arg):
 104     (left, right) = arg
 105     return ast.Compare(left, [ast.Eq()], [right])
 106
 107
 108 binary_ops = {
 109     "&": ast.BitAnd(),
 110     "|": ast.BitOr(),
 111     "+": ast.Add(),
 112     "-": ast.Sub(),
 113     "*": ast.Mult(),
 114     "/": ast.Div(),
 115     "<=": make_le_compare,
 116     ">=": make_ge_compare,
 117     "<": make_lt_compare,
 118     ">": make_gt_compare,
 119     "=": make_eq_compare,
 120 }
 121 unary_ops = {
 122     "+": ast.UAdd(),
 123     "-": ast.USub(),
 124     "¬": ast.Invert(),
 125 }
 126
 127
 128 def check_concat(node):  # checks if the comparison is already a concat
 129     print("check concat", node)
 130     if not isinstance(node, ast.Call):
 131         return [node]
 132     print("func", node.func.id)
 133     if node.func.id != 'concat':
 134         return [node]
 135     return node.args
 136
 137
 138 ##########   Parser (tokens -> AST) ######
 139
 140 # also part of Ply
 141 #import yacc
 142
 143 class PowerParser:
 144
 145     precedence = (
 146         ("left", "EQ", "GT", "LT", "LE", "GE", "LTU", "GTU"),
 147         ("left", "BITOR"),
 148         ("left", "BITAND"),
 149         ("left", "PLUS", "MINUS"),
 150         ("left", "MULT", "DIV"),
 151         ("left", "INVERT"),
 152     )
 153
 154     def __init__(self):
 155         self.gprs = {}
 156         for rname in ['RA', 'RB', 'RC', 'RT', 'RS']:
 157             self.gprs[rname] = None
 158         self.read_regs = []
 159         self.uninit_regs = []
 160         self.write_regs = []
 161
 162     # The grammar comments come from Python's Grammar/Grammar file
 163
 164     # NB: compound_stmt in single_input is followed by extra NEWLINE!
 165     # file_input: (NEWLINE | stmt)* ENDMARKER
 166
 167     def p_file_input_end(self, p):
 168         """file_input_end : file_input ENDMARKER"""
 169         print("end", p[1])
 170         p[0] = p[1]
 171
 172     def p_file_input(self, p):
 173         """file_input : file_input NEWLINE
 174                       | file_input stmt
 175                       | NEWLINE
 176                       | stmt"""
 177         if isinstance(p[len(p)-1], str):
 178             if len(p) == 3:
 179                 p[0] = p[1]
 180             else:
 181                 p[0] = []  # p == 2 --> only a blank line
 182         else:
 183             if len(p) == 3:
 184                 p[0] = p[1] + p[2]
 185             else:
 186                 p[0] = p[1]
 187
 188     # funcdef: [decorators] 'def' NAME parameters ':' suite
 189     # ignoring decorators
 190
 191     def p_funcdef(self, p):
 192         "funcdef : DEF NAME parameters COLON suite"
 193         p[0] = ast.FunctionDef(p[2], p[3], p[5], ())
 194
 195     # parameters: '(' [varargslist] ')'
 196     def p_parameters(self, p):
 197         """parameters : LPAR RPAR
 198                       | LPAR varargslist RPAR"""
 199         if len(p) == 3:
 200             args = []
 201         else:
 202             args = p[2]
 203         p[0] = ast.arguments(args=args, vararg=None, kwarg=None, defaults=[])
 204
 205     # varargslist: (fpdef ['=' test] ',')* ('*' NAME [',' '**' NAME] |
 206     # '**' NAME) |
 207     # highly simplified
 208
 209     def p_varargslist(self, p):
 210         """varargslist : varargslist COMMA NAME
 211                        | NAME"""
 212         if len(p) == 4:
 213             p[0] = p[1] + p[3]
 214         else:
 215             p[0] = [p[1]]
 216
 217     # stmt: simple_stmt | compound_stmt
 218     def p_stmt_simple(self, p):
 219         """stmt : simple_stmt"""
 220         # simple_stmt is a list
 221         p[0] = p[1]
 222
 223     def p_stmt_compound(self, p):
 224         """stmt : compound_stmt"""
 225         p[0] = [p[1]]
 226
 227     # simple_stmt: small_stmt (';' small_stmt)* [';'] NEWLINE
 228     def p_simple_stmt(self, p):
 229         """simple_stmt : small_stmts NEWLINE
 230                        | small_stmts SEMICOLON NEWLINE"""
 231         p[0] = p[1]
 232
 233     def p_small_stmts(self, p):
 234         """small_stmts : small_stmts SEMICOLON small_stmt
 235                        | small_stmt"""
 236         if len(p) == 4:
 237             p[0] = p[1] + [p[3]]
 238         else:
 239             p[0] = [p[1]]
 240
 241     # small_stmt: expr_stmt | print_stmt  | del_stmt | pass_stmt | flow_stmt |
 242     #    import_stmt | global_stmt | exec_stmt | assert_stmt
 243     def p_small_stmt(self, p):
 244         """small_stmt : flow_stmt
 245                       | break_stmt
 246                       | expr_stmt"""
 247         if isinstance(p[1], ast.Call):
 248             p[0] = ast.Expr(p[1])
 249         else:
 250             p[0] = p[1]
 251
 252     # expr_stmt: testlist (augassign (yield_expr|testlist) |
 253     #                      ('=' (yield_expr|testlist))*)
 254     # augassign: ('+=' | '-=' | '*=' | '/=' | '%=' | '&=' | '|=' | '^=' |
 255     #             '<<=' | '>>=' | '**=' | '//=')
 256     def p_expr_stmt(self, p):
 257         """expr_stmt : testlist ASSIGN testlist
 258                      | testlist """
 259         print("expr_stmt", p)
 260         if len(p) == 2:
 261             # a list of expressions
 262             #p[0] = ast.Discard(p[1])
 263             p[0] = p[1]
 264         else:
 265             if isinstance(p[1], ast.Name):
 266                 name = p[1].id
 267             elif isinstance(p[1], ast.Subscript):
 268                 name = p[1].value.id
 269                 if name in self.gprs:
 270                     # add to list of uninitialised
 271                     self.uninit_regs.append(name)
 272             print("expr assign", name, p[1])
 273             if name in self.gprs:
 274                 self.write_regs.append(name)  # add to list of regs to write
 275             p[0] = Assign(p[1], p[3])
 276
 277     def p_flow_stmt(self, p):
 278         "flow_stmt : return_stmt"
 279         p[0] = p[1]
 280
 281     # return_stmt: 'return' [testlist]
 282     def p_return_stmt(self, p):
 283         "return_stmt : RETURN testlist"
 284         p[0] = ast.Return(p[2])
 285
 286     def p_compound_stmt(self, p):
 287         """compound_stmt : if_stmt
 288                          | while_stmt
 289                          | for_stmt
 290                          | funcdef
 291         """
 292         p[0] = p[1]
 293
 294     def p_break_stmt(self, p):
 295         """break_stmt : BREAK
 296         """
 297         p[0] = ast.Break()
 298
 299     def p_for_stmt(self, p):
 300         """for_stmt : FOR test EQ test TO test COLON suite
 301         """
 302         p[0] = ast.While(p[2], p[4], [])
 303         # auto-add-one (sigh) due to python range
 304         start = p[4]
 305         end = ast.BinOp(p[6], ast.Add(), ast.Constant(1))
 306         it = ast.Call(ast.Name("range"), [start, end], [])
 307         p[0] = ast.For(p[2], it, p[8], [])
 308
 309     def p_while_stmt(self, p):
 310         """while_stmt : DO WHILE test COLON suite ELSE COLON suite
 311                       | DO WHILE test COLON suite
 312         """
 313         if len(p) == 6:
 314             p[0] = ast.While(p[3], p[5], [])
 315         else:
 316             p[0] = ast.While(p[3], p[5], p[8])
 317
 318     def p_if_stmt(self, p):
 319         """if_stmt : IF test COLON suite ELSE COLON if_stmt
 320                    | IF test COLON suite ELSE COLON suite
 321                    | IF test COLON suite
 322         """
 323         if len(p) == 8 and isinstance(p[7], ast.If):
 324             p[0] = ast.If(p[2], p[4], [p[7]])
 325         elif len(p) == 5:
 326             p[0] = ast.If(p[2], p[4], [])
 327         else:
 328             p[0] = ast.If(p[2], p[4], p[7])
 329
 330     def p_suite(self, p):
 331         """suite : simple_stmt
 332                  | NEWLINE INDENT stmts DEDENT"""
 333         if len(p) == 2:
 334             p[0] = p[1]
 335         else:
 336             p[0] = p[3]
 337
 338     def p_stmts(self, p):
 339         """stmts : stmts stmt
 340                  | stmt"""
 341         if len(p) == 3:
 342             p[0] = p[1] + p[2]
 343         else:
 344             p[0] = p[1]
 345
 346     def p_comparison(self, p):
 347         """comparison : comparison PLUS comparison
 348                       | comparison MINUS comparison
 349                       | comparison MULT comparison
 350                       | comparison DIV comparison
 351                       | comparison EQ comparison
 352                       | comparison LE comparison
 353                       | comparison GE comparison
 354                       | comparison LTU comparison
 355                       | comparison GTU comparison
 356                       | comparison LT comparison
 357                       | comparison GT comparison
 358                       | comparison BITOR comparison
 359                       | comparison BITAND comparison
 360                       | PLUS comparison
 361                       | comparison MINUS
 362                       | INVERT comparison
 363                       | comparison APPEND comparison
 364                       | power"""
 365         if len(p) == 4:
 366             print(list(p))
 367             if p[2] == '<u':
 368                 p[0] = ast.Call(ast.Name("ltu"), (p[1], p[3]), [])
 369             elif p[2] == '>u':
 370                 p[0] = ast.Call(ast.Name("gtu"), (p[1], p[3]), [])
 371             elif p[2] == '||':
 372                 l = check_concat(p[1]) + check_concat(p[3])
 373                 p[0] = ast.Call(ast.Name("concat"), l, [])
 374             elif p[2] in ['<', '>', '=', '<=', '>=']:
 375                 p[0] = binary_ops[p[2]]((p[1], p[3]))
 376             else:
 377                 p[0] = ast.BinOp(p[1], binary_ops[p[2]], p[3])
 378         elif len(p) == 3:
 379             if isinstance(p[2], str) and p[2] == '-':
 380                 p[0] = ast.UnaryOp(unary_ops[p[2]], p[1])
 381             else:
 382                 p[0] = ast.UnaryOp(unary_ops[p[1]], p[2])
 383         else:
 384             p[0] = p[1]
 385
 386     # power: atom trailer* ['**' factor]
 387     # trailers enables function calls (and subscripts).
 388     # I only allow one level of calls
 389     # so this is 'trailer'
 390     def p_power(self, p):
 391         """power : atom
 392                  | atom trailer"""
 393         if len(p) == 2:
 394             p[0] = p[1]
 395         else:
 396             if p[2][0] == "CALL":
 397                 #p[0] = ast.Expr(ast.Call(p[1], p[2][1], []))
 398                 p[0] = ast.Call(p[1], p[2][1], [])
 399                 # if p[1].id == 'print':
 400                 #    p[0] = ast.Printnl(ast.Tuple(p[2][1]), None, None)
 401                 # else:
 402                 #    p[0] = ast.CallFunc(p[1], p[2][1], None, None)
 403             else:
 404                 print("subscript atom", p[2][1])
 405                 #raise AssertionError("not implemented %s" % p[2][0])
 406                 subs = p[2][1]
 407                 if len(subs) == 1:
 408                     idx = subs[0]
 409                 else:
 410                     idx = ast.Slice(subs[0], subs[1], None)
 411                 p[0] = ast.Subscript(p[1], idx, ast.Load())
 412
 413     def p_atom_name(self, p):
 414         """atom : NAME"""
 415         p[0] = ast.Name(id=p[1], ctx=ast.Load())
 416
 417     def p_atom_number(self, p):
 418         """atom : BINARY
 419                 | NUMBER
 420                 | STRING"""
 421         p[0] = ast.Constant(p[1])
 422
 423     # '[' [listmaker] ']' |
 424
 425     def p_atom_listmaker(self, p):
 426         """atom : LBRACK listmaker RBRACK"""
 427         p[0] = p[2]
 428
 429     def p_listmaker(self, p):
 430         """listmaker : test COMMA listmaker
 431                      | test
 432         """
 433         if len(p) == 2:
 434             p[0] = ast.List([p[1]])
 435         else:
 436             p[0] = ast.List([p[1]] + p[3].nodes)
 437
 438     def p_atom_tuple(self, p):
 439         """atom : LPAR testlist RPAR"""
 440         print("tuple", p[2])
 441         if isinstance(p[2], ast.Name):
 442             print("tuple name", p[2].id)
 443             if p[2].id in self.gprs:
 444                 self.read_regs.append(p[2].id)  # add to list of regs to read
 445                 #p[0] = ast.Subscript(ast.Name("GPR"), ast.Str(p[2].id))
 446                 # return
 447         p[0] = p[2]
 448
 449     # trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
 450     def p_trailer(self, p):
 451         """trailer : trailer_arglist
 452                    | trailer_subscript
 453         """
 454         p[0] = p[1]
 455
 456     def p_trailer_arglist(self, p):
 457         "trailer_arglist : LPAR arglist RPAR"
 458         p[0] = ("CALL", p[2])
 459
 460     def p_trailer_subscript(self, p):
 461         "trailer_subscript : LBRACK subscript RBRACK"
 462         p[0] = ("SUBS", p[2])
 463
 464     # subscript: '.' '.' '.' | test | [test] ':' [test]
 465
 466     def p_subscript(self, p):
 467         """subscript : test COLON test
 468                      | test
 469         """
 470         if len(p) == 4:
 471             # add one to end
 472             if isinstance(p[3], ast.Constant):
 473                 end = ast.Constant(p[3].value+1)
 474             else:
 475                 end = ast.BinOp(p[3], ast.Add(), ast.Constant(1))
 476             p[0] = [p[1], end]
 477         else:
 478             p[0] = [p[1]]
 479
 480     # testlist: test (',' test)* [',']
 481     # Contains shift/reduce error
 482
 483     def p_testlist(self, p):
 484         """testlist : testlist_multi COMMA
 485                     | testlist_multi """
 486         if len(p) == 2:
 487             p[0] = p[1]
 488         else:
 489             # May need to promote singleton to tuple
 490             if isinstance(p[1], list):
 491                 p[0] = p[1]
 492             else:
 493                 p[0] = [p[1]]
 494         # Convert into a tuple?
 495         if isinstance(p[0], list):
 496             p[0] = ast.Tuple(p[0])
 497
 498     def p_testlist_multi(self, p):
 499         """testlist_multi : testlist_multi COMMA test
 500                           | test"""
 501         if len(p) == 2:
 502             # singleton
 503             p[0] = p[1]
 504         else:
 505             if isinstance(p[1], list):
 506                 p[0] = p[1] + [p[3]]
 507             else:
 508                 # singleton -> tuple
 509                 p[0] = [p[1], p[3]]
 510
 511     # test: or_test ['if' or_test 'else' test] | lambdef
 512     #  as I don't support 'and', 'or', and 'not' this works down to 'comparison'
 513
 514     def p_test(self, p):
 515         "test : comparison"
 516         p[0] = p[1]
 517
 518     # arglist: (argument ',')* (argument [',']| '*' test [',' '**' test]
 519     # | '**' test)
 520     # XXX INCOMPLETE: this doesn't allow the trailing comma
 521
 522     def p_arglist(self, p):
 523         """arglist : arglist COMMA argument
 524                    | argument"""
 525         if len(p) == 4:
 526             p[0] = p[1] + [p[3]]
 527         else:
 528             p[0] = [p[1]]
 529
 530     # argument: test [gen_for] | test '=' test  # Really [keyword '='] test
 531     def p_argument(self, p):
 532         "argument : test"
 533         p[0] = p[1]
 534
 535     def p_error(self, p):
 536         # print "Error!", repr(p)
 537         raise SyntaxError(p)
 538
 539
 540 class GardenSnakeParser(PowerParser):
 541     def __init__(self, lexer=None):
 542         PowerParser.__init__(self)
 543         if lexer is None:
 544             lexer = IndentLexer(debug=0)
 545         self.lexer = lexer
 546         self.tokens = lexer.tokens
 547         self.parser = yacc.yacc(module=self, start="file_input_end",
 548                                 debug=False, write_tables=False)
 549
 550         self.sd = create_pdecode()
 551
 552     def parse(self, code):
 553         # self.lexer.input(code)
 554         result = self.parser.parse(code, lexer=self.lexer, debug=False)
 555         return ast.Module(result)
 556
 557
 558 ###### Code generation ######
 559
 560 #from compiler import misc, syntax, pycodegen
 561
 562 class GardenSnakeCompiler(object):
 563     def __init__(self):
 564         self.parser = GardenSnakeParser()
 565
 566     def compile(self, code, mode="exec", filename="<string>"):
 567         tree = self.parser.parse(code)
 568         print("snake")
 569         pprint(tree)
 570         return tree
 571         #misc.set_filename(filename, tree)
 572         return compile(tree, mode="exec", filename="<string>")
 573         # syntax.check(tree)
 574         gen = pycodegen.ModuleCodeGenerator(tree)
 575         code = gen.getCode()
 576         return code