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     "%": ast.Mod(),
 116     "<=": make_le_compare,
 117     ">=": make_ge_compare,
 118     "<": make_lt_compare,
 119     ">": make_gt_compare,
 120     "=": make_eq_compare,
 121 }
 122 unary_ops = {
 123     "+": ast.UAdd(),
 124     "-": ast.USub(),
 125     "¬": ast.Invert(),
 126 }
 127
 128
 129 def check_concat(node):  # checks if the comparison is already a concat
 130     print("check concat", node)
 131     if not isinstance(node, ast.Call):
 132         return [node]
 133     print("func", node.func.id)
 134     if node.func.id != 'concat':
 135         return [node]
 136     return node.args
 137
 138
 139 ##########   Parser (tokens -> AST) ######
 140
 141 # also part of Ply
 142 #import yacc
 143
 144 # https://www.mathcs.emory.edu/~valerie/courses/fall10/155/resources/op_precedence.html
 145 # python operator precedence
 146 # Highest precedence at top, lowest at bottom.
 147 # Operators in the same box evaluate left to right.
 148 #
 149 # Operator Description
 150 # ()                                                     Parentheses (grouping)
 151 # f(args...)                                             Function call
 152 # x[index:index]                                         Slicing
 153 # x[index]                                               Subscription
 154 # x.attribute                                            Attribute reference
 155 # **                                                     Exponentiation
 156 # ~x                                                     Bitwise not
 157 # +x, -x                                                 Positive, negative
 158 # *, /, %                                                mul, div, remainder
 159 # +, -                                                   Addition, subtraction
 160 # <<, >>                                                 Bitwise shifts
 161 # &                                                      Bitwise AND
 162 # ^                                                      Bitwise XOR
 163 # |                                                      Bitwise OR
 164 # in, not in, is, is not, <, <=,  >,  >=, <>, !=, ==     comp, membership, ident
 165 # not x                                                  Boolean NOT
 166 # and                                                    Boolean AND
 167 # or                                                     Boolean OR
 168 # lambda                                                 Lambda expression
 169
 170 class PowerParser:
 171
 172     precedence = (
 173         ("left", "EQ", "GT", "LT", "LE", "GE", "LTU", "GTU"),
 174         ("left", "BITOR"),
 175         ("left", "BITAND"),
 176         ("left", "PLUS", "MINUS"),
 177         ("left", "MULT", "DIV", "MOD"),
 178         ("left", "INVERT"),
 179     )
 180
 181     def __init__(self):
 182         self.gprs = {}
 183         for rname in ['RA', 'RB', 'RC', 'RT', 'RS']:
 184             self.gprs[rname] = None
 185         self.read_regs = []
 186         self.uninit_regs = []
 187         self.write_regs = []
 188
 189     # The grammar comments come from Python's Grammar/Grammar file
 190
 191     # NB: compound_stmt in single_input is followed by extra NEWLINE!
 192     # file_input: (NEWLINE | stmt)* ENDMARKER
 193
 194     def p_file_input_end(self, p):
 195         """file_input_end : file_input ENDMARKER"""
 196         print("end", p[1])
 197         p[0] = p[1]
 198
 199     def p_file_input(self, p):
 200         """file_input : file_input NEWLINE
 201                       | file_input stmt
 202                       | NEWLINE
 203                       | stmt"""
 204         if isinstance(p[len(p)-1], str):
 205             if len(p) == 3:
 206                 p[0] = p[1]
 207             else:
 208                 p[0] = []  # p == 2 --> only a blank line
 209         else:
 210             if len(p) == 3:
 211                 p[0] = p[1] + p[2]
 212             else:
 213                 p[0] = p[1]
 214
 215     # funcdef: [decorators] 'def' NAME parameters ':' suite
 216     # ignoring decorators
 217
 218     def p_funcdef(self, p):
 219         "funcdef : DEF NAME parameters COLON suite"
 220         p[0] = ast.FunctionDef(p[2], p[3], p[5], ())
 221
 222     # parameters: '(' [varargslist] ')'
 223     def p_parameters(self, p):
 224         """parameters : LPAR RPAR
 225                       | LPAR varargslist RPAR"""
 226         if len(p) == 3:
 227             args = []
 228         else:
 229             args = p[2]
 230         p[0] = ast.arguments(args=args, vararg=None, kwarg=None, defaults=[])
 231
 232     # varargslist: (fpdef ['=' test] ',')* ('*' NAME [',' '**' NAME] |
 233     # '**' NAME) |
 234     # highly simplified
 235
 236     def p_varargslist(self, p):
 237         """varargslist : varargslist COMMA NAME
 238                        | NAME"""
 239         if len(p) == 4:
 240             p[0] = p[1] + p[3]
 241         else:
 242             p[0] = [p[1]]
 243
 244     # stmt: simple_stmt | compound_stmt
 245     def p_stmt_simple(self, p):
 246         """stmt : simple_stmt"""
 247         # simple_stmt is a list
 248         p[0] = p[1]
 249
 250     def p_stmt_compound(self, p):
 251         """stmt : compound_stmt"""
 252         p[0] = [p[1]]
 253
 254     # simple_stmt: small_stmt (';' small_stmt)* [';'] NEWLINE
 255     def p_simple_stmt(self, p):
 256         """simple_stmt : small_stmts NEWLINE
 257                        | small_stmts SEMICOLON NEWLINE"""
 258         p[0] = p[1]
 259
 260     def p_small_stmts(self, p):
 261         """small_stmts : small_stmts SEMICOLON small_stmt
 262                        | small_stmt"""
 263         if len(p) == 4:
 264             p[0] = p[1] + [p[3]]
 265         else:
 266             p[0] = [p[1]]
 267
 268     # small_stmt: expr_stmt | print_stmt  | del_stmt | pass_stmt | flow_stmt |
 269     #    import_stmt | global_stmt | exec_stmt | assert_stmt
 270     def p_small_stmt(self, p):
 271         """small_stmt : flow_stmt
 272                       | break_stmt
 273                       | expr_stmt"""
 274         if isinstance(p[1], ast.Call):
 275             p[0] = ast.Expr(p[1])
 276         else:
 277             p[0] = p[1]
 278
 279     # expr_stmt: testlist (augassign (yield_expr|testlist) |
 280     #                      ('=' (yield_expr|testlist))*)
 281     # augassign: ('+=' | '-=' | '*=' | '/=' | '%=' | '&=' | '|=' | '^=' |
 282     #             '<<=' | '>>=' | '**=' | '//=')
 283     def p_expr_stmt(self, p):
 284         """expr_stmt : testlist ASSIGN testlist
 285                      | testlist """
 286         print("expr_stmt", p)
 287         if len(p) == 2:
 288             # a list of expressions
 289             #p[0] = ast.Discard(p[1])
 290             p[0] = p[1]
 291         else:
 292             if isinstance(p[1], ast.Name):
 293                 name = p[1].id
 294             elif isinstance(p[1], ast.Subscript):
 295                 name = p[1].value.id
 296                 if name in self.gprs:
 297                     # add to list of uninitialised
 298                     self.uninit_regs.append(name)
 299             print("expr assign", name, p[1])
 300             if name in self.gprs:
 301                 self.write_regs.append(name)  # add to list of regs to write
 302             p[0] = Assign(p[1], p[3])
 303
 304     def p_flow_stmt(self, p):
 305         "flow_stmt : return_stmt"
 306         p[0] = p[1]
 307
 308     # return_stmt: 'return' [testlist]
 309     def p_return_stmt(self, p):
 310         "return_stmt : RETURN testlist"
 311         p[0] = ast.Return(p[2])
 312
 313     def p_compound_stmt(self, p):
 314         """compound_stmt : if_stmt
 315                          | while_stmt
 316                          | for_stmt
 317                          | funcdef
 318         """
 319         p[0] = p[1]
 320
 321     def p_break_stmt(self, p):
 322         """break_stmt : BREAK
 323         """
 324         p[0] = ast.Break()
 325
 326     def p_for_stmt(self, p):
 327         """for_stmt : FOR test EQ test TO test COLON suite
 328         """
 329         p[0] = ast.While(p[2], p[4], [])
 330         # auto-add-one (sigh) due to python range
 331         start = p[4]
 332         end = ast.BinOp(p[6], ast.Add(), ast.Constant(1))
 333         it = ast.Call(ast.Name("range"), [start, end], [])
 334         p[0] = ast.For(p[2], it, p[8], [])
 335
 336     def p_while_stmt(self, p):
 337         """while_stmt : DO WHILE test COLON suite ELSE COLON suite
 338                       | DO WHILE test COLON suite
 339         """
 340         if len(p) == 6:
 341             p[0] = ast.While(p[3], p[5], [])
 342         else:
 343             p[0] = ast.While(p[3], p[5], p[8])
 344
 345     def p_if_stmt(self, p):
 346         """if_stmt : IF test COLON suite ELSE COLON if_stmt
 347                    | IF test COLON suite ELSE COLON suite
 348                    | IF test COLON suite
 349         """
 350         if len(p) == 8 and isinstance(p[7], ast.If):
 351             p[0] = ast.If(p[2], p[4], [p[7]])
 352         elif len(p) == 5:
 353             p[0] = ast.If(p[2], p[4], [])
 354         else:
 355             p[0] = ast.If(p[2], p[4], p[7])
 356
 357     def p_suite(self, p):
 358         """suite : simple_stmt
 359                  | NEWLINE INDENT stmts DEDENT"""
 360         if len(p) == 2:
 361             p[0] = p[1]
 362         else:
 363             p[0] = p[3]
 364
 365     def p_stmts(self, p):
 366         """stmts : stmts stmt
 367                  | stmt"""
 368         if len(p) == 3:
 369             p[0] = p[1] + p[2]
 370         else:
 371             p[0] = p[1]
 372
 373     def p_comparison(self, p):
 374         """comparison : comparison PLUS comparison
 375                       | comparison MINUS comparison
 376                       | comparison MULT comparison
 377                       | comparison DIV comparison
 378                       | comparison MOD comparison
 379                       | comparison EQ comparison
 380                       | comparison LE comparison
 381                       | comparison GE comparison
 382                       | comparison LTU comparison
 383                       | comparison GTU comparison
 384                       | comparison LT comparison
 385                       | comparison GT comparison
 386                       | comparison BITOR comparison
 387                       | comparison BITAND comparison
 388                       | PLUS comparison
 389                       | comparison MINUS
 390                       | INVERT comparison
 391                       | comparison APPEND comparison
 392                       | power"""
 393         if len(p) == 4:
 394             print(list(p))
 395             if p[2] == '<u':
 396                 p[0] = ast.Call(ast.Name("ltu"), (p[1], p[3]), [])
 397             elif p[2] == '>u':
 398                 p[0] = ast.Call(ast.Name("gtu"), (p[1], p[3]), [])
 399             elif p[2] == '||':
 400                 l = check_concat(p[1]) + check_concat(p[3])
 401                 p[0] = ast.Call(ast.Name("concat"), l, [])
 402             elif p[2] in ['<', '>', '=', '<=', '>=']:
 403                 p[0] = binary_ops[p[2]]((p[1], p[3]))
 404             else:
 405                 p[0] = ast.BinOp(p[1], binary_ops[p[2]], p[3])
 406         elif len(p) == 3:
 407             if isinstance(p[2], str) and p[2] == '-':
 408                 p[0] = ast.UnaryOp(unary_ops[p[2]], p[1])
 409             else:
 410                 p[0] = ast.UnaryOp(unary_ops[p[1]], p[2])
 411         else:
 412             p[0] = p[1]
 413
 414     # power: atom trailer* ['**' factor]
 415     # trailers enables function calls (and subscripts).
 416     # I only allow one level of calls
 417     # so this is 'trailer'
 418     def p_power(self, p):
 419         """power : atom
 420                  | atom trailer"""
 421         if len(p) == 2:
 422             p[0] = p[1]
 423         else:
 424             if p[2][0] == "CALL":
 425                 #p[0] = ast.Expr(ast.Call(p[1], p[2][1], []))
 426                 p[0] = ast.Call(p[1], p[2][1], [])
 427                 # if p[1].id == 'print':
 428                 #    p[0] = ast.Printnl(ast.Tuple(p[2][1]), None, None)
 429                 # else:
 430                 #    p[0] = ast.CallFunc(p[1], p[2][1], None, None)
 431             else:
 432                 print("subscript atom", p[2][1])
 433                 #raise AssertionError("not implemented %s" % p[2][0])
 434                 subs = p[2][1]
 435                 if len(subs) == 1:
 436                     idx = subs[0]
 437                 else:
 438                     idx = ast.Slice(subs[0], subs[1], None)
 439                 p[0] = ast.Subscript(p[1], idx, ast.Load())
 440
 441     def p_atom_name(self, p):
 442         """atom : NAME"""
 443         p[0] = ast.Name(id=p[1], ctx=ast.Load())
 444
 445     def p_atom_number(self, p):
 446         """atom : BINARY
 447                 | NUMBER
 448                 | STRING"""
 449         p[0] = ast.Constant(p[1])
 450
 451     # '[' [listmaker] ']' |
 452
 453     def p_atom_listmaker(self, p):
 454         """atom : LBRACK listmaker RBRACK"""
 455         p[0] = p[2]
 456
 457     def p_listmaker(self, p):
 458         """listmaker : test COMMA listmaker
 459                      | test
 460         """
 461         if len(p) == 2:
 462             p[0] = ast.List([p[1]])
 463         else:
 464             p[0] = ast.List([p[1]] + p[3].nodes)
 465
 466     def p_atom_tuple(self, p):
 467         """atom : LPAR testlist RPAR"""
 468         print("tuple", p[2])
 469         if isinstance(p[2], ast.Name):
 470             print("tuple name", p[2].id)
 471             if p[2].id in self.gprs:
 472                 self.read_regs.append(p[2].id)  # add to list of regs to read
 473                 #p[0] = ast.Subscript(ast.Name("GPR"), ast.Str(p[2].id))
 474                 # return
 475         p[0] = p[2]
 476
 477     # trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
 478     def p_trailer(self, p):
 479         """trailer : trailer_arglist
 480                    | trailer_subscript
 481         """
 482         p[0] = p[1]
 483
 484     def p_trailer_arglist(self, p):
 485         "trailer_arglist : LPAR arglist RPAR"
 486         p[0] = ("CALL", p[2])
 487
 488     def p_trailer_subscript(self, p):
 489         "trailer_subscript : LBRACK subscript RBRACK"
 490         p[0] = ("SUBS", p[2])
 491
 492     # subscript: '.' '.' '.' | test | [test] ':' [test]
 493
 494     def p_subscript(self, p):
 495         """subscript : test COLON test
 496                      | test
 497         """
 498         if len(p) == 4:
 499             # add one to end
 500             if isinstance(p[3], ast.Constant):
 501                 end = ast.Constant(p[3].value+1)
 502             else:
 503                 end = ast.BinOp(p[3], ast.Add(), ast.Constant(1))
 504             p[0] = [p[1], end]
 505         else:
 506             p[0] = [p[1]]
 507
 508     # testlist: test (',' test)* [',']
 509     # Contains shift/reduce error
 510
 511     def p_testlist(self, p):
 512         """testlist : testlist_multi COMMA
 513                     | testlist_multi """
 514         if len(p) == 2:
 515             p[0] = p[1]
 516         else:
 517             # May need to promote singleton to tuple
 518             if isinstance(p[1], list):
 519                 p[0] = p[1]
 520             else:
 521                 p[0] = [p[1]]
 522         # Convert into a tuple?
 523         if isinstance(p[0], list):
 524             p[0] = ast.Tuple(p[0])
 525
 526     def p_testlist_multi(self, p):
 527         """testlist_multi : testlist_multi COMMA test
 528                           | test"""
 529         if len(p) == 2:
 530             # singleton
 531             p[0] = p[1]
 532         else:
 533             if isinstance(p[1], list):
 534                 p[0] = p[1] + [p[3]]
 535             else:
 536                 # singleton -> tuple
 537                 p[0] = [p[1], p[3]]
 538
 539     # test: or_test ['if' or_test 'else' test] | lambdef
 540     #  as I don't support 'and', 'or', and 'not' this works down to 'comparison'
 541
 542     def p_test(self, p):
 543         "test : comparison"
 544         p[0] = p[1]
 545
 546     # arglist: (argument ',')* (argument [',']| '*' test [',' '**' test]
 547     # | '**' test)
 548     # XXX INCOMPLETE: this doesn't allow the trailing comma
 549
 550     def p_arglist(self, p):
 551         """arglist : arglist COMMA argument
 552                    | argument"""
 553         if len(p) == 4:
 554             p[0] = p[1] + [p[3]]
 555         else:
 556             p[0] = [p[1]]
 557
 558     # argument: test [gen_for] | test '=' test  # Really [keyword '='] test
 559     def p_argument(self, p):
 560         "argument : test"
 561         p[0] = p[1]
 562
 563     def p_error(self, p):
 564         # print "Error!", repr(p)
 565         raise SyntaxError(p)
 566
 567
 568 class GardenSnakeParser(PowerParser):
 569     def __init__(self, lexer=None):
 570         PowerParser.__init__(self)
 571         if lexer is None:
 572             lexer = IndentLexer(debug=0)
 573         self.lexer = lexer
 574         self.tokens = lexer.tokens
 575         self.parser = yacc.yacc(module=self, start="file_input_end",
 576                                 debug=False, write_tables=False)
 577
 578         self.sd = create_pdecode()
 579
 580     def parse(self, code):
 581         # self.lexer.input(code)
 582         result = self.parser.parse(code, lexer=self.lexer, debug=False)
 583         return ast.Module(result)
 584
 585
 586 ###### Code generation ######
 587
 588 #from compiler import misc, syntax, pycodegen
 589
 590 class GardenSnakeCompiler(object):
 591     def __init__(self):
 592         self.parser = GardenSnakeParser()
 593
 594     def compile(self, code, mode="exec", filename="<string>"):
 595         tree = self.parser.parse(code)
 596         print("snake")
 597         pprint(tree)
 598         return tree
 599         #misc.set_filename(filename, tree)
 600         return compile(tree, mode="exec", filename="<string>")
 601         # syntax.check(tree)
 602         gen = pycodegen.ModuleCodeGenerator(tree)
 603         code = gen.getCode()
 604         return code