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     if node.keywords: # a repeated list-constant, don't optimise
 137         return [node]
 138     return node.args
 139
 140
 141 # identify SelectableInt pattern
 142 def identify_sint_mul_pattern(p):
 143     if not isinstance(p[3], ast.Constant):
 144         return False
 145     if not isinstance(p[1], ast.List):
 146         return False
 147     l = p[1].elts
 148     if len(l) != 1:
 149         return False
 150     elt = l[0]
 151     return isinstance(elt, ast.Constant)
 152
 153
 154 ##########   Parser (tokens -> AST) ######
 155
 156 # also part of Ply
 157 #import yacc
 158
 159 # https://www.mathcs.emory.edu/~valerie/courses/fall10/155/resources/op_precedence.html
 160 # python operator precedence
 161 # Highest precedence at top, lowest at bottom.
 162 # Operators in the same box evaluate left to right.
 163 #
 164 # Operator Description
 165 # ()                                                     Parentheses (grouping)
 166 # f(args...)                                             Function call
 167 # x[index:index]                                         Slicing
 168 # x[index]                                               Subscription
 169 # x.attribute                                            Attribute reference
 170 # **                                                     Exponentiation
 171 # ~x                                                     Bitwise not
 172 # +x, -x                                                 Positive, negative
 173 # *, /, %                                                mul, div, remainder
 174 # +, -                                                   Addition, subtraction
 175 # <<, >>                                                 Bitwise shifts
 176 # &                                                      Bitwise AND
 177 # ^                                                      Bitwise XOR
 178 # |                                                      Bitwise OR
 179 # in, not in, is, is not, <, <=,  >,  >=, <>, !=, ==     comp, membership, ident
 180 # not x                                                  Boolean NOT
 181 # and                                                    Boolean AND
 182 # or                                                     Boolean OR
 183 # lambda                                                 Lambda expression
 184
 185 class PowerParser:
 186
 187     precedence = (
 188         ("left", "EQ", "GT", "LT", "LE", "GE", "LTU", "GTU"),
 189         ("left", "BITOR"),
 190         ("left", "BITAND"),
 191         ("left", "PLUS", "MINUS"),
 192         ("left", "MULT", "DIV", "MOD"),
 193         ("left", "INVERT"),
 194     )
 195
 196     def __init__(self):
 197         self.gprs = {}
 198         for rname in ['RA', 'RB', 'RC', 'RT', 'RS']:
 199             self.gprs[rname] = None
 200         self.read_regs = []
 201         self.uninit_regs = []
 202         self.write_regs = []
 203
 204     # The grammar comments come from Python's Grammar/Grammar file
 205
 206     # NB: compound_stmt in single_input is followed by extra NEWLINE!
 207     # file_input: (NEWLINE | stmt)* ENDMARKER
 208
 209     def p_file_input_end(self, p):
 210         """file_input_end : file_input ENDMARKER"""
 211         print("end", p[1])
 212         p[0] = p[1]
 213
 214     def p_file_input(self, p):
 215         """file_input : file_input NEWLINE
 216                       | file_input stmt
 217                       | NEWLINE
 218                       | stmt"""
 219         if isinstance(p[len(p)-1], str):
 220             if len(p) == 3:
 221                 p[0] = p[1]
 222             else:
 223                 p[0] = []  # p == 2 --> only a blank line
 224         else:
 225             if len(p) == 3:
 226                 p[0] = p[1] + p[2]
 227             else:
 228                 p[0] = p[1]
 229
 230     # funcdef: [decorators] 'def' NAME parameters ':' suite
 231     # ignoring decorators
 232
 233     def p_funcdef(self, p):
 234         "funcdef : DEF NAME parameters COLON suite"
 235         p[0] = ast.FunctionDef(p[2], p[3], p[5], ())
 236
 237     # parameters: '(' [varargslist] ')'
 238     def p_parameters(self, p):
 239         """parameters : LPAR RPAR
 240                       | LPAR varargslist RPAR"""
 241         if len(p) == 3:
 242             args = []
 243         else:
 244             args = p[2]
 245         p[0] = ast.arguments(args=args, vararg=None, kwarg=None, defaults=[])
 246
 247     # varargslist: (fpdef ['=' test] ',')* ('*' NAME [',' '**' NAME] |
 248     # '**' NAME) |
 249     # highly simplified
 250
 251     def p_varargslist(self, p):
 252         """varargslist : varargslist COMMA NAME
 253                        | NAME"""
 254         if len(p) == 4:
 255             p[0] = p[1] + p[3]
 256         else:
 257             p[0] = [p[1]]
 258
 259     # stmt: simple_stmt | compound_stmt
 260     def p_stmt_simple(self, p):
 261         """stmt : simple_stmt"""
 262         # simple_stmt is a list
 263         p[0] = p[1]
 264
 265     def p_stmt_compound(self, p):
 266         """stmt : compound_stmt"""
 267         p[0] = [p[1]]
 268
 269     # simple_stmt: small_stmt (';' small_stmt)* [';'] NEWLINE
 270     def p_simple_stmt(self, p):
 271         """simple_stmt : small_stmts NEWLINE
 272                        | small_stmts SEMICOLON NEWLINE"""
 273         p[0] = p[1]
 274
 275     def p_small_stmts(self, p):
 276         """small_stmts : small_stmts SEMICOLON small_stmt
 277                        | small_stmt"""
 278         if len(p) == 4:
 279             p[0] = p[1] + [p[3]]
 280         else:
 281             p[0] = [p[1]]
 282
 283     # small_stmt: expr_stmt | print_stmt  | del_stmt | pass_stmt | flow_stmt |
 284     #    import_stmt | global_stmt | exec_stmt | assert_stmt
 285     def p_small_stmt(self, p):
 286         """small_stmt : flow_stmt
 287                       | break_stmt
 288                       | expr_stmt"""
 289         if isinstance(p[1], ast.Call):
 290             p[0] = ast.Expr(p[1])
 291         else:
 292             p[0] = p[1]
 293
 294     # expr_stmt: testlist (augassign (yield_expr|testlist) |
 295     #                      ('=' (yield_expr|testlist))*)
 296     # augassign: ('+=' | '-=' | '*=' | '/=' | '%=' | '&=' | '|=' | '^=' |
 297     #             '<<=' | '>>=' | '**=' | '//=')
 298     def p_expr_stmt(self, p):
 299         """expr_stmt : testlist ASSIGN testlist
 300                      | testlist """
 301         print("expr_stmt", p)
 302         if len(p) == 2:
 303             # a list of expressions
 304             #p[0] = ast.Discard(p[1])
 305             p[0] = p[1]
 306         else:
 307             if isinstance(p[1], ast.Name):
 308                 name = p[1].id
 309             elif isinstance(p[1], ast.Subscript):
 310                 name = p[1].value.id
 311                 if name in self.gprs:
 312                     # add to list of uninitialised
 313                     self.uninit_regs.append(name)
 314             print("expr assign", name, p[1])
 315             if name in self.gprs:
 316                 self.write_regs.append(name)  # add to list of regs to write
 317             p[0] = Assign(p[1], p[3])
 318
 319     def p_flow_stmt(self, p):
 320         "flow_stmt : return_stmt"
 321         p[0] = p[1]
 322
 323     # return_stmt: 'return' [testlist]
 324     def p_return_stmt(self, p):
 325         "return_stmt : RETURN testlist"
 326         p[0] = ast.Return(p[2])
 327
 328     def p_compound_stmt(self, p):
 329         """compound_stmt : if_stmt
 330                          | while_stmt
 331                          | for_stmt
 332                          | funcdef
 333         """
 334         p[0] = p[1]
 335
 336     def p_break_stmt(self, p):
 337         """break_stmt : BREAK
 338         """
 339         p[0] = ast.Break()
 340
 341     def p_for_stmt(self, p):
 342         """for_stmt : FOR test EQ test TO test COLON suite
 343         """
 344         p[0] = ast.While(p[2], p[4], [])
 345         # auto-add-one (sigh) due to python range
 346         start = p[4]
 347         end = ast.BinOp(p[6], ast.Add(), ast.Constant(1))
 348         it = ast.Call(ast.Name("range"), [start, end], [])
 349         p[0] = ast.For(p[2], it, p[8], [])
 350
 351     def p_while_stmt(self, p):
 352         """while_stmt : DO WHILE test COLON suite ELSE COLON suite
 353                       | DO WHILE test COLON suite
 354         """
 355         if len(p) == 6:
 356             p[0] = ast.While(p[3], p[5], [])
 357         else:
 358             p[0] = ast.While(p[3], p[5], p[8])
 359
 360     def p_if_stmt(self, p):
 361         """if_stmt : IF test COLON suite ELSE COLON if_stmt
 362                    | IF test COLON suite ELSE COLON suite
 363                    | IF test COLON suite
 364         """
 365         if len(p) == 8 and isinstance(p[7], ast.If):
 366             p[0] = ast.If(p[2], p[4], [p[7]])
 367         elif len(p) == 5:
 368             p[0] = ast.If(p[2], p[4], [])
 369         else:
 370             p[0] = ast.If(p[2], p[4], p[7])
 371
 372     def p_suite(self, p):
 373         """suite : simple_stmt
 374                  | NEWLINE INDENT stmts DEDENT"""
 375         if len(p) == 2:
 376             p[0] = p[1]
 377         else:
 378             p[0] = p[3]
 379
 380     def p_stmts(self, p):
 381         """stmts : stmts stmt
 382                  | stmt"""
 383         if len(p) == 3:
 384             p[0] = p[1] + p[2]
 385         else:
 386             p[0] = p[1]
 387
 388     def p_comparison(self, p):
 389         """comparison : comparison PLUS comparison
 390                       | comparison MINUS comparison
 391                       | comparison MULT comparison
 392                       | comparison DIV comparison
 393                       | comparison MOD comparison
 394                       | comparison EQ comparison
 395                       | comparison LE comparison
 396                       | comparison GE comparison
 397                       | comparison LTU comparison
 398                       | comparison GTU comparison
 399                       | comparison LT comparison
 400                       | comparison GT comparison
 401                       | comparison BITOR comparison
 402                       | comparison BITAND comparison
 403                       | PLUS comparison
 404                       | comparison MINUS
 405                       | INVERT comparison
 406                       | comparison APPEND comparison
 407                       | power"""
 408         if len(p) == 4:
 409             print(list(p))
 410             if p[2] == '<u':
 411                 p[0] = ast.Call(ast.Name("ltu"), (p[1], p[3]), [])
 412             elif p[2] == '>u':
 413                 p[0] = ast.Call(ast.Name("gtu"), (p[1], p[3]), [])
 414             elif p[2] == '||':
 415                 l = check_concat(p[1]) + check_concat(p[3])
 416                 p[0] = ast.Call(ast.Name("concat"), l, [])
 417             elif p[2] in ['<', '>', '=', '<=', '>=']:
 418                 p[0] = binary_ops[p[2]]((p[1], p[3]))
 419             elif identify_sint_mul_pattern(p):
 420                 keywords=[ast.keyword(arg='repeat', value=p[3])]
 421                 l = p[1].elts
 422                 p[0] = ast.Call(ast.Name("concat"), l, keywords)
 423             else:
 424                 p[0] = ast.BinOp(p[1], binary_ops[p[2]], p[3])
 425         elif len(p) == 3:
 426             if isinstance(p[2], str) and p[2] == '-':
 427                 p[0] = ast.UnaryOp(unary_ops[p[2]], p[1])
 428             else:
 429                 p[0] = ast.UnaryOp(unary_ops[p[1]], p[2])
 430         else:
 431             p[0] = p[1]
 432
 433     # power: atom trailer* ['**' factor]
 434     # trailers enables function calls (and subscripts).
 435     # I only allow one level of calls
 436     # so this is 'trailer'
 437     def p_power(self, p):
 438         """power : atom
 439                  | atom trailer"""
 440         if len(p) == 2:
 441             p[0] = p[1]
 442         else:
 443             if p[2][0] == "CALL":
 444                 #p[0] = ast.Expr(ast.Call(p[1], p[2][1], []))
 445                 p[0] = ast.Call(p[1], p[2][1], [])
 446                 # if p[1].id == 'print':
 447                 #    p[0] = ast.Printnl(ast.Tuple(p[2][1]), None, None)
 448                 # else:
 449                 #    p[0] = ast.CallFunc(p[1], p[2][1], None, None)
 450             else:
 451                 print("subscript atom", p[2][1])
 452                 #raise AssertionError("not implemented %s" % p[2][0])
 453                 subs = p[2][1]
 454                 if len(subs) == 1:
 455                     idx = subs[0]
 456                 else:
 457                     idx = ast.Slice(subs[0], subs[1], None)
 458                 p[0] = ast.Subscript(p[1], idx, ast.Load())
 459
 460     def p_atom_name(self, p):
 461         """atom : NAME"""
 462         p[0] = ast.Name(id=p[1], ctx=ast.Load())
 463
 464     def p_atom_number(self, p):
 465         """atom : BINARY
 466                 | NUMBER
 467                 | STRING"""
 468         p[0] = ast.Constant(p[1])
 469
 470     # '[' [listmaker] ']' |
 471
 472     def p_atom_listmaker(self, p):
 473         """atom : LBRACK listmaker RBRACK"""
 474         p[0] = p[2]
 475
 476     def p_listmaker(self, p):
 477         """listmaker : test COMMA listmaker
 478                      | test
 479         """
 480         if len(p) == 2:
 481             p[0] = ast.List([p[1]])
 482         else:
 483             p[0] = ast.List([p[1]] + p[3].nodes)
 484
 485     def p_atom_tuple(self, p):
 486         """atom : LPAR testlist RPAR"""
 487         print("tuple", p[2])
 488         if isinstance(p[2], ast.Name):
 489             print("tuple name", p[2].id)
 490             if p[2].id in self.gprs:
 491                 self.read_regs.append(p[2].id)  # add to list of regs to read
 492                 #p[0] = ast.Subscript(ast.Name("GPR"), ast.Str(p[2].id))
 493                 # return
 494         p[0] = p[2]
 495
 496     # trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
 497     def p_trailer(self, p):
 498         """trailer : trailer_arglist
 499                    | trailer_subscript
 500         """
 501         p[0] = p[1]
 502
 503     def p_trailer_arglist(self, p):
 504         "trailer_arglist : LPAR arglist RPAR"
 505         p[0] = ("CALL", p[2])
 506
 507     def p_trailer_subscript(self, p):
 508         "trailer_subscript : LBRACK subscript RBRACK"
 509         p[0] = ("SUBS", p[2])
 510
 511     # subscript: '.' '.' '.' | test | [test] ':' [test]
 512
 513     def p_subscript(self, p):
 514         """subscript : test COLON test
 515                      | test
 516         """
 517         if len(p) == 4:
 518             # add one to end
 519             if isinstance(p[3], ast.Constant):
 520                 end = ast.Constant(p[3].value+1)
 521             else:
 522                 end = ast.BinOp(p[3], ast.Add(), ast.Constant(1))
 523             p[0] = [p[1], end]
 524         else:
 525             p[0] = [p[1]]
 526
 527     # testlist: test (',' test)* [',']
 528     # Contains shift/reduce error
 529
 530     def p_testlist(self, p):
 531         """testlist : testlist_multi COMMA
 532                     | testlist_multi """
 533         if len(p) == 2:
 534             p[0] = p[1]
 535         else:
 536             # May need to promote singleton to tuple
 537             if isinstance(p[1], list):
 538                 p[0] = p[1]
 539             else:
 540                 p[0] = [p[1]]
 541         # Convert into a tuple?
 542         if isinstance(p[0], list):
 543             p[0] = ast.Tuple(p[0])
 544
 545     def p_testlist_multi(self, p):
 546         """testlist_multi : testlist_multi COMMA test
 547                           | test"""
 548         if len(p) == 2:
 549             # singleton
 550             p[0] = p[1]
 551         else:
 552             if isinstance(p[1], list):
 553                 p[0] = p[1] + [p[3]]
 554             else:
 555                 # singleton -> tuple
 556                 p[0] = [p[1], p[3]]
 557
 558     # test: or_test ['if' or_test 'else' test] | lambdef
 559     #  as I don't support 'and', 'or', and 'not' this works down to 'comparison'
 560
 561     def p_test(self, p):
 562         "test : comparison"
 563         p[0] = p[1]
 564
 565     # arglist: (argument ',')* (argument [',']| '*' test [',' '**' test]
 566     # | '**' test)
 567     # XXX INCOMPLETE: this doesn't allow the trailing comma
 568
 569     def p_arglist(self, p):
 570         """arglist : arglist COMMA argument
 571                    | argument"""
 572         if len(p) == 4:
 573             p[0] = p[1] + [p[3]]
 574         else:
 575             p[0] = [p[1]]
 576
 577     # argument: test [gen_for] | test '=' test  # Really [keyword '='] test
 578     def p_argument(self, p):
 579         "argument : test"
 580         p[0] = p[1]
 581
 582     def p_error(self, p):
 583         # print "Error!", repr(p)
 584         raise SyntaxError(p)
 585
 586
 587 class GardenSnakeParser(PowerParser):
 588     def __init__(self, lexer=None):
 589         PowerParser.__init__(self)
 590         if lexer is None:
 591             lexer = IndentLexer(debug=0)
 592         self.lexer = lexer
 593         self.tokens = lexer.tokens
 594         self.parser = yacc.yacc(module=self, start="file_input_end",
 595                                 debug=False, write_tables=False)
 596
 597         self.sd = create_pdecode()
 598
 599     def parse(self, code):
 600         # self.lexer.input(code)
 601         result = self.parser.parse(code, lexer=self.lexer, debug=False)
 602         return ast.Module(result)
 603
 604
 605 ###### Code generation ######
 606
 607 #from compiler import misc, syntax, pycodegen
 608
 609 class GardenSnakeCompiler(object):
 610     def __init__(self):
 611         self.parser = GardenSnakeParser()
 612
 613     def compile(self, code, mode="exec", filename="<string>"):
 614         tree = self.parser.parse(code)
 615         print("snake")
 616         pprint(tree)
 617         return tree
 618         #misc.set_filename(filename, tree)
 619         return compile(tree, mode="exec", filename="<string>")
 620         # syntax.check(tree)
 621         gen = pycodegen.ModuleCodeGenerator(tree)
 622         code = gen.getCode()
 623         return code