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