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