1 # Based on GardenSnake - a parser generator demonstration program
2 # GardenSnake was released into the Public Domain by Andrew Dalke.
4 # Portions of this work are derived from Python's Grammar definition
5 # and may be covered under the Python copyright and license
7 # Andrew Dalke / Dalke Scientific Software, LLC
8 # 30 August 2006 / Cape Town, South Africa
10 # Modifications for inclusion in PLY distribution
11 from pprint
import pprint
12 from ply
import lex
, yacc
15 from soc
.decoder
.power_decoder
import create_pdecode
16 from soc
.decoder
.pseudo
.lexer
import IndentLexer
18 # I use the Python AST
19 #from compiler import ast
23 def Assign(left
, right
):
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
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...
46 if isinstance(ls
, ast
.Slice
):
47 lower
, upper
, step
= ls
.lower
, ls
.upper
, ls
.step
48 print ("slice assign", lower
, upper
, step
)
50 ls
= (lower
, upper
, None)
52 ls
= (lower
, upper
, step
)
54 return ast
.Call(ast
.Name("selectassign"),
55 [left
.value
, ls
, right
], [])
58 raise SyntaxError("Can't do that yet")
61 ## I implemented INDENT / DEDENT generation as a post-processing filter
63 # The original lex token stream contains WS and NEWLINE characters.
64 # WS will only occur before any other tokens on a line.
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.
73 ## No using Python's approach because Ply supports precedence
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'
81 def make_le_compare(arg
):
83 return ast
.Compare(left
, [ast
.LtE()], [right
])
84 def make_ge_compare(arg
):
86 return ast
.Compare(left
, [ast
.GtE()], [right
])
87 def make_lt_compare(arg
):
89 return ast
.Compare(left
, [ast
.Lt()], [right
])
90 def make_gt_compare(arg
):
92 return ast
.Compare(left
, [ast
.Gt()], [right
])
93 def make_eq_compare(arg
):
95 return ast
.Compare(left
, [ast
.Eq()], [right
])
104 "<=": make_le_compare
,
105 ">=": make_ge_compare
,
106 "<": make_lt_compare
,
107 ">": make_gt_compare
,
108 "=": make_eq_compare
,
116 def check_concat(node
): # checks if the comparison is already a concat
117 print ("check concat", node
)
118 if not isinstance(node
, ast
.Call
):
120 print ("func", node
.func
.id)
121 if node
.func
.id != 'concat':
126 ########## Parser (tokens -> AST) ######
134 ("left", "BITOR", "BITAND"),
135 ("left", "EQ", "GT", "LT", "LE", "GE", "LTU", "GTU"),
136 ("left", "PLUS", "MINUS"),
137 ("left", "MULT", "DIV"),
143 for rname
in ['RA', 'RB', 'RC', 'RT', 'RS']:
144 self
.gprs
[rname
] = None
146 self
.uninit_regs
= []
149 # The grammar comments come from Python's Grammar/Grammar file
151 ## NB: compound_stmt in single_input is followed by extra NEWLINE!
152 # file_input: (NEWLINE | stmt)* ENDMARKER
154 def p_file_input_end(self
, p
):
155 """file_input_end : file_input ENDMARKER"""
159 def p_file_input(self
, p
):
160 """file_input : file_input NEWLINE
164 if isinstance(p
[len(p
)-1], str):
168 p
[0] = [] # p == 2 --> only a blank line
176 # funcdef: [decorators] 'def' NAME parameters ':' suite
177 # ignoring decorators
178 def p_funcdef(self
, p
):
179 "funcdef : DEF NAME parameters COLON suite"
180 p
[0] = ast
.FunctionDef(p
[2], p
[3], p
[5], ())
182 # parameters: '(' [varargslist] ')'
183 def p_parameters(self
, p
):
184 """parameters : LPAR RPAR
185 | LPAR varargslist RPAR"""
190 p
[0] = ast
.arguments(args
=args
, vararg
=None, kwarg
=None, defaults
=[])
193 # varargslist: (fpdef ['=' test] ',')* ('*' NAME [',' '**' NAME] |
196 def p_varargslist(self
, p
):
197 """varargslist : varargslist COMMA NAME
204 # stmt: simple_stmt | compound_stmt
205 def p_stmt_simple(self
, p
):
206 """stmt : simple_stmt"""
207 # simple_stmt is a list
210 def p_stmt_compound(self
, p
):
211 """stmt : compound_stmt"""
214 # simple_stmt: small_stmt (';' small_stmt)* [';'] NEWLINE
215 def p_simple_stmt(self
, p
):
216 """simple_stmt : small_stmts NEWLINE
217 | small_stmts SEMICOLON NEWLINE"""
220 def p_small_stmts(self
, p
):
221 """small_stmts : small_stmts SEMICOLON small_stmt
228 # small_stmt: expr_stmt | print_stmt | del_stmt | pass_stmt | flow_stmt |
229 # import_stmt | global_stmt | exec_stmt | assert_stmt
230 def p_small_stmt(self
, p
):
231 """small_stmt : flow_stmt
234 if isinstance(p
[1], ast
.Call
):
235 p
[0] = ast
.Expr(p
[1])
239 # expr_stmt: testlist (augassign (yield_expr|testlist) |
240 # ('=' (yield_expr|testlist))*)
241 # augassign: ('+=' | '-=' | '*=' | '/=' | '%=' | '&=' | '|=' | '^=' |
242 # '<<=' | '>>=' | '**=' | '//=')
243 def p_expr_stmt(self
, p
):
244 """expr_stmt : testlist ASSIGN testlist
246 print ("expr_stmt", p
)
248 # a list of expressions
249 #p[0] = ast.Discard(p[1])
252 if isinstance(p
[1], ast
.Name
):
254 elif isinstance(p
[1], ast
.Subscript
):
256 if name
in self
.gprs
:
257 self
.uninit_regs
.append(name
) # add to list of uninitialised
258 print ("expr assign", name
, p
[1])
259 if name
in self
.gprs
:
260 self
.write_regs
.append(name
) # add to list of regs to write
261 p
[0] = Assign(p
[1], p
[3])
263 def p_flow_stmt(self
, p
):
264 "flow_stmt : return_stmt"
267 # return_stmt: 'return' [testlist]
268 def p_return_stmt(self
, p
):
269 "return_stmt : RETURN testlist"
270 p
[0] = ast
.Return(p
[2])
273 def p_compound_stmt(self
, p
):
274 """compound_stmt : if_stmt
281 def p_break_stmt(self
, p
):
282 """break_stmt : BREAK
286 def p_for_stmt(self
, p
):
287 """for_stmt : FOR test EQ test TO test COLON suite
289 p
[0] = ast
.While(p
[2], p
[4], [])
290 # auto-add-one (sigh) due to python range
292 end
= ast
.BinOp(p
[6], ast
.Add(), ast
.Constant(1))
293 it
= ast
.Call(ast
.Name("range"), [start
, end
], [])
294 p
[0] = ast
.For(p
[2], it
, p
[8], [])
296 def p_while_stmt(self
, p
):
297 """while_stmt : DO WHILE test COLON suite ELSE COLON suite
298 | DO WHILE test COLON suite
301 p
[0] = ast
.While(p
[3], p
[5], [])
303 p
[0] = ast
.While(p
[3], p
[5], p
[8])
305 def p_if_stmt(self
, p
):
306 """if_stmt : IF test COLON suite ELSE COLON if_stmt
307 | IF test COLON suite ELSE COLON suite
308 | IF test COLON suite
310 if len(p
) == 8 and isinstance(p
[7], ast
.If
):
311 p
[0] = ast
.If(p
[2], p
[4], [p
[7]])
313 p
[0] = ast
.If(p
[2], p
[4], [])
315 p
[0] = ast
.If(p
[2], p
[4], p
[7])
317 def p_suite(self
, p
):
318 """suite : simple_stmt
319 | NEWLINE INDENT stmts DEDENT"""
326 def p_stmts(self
, p
):
327 """stmts : stmts stmt
334 def p_comparison(self
, p
):
335 """comparison : comparison PLUS comparison
336 | comparison MINUS comparison
337 | comparison MULT comparison
338 | comparison DIV comparison
339 | comparison EQ comparison
340 | comparison LE comparison
341 | comparison GE comparison
342 | comparison LTU comparison
343 | comparison GTU comparison
344 | comparison LT comparison
345 | comparison GT comparison
346 | comparison BITOR comparison
347 | comparison BITAND comparison
351 | comparison APPEND comparison
356 p
[0] = ast
.Call(ast
.Name("ltu"), (p
[1], p
[3]), [])
358 p
[0] = ast
.Call(ast
.Name("gtu"), (p
[1], p
[3]), [])
360 l
= check_concat(p
[1]) + check_concat(p
[3])
361 p
[0] = ast
.Call(ast
.Name("concat"), l
, [])
362 elif p
[2] in ['<', '>', '=', '<=', '>=']:
363 p
[0] = binary_ops
[p
[2]]((p
[1],p
[3]))
365 p
[0] = ast
.BinOp(p
[1], binary_ops
[p
[2]], p
[3])
367 p
[0] = ast
.UnaryOp(unary_ops
[p
[1]], p
[2])
371 # power: atom trailer* ['**' factor]
372 # trailers enables function calls (and subscripts).
373 # I only allow one level of calls
374 # so this is 'trailer'
375 def p_power(self
, p
):
381 if p
[2][0] == "CALL":
382 #p[0] = ast.Expr(ast.Call(p[1], p[2][1], []))
383 p
[0] = ast
.Call(p
[1], p
[2][1], [])
384 #if p[1].id == 'print':
385 # p[0] = ast.Printnl(ast.Tuple(p[2][1]), None, None)
387 # p[0] = ast.CallFunc(p[1], p[2][1], None, None)
389 print ("subscript atom", p
[2][1])
390 #raise AssertionError("not implemented %s" % p[2][0])
395 idx
= ast
.Slice(subs
[0], subs
[1], None)
396 p
[0] = ast
.Subscript(p
[1], idx
)
398 def p_atom_name(self
, p
):
400 p
[0] = ast
.Name(p
[1], ctx
=ast
.Load())
402 def p_atom_number(self
, p
):
406 p
[0] = ast
.Constant(p
[1])
408 #'[' [listmaker] ']' |
410 def p_atom_listmaker(self
, p
):
411 """atom : LBRACK listmaker RBRACK"""
414 def p_listmaker(self
, p
):
415 """listmaker : test COMMA listmaker
419 p
[0] = ast
.List([p
[1]])
421 p
[0] = ast
.List([p
[1]] + p
[3].nodes
)
423 def p_atom_tuple(self
, p
):
424 """atom : LPAR testlist RPAR"""
425 print ("tuple", p
[2])
426 if isinstance(p
[2], ast
.Name
):
427 print ("tuple name", p
[2].id)
428 if p
[2].id in self
.gprs
:
429 self
.read_regs
.append(p
[2].id) # add to list of regs to read
430 #p[0] = ast.Subscript(ast.Name("GPR"), ast.Str(p[2].id))
434 # trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
435 def p_trailer(self
, p
):
436 """trailer : trailer_arglist
441 def p_trailer_arglist(self
, p
):
442 "trailer_arglist : LPAR arglist RPAR"
443 p
[0] = ("CALL", p
[2])
445 def p_trailer_subscript(self
, p
):
446 "trailer_subscript : LBRACK subscript RBRACK"
447 p
[0] = ("SUBS", p
[2])
449 #subscript: '.' '.' '.' | test | [test] ':' [test]
451 def p_subscript(self
, p
):
452 """subscript : test COLON test
457 if isinstance(p
[3], ast
.Constant
):
458 end
= ast
.Constant(p
[3].value
+1)
460 end
= ast
.BinOp(p
[3], ast
.Add(), ast
.Constant(1))
466 # testlist: test (',' test)* [',']
467 # Contains shift/reduce error
468 def p_testlist(self
, p
):
469 """testlist : testlist_multi COMMA
474 # May need to promote singleton to tuple
475 if isinstance(p
[1], list):
479 # Convert into a tuple?
480 if isinstance(p
[0], list):
481 p
[0] = ast
.Tuple(p
[0])
483 def p_testlist_multi(self
, p
):
484 """testlist_multi : testlist_multi COMMA test
490 if isinstance(p
[1], list):
497 # test: or_test ['if' or_test 'else' test] | lambdef
498 # as I don't support 'and', 'or', and 'not' this works down to 'comparison'
505 # arglist: (argument ',')* (argument [',']| '*' test [',' '**' test]
507 # XXX INCOMPLETE: this doesn't allow the trailing comma
508 def p_arglist(self
, p
):
509 """arglist : arglist COMMA argument
516 # argument: test [gen_for] | test '=' test # Really [keyword '='] test
517 def p_argument(self
, p
):
521 def p_error(self
, p
):
522 #print "Error!", repr(p)
526 class GardenSnakeParser(PowerParser
):
527 def __init__(self
, lexer
= None):
528 PowerParser
.__init
__(self
)
530 lexer
= IndentLexer(debug
=0)
532 self
.tokens
= lexer
.tokens
533 self
.parser
= yacc
.yacc(module
=self
, start
="file_input_end",
534 debug
=False, write_tables
=False)
536 self
.sd
= create_pdecode()
538 def parse(self
, code
):
539 #self.lexer.input(code)
540 result
= self
.parser
.parse(code
, lexer
=self
.lexer
, debug
=False)
541 return ast
.Module(result
)
544 ###### Code generation ######
546 #from compiler import misc, syntax, pycodegen
548 class GardenSnakeCompiler(object):
550 self
.parser
= GardenSnakeParser()
552 def compile(self
, code
, mode
="exec", filename
="<string>"):
553 tree
= self
.parser
.parse(code
)
557 #misc.set_filename(filename, tree)
558 return compile(tree
, mode
="exec", filename
="<string>")
560 gen
= pycodegen
.ModuleCodeGenerator(tree
)