from soc.decoder.power_decoder import create_pdecode
from soc.decoder.pseudo.lexer import IndentLexer
+from soc.decoder.orderedset import OrderedSet
# I use the Python AST
#from compiler import ast
import ast
# Helper function
-def Assign(left, right):
+
+
+def Assign(left, right, iea_mode):
names = []
- print ("Assign", left, right)
+ print("Assign", left, right)
if isinstance(left, ast.Name):
# Single assignment on left
# XXX when doing IntClass, which will have an "eq" function,
ls = left.slice
if isinstance(ls, ast.Slice):
lower, upper, step = ls.lower, ls.upper, ls.step
- print ("slice assign", lower, upper, step)
+ print("slice assign", lower, upper, step)
if step is None:
ls = (lower, upper, None)
else:
return ast.Call(ast.Name("selectassign"),
[left.value, ls, right], [])
else:
- print ("Assign fail")
+ print("Assign fail")
raise SyntaxError("Can't do that yet")
-## I implemented INDENT / DEDENT generation as a post-processing filter
+# I implemented INDENT / DEDENT generation as a post-processing filter
# The original lex token stream contains WS and NEWLINE characters.
# WS will only occur before any other tokens on a line.
# see if the new line has changed indication level.
-## No using Python's approach because Ply supports precedence
+# No using Python's approach because Ply supports precedence
# comparison: expr (comp_op expr)*
# arith_expr: term (('+'|'-') term)*
def make_le_compare(arg):
(left, right) = arg
return ast.Compare(left, [ast.LtE()], [right])
+
+
def make_ge_compare(arg):
(left, right) = arg
return ast.Compare(left, [ast.GtE()], [right])
+
+
def make_lt_compare(arg):
(left, right) = arg
return ast.Compare(left, [ast.Lt()], [right])
+
+
def make_gt_compare(arg):
(left, right) = arg
return ast.Compare(left, [ast.Gt()], [right])
+
+
def make_eq_compare(arg):
(left, right) = arg
return ast.Compare(left, [ast.Eq()], [right])
+
binary_ops = {
+ "^": ast.BitXor(),
"&": ast.BitAnd(),
"|": ast.BitOr(),
"+": ast.Add(),
"-": ast.Sub(),
"*": ast.Mult(),
"/": ast.Div(),
+ "%": ast.Mod(),
"<=": make_le_compare,
">=": make_ge_compare,
"<": make_lt_compare,
"+": ast.UAdd(),
"-": ast.USub(),
"¬": ast.Invert(),
- }
+}
-def check_concat(node): # checks if the comparison is already a concat
- print ("check concat", node)
+
+def check_concat(node): # checks if the comparison is already a concat
+ print("check concat", node)
if not isinstance(node, ast.Call):
return [node]
- print ("func", node.func.id)
+ print("func", node.func.id)
if node.func.id != 'concat':
return [node]
+ if node.keywords: # a repeated list-constant, don't optimise
+ return [node]
return node.args
+# identify SelectableInt pattern
+def identify_sint_mul_pattern(p):
+ if not isinstance(p[3], ast.Constant):
+ return False
+ if not isinstance(p[1], ast.List):
+ return False
+ l = p[1].elts
+ if len(l) != 1:
+ return False
+ elt = l[0]
+ return isinstance(elt, ast.Constant)
+
+def apply_trailer(atom, trailer):
+ if trailer[0] == "TLIST":
+ # assume depth of one
+ atom = apply_trailer(atom, trailer[1])
+ trailer = trailer[2]
+ if trailer[0] == "CALL":
+ #p[0] = ast.Expr(ast.Call(p[1], p[2][1], []))
+ return ast.Call(atom, trailer[1], [])
+ # if p[1].id == 'print':
+ # p[0] = ast.Printnl(ast.Tuple(p[2][1]), None, None)
+ # else:
+ # p[0] = ast.CallFunc(p[1], p[2][1], None, None)
+ else:
+ print("subscript atom", trailer[1])
+ #raise AssertionError("not implemented %s" % p[2][0])
+ subs = trailer[1]
+ if len(subs) == 1:
+ idx = subs[0]
+ else:
+ idx = ast.Slice(subs[0], subs[1], None)
+ return ast.Subscript(atom, idx, ast.Load())
+
########## Parser (tokens -> AST) ######
# also part of Ply
#import yacc
+# https://www.mathcs.emory.edu/~valerie/courses/fall10/155/resources/op_precedence.html
+# python operator precedence
+# Highest precedence at top, lowest at bottom.
+# Operators in the same box evaluate left to right.
+#
+# Operator Description
+# () Parentheses (grouping)
+# f(args...) Function call
+# x[index:index] Slicing
+# x[index] Subscription
+# x.attribute Attribute reference
+# ** Exponentiation
+# ~x Bitwise not
+# +x, -x Positive, negative
+# *, /, % mul, div, remainder
+# +, - Addition, subtraction
+# <<, >> Bitwise shifts
+# & Bitwise AND
+# ^ Bitwise XOR
+# | Bitwise OR
+# in, not in, is, is not, <, <=, >, >=, <>, !=, == comp, membership, ident
+# not x Boolean NOT
+# and Boolean AND
+# or Boolean OR
+# lambda Lambda expression
+
class PowerParser:
precedence = (
- ("left", "BITOR", "BITAND"),
("left", "EQ", "GT", "LT", "LE", "GE", "LTU", "GTU"),
+ ("left", "BITOR"),
+ ("left", "BITXOR"),
+ ("left", "BITAND"),
("left", "PLUS", "MINUS"),
- ("left", "MULT", "DIV"),
+ ("left", "MULT", "DIV", "MOD"),
("left", "INVERT"),
- )
+ )
def __init__(self):
self.gprs = {}
for rname in ['RA', 'RB', 'RC', 'RT', 'RS']:
self.gprs[rname] = None
- self.read_regs = []
- self.uninit_regs = []
- self.write_regs = []
+ self.read_regs = OrderedSet()
+ self.uninit_regs = OrderedSet()
+ self.write_regs = OrderedSet()
# The grammar comments come from Python's Grammar/Grammar file
- ## NB: compound_stmt in single_input is followed by extra NEWLINE!
+ # NB: compound_stmt in single_input is followed by extra NEWLINE!
# file_input: (NEWLINE | stmt)* ENDMARKER
def p_file_input_end(self, p):
"""file_input_end : file_input ENDMARKER"""
- print ("end", p[1])
+ print("end", p[1])
p[0] = p[1]
def p_file_input(self, p):
if len(p) == 3:
p[0] = p[1]
else:
- p[0] = [] # p == 2 --> only a blank line
+ p[0] = [] # p == 2 --> only a blank line
else:
if len(p) == 3:
p[0] = p[1] + p[2]
else:
p[0] = p[1]
-
# funcdef: [decorators] 'def' NAME parameters ':' suite
# ignoring decorators
+
def p_funcdef(self, p):
"funcdef : DEF NAME parameters COLON suite"
p[0] = ast.FunctionDef(p[2], p[3], p[5], ())
"""parameters : LPAR RPAR
| LPAR varargslist RPAR"""
if len(p) == 3:
- args=[]
+ args = []
else:
args = p[2]
p[0] = ast.arguments(args=args, vararg=None, kwarg=None, defaults=[])
-
# varargslist: (fpdef ['=' test] ',')* ('*' NAME [',' '**' NAME] |
# '**' NAME) |
# highly simplified
+
def p_varargslist(self, p):
"""varargslist : varargslist COMMA NAME
| NAME"""
# augassign: ('+=' | '-=' | '*=' | '/=' | '%=' | '&=' | '|=' | '^=' |
# '<<=' | '>>=' | '**=' | '//=')
def p_expr_stmt(self, p):
- """expr_stmt : testlist ASSIGN testlist
+ """expr_stmt : testlist ASSIGNEA testlist
+ | testlist ASSIGN testlist
| testlist """
- print ("expr_stmt", p)
+ print("expr_stmt", p)
if len(p) == 2:
# a list of expressions
#p[0] = ast.Discard(p[1])
p[0] = p[1]
else:
+ iea_mode = p[2] == '<-iea'
+ name = None
if isinstance(p[1], ast.Name):
name = p[1].id
elif isinstance(p[1], ast.Subscript):
name = p[1].value.id
if name in self.gprs:
- self.uninit_regs.append(name) # add to list of uninitialised
- print ("expr assign", name, p[1])
- if name in self.gprs:
- self.write_regs.append(name) # add to list of regs to write
- p[0] = Assign(p[1], p[3])
+ # add to list of uninitialised
+ self.uninit_regs.add(name)
+ elif isinstance(p[1], ast.Call) and p[1].func.id == 'GPR':
+ print(astor.dump_tree(p[1]))
+ # replace GPR(x) with GPR[x]
+ idx = p[1].args[0]
+ p[1] = ast.Subscript(p[1].func, idx)
+ elif isinstance(p[1], ast.Call) and p[1].func.id == 'MEM':
+ print ("mem assign")
+ print(astor.dump_tree(p[1]))
+ p[1].func.id = "memassign" # change function name to set
+ p[1].args.append(p[3])
+ p[0] = p[1]
+ print ("mem rewrite")
+ print(astor.dump_tree(p[0]))
+ return
+ else:
+ print ("help, help")
+ print(astor.dump_tree(p[1]))
+ print("expr assign", name, p[1])
+ if name and name in self.gprs:
+ self.write_regs.add(name) # add to list of regs to write
+ p[0] = Assign(p[1], p[3], iea_mode)
def p_flow_stmt(self, p):
"flow_stmt : return_stmt"
"return_stmt : RETURN testlist"
p[0] = ast.Return(p[2])
-
def p_compound_stmt(self, p):
"""compound_stmt : if_stmt
| while_stmt
p[0] = ast.Break()
def p_for_stmt(self, p):
- """for_stmt : FOR test EQ test TO test COLON suite
+ """for_stmt : FOR atom EQ test TO test COLON suite
+ | DO atom EQ test TO test COLON suite
"""
- p[0] = ast.While(p[2], p[4], [])
# auto-add-one (sigh) due to python range
start = p[4]
end = ast.BinOp(p[6], ast.Add(), ast.Constant(1))
else:
p[0] = p[3]
-
def p_stmts(self, p):
"""stmts : stmts stmt
| stmt"""
| comparison MINUS comparison
| comparison MULT comparison
| comparison DIV comparison
+ | comparison MOD comparison
| comparison EQ comparison
| comparison LE comparison
| comparison GE comparison
| comparison LT comparison
| comparison GT comparison
| comparison BITOR comparison
+ | comparison BITXOR comparison
| comparison BITAND comparison
| PLUS comparison
| comparison MINUS
| comparison APPEND comparison
| power"""
if len(p) == 4:
- print (list(p))
+ print(list(p))
if p[2] == '<u':
p[0] = ast.Call(ast.Name("ltu"), (p[1], p[3]), [])
elif p[2] == '>u':
l = check_concat(p[1]) + check_concat(p[3])
p[0] = ast.Call(ast.Name("concat"), l, [])
elif p[2] in ['<', '>', '=', '<=', '>=']:
- p[0] = binary_ops[p[2]]((p[1],p[3]))
+ p[0] = binary_ops[p[2]]((p[1], p[3]))
+ elif identify_sint_mul_pattern(p):
+ keywords=[ast.keyword(arg='repeat', value=p[3])]
+ l = p[1].elts
+ p[0] = ast.Call(ast.Name("concat"), l, keywords)
else:
p[0] = ast.BinOp(p[1], binary_ops[p[2]], p[3])
elif len(p) == 3:
# power: atom trailer* ['**' factor]
# trailers enables function calls (and subscripts).
- # I only allow one level of calls
- # so this is 'trailer'
+ # so this is 'trailerlist'
def p_power(self, p):
"""power : atom
- | atom trailer"""
+ | atom trailerlist"""
if len(p) == 2:
p[0] = p[1]
else:
- if p[2][0] == "CALL":
- #p[0] = ast.Expr(ast.Call(p[1], p[2][1], []))
- p[0] = ast.Call(p[1], p[2][1], [])
- #if p[1].id == 'print':
- # p[0] = ast.Printnl(ast.Tuple(p[2][1]), None, None)
- #else:
- # p[0] = ast.CallFunc(p[1], p[2][1], None, None)
- else:
- print ("subscript atom", p[2][1])
- #raise AssertionError("not implemented %s" % p[2][0])
- subs = p[2][1]
- if len(subs) == 1:
- idx = subs[0]
- else:
- idx = ast.Slice(subs[0], subs[1], None)
- p[0] = ast.Subscript(p[1], idx)
+ print("power dump atom")
+ print(astor.dump_tree(p[1]))
+ print("power dump trailerlist")
+ print(astor.dump_tree(p[2]))
+ p[0] = apply_trailer(p[1], p[2])
def p_atom_name(self, p):
"""atom : NAME"""
- p[0] = ast.Name(p[1], ctx=ast.Load())
+ p[0] = ast.Name(id=p[1], ctx=ast.Load())
def p_atom_number(self, p):
"""atom : BINARY
| STRING"""
p[0] = ast.Constant(p[1])
- #'[' [listmaker] ']' |
+ # '[' [listmaker] ']' |
def p_atom_listmaker(self, p):
"""atom : LBRACK listmaker RBRACK"""
def p_atom_tuple(self, p):
"""atom : LPAR testlist RPAR"""
- print ("tuple", p[2])
+ print("tuple", p[2])
+ print("astor dump")
+ print(astor.dump_tree(p[2]))
+
if isinstance(p[2], ast.Name):
- print ("tuple name", p[2].id)
+ print("tuple name", p[2].id)
if p[2].id in self.gprs:
- self.read_regs.append(p[2].id) # add to list of regs to read
+ self.read_regs.add(p[2].id) # add to list of regs to read
#p[0] = ast.Subscript(ast.Name("GPR"), ast.Str(p[2].id))
- #return
- p[0] = p[2]
+ # return
+ p[0] = p[2]
+ elif isinstance(p[2], ast.BinOp):
+ if isinstance(p[2].left, ast.Name) and \
+ isinstance(p[2].right, ast.Constant) and \
+ p[2].right.value == 0 and \
+ p[2].left.id in self.gprs:
+ rid = p[2].left.id
+ self.read_regs.add(rid) # add to list of regs to read
+ # create special call to GPR.getz
+ gprz = ast.Name("GPR")
+ gprz = ast.Attribute(gprz, "getz") # get testzero function
+ # *sigh* see class GPR. we need index itself not reg value
+ ridx = ast.Name("_%s" % rid)
+ p[0] = ast.Call(gprz, [ridx], [])
+ print("tree", astor.dump_tree(p[0]))
+ else:
+ p[0] = p[2]
+ else:
+ p[0] = p[2]
+
+ def p_trailerlist(self, p):
+ """trailerlist : trailer trailerlist
+ | trailer
+ """
+ if len(p) == 2:
+ p[0] = p[1]
+ else:
+ p[0] = ("TLIST", p[1], p[2])
# trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
def p_trailer(self, p):
"trailer_subscript : LBRACK subscript RBRACK"
p[0] = ("SUBS", p[2])
- #subscript: '.' '.' '.' | test | [test] ':' [test]
+ # subscript: '.' '.' '.' | test | [test] ':' [test]
def p_subscript(self, p):
"""subscript : test COLON test
else:
p[0] = [p[1]]
-
# testlist: test (',' test)* [',']
# Contains shift/reduce error
+
def p_testlist(self, p):
"""testlist : testlist_multi COMMA
| testlist_multi """
# singleton -> tuple
p[0] = [p[1], p[3]]
-
# test: or_test ['if' or_test 'else' test] | lambdef
# as I don't support 'and', 'or', and 'not' this works down to 'comparison'
+
def p_test(self, p):
"test : comparison"
p[0] = p[1]
-
-
# arglist: (argument ',')* (argument [',']| '*' test [',' '**' test]
# | '**' test)
# XXX INCOMPLETE: this doesn't allow the trailing comma
+
def p_arglist(self, p):
"""arglist : arglist COMMA argument
| argument"""
p[0] = p[1]
def p_error(self, p):
- #print "Error!", repr(p)
+ # print "Error!", repr(p)
raise SyntaxError(p)
class GardenSnakeParser(PowerParser):
- def __init__(self, lexer = None):
+ def __init__(self, lexer=None, debug=False):
PowerParser.__init__(self)
+ self.debug = debug
if lexer is None:
lexer = IndentLexer(debug=0)
self.lexer = lexer
self.tokens = lexer.tokens
self.parser = yacc.yacc(module=self, start="file_input_end",
- debug=False, write_tables=False)
+ debug=debug, write_tables=False)
self.sd = create_pdecode()
def parse(self, code):
- #self.lexer.input(code)
- result = self.parser.parse(code, lexer=self.lexer, debug=False)
+ # self.lexer.input(code)
+ result = self.parser.parse(code, lexer=self.lexer, debug=self.debug)
return ast.Module(result)
#from compiler import misc, syntax, pycodegen
class GardenSnakeCompiler(object):
- def __init__(self):
- self.parser = GardenSnakeParser()
+ def __init__(self, debug=False):
+ self.parser = GardenSnakeParser(debug=debug)
def compile(self, code, mode="exec", filename="<string>"):
tree = self.parser.parse(code)
- print ("snake")
+ print("snake")
pprint(tree)
return tree
#misc.set_filename(filename, tree)
return compile(tree, mode="exec", filename="<string>")
- #syntax.check(tree)
+ # syntax.check(tree)
gen = pycodegen.ModuleCodeGenerator(tree)
code = gen.getCode()
return code
-