| 1 | #!/usr/bin/env python2
|
| 2 | """
|
| 3 | parse.py
|
| 4 | """
|
| 5 | from __future__ import print_function
|
| 6 |
|
| 7 | import os
|
| 8 | import sys
|
| 9 |
|
| 10 | # PYTHONPATH=$REPO_ROOT/vendor
|
| 11 | from typing import Tuple, Optional, cast
|
| 12 |
|
| 13 | # PYTHONPATH=$REPO_ROOT/mycpp
|
| 14 | from mycpp.mylib import log, tagswitch
|
| 15 | from mycpp import mylib
|
| 16 | from _devbuild.gen.expr_asdl import expr, expr_e, expr_t, tok_e, tok_t
|
| 17 |
|
| 18 | # PYTHONPATH=$REPO_ROOT
|
| 19 | from asdl import format as fmt
|
| 20 |
|
| 21 |
|
| 22 | class Lexer(object):
|
| 23 | # "type declaration" of members
|
| 24 |
|
| 25 | def __init__(self, s):
|
| 26 | # type: (str) -> None
|
| 27 | self.s = s
|
| 28 | self.i = 0
|
| 29 | self.n = len(s)
|
| 30 |
|
| 31 | def Read(self):
|
| 32 | # type: () -> Tuple[tok_t, str]
|
| 33 | if self.i >= self.n:
|
| 34 | return tok_e.Eof, '' # sentinel
|
| 35 |
|
| 36 | tok = self.s[self.i]
|
| 37 | self.i += 1
|
| 38 |
|
| 39 | if tok.isdigit():
|
| 40 | return tok_e.Const, tok
|
| 41 | if tok.isalpha():
|
| 42 | return tok_e.Var, tok
|
| 43 | if tok in '+-':
|
| 44 | return tok_e.Op1, tok # lower precedence
|
| 45 | if tok in '*/':
|
| 46 | return tok_e.Op2, tok # higher precedence
|
| 47 | if tok in '()':
|
| 48 | return tok_e.Paren, tok
|
| 49 |
|
| 50 | return tok_e.Invalid, tok
|
| 51 |
|
| 52 | def _MethodCallingOtherMethod(self):
|
| 53 | # type: () -> None
|
| 54 | # Make sure we don't get a member var called Read!
|
| 55 | # This is easy because we're only checking assignment statements.
|
| 56 | self.Read()
|
| 57 |
|
| 58 |
|
| 59 | class ParseError(Exception):
|
| 60 |
|
| 61 | def __init__(self, msg):
|
| 62 | # type: (str) -> None
|
| 63 | self.msg = msg
|
| 64 |
|
| 65 |
|
| 66 | class Parser(object):
|
| 67 | """
|
| 68 | Grammar from http://compilers.iecc.com/crenshaw/tutor6.txt, adapted to ANTLR
|
| 69 | syntax.
|
| 70 |
|
| 71 | Expr = Term ('+' Term)*
|
| 72 | Term = Factor ('*' Factor)*
|
| 73 | Factor = VAR
|
| 74 | | CONST
|
| 75 | | '(' Expr ')'
|
| 76 | """
|
| 77 |
|
| 78 | def __init__(self, lexer):
|
| 79 | # type: (Lexer) -> None
|
| 80 | self.lexer = lexer
|
| 81 | self.tok_type = tok_e.Eof # type: tok_t
|
| 82 | self.tok_val = ''
|
| 83 |
|
| 84 | def Next(self):
|
| 85 | # type: () -> None
|
| 86 | self.tok_type, self.tok_val = self.lexer.Read()
|
| 87 | #log('-- %r %r', self.tok_type, self.tok_val)
|
| 88 |
|
| 89 | def Eat(self, tok_val):
|
| 90 | # type: (str) -> None
|
| 91 | if self.tok_val != tok_val:
|
| 92 | #raise ParseError('Expected %r, got %r' % (tok_val, self.tok_val))
|
| 93 | raise ParseError('Expected ' + tok_val)
|
| 94 | self.Next()
|
| 95 |
|
| 96 | def ParseFactor(self):
|
| 97 | # type: () -> expr_t
|
| 98 |
|
| 99 | #log('ParseFactor')
|
| 100 | if self.tok_type == tok_e.Var:
|
| 101 | n1 = expr.Var(self.tok_val)
|
| 102 | self.Next()
|
| 103 | return n1
|
| 104 |
|
| 105 | if self.tok_type == tok_e.Const:
|
| 106 | n2 = expr.Const(int(self.tok_val))
|
| 107 | self.Next()
|
| 108 | return n2
|
| 109 |
|
| 110 | if self.tok_type == tok_e.Paren:
|
| 111 | self.Eat('(')
|
| 112 | n3 = self.ParseExpr()
|
| 113 | self.Eat(')')
|
| 114 | return n3
|
| 115 |
|
| 116 | #raise ParseError('Unexpected token %s %s' % (self.tok_type, self.tok_val))
|
| 117 | raise ParseError('Unexpected token ' + self.tok_val)
|
| 118 |
|
| 119 | def ParseTerm(self):
|
| 120 | # type: () -> expr_t
|
| 121 |
|
| 122 | #log('ParseTerm')
|
| 123 | node = self.ParseFactor()
|
| 124 |
|
| 125 | # TODO: Iterate and create nodes
|
| 126 | while self.tok_type == tok_e.Op2:
|
| 127 | op = self.tok_val
|
| 128 | self.Next()
|
| 129 | n2 = self.ParseFactor()
|
| 130 | node = expr.Binary(op, node, n2)
|
| 131 | return node
|
| 132 |
|
| 133 | def ParseExpr(self):
|
| 134 | # type: () -> expr_t
|
| 135 |
|
| 136 | #log('ParseExpr')
|
| 137 | node = self.ParseTerm()
|
| 138 |
|
| 139 | while self.tok_type == tok_e.Op1:
|
| 140 | op = self.tok_val
|
| 141 | self.Next()
|
| 142 | n2 = self.ParseTerm()
|
| 143 | node = expr.Binary(op, node, n2)
|
| 144 | return node
|
| 145 |
|
| 146 | def Parse(self):
|
| 147 | # type: () -> expr_t
|
| 148 | self.Next()
|
| 149 | return self.ParseExpr()
|
| 150 |
|
| 151 |
|
| 152 | def TestParse():
|
| 153 | # type: () -> None
|
| 154 |
|
| 155 | lex = Lexer('abc')
|
| 156 | while True:
|
| 157 | tok_type, tok_val = lex.Read()
|
| 158 | if tok_type == tok_e.Eof:
|
| 159 | break
|
| 160 | #print('%s %s' % (tok_type, tok_val))
|
| 161 | log('tok_val %s', tok_val)
|
| 162 |
|
| 163 | CASES = [
|
| 164 | '1+2',
|
| 165 | '1+2*3',
|
| 166 | '1*2+3',
|
| 167 | '(1+2)*3',
|
| 168 | 'a+b+c+d',
|
| 169 | 'a*b*3*4',
|
| 170 | '1',
|
| 171 | 'a',
|
| 172 |
|
| 173 | # expect errors here:
|
| 174 | '(',
|
| 175 | ')',
|
| 176 | '(a+b',
|
| 177 | ' ',
|
| 178 | ' $$ ',
|
| 179 | ]
|
| 180 | for expr_ in CASES:
|
| 181 | lex = Lexer(expr_)
|
| 182 | p = Parser(lex)
|
| 183 | log('')
|
| 184 | log('--')
|
| 185 | log('%s =>', expr_)
|
| 186 |
|
| 187 | node = None # type: Optional[expr_t]
|
| 188 | try:
|
| 189 | node = p.Parse()
|
| 190 | except ParseError as e:
|
| 191 | log('Parse error: %s', e.msg)
|
| 192 | continue
|
| 193 |
|
| 194 | #log('%s', tree)
|
| 195 |
|
| 196 | htree = node.PrettyTree()
|
| 197 | ast_f = fmt.AnsiOutput(mylib.Stdout())
|
| 198 |
|
| 199 | fmt.PrintTree(htree, ast_f)
|
| 200 | ast_f.write('\n')
|
| 201 |
|
| 202 | UP_node = node
|
| 203 | with tagswitch(UP_node) as case:
|
| 204 | if case(expr_e.Const):
|
| 205 | node = cast(expr.Const, UP_node)
|
| 206 | log('Const %d', node.i)
|
| 207 |
|
| 208 | elif case(expr_e.Var):
|
| 209 | node = cast(expr.Var, UP_node)
|
| 210 | log('Var %s', node.name)
|
| 211 |
|
| 212 | else:
|
| 213 | log('Other')
|
| 214 |
|
| 215 |
|
| 216 | def TestCreateNull():
|
| 217 | # type: () -> None
|
| 218 |
|
| 219 | c = expr.Const.CreateNull(alloc_lists=True)
|
| 220 | log('c.i %d', c.i)
|
| 221 |
|
| 222 | v = expr.Var.CreateNull(alloc_lists=True)
|
| 223 | log('v.name %r', v.name)
|
| 224 |
|
| 225 | b = expr.Binary.CreateNull(alloc_lists=True)
|
| 226 | log('b.op %r', b.op)
|
| 227 | b.op = '+'
|
| 228 |
|
| 229 | # Must assign these
|
| 230 | b.left = c
|
| 231 | b.right = v
|
| 232 |
|
| 233 | htree = b.PrettyTree()
|
| 234 | ast_f = fmt.AnsiOutput(mylib.Stdout())
|
| 235 |
|
| 236 | fmt.PrintTree(htree, ast_f)
|
| 237 | ast_f.write('\n')
|
| 238 |
|
| 239 |
|
| 240 | def run_tests():
|
| 241 | # type: () -> None
|
| 242 |
|
| 243 | TestParse()
|
| 244 | TestCreateNull()
|
| 245 |
|
| 246 |
|
| 247 | def run_benchmarks():
|
| 248 | # type: () -> None
|
| 249 | n = 100000
|
| 250 |
|
| 251 | result = 0
|
| 252 | i = 0
|
| 253 | while i < n:
|
| 254 | lex = Lexer('a*b*3*4')
|
| 255 | p = Parser(lex)
|
| 256 | tree = p.Parse()
|
| 257 |
|
| 258 | i += 1
|
| 259 |
|
| 260 | mylib.MaybeCollect() # manual GC point
|
| 261 |
|
| 262 | log('result = %d', result)
|
| 263 | log('iterations = %d', n)
|
| 264 |
|
| 265 |
|
| 266 | if __name__ == '__main__':
|
| 267 | if os.getenv('BENCHMARK'):
|
| 268 | log('Benchmarking...')
|
| 269 | run_benchmarks()
|
| 270 | else:
|
| 271 | run_tests()
|