Mercurial > lcfOS

comparison python/pyyacc.py @ 192:6cd6260789a1

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Added more tests for parser generator
author Windel Bouwman
date Sun, 26 May 2013 23:19:27 +0200
parents 6b2bec5653f1
children b01429a5d695
comparison
equal deleted inserted replaced
191:6b2bec5653f1 192:6cd6260789a1
30 30
31 def add_production(self, name, symbols): 31 def add_production(self, name, symbols):
32 """ Add a production rule to the grammar """ 32 """ Add a production rule to the grammar """
33 production = Production(name, symbols) 33 production = Production(name, symbols)
34 self.productions.append(production) 34 self.productions.append(production)
35 assert not name in self.terminals, "Cannot redefine terminal" 35 if name in self.terminals:
36 raise ParserGenerationException("Cannot redefine terminal {0}".format(name))
36 if not name in self.nonterminals: 37 if not name in self.nonterminals:
37 self.nonterminals.append(name) 38 self.nonterminals.append(name)
38 39
39 def productionsForName(self, name): 40 def productionsForName(self, name):
40 """ Retrieve all productions for a non terminal """ 41 """ Retrieve all productions for a non terminal """
148 if not nis: 149 if not nis:
149 continue 150 continue
150 addSt(nis) 151 addSt(nis)
151 transitions[(states.index(itemset), symbol)] = states.index(nis) 152 transitions[(states.index(itemset), symbol)] = states.index(nis)
152 return states, transitions 153 return states, transitions
153 154
155 def checkSymbols(self):
156 """ Checks no symbols are undefined """
157 for production in self.productions:
158 for symbol in production.symbols:
159 if symbol not in self.Symbols:
160 raise ParserGenerationException('Symbol {0} undefined'.format(symbol))
161
162
154 def genParser(self): 163 def genParser(self):
155 """ Generates a parser from the grammar """ 164 """ Generates a parser from the grammar """
165 self.checkSymbols()
156 action_table = {} 166 action_table = {}
157 goto_table = {} 167 goto_table = {}
158 iis = self.initialItemSet() 168 iis = self.initialItemSet()
159 169
160 # First generate all item sets by using the nextItemset function: 170 # First generate all item sets by using the nextItemset function:
301 look_ahead = Token(EOF, EOF, 0) 311 look_ahead = Token(EOF, EOF, 0)
302 assert type(look_ahead) is Token 312 assert type(look_ahead) is Token
303 elif action == ACCEPT: 313 elif action == ACCEPT:
304 break 314 break
305 315
316
306 def testSimpleGrammar(): 317 def testSimpleGrammar():
307 # 1. define a simple grammar: 318 # 1. define a simple grammar:
308 g = Grammar(['EOF', 'identifier', '(', ')', '+', '*']) 319 g = Grammar(['EOF', 'identifier', '(', ')', '+', '*'])
309 g.add_production('input', ['expression']) 320 g.add_production('input', ['expression'])
310 g.add_production('expression', ['term']) 321 g.add_production('expression', ['term'])
313 g.add_production('term', ['term', '*', 'factor']) 324 g.add_production('term', ['term', '*', 'factor'])
314 g.add_production('factor', ['(', 'expression', ')']) 325 g.add_production('factor', ['(', 'expression', ')'])
315 g.add_production('factor', ['identifier']) 326 g.add_production('factor', ['identifier'])
316 g.start_symbol = 'input' 327 g.start_symbol = 'input'
317 # 2. define input: 328 # 2. define input:
318 tokens = ['identifier', '+', 'identifier', '+', 'identifier', 'EOF'] 329 tokens = ['identifier', '+', 'identifier', '+', 'identifier']
319 # 3. build parser: 330 # 3. build parser:
320 p = g.genParser() 331 p = g.genParser()
321 # 4. feed input: 332 # 4. feed input:
322 p.parse(tokens) 333 def genTokens(lst):
334 for t in lst:
335 yield Token(t, t, 0)
336 p.parse(genTokens(tokens))
323 337
324 338
325 if __name__ == '__main__': 339 if __name__ == '__main__':
326 testSimpleGrammar() 340 testSimpleGrammar()
341