Updates to Funl parser assignment:
- The definition of the
valexpression has changed. - I have provided some unit tests, as ParserTest.java.
- I have provided a tree parser, as tree-parser.java.
- The tokenizer.zip that I provided for your use does not tokenize quoted strings. As the upgrade requires only four lines (one to say you want quoted strings, three to handle that case), I'll leave that fix up to you. (Additional hint: Read the StreamTokenizer documentation!)
implies" < "or" < "and" < "not"), and has no remaining (known) bugs.
program ::= sentence { sentence }
sentence ::= fact | query
fact ::= "!" implication | "Fact:" implication
query ::= "?" implication | "Query:" implication
implication ::= disjunction [ "implies" disjunction ]
disjunction ::= conjunction { "or" conjunction }
conjunction ::= basis { "and" basis }
basis ::= [ "not" ] variable | [ "not" ] "(" implication ")"
If you are having trouble with recursive descent parsing, I recommend that you look over this code (particularly the LittleLogicLanguageParser class). A number of people have tried to do Part 2 of the BNF Tokenizer and Parser assignment without using the method of recursive descent. I've written an example recursive descent parser for a "little logical language", with syntax as follows:
program ::= sentence { sentence }
sentence ::= fact | query
fact ::= "!" expression
query ::= "?" expression
expression ::= [ "not" ] disjunct [ "or" expression ]
disjunct ::= conjunct { "and" conjunct }
conjunct ::= variable | "(" expression ")"
The parser itself is about 225 lines, but there are supporting classes (basic tree and token classes, and a tokenizer). There are still some bugs, but the parser is working well enough to post it as an example of how recursive descent should be done. Here it is: logic_language.zip. I expect to post a final version sometime later.
I have tried to clarify the BNF assignment by means of stepping through a small example.
The bottom line is that the assignment is correct as written, so I don't see any need to extend the deadline. Sorry!
In addition to textbooks, you should purchase a TurningPoint ResponseCard (or "clicker"). These are approximately $50 and are available at the purchase desk in the Penn Bookstore.