vernalhav / pyrser Goto Github PK

Could be based off of important nonterminals (either specified by the user or inferred) as described in page 283 (306) of the Dragon Book.

The user can manually override parsing table entries with custom error recovery routines that would allow them to operate the parser as desired. Examples can be found in page 284 (307) of the Dragon Book.

The procedure used by Yacc is briedly described in Dragon Book page 293 (316). The biggest design challenge is to understand where and how the user would specify this precedence.

At the parser level:

• The parsing table would contain Conflict entries and the parser would apply the supplied rules while parsing to disambiguate.
• Or the parser could "manually" override the Conflict entries as a pre-processing step. This would result in lower overhead while parsing.
• This would decouple the automata from the precedence and associativity rules.

At the table creation level:

• The automata would have access to the given precedences and associativites to transform Conflict entries into Shift or Goto.

Should this be solved at the parsing table level (as in there always being exactly one Action for each transition) or at the parser level (as in having a actions.Conflict action that stores all possibilities), to be disambiguated by the parser?