magne / dfalex.net Goto Github PK

Currently, RegexParser takes a bool caseIndependent parameter. Instead, make it take some options object. This will make it easier to add options later.

We need to create a DFA that considers the prioritized transitions from the new NFA produced by the lazy quantifiers.

This can probably be done by matching NFA states in a DFA state as an ordered list, not as a set. The ordered list of NFA states must be produced by simulating the NFA with a low and a normal stack. See the nes1983 code in the alternate-implementations branch.

Currently, RegexParser produces an IMatchable structure. I would like it to be able to produce arbitrary structures.

This will require extracting the different actions that produce IMatchables to a support class, make that class implement an interface, and initialize the RegexParser with implementations of that interface. RegexParser must also be extended to produce generic structures, as produced by the support classes.

We should be able to minimize the DFA created from the NFA with prioritized transitions.

This probably involves the mappable concept described by Laurikari (see references in the Wiki), and the nes1983 code in the alternate-implementations branch has some code that might be useful.

Add parsing of the non-greedy quantifiers (??, *?, +?) to regex parsing.

• Add priority to NFA transitions (Normal and Low).
• Update existing Patterns with priority transitions.
• Add non-greedy quantifiers to Pattern.