feat/items-grammar, closures and goto

src/grammar/consts.rs

@@ -0,0 +1,7 @@
+
+pub const EPSILON: char = 'ε';
+pub const STRING_END: char = '$';
+
+// NOTE: this could be in conflict with the Terminal symbols, so
+// it is MANDATORY that the Terminal doesn´t have dots in it!
+pub const ITEM_SEP: char = '.';

src/grammar/grammar.rs

@@ -1,6 +1,9 @@
 use std::collections::{BTreeSet, BTreeMap};
 
 use crate::automata::DFA;
+use crate::grammar::consts::{EPSILON, STRING_END};
+
+use super::item::Item;
 
 pub type NonTerminal = usize;
 pub type Terminal = char;
@@ -11,10 +14,10 @@ pub enum Letter {
     Terminal(Terminal),
 }
 
-#[derive(Debug, PartialEq)]
+#[derive(Debug, Clone, PartialEq, PartialOrd, Eq, Ord)]
 pub struct Production {
-    lhs: NonTerminal,
-    rhs: Vec<Letter>,
+    pub lhs: NonTerminal,
+    pub rhs: Vec<Letter>,
 }
 
 #[derive(Debug, PartialEq)]
@@ -28,14 +31,24 @@ pub struct Grammar {
     nullable: Option<BTreeSet<NonTerminal>>,
 }
 
-const EPSILON: char = 'ε';
-const STRING_END: char = '$';
-
-// NOTE: this could be in conflict with the Terminal symbols, so
-// it is MANDATORY that the Terminal doesn´t have dots in it!
-const ITEM_SEP: char = '.';
 
 impl Grammar {
+    pub fn new(start_symbol: NonTerminal, productions: Vec<Production>) -> Self {
+        Grammar {
+            start_symbol,
+            productions,
+            nullable: None,
+        }
+    }
+
+    pub fn get_start_symbol(&self) -> NonTerminal {
+        self.start_symbol
+    }
+
+    pub fn get_productions(&self) -> &Vec<Production> {
+        &self.productions
+    }
+
     pub fn first(&mut self, letter: &Letter) -> BTreeSet<Terminal> {
         if let None = self.nullable {
             self.nullable = Some(self.get_nullable());
@@ -352,7 +365,7 @@ impl Grammar {
         });
 
         // add corresponding productions 
-        let mut adj_list = self.transitions_to_adj_list();
+        let mut adj_list = self.productions_to_adj_list();
         for unitary_couple in unitary_couples.iter() {
             if unitary_couple.0 == unitary_couple.1 {
                 continue;
@@ -381,7 +394,7 @@ impl Grammar {
         self.nullable = None;
     }
 
-    fn transitions_to_adj_list(&self) -> BTreeMap<NonTerminal, BTreeSet<Vec<Letter>>> {
+    pub fn productions_to_adj_list(&self) -> BTreeMap<NonTerminal, BTreeSet<Vec<Letter>>> {
         let mut adj_list: BTreeMap<NonTerminal, BTreeSet<Vec<Letter>>> = BTreeMap::new();
         for production in self.productions.iter() {
             adj_list.entry(production.lhs)
@@ -392,29 +405,14 @@ impl Grammar {
         adj_list
     }
 
-    pub fn get_itemization(&self) -> Vec<Production> {
-        let mut itemized_transitions = vec![];
-        for production in self.productions.iter() {
-            if production.rhs.len() == 1 && production.rhs[0] == Letter::Terminal(EPSILON) {
-                itemized_transitions.push(Production {
-                    lhs: production.lhs,
-                    rhs: vec![Letter::Terminal(ITEM_SEP)]
-                });
-                continue;
-            }
-
-            for i in 0..=production.rhs.len() {
-                let mut rhs = production.rhs.clone();
-
-                rhs.insert(i, Letter::Terminal(ITEM_SEP));
-                itemized_transitions.push(Production {
-                    lhs: production.lhs,
-                    rhs: rhs
-                });
-            }
-        }
+    pub fn add_fake_initial_state(&mut self) -> () {
+        let new_state = self.get_non_terminal().iter().max().unwrap() + 1;
+        self.productions.push(Production {
+            lhs: new_state,
+            rhs: vec![Letter::NonTerminal(self.start_symbol)]
+        });
 
-        itemized_transitions
+        self.start_symbol = new_state;
     }
 }
 
@@ -625,28 +623,4 @@ mod test {
 
         assert_eq!(grammar, result);
     }
-
-    #[test]
-    fn test_itemization() {
-        let grammar = get_test_grammar();
-
-        let items = grammar.get_itemization();
-
-        let result_productions = vec![
-            Production { lhs: 0, rhs: vec![Letter::Terminal(ITEM_SEP), Letter::NonTerminal(1), Letter::Terminal('b')] },
-            Production { lhs: 0, rhs: vec![Letter::NonTerminal(1), Letter::Terminal(ITEM_SEP), Letter::Terminal('b')] },
-            Production { lhs: 0, rhs: vec![Letter::NonTerminal(1), Letter::Terminal('b'), Letter::Terminal(ITEM_SEP)] },
-
-            Production { lhs: 0, rhs: vec![Letter::Terminal(ITEM_SEP), Letter::Terminal('c')] },
-            Production { lhs: 0, rhs: vec![Letter::Terminal('c'), Letter::Terminal(ITEM_SEP)] },
-
-            Production { lhs: 1, rhs: vec![Letter::Terminal(ITEM_SEP), Letter::Terminal('a'), Letter::NonTerminal(1)] },
-            Production { lhs: 1, rhs: vec![Letter::Terminal('a'), Letter::Terminal(ITEM_SEP), Letter::NonTerminal(1)] },
-            Production { lhs: 1, rhs: vec![Letter::Terminal('a'), Letter::NonTerminal(1), Letter::Terminal(ITEM_SEP)] },
-            Production { lhs: 1, rhs: vec![Letter::Terminal(ITEM_SEP)] },
-        ];
-
-        assert!(items.iter().all(|item| result_productions.contains(item)));
-        assert!(result_productions.iter().all(|item| items.contains(item)));
-    }
 }