slr parser

src/bin/book.rs

@@ -69,8 +69,8 @@ fn grammer() -> Grammar<NoneTerminals, BareTokens> {
 }
 
 fn main() {
-    //let code = String::from("a = b()+c+(d+e())");
-    //let mut m = Scanner::<Tokens>::new(code).with_skipping(Tokens::WhiteSpace);
+    let code = String::from("a = b(f)+c+(d+e(f))");
+    let mut m = Scanner::<Tokens>::new(code).with_skipping(Tokens::WhiteSpace);
 
     let mut grammar = grammer();
     grammar.gen_follow();
@@ -78,7 +78,10 @@ fn main() {
     println!("follow: {:?}", grammar.follow);
     grammar.gen_lr0_automaton();
     println!("automaton: {:?}", grammar.lr0_automaton);
-    grammar.gen_slr_parse_table();
+    println!("conflict: {}", grammar.gen_slr_parse_table());
     println!("parse_table: {:?}", grammar.slr_parse_table);
-
+    println!(
+        "parsed: {:?}",
+        grammar.slr_parser(&mut m.iter_mut()).parse()
+    )
 }

src/cfg/ll_grammar.rs

@@ -1,6 +1,6 @@
 use std::{collections::HashMap, fmt::Debug, hash::Hash};
 
-use super::{Grammar, Sentential};
+use super::{Grammar, NodeChild, ParseTree, Sentential};
 
 impl<N: PartialEq + Eq + Hash + Clone, T: PartialEq + Eq + Hash + Clone> Grammar<N, T> {
     pub fn gen_ll_parse_table(&mut self) -> bool {
@@ -15,7 +15,7 @@ impl<N: PartialEq + Eq + Hash + Clone, T: PartialEq + Eq + Hash + Clone> Grammar
         }
         let mut conflict = false;
         // left derivation
-        // when hidding N and T is next (None = e)
+        // when hiding N and T is next (None = e)
         // then the n'th (usize) variant of N can be used.
         let mut parse_table: HashMap<(N, Option<T>), usize> = HashMap::new();
         for (from, to) in self.rules.iter() {
@@ -85,23 +85,21 @@ impl<N: PartialEq + Eq + Hash + Clone, T: PartialEq + Eq + Hash + Clone> Grammar
     }
 }
 
-/// Just checks a program. Does not generates output.
-pub struct LLTabelParser<
-    'a,
+pub struct LLTabelParser<'a, N, T, S>
+where
     N: PartialEq + Eq + Hash + Clone,
     T: PartialEq + Eq + Hash + Clone,
     S: Into<T> + PartialEq<T> + Clone,
-> {
+{
     grammar: &'a Grammar<N, T>,
     input: &'a mut dyn Iterator<Item = Result<S, String>>,
 }
 
-impl<
-        'a,
-        N: PartialEq + Eq + Hash + Clone + Debug,
-        T: PartialEq + Eq + Hash + Clone + Debug,
-        S: Into<T> + PartialEq<T> + Clone + Debug,
-    > LLTabelParser<'a, N, T, S>
+impl<'a, N, T, S> LLTabelParser<'a, N, T, S>
+where
+    N: PartialEq + Eq + Hash + Clone + Debug,
+    T: PartialEq + Eq + Hash + Clone + Debug,
+    S: Into<T> + PartialEq<T> + Clone + Debug,
 {
     pub fn parse(&mut self) -> Result<ParseTree<N, S>, String> {
         // stack of table driven parser
@@ -205,29 +203,6 @@ pub trait Skippable {
         false
     }
 }
-
-#[derive(Debug, Clone)]
-pub enum NodeChild<N, S> {
-    Child(ParseTree<N, S>),
-    Data(S),
-}
-
-#[derive(Debug, Clone)]
-#[allow(dead_code)]
-pub struct ParseTree<N, S> {
-    pub rule: Option<(N, usize)>,
-    pub childs: Vec<NodeChild<N, S>>,
-}
-
-impl<N, S> ParseTree<N, S> {
-    pub fn new(rule: Option<(N, usize)>) -> Self {
-        Self {
-            rule,
-            childs: Vec::new(),
-        }
-    }
-}
-
 impl<N: Skippable + Debug, S: Debug> ParseTree<N, S> {
     /// cleanup the parse tree
     /// does not work on a subtree

src/cfg/lr0_grammar.rs

@@ -1,16 +1,18 @@
 use std::{
     collections::{HashMap, HashSet},
+    fmt::Debug,
     hash::{Hash, Hasher},
     rc::{Rc, Weak},
 };
 
-use super::{Grammar, RuleIndex, Sentential};
+use super::{Grammar, NodeChild, ParseTree, RuleIndex, Sentential};
 
 pub type RL0Automaton<N, T> =
     HashMap<Rc<LR0State<N>>, HashMap<Sentential<N, T>, Weak<LR0State<N>>>>;
 
 #[derive(Debug, Eq, PartialEq)]
 pub struct LR0State<N: Hash + Eq>(HashSet<(RuleIndex<N>, usize)>);
+pub type LR0Start<N> = Weak<LR0State<N>>;
 
 impl<N: Hash + Eq + Ord> Hash for LR0State<N> {
     fn hash<H: Hasher>(&self, state: &mut H) {
@@ -27,9 +29,10 @@ pub enum SlrAction<Shift, Reduce> {
     Reduce(Reduce),
 }
 
+/// None is $
 pub type SlrActionTable<N, T> = HashMap<(usize, Option<T>), SlrAction<usize, RuleIndex<N>>>;
 pub type SlrGotoTable<N> = HashMap<(usize, N), usize>;
-pub type SlrParseTable<N, T> = (SlrActionTable<N, T>, SlrGotoTable<N>);
+pub type SlrParseTable<N, T> = (SlrActionTable<N, T>, SlrGotoTable<N>, usize);
 
 impl<N, T> Grammar<N, T>
 where
@@ -100,6 +103,7 @@ where
         }
         // add state to graph and mark for todo
         let rc = Rc::new(self.lr0_clozure(start_state));
+        let start = Rc::downgrade(&rc);
         let mut todo = vec![Rc::downgrade(&rc)];
         out.insert(rc, HashMap::new());
 
@@ -128,10 +132,10 @@ where
                 });
             }
         }
-        self.lr0_automaton = Some(out);
+        self.lr0_automaton = Some((out, start));
     }
 
-    pub fn gen_slr_parse_table(&mut self) {
+    pub fn gen_slr_parse_table(&mut self) -> bool {
         if self.follow.is_none() {
             self.gen_follow();
         }
@@ -143,15 +147,21 @@ where
 
         let ids: HashMap<Rc<LR0State<N>>, usize> = HashMap::from_iter(
             lr0_automaton
+                .0
                 .iter()
                 .enumerate()
                 .map(|(id, (a, _))| (a.clone(), id)),
         );
 
-        // none is $
+        let start = *lr0_automaton
+            .1
+            .upgrade()
+            .and_then(|rc| ids.get(&rc))
+            .expect("Found broken state in slr parse table gen.");
         let mut action: SlrActionTable<N, T> = HashMap::new();
         let mut goto: SlrGotoTable<N> = HashMap::new();
-        for (state, to) in lr0_automaton {
+        let mut conflict = false;
+        for (state, to) in lr0_automaton.0.iter() {
             let id = ids
                 .get(state)
                 .expect("Found broken state in slr parse table gen.");
@@ -167,10 +177,12 @@ where
 
                 match go {
                     Sentential::Terminal(t) => {
-                        action.insert((*id, Some(t)), SlrAction::Shift(*to_id));
+                        conflict |= action
+                            .insert((*id, Some(t)), SlrAction::Shift(*to_id))
+                            .is_some();
                     }
                     Sentential::NoneTerminal(nt) => {
-                        goto.insert((*id, nt), *to_id);
+                        conflict |= goto.insert((*id, nt), *to_id).is_some();
                     }
                 };
             }
@@ -181,11 +193,151 @@ where
                 };
                 if to.len() <= *dot {
                     for follow in self.follow(from) {
-                        action.insert((*id, follow), SlrAction::Reduce((from.clone(), *rule_id)));
+                        conflict |= action
+                            .insert((*id, follow), SlrAction::Reduce((from.clone(), *rule_id)))
+                            .is_some();
                     }
                 }
             }
         }
-        self.slr_parse_table = Some((action, goto));
+        self.slr_parse_table = Some((action, goto, start));
+        conflict
+    }
+
+    pub fn slr_parser<'a, S: Into<T> + PartialEq<T> + Clone>(
+        &'a self,
+        iter: &'a mut dyn Iterator<Item = Result<S, String>>,
+    ) -> SLRTabelParser<N, T, S> {
+        assert!(
+            self.slr_parse_table.is_some(),
+            "Please call gen_slr_parse_table before this!"
+        );
+        SLRTabelParser {
+            input: iter,
+            grammar: self,
+        }
+    }
+    pub fn get_slr_action(
+        &self,
+        state: &usize,
+        next: &Option<T>,
+    ) -> Option<&SlrAction<usize, (N, usize)>> {
+        assert!(
+            self.slr_parse_table.is_some(),
+            "Please call gen_slr_parse_table before this!"
+        );
+        self.slr_parse_table
+            .as_ref()
+            .unwrap()
+            .0
+            .get(&(*state, next.clone()))
+    }
+    pub fn get_slr_goto(&self, state: &usize, next: &N) -> Option<&usize> {
+        assert!(
+            self.slr_parse_table.is_some(),
+            "Please call gen_slr_parse_table before this!"
+        );
+        self.slr_parse_table
+            .as_ref()
+            .unwrap()
+            .1
+            .get(&(*state, next.clone()))
+    }
+}
+
+pub struct SLRTabelParser<'a, N, T, S>
+where
+    N: PartialEq + Eq + Hash + Clone,
+    T: PartialEq + Eq + Hash + Clone,
+    S: Into<T> + PartialEq<T> + Clone,
+{
+    grammar: &'a Grammar<N, T>,
+    input: &'a mut dyn Iterator<Item = Result<S, String>>,
+}
+
+impl<'a, N, T, S> SLRTabelParser<'a, N, T, S>
+where
+    N: PartialEq + Eq + Hash + Clone + Debug + Ord,
+    T: PartialEq + Eq + Hash + Clone + Debug + Ord,
+    S: Into<T> + PartialEq<T> + Clone + Debug,
+{
+    pub fn parse(&mut self) -> Result<ParseTree<N, S>, String> {
+        let mut stack: Vec<(NodeChild<N, S>, usize)> = Vec::new();
+
+        let mut next = match self.input.next() {
+            Some(Ok(d)) => Some(d),
+            Some(Err(err)) => return Err(format!("Invalid token: {}", err)),
+            None => None,
+        };
+        let mut first = true;
+        loop {
+            let state = if first {
+                self.grammar.slr_parse_table.as_ref().unwrap().2
+            } else {
+                let Some(state) = stack.last() else {
+                    return Err("Unexpected EOS".into());
+                };
+                state.1
+            };
+            first = false;
+            let Some(current_state) = self
+                .grammar
+                .get_slr_action(&state, &next.as_ref().map(|f| f.clone().into()))
+            else {
+                return Err(format!(
+                    "Unexpected Token: {}",
+                    next.map(|f| format!("{f:?}"))
+                        .unwrap_or("end of file".to_string())
+                ));
+            };
+
+            match current_state {
+                SlrAction::Shift(to) => {
+                    stack.push((NodeChild::Data(next.expect("Can not shift on EOF.")), *to));
+                    next = match self.input.next() {
+                        Some(Ok(d)) => Some(d),
+                        Some(Err(err)) => return Err(format!("Invalid token: {}", err)),
+                        None => None,
+                    };
+                }
+                SlrAction::Reduce((rule, ind)) => {
+                    let Some(r) = self.grammar.rules.get(rule).and_then(|e| e.get(*ind)) else {
+                        return Err(format!("Invalid rule: {:?}-{}", rule, ind));
+                    };
+                    let mut childs = Vec::new();
+                    for elem in r.iter().rev() {
+                        let Some(last) = stack.pop() else {
+                            return Err("Unexpected EOF".into());
+                        };
+                        if last.0 != *elem {
+                            return Err(format!("Broken Stack: {:?} and {:?}", last.0, elem));
+                        }
+                        childs.push(last);
+                    }
+                    if self.grammar.start == *rule {
+                        return Ok(ParseTree {
+                            rule: Some((rule.clone(), *ind)),
+                            childs: childs.into_iter().rev().map(|(a, _)| a.clone()).collect(),
+                        });
+                    }
+                    let Some(state) = stack.last() else {
+                        return Err("Unexpected EOS".into());
+                    };
+                    let Some(next) = self.grammar.get_slr_goto(&state.1, rule) else {
+                        return Err(format!(
+                            "Invalid reduction: state: {} rule: {:?}",
+                            state.1, rule
+                        ));
+                    };
+                    stack.push((
+                        NodeChild::Child(ParseTree {
+                            rule: Some((rule.clone(), *ind)),
+                            childs: childs.into_iter().rev().map(|(d, _)| d).collect(),
+                        }),
+                        *next,
+                    ));
+                }
+            }
+        }
     }
 }

src/cfg/mod.rs

@@ -1,12 +1,13 @@
 use std::{
     collections::{HashMap, HashSet},
+    fmt::Debug,
     hash::Hash,
 };
 
-use lr0::{RL0Automaton, SlrParseTable};
+use lr0_grammar::{LR0Start, RL0Automaton, SlrParseTable};
 
 pub mod ll_grammar;
-pub mod lr0;
+pub mod lr0_grammar;
 
 #[macro_export]
 macro_rules! cfg_grammar {
@@ -85,7 +86,7 @@ pub struct Grammar<N: PartialEq + Eq + Hash + Clone, T: PartialEq + Eq + Hash +
     /// Graph, defined throw this adjacent list.
     /// - key: states
     /// - value: list with read symbol and linked node.
-    pub lr0_automaton: Option<RL0Automaton<N, T>>,
+    pub lr0_automaton: Option<(RL0Automaton<N, T>, LR0Start<N>)>,
 
     ///
     pub slr_parse_table: Option<SlrParseTable<N, T>>,
@@ -278,3 +279,48 @@ impl<N: PartialEq + Eq + Hash + Clone, T: PartialEq + Eq + Hash + Clone> Grammar
             .unwrap_or(HashSet::new())
     }
 }
+
+#[derive(Debug, Clone)]
+pub enum NodeChild<N, S> {
+    Child(ParseTree<N, S>),
+    Data(S),
+}
+
+#[derive(Debug, Clone)]
+#[allow(dead_code)]
+pub struct ParseTree<N, S> {
+    pub rule: Option<RuleIndex<N>>,
+    pub childs: Vec<NodeChild<N, S>>,
+}
+
+impl<N, S> ParseTree<N, S> {
+    pub fn new(rule: Option<RuleIndex<N>>) -> Self {
+        Self {
+            rule,
+            childs: Vec::new(),
+        }
+    }
+}
+
+impl<N, T, S> PartialEq<Sentential<N, T>> for NodeChild<N, S>
+where
+    N: PartialEq + Eq,
+    T: PartialEq + Eq,
+    S: PartialEq<T>,
+{
+    fn eq(&self, other: &Sentential<N, T>) -> bool {
+        use NodeChild::*;
+        use Sentential::*;
+        match (self, other) {
+            (Data(s), Terminal(t)) if *s == *t => true,
+            (
+                Child(ParseTree {
+                    rule: Some((rule, _)),
+                    childs: _,
+                }),
+                NoneTerminal(nt),
+            ) if *rule == *nt => true,
+            _ => false,
+        }
+    }
+}

src/lib.rs

@@ -5,7 +5,7 @@ pub mod scanner;
 pub mod prelude {
     pub use crate::cfg::*;
     pub use crate::cfg::ll_grammar::*;
-    pub use crate::cfg::lr0::*;
+    pub use crate::cfg::lr0_grammar::*;
     pub use crate::cfg_grammar;
     pub use crate::double_enum;
     pub use crate::scanner::*;