optimized up to section 5.11

src/prolog/codegen.rs

@@ -3,7 +3,8 @@ use prolog::iterators::{FactIterator, QueryIterator};
 
 use std::cell::Cell;
 use std::cmp::max;
-use std::collections::HashMap;
+use std::collections::{HashMap, VecDeque};
+use std::hash::Hash;
 use std::vec::Vec;
 
 trait CompilationTarget<'a> {
@@ -244,6 +245,239 @@ impl<'a> TermMarker<'a> {
     }
 }
 
+#[derive(Clone, Copy)]
+enum IntIndex {
+    External(usize), Fail, Internal(usize)
+}
+
+impl Into<usize> for IntIndex {
+    fn into(self) -> usize {
+        match self {
+            IntIndex::Internal(i) => i,
+            _ => 0
+        }
+    }
+}
+
+struct CodeOffsets {
+    constants:  HashMap<Constant, ThirdLevelIndex>,
+    lists: ThirdLevelIndex,
+    structures: HashMap<(Atom, usize), ThirdLevelIndex>
+}
+
+impl CodeOffsets {
+    fn new() -> Self {
+        CodeOffsets {
+            constants: HashMap::new(),
+            lists: Vec::new(),
+            structures: HashMap::new()
+        }
+    }
+
+    fn cap_choice_seq_with_trust(prelude: &mut ThirdLevelIndex) {
+        prelude.last_mut().map(|instr| {
+            match instr {
+                &mut IndexedChoiceInstruction::Retry(i) =>
+                    *instr = IndexedChoiceInstruction::Trust(i),
+                _ => {}
+            };
+        });
+    }
+
+    fn add_index(is_first_index: bool, index: usize) -> IndexedChoiceInstruction {
+        if is_first_index {
+            IndexedChoiceInstruction::Try(index)
+        } else {
+            IndexedChoiceInstruction::Retry(index)
+        }
+    }
+
+    fn index_term(&mut self, first_arg: &Term, index: usize) {
+        match first_arg {
+            &Term::Clause(_, ref name, ref terms) => {
+                let code = self.structures.entry((name.clone(), terms.len()))
+                               .or_insert(Vec::new());
+
+                let is_initial_index = code.is_empty();
+                code.push(Self::add_index(is_initial_index, index));
+            },
+            &Term::Cons(_, _, _) => {
+                let is_initial_index = self.lists.is_empty();
+                self.lists.push(Self::add_index(is_initial_index, index));
+            },
+            &Term::Constant(_, ref constant) => {
+                let code = self.constants.entry(constant.clone())
+                               .or_insert(Vec::new());
+
+                let is_initial_index = code.is_empty();
+                code.push(Self::add_index(is_initial_index, index));
+            },
+            _ => {}
+        };
+    }
+
+    fn second_level_index<Index>(indices: HashMap<Index, ThirdLevelIndex>,
+                                 prelude: &mut CodeDeque)
+                                 -> HashMap<Index, IntIndex>
+        where Index: Eq + Hash
+    {
+        let mut index_locs = HashMap::new();
+
+        for (key, mut code) in indices.into_iter() {
+            if code.len() > 1 {
+                index_locs.insert(key, IntIndex::Internal(prelude.len()));
+                Self::cap_choice_seq_with_trust(&mut code);
+                prelude.extend(code.into_iter().map(|code| Line::from(code)));
+            } else {
+                code.first().map(|i| {
+                    index_locs.insert(key, IntIndex::External(i.offset()));
+                });
+            }
+        }
+
+        index_locs
+    }
+
+    fn no_indices(&self) -> bool {
+        let no_constants = self.constants.is_empty();
+        let no_structures = self.structures.is_empty();
+        let no_lists = self.lists.is_empty();
+
+        no_constants && no_structures && no_lists
+    }
+
+    fn flatten_index<Index>(index: HashMap<Index, IntIndex>, len: usize)
+                            -> HashMap<Index, usize>
+        where Index: Eq + Hash
+    {
+        let mut flattened_index = HashMap::new();
+
+        for (key, int_index) in index.into_iter() {
+            match int_index {
+                IntIndex::External(offset) => {
+                    flattened_index.insert(key, offset + len + 1);
+                },
+                IntIndex::Internal(offset) => {
+                    flattened_index.insert(key, offset + 1);
+                },
+                _ => {}
+            };
+        }
+
+        flattened_index
+    }
+
+    fn switch_on_constant(con_ind: HashMap<Constant, ThirdLevelIndex>,
+                          prelude: &mut CodeDeque)
+                          -> IntIndex
+    {
+        let con_ind = Self::second_level_index(con_ind, prelude);
+
+        if con_ind.len() > 1 {
+            let index = Self::flatten_index(con_ind, prelude.len());
+            let instr = IndexingInstruction::SwitchOnConstant(index.len(), index);
+
+            prelude.push_front(Line::from(instr));
+
+            IntIndex::Internal(1)
+        } else {
+            con_ind.values().next()
+                   .map(|i| *i)
+                   .unwrap_or(IntIndex::Fail)
+        }
+    }
+
+    fn switch_on_list(mut lists: ThirdLevelIndex, prelude: &mut CodeDeque) -> IntIndex
+    {
+        if lists.len() > 1 {
+            Self::cap_choice_seq_with_trust(&mut lists);
+            prelude.extend(lists.into_iter().map(|i| Line::from(i)));
+            IntIndex::Internal(0)
+        } else {
+            lists.first()
+                 .map(|i| IntIndex::External(i.offset()))
+                 .unwrap_or(IntIndex::Fail)
+        }
+    }
+
+    fn switch_on_structure(str_ind: HashMap<(Atom, usize), ThirdLevelIndex>,
+                           prelude: &mut CodeDeque)
+                           -> IntIndex
+    {
+        let str_ind = Self::second_level_index(str_ind, prelude);
+
+        if str_ind.len() > 1 {
+            let index = Self::flatten_index(str_ind, prelude.len());
+            let instr = IndexingInstruction::SwitchOnStructure(index.len(), index);
+
+            prelude.push_front(Line::from(instr));
+
+            IntIndex::Internal(1)
+        } else {
+            str_ind.values().next()
+                   .map(|i| *i)
+                   .unwrap_or(IntIndex::Fail)
+        }
+    }
+
+    fn add_indices(self, code: &mut Code, mut code_body: Code)
+    {
+        if self.no_indices() {
+            *code = code_body;
+            return;
+        }
+
+        let mut prelude = VecDeque::new();
+
+        let lst_step = Self::switch_on_list(self.lists, &mut prelude);
+        let lst_offset = prelude.len();
+
+        let str_step = Self::switch_on_structure(self.structures, &mut prelude);
+        let con_step = Self::switch_on_constant(self.constants, &mut prelude);
+
+        let prelude_length = prelude.len();
+
+        for (index, line) in prelude.iter_mut().enumerate() {
+            match line {
+                  &mut Line::IndexedChoice(IndexedChoiceInstruction::Try(ref mut i))
+                | &mut Line::IndexedChoice(IndexedChoiceInstruction::Retry(ref mut i))
+                | &mut Line::IndexedChoice(IndexedChoiceInstruction::Trust(ref mut i)) =>
+                    *i += prelude_length - index,
+                _ => {}
+            }
+        }
+
+        let str_step = match str_step {
+            IntIndex::External(o) => o + prelude.len() + 1,
+            IntIndex::Fail => 0,
+            IntIndex::Internal(_) => match con_step {
+                IntIndex::Internal(_) => 2,
+                _ => 1
+            }
+        };
+        let con_step = match con_step {
+            IntIndex::External(offset) => offset + prelude.len() + 1,
+            IntIndex::Fail => 0,
+            IntIndex::Internal(offset) => offset,
+        };
+        let lst_step = match lst_step {
+            IntIndex::External(o) => o + prelude.len() + 1,
+            IntIndex::Fail => 0,
+            IntIndex::Internal(_) => prelude.len() - lst_offset + 1
+        };
+
+        let switch_instr = IndexingInstruction::SwitchOnTerm(prelude.len() + 1,
+                                                             con_step,
+                                                             lst_step,
+                                                             str_step);
+
+        prelude.push_front(Line::from(switch_instr));
+
+        *code = Vec::from(prelude);
+        code.append(&mut code_body);
+    }
+}
+
 #[derive(Copy, Clone)]
 enum VarStatus {
     New, Old, Permanent(usize)
@@ -643,7 +877,7 @@ impl<'a> CodeGenerator<'a> {
         Self::add_conditional_call(&mut body, p1, perm_vars);
 
         body = clauses.iter().enumerate()
-            .map(|(i, ref term)| {
+            .map(|(i, term)| {
                 let num_vars = Self::vars_above_threshold(&vs, i+1);
                 self.compile_internal_query(term, num_vars)
             })
@@ -705,13 +939,13 @@ impl<'a> CodeGenerator<'a> {
         self.update_var_count(term.breadth_first_iter());
 
         let mut code = Vec::new();
-        
+
         if term.is_clause() {
-            let mut compiled_fact = self.compile_target(term, false);            
+            let mut compiled_fact = self.compile_target(term, false);
             Self::mark_unsafe_fact_vars(&mut compiled_fact, self.vars());
             code.push(Line::Fact(compiled_fact));
         }
-        
+
         let proceed = Line::Control(ControlInstruction::Proceed);
 
         code.push(proceed);
@@ -725,11 +959,11 @@ impl<'a> CodeGenerator<'a> {
 
         let mut code = Vec::new();
 
-        if term.is_clause() {                
+        if term.is_clause() {
             let compiled_query = Line::Query(self.compile_target(term, false));
             code.push(compiled_query);
         }
-        
+
         Self::add_conditional_call(&mut code, term, index);
 
         code
@@ -741,19 +975,45 @@ impl<'a> CodeGenerator<'a> {
 
         let mut code = Vec::new();
 
-        if term.is_clause() {                
+        if term.is_clause() {
             let compiled_query = Line::Query(self.compile_target(term, false));
             code.push(compiled_query);
         }
-        
+
         Self::add_conditional_call(&mut code, term, 0);
 
         code
     }
 
-    pub fn compile_predicate(&mut self, clauses: &'a Vec<PredicateClause>) -> Code
+    fn split_predicate(clauses: &Vec<PredicateClause>) -> Vec<(usize, usize)>
     {
-        let mut code = Vec::new();
+        let mut subseqs = Vec::new();
+        let mut left_index = 0;
+
+        for (right_index, clause) in clauses.iter().enumerate() {
+            if let Some(&Term::Var(_, _)) = clause.first_arg() {
+                if left_index < right_index {
+                    subseqs.push((left_index, right_index));
+                }
+
+                subseqs.push((right_index, right_index + 1));
+                left_index = right_index + 1;
+            }
+        }
+
+        if left_index < clauses.len() {
+            subseqs.push((left_index, clauses.len()));
+        }
+
+        subseqs
+    }
+
+    fn compile_pred_subseq(&mut self, clauses: &'a [PredicateClause]) -> Code
+    {
+        let mut code_body = Vec::new();
+        let mut code_offsets = CodeOffsets::new();
+
+        let multi_clause = clauses.len() > 1;
 
         for (i, clause) in clauses.iter().enumerate() {
             self.marker.reset();
@@ -765,14 +1025,50 @@ impl<'a> CodeGenerator<'a> {
                     self.compile_rule(rule)
             };
 
-            let choice = match i {
-                0 => ChoiceInstruction::TryMeElse(clause_code.len() + 1),
-                _ if i == clauses.len() - 1 => ChoiceInstruction::TrustMe,
-                _ => ChoiceInstruction::RetryMeElse(clause_code.len() + 1)
-            };
+            if multi_clause {
+                let choice = match i {
+                    0 => ChoiceInstruction::TryMeElse(clause_code.len() + 1),
+                    _ if i == clauses.len() - 1 => ChoiceInstruction::TrustMe,
+                    _ => ChoiceInstruction::RetryMeElse(clause_code.len() + 1)
+                };
 
-            code.push(Line::Choice(choice));
-            code.append(&mut clause_code);
+                code_body.push(Line::Choice(choice));
+            }
+
+            clause.first_arg().map(|arg| {
+                let index = code_body.len();
+                code_offsets.index_term(arg, index);
+            });
+
+            code_body.append(&mut clause_code);
+        }
+
+        let mut code = Vec::new();
+
+        code_offsets.add_indices(&mut code, code_body);
+        code
+    }
+
+    pub fn compile_predicate(&mut self, clauses: &'a Vec<PredicateClause>) -> Code
+    {
+        let mut code = Vec::new();
+        let split_pred = Self::split_predicate(clauses);
+        let multi_seq  = split_pred.len() > 1;
+
+        for &(l, r) in split_pred.iter() {
+            let mut code_segment = self.compile_pred_subseq(&clauses[l .. r]);
+
+            if multi_seq {
+                let choice = match l {
+                    0 => ChoiceInstruction::TryMeElse(code_segment.len() + 1),
+                    _ if r == clauses.len() => ChoiceInstruction::TrustMe,
+                    _ => ChoiceInstruction::RetryMeElse(code_segment.len() + 1)
+                };
+
+                code.push(Line::Choice(choice));
+            }
+
+            code.append(&mut code_segment);
         }
 
         code