fix: Remove stop-words from the serve examples

src/automaton.rs

@@ -1,4 +1,5 @@
 use std::ops::Deref;
+
 use fst::Automaton;
 use levenshtein_automata::{
     LevenshteinAutomatonBuilder as LevBuilder,
@@ -50,16 +51,40 @@ impl AutomatonExt for DfaExt {
     }
 }
 
-pub fn build(query: &str) -> DfaExt {
+enum PrefixSetting {
+    Prefix,
+    NoPrefix,
+}
+
+fn build_dfa_with_setting(query: &str, setting: PrefixSetting) -> DfaExt {
+    use self::PrefixSetting::{Prefix, NoPrefix};
+
     let dfa = match query.len() {
-        0 ..= 4 => LEVDIST0.build_prefix_dfa(query),
-        5 ..= 8 => LEVDIST1.build_prefix_dfa(query),
-        _       => LEVDIST2.build_prefix_dfa(query),
+        0 ..= 4 => match setting {
+            Prefix   => LEVDIST0.build_prefix_dfa(query),
+            NoPrefix => LEVDIST0.build_dfa(query),
+        },
+        5 ..= 8 => match setting {
+            Prefix   => LEVDIST1.build_prefix_dfa(query),
+            NoPrefix => LEVDIST1.build_dfa(query),
+        },
+        _ => match setting {
+            Prefix   => LEVDIST2.build_prefix_dfa(query),
+            NoPrefix => LEVDIST2.build_dfa(query),
+        },
     };
 
     DfaExt { query_len: query.len(), automaton: dfa }
 }
 
+pub fn build_prefix_dfa(query: &str) -> DfaExt {
+    build_dfa_with_setting(query, PrefixSetting::Prefix)
+}
+
+pub fn build_dfa(query: &str) -> DfaExt {
+    build_dfa_with_setting(query, PrefixSetting::NoPrefix)
+}
+
 pub trait AutomatonExt: Automaton {
     fn eval<B: AsRef<[u8]>>(&self, s: B) -> Distance;
     fn query_len(&self) -> usize;

src/rank/criterion/mod.rs

@@ -57,6 +57,8 @@ impl Criterion for DocumentId {
     }
 }
 
+// TODO there is too much Box here, can we use
+//      static references or static closures
 pub fn default() -> Vec<Box<dyn Criterion>> {
     vec![
         Box::new(SumOfTypos),

src/rank/mod.rs

@@ -21,6 +21,11 @@ impl Document {
         unsafe { Self::from_sorted_matches(doc, vec![match_]) }
     }
 
+    pub fn from_matches(doc: DocumentId, mut matches: Vec<Match>) -> Self {
+        matches.sort_unstable();
+        unsafe { Self::from_sorted_matches(doc, matches) }
+    }
+
     pub unsafe fn from_sorted_matches(id: DocumentId, matches: Vec<Match>) -> Self {
         Self { id, matches }
     }

src/rank/ranked_stream.rs

@@ -2,7 +2,7 @@ use std::collections::HashMap;
 use std::hash::Hash;
 use std::ops::Range;
 use std::rc::Rc;
-use std::{mem, vec, cmp};
+use std::{mem, vec};
 
 use fnv::FnvHashMap;
 use fst::Streamer;
@@ -11,10 +11,17 @@ use group_by::GroupByMut;
 use crate::automaton::{DfaExt, AutomatonExt};
 use crate::metadata::Metadata;
 use crate::metadata::ops::OpBuilder;
-use crate::rank::criterion::{self, Criterion};
+use crate::rank::criterion::Criterion;
 use crate::rank::Document;
 use crate::{Match, DocumentId};
 
+fn clamp_range<T: Copy + Ord>(range: Range<T>, big: Range<T>) -> Range<T> {
+    Range {
+        start: range.start.min(big.end).max(big.start),
+        end: range.end.min(big.end).max(big.start),
+    }
+}
+
 pub struct Config<'m, C, F> {
     pub metadata: &'m Metadata,
     pub automatons: Vec<DfaExt>,
@@ -67,10 +74,7 @@ impl<'m, C, F> RankedStream<'m, C, F> {
             }
         }
 
-        matches.into_iter().map(|(id, mut matches)| {
-            matches.sort_unstable();
-            unsafe { Document::from_sorted_matches(id, matches) }
-        }).collect()
+        matches.into_iter().map(|(id, matches)| Document::from_matches(id, matches)).collect()
     }
 }
 
@@ -92,10 +96,7 @@ where C: Criterion
             }
         }
 
-        let range = Range {
-            start: cmp::min(range.start, documents.len()),
-            end: cmp::min(range.end, documents.len()),
-        };
+        let range = clamp_range(range, 0..documents.len());
         documents[range].to_vec()
     }