fix: total ordering for Priority

Author: wash2

service/src/lib.rs

@@ -572,8 +572,8 @@ impl<O: futures::Sink<Response> + Unpin> Service<O> {
                     Priority {
                         plugin_priority: plg.config.query.priority,
                         match_score: calculate_weight(sr, query),
-                        recent_use_index: ex.as_ref().map(|s| recent.get_recent(s)).unwrap_or(0),
-                        use_freq: ex.as_ref().map(|s| recent.get_freq(s)).unwrap_or(0),
+                        recent_score: ex.as_ref().map(|s| recent.get_recent(s)).unwrap_or(0.),
+                        freq_score: ex.as_ref().map(|s| recent.get_freq(s)).unwrap_or(0.),
                         execlen: sr.name.len(),
                     }
                 };

service/src/priority.rs

@@ -6,62 +6,50 @@ use crate::PluginPriority;
 pub struct Priority {
     pub plugin_priority: PluginPriority,
     pub match_score: f64,
-    pub recent_use_index: usize,
-    pub use_freq: usize,
+    pub recent_score: f64,
+    pub freq_score: f64,
     pub execlen: usize,
 }
 
-fn signum(val: i32) -> f64 {
-    if val > 0 {
-        return 1.0;
-    }
-    if val < 0 {
-        return -1.0;
-    }
-    0.0
+fn falloff(x: f64) -> f64 {
+    x.clamp(0., 1.).powi(3)
 }
 
 impl Priority {
-    fn compute_value(&self, other: &Self) -> f64 {
-        // increases compared jw-score if this search result
-        // was activated more frequent or recent by constant values
-        let score = self.match_score
-            + 0.06 * signum(self.recent_use_index as i32 - other.recent_use_index as i32)
-            + 0.03 * signum(self.use_freq as i32 - other.use_freq as i32);
-        // score cannot surpass exact matches
-        if self.match_score < 1.0 {
-            return score.min(0.99);
-        }
-
-        score
+    fn compute_value(&self) -> f64 {
+        let score = if self.match_score > 1. {
+            self.match_score + 0.1
+        } else {
+            self.match_score
+        };
+        score + 0.06 * falloff(self.recent_score) + 0.03 * falloff(self.freq_score)
     }
 }
 
 impl PartialEq for Priority {
     fn eq(&self, other: &Self) -> bool {
         self.plugin_priority == other.plugin_priority
-            && self.compute_value(other) == other.match_score
+            && self.compute_value() == other.compute_value()
             && self.execlen == other.execlen
     }
 }
 
 impl Eq for Priority {}
 
 impl PartialOrd for Priority {
-    #[allow(clippy::non_canonical_partial_ord_impl)]
     fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
-        // todo: what is going on here ?
-        (
-            other.plugin_priority,
-            self.compute_value(other),
-            self.execlen,
-        )
-            .partial_cmp(&(self.plugin_priority, other.match_score, other.execlen))
+        Some(self.cmp(other))
     }
 }
 
 impl Ord for Priority {
     fn cmp(&self, other: &Self) -> Ordering {
-        self.partial_cmp(other).unwrap()
+        match other.plugin_priority.cmp(&self.plugin_priority) {
+            Ordering::Equal => match self.compute_value().total_cmp(&other.compute_value()) {
+                Ordering::Equal => self.execlen.cmp(&other.execlen),
+                p => p,
+            },
+            p => p,
+        }
     }
 }

service/src/recent.rs

@@ -15,6 +15,10 @@ const LONGTERM_CAP: usize = 100;
 pub struct RecentUseStorage {
     long_term: HashMap<u64, usize>,
     short_term: HashMap<u64, usize>,
+    /// used for normalizing individual scores
+    max_long_term: usize,
+    /// used for normalizing individual scores
+    max_short_term: usize,
 }
 
 fn hash_key<K: Hash>(key: K) -> u64 {
@@ -26,8 +30,11 @@ fn hash_key<K: Hash>(key: K) -> u64 {
 impl RecentUseStorage {
     pub fn add<K: Hash>(&mut self, exec: &K) {
         let key = hash_key(exec);
-        *self.long_term.entry(key).or_insert(0) += 1;
+        let entry = self.long_term.entry(key).or_insert(0);
+        *entry += 1;
+        self.max_long_term = self.max_long_term.max(*entry);
         let short_term_idx = self.short_term.values().max().unwrap_or(&0) + 1;
+        self.max_short_term = self.max_short_term.max(short_term_idx);
         self.short_term.insert(key, short_term_idx);
         self.trim();
     }
@@ -39,6 +46,8 @@ impl RecentUseStorage {
         }
 
         while self.long_term.values().sum::<usize>() > LONGTERM_CAP {
+            self.max_long_term /= 2;
+
             let mut delete_keys = Vec::new();
             for (k, v) in &mut self.long_term {
                 *v /= 2;
@@ -52,12 +61,14 @@ impl RecentUseStorage {
         }
     }
 
-    pub fn get_recent<K: Hash>(&self, exec: &K) -> usize {
-        self.short_term.get(&hash_key(exec)).copied().unwrap_or(0)
+    pub fn get_recent<K: Hash>(&self, exec: &K) -> f64 {
+        self.short_term.get(&hash_key(exec)).copied().unwrap_or(0) as f64
+            / (self.max_short_term.max(1) as f64)
     }
 
-    pub fn get_freq<K: Hash>(&self, exec: &K) -> usize {
-        self.long_term.get(&hash_key(exec)).copied().unwrap_or(0)
+    pub fn get_freq<K: Hash>(&self, exec: &K) -> f64 {
+        self.long_term.get(&hash_key(exec)).copied().unwrap_or(0) as f64
+            / (self.max_long_term.max(1) as f64)
     }
 }
 
@@ -80,9 +91,13 @@ impl<'de> Deserialize<'de> for RecentUseStorage {
         type SerType = (HashMap<u64, usize>, Vec<u64>);
         let (long_term, stv) = SerType::deserialize(deserializer)?;
         let short_term: HashMap<_, _> = stv.into_iter().enumerate().map(|(v, k)| (k, v)).collect();
+        let max_long_term = long_term.values().max().copied().unwrap_or(1);
+        let max_short_term = short_term.values().max().copied().unwrap_or(1);
         Ok(RecentUseStorage {
             long_term,
             short_term,
+            max_long_term,
+            max_short_term,
         })
     }
 }