fix slop (#2031)

Fix slop by carrying slop so far for multiterms.
Define slop contract in the API

Author: PSeitz

src/query/phrase_query/mod.rs

@@ -184,6 +184,20 @@ pub mod tests {
         Ok(())
     }
 
+    #[test]
+    pub fn test_phrase_score_with_slop_bug_2() -> crate::Result<()> {
+        // fails
+        let index = create_index(&["a x b x c", "a a c"])?;
+        let scores = test_query(2, &index, vec!["a", "b", "c"]);
+        assert_eq!(scores.len(), 1);
+
+        let index = create_index(&["a x b x c", "b c c"])?;
+        let scores = test_query(2, &index, vec!["a", "b", "c"]);
+        assert_eq!(scores.len(), 1);
+
+        Ok(())
+    }
+
     fn test_query(slop: u32, index: &Index, texts: Vec<&str>) -> Vec<f32> {
         let text_field = index.schema().get_field("text").unwrap();
         let searcher = index.reader().unwrap().searcher();
@@ -212,11 +226,33 @@ pub mod tests {
     pub fn test_phrase_score_with_slop_size() -> crate::Result<()> {
         let index = create_index(&["a b e c", "a e e e c", "a e e e e c"])?;
         let scores = test_query(3, &index, vec!["a", "c"]);
+        assert_eq!(scores.len(), 2);
         assert_nearly_equals!(scores[0], 0.29086056);
         assert_nearly_equals!(scores[1], 0.26706287);
         Ok(())
     }
 
+    #[test]
+    pub fn test_phrase_slop() -> crate::Result<()> {
+        let index = create_index(&["a x b c"])?;
+        let scores = test_query(1, &index, vec!["a", "b", "c"]);
+        assert_eq!(scores.len(), 1);
+
+        let index = create_index(&["a x b x c"])?;
+        let scores = test_query(1, &index, vec!["a", "b", "c"]);
+        assert_eq!(scores.len(), 0);
+
+        let index = create_index(&["a b"])?;
+        let scores = test_query(1, &index, vec!["b", "a"]);
+        assert_eq!(scores.len(), 0);
+
+        let index = create_index(&["a b"])?;
+        let scores = test_query(2, &index, vec!["b", "a"]);
+        assert_eq!(scores.len(), 1);
+
+        Ok(())
+    }
+
     #[test]
     pub fn test_phrase_score_with_slop_ordering() -> crate::Result<()> {
         let index = create_index(&[

src/query/phrase_query/phrase_query.rs

@@ -66,6 +66,16 @@ impl PhraseQuery {
     /// Slop allowed for the phrase.
     ///
     /// The query will match if its terms are separated by `slop` terms at most.
+    /// The slop can be considered a budget between all terms.
+    /// E.g. "A B C" with slop 1 allows "A X B C", "A B X C", but not "A X B X C".
+    ///
+    /// Transposition costs 2, e.g. "A B" with slop 1 will not match "B A" but it would with slop 2
+    /// Transposition is not a special case, in the example above A is moved 1 position and B is
+    /// moved 1 position, so the slop is 2.
+    ///
+    /// As a result slop works in both directions, so the order of the terms may changed as long as
+    /// they respect the slop.
+    ///
     /// By default the slop is 0 meaning query terms need to be adjacent.
     pub fn set_slop(&mut self, value: u32) {
         self.slop = value;