Compute max_score using only matching documents

postgres/src/lib.rs

@@ -266,6 +266,43 @@ mod tests {
         assert_eq!(inspected, Some(vec!["rare".to_owned()]));
     }
 
+    #[pg_test]
+    fn max_score_excludes_nonmatching_documents() {
+        for (name, matching, nonmatching, query) in [
+            (
+                "boolean",
+                "beer wine",
+                "beer beer beer beer beer",
+                "beer^1 AND wine^0",
+            ),
+            (
+                "phrase",
+                "beer beer wine",
+                "beer noise beer noise beer noise beer noise wine",
+                "\"beer beer\"^1 AND wine^0",
+            ),
+            (
+                "positional",
+                "beer wine",
+                "wine beer beer beer beer beer",
+                "beer BEFORE wine^0",
+            ),
+        ] {
+            Spi::run(&format!(
+                "CREATE TABLE lite_max_score_{name} (body text);
+                 INSERT INTO lite_max_score_{name} VALUES ('{matching}'), ('{nonmatching}');
+                 CREATE INDEX ON lite_max_score_{name} USING tin (body);"
+            ))
+            .unwrap();
+            let (score, max) = Spi::get_two::<f32, f32>(&format!(
+                "SELECT tin.full_score(ctid), tin.max_score(ctid)
+                 FROM lite_max_score_{name} WHERE body ==> '{query}'"
+            ))
+            .unwrap();
+            assert_eq!(max, score, "{name}");
+        }
+    }
+
     #[pg_test]
     fn full_score_normalization_matches_tin() {
         Spi::run(

postgres/src/score.rs

@@ -26,7 +26,7 @@ use pgrx::{
 use rustc_hash::FxHashMap;
 use std::cell::RefCell;
 use std::ffi::{CStr, CString, c_void};
-use tinql::runtime::{Query, SpanTermSlot, parse_tinql_to_query};
+use tinql::runtime::{Query, SpanTermSlot, evaluate, parse_tinql_to_query, tokenize_doc};
 use tokenizer::{CompiledTokenizerPipeline, Tokenizer};
 
 #[derive(Clone, Debug, Eq, PartialEq)]
@@ -220,7 +220,12 @@ fn build_corpus(
                 scorer.score_count(tf, tokens.len() as u32)
             }
         }));
-        max = max.max(score);
+        let matched = evaluate(&query, &tokenize_doc(&document, &tokenizer))
+            .unwrap_or_else(|error| pgrx::error!("tin score query evaluation failed: {error}"))
+            .matched;
+        if matched {
+            max = max.max(score);
+        }
         by_document.insert(document, score);
     }
     ScoreCorpus {