[feat] 0.23.8

izihawa · May 11, 2024 · 815c32f · 815c32f
1 parent 8b06507
commit 815c32f
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 13 deletions.
diff --git a/summa-core/src/components/query_parser/morphology/english.rs b/summa-core/src/components/query_parser/morphology/english.rs
@@ -42,7 +42,7 @@ impl Default for EnglishMorphology {
 }
 
 impl Morphology for EnglishMorphology {
-    fn derive_tenses(&self, word: &str) -> Option<String> {
+    fn derive_tenses(&self, word: &str) -> Option<(String, String)> {
         thread_local! {
             static NOT_A_NOUN: (RegexSet, HashSet<&'static str>) = (RegexSet::new([
                 r"\d$",
@@ -56,9 +56,9 @@ impl Morphology for EnglishMorphology {
             let is_singular = pluralize_rs::is_singular(word);
             let is_plural = pluralize_rs::is_plural(word);
             if is_singular {
-                Some(pluralize_rs::to_plural(word))
+                Some((word.to_string(), pluralize_rs::to_plural(word)))
             } else if is_plural {
-                Some(pluralize_rs::to_singular(word))
+                Some((pluralize_rs::to_singular(word), word.to_string()))
             } else {
                 None
             }

diff --git a/summa-core/src/components/query_parser/morphology/mod.rs b/summa-core/src/components/query_parser/morphology/mod.rs
@@ -1,4 +1,3 @@
-use std::collections::HashSet;
 use summa_proto::proto;
 use tantivy::query::{DisjunctionMaxQuery, Query, TermQuery};
 use tantivy::schema::{Field, FieldType, IndexRecordOption};
@@ -11,7 +10,7 @@ pub use manager::MorphologyManager;
 use crate::components::query_parser::utils::cast_field_to_term;
 
 pub trait Morphology: MorphologyClone + Send + Sync {
-    fn derive_tenses(&self, word: &str) -> Option<String>;
+    fn derive_tenses(&self, word: &str) -> Option<(String, String)>;
     fn derive_spelling(&self, word: &str) -> Option<String>;
 
     fn derive_query(&self, config: proto::MorphologyConfig, field: &Field, full_path: &str, field_type: &FieldType, text: &str) -> Box<dyn Query> {
@@ -21,14 +20,21 @@ pub trait Morphology: MorphologyClone + Send + Sync {
             let term = cast_field_to_term(field, full_path, field_type, text, false);
             return Box::new(TermQuery::new(term, IndexRecordOption::WithFreqs)) as Box<dyn Query>;
         };
-        let mut terms = HashSet::new();
-        terms.insert(text.to_string());
-        if let Some(other_tense_text) = self.derive_tenses(text) {
-            terms.insert(other_tense_text);
+        let mut terms = vec![];
+        if let Some((singular, plural)) = self.derive_tenses(text) {
+            terms.push(singular);
+            terms.push(plural);
+        } else {
+            terms.push(text.to_string());
+        }
+        if let Some(spelling) = self.derive_spelling(&terms[0]) {
+            if let Some((spelling_tense_singular, spelling_tense_plural)) = self.derive_tenses(&spelling) {
+                terms.push(spelling_tense_singular);
+                terms.push(spelling_tense_plural);
+            } else {
+                terms.push(spelling);
+            }
         }
-        terms.extend(terms.iter().filter_map(|t| self.derive_spelling(t)).collect::<Vec<_>>());
-        terms.extend(terms.iter().filter_map(|t| self.derive_tenses(t)).collect::<Vec<_>>());
-        let terms = Vec::from_iter(terms.into_iter());
         if terms.len() == 1 {
             Box::new(TermQuery::new(
                 cast_field_to_term(field, full_path, field_type, &terms[0], false),

diff --git a/summa-server/Cargo.toml b/summa-server/Cargo.toml
@@ -1,7 +1,7 @@
 [package]
 edition = "2021"
 name = "summa-server"
-version = "0.23.7"
+version = "0.23.8"
 license-file = "LICENSE"
 description = "Fast full-text search server"
 homepage = "https://github.com/izihawa/summa"