feature: recognize domain specific words

frontends/search/src/components/ResultsPage.tsx

@@ -294,6 +294,8 @@ const ResultsPage = (props: ResultsPageProps) => {
             max: props.search.debounced.twoTypoWordRangeMax,
           },
           disable_on_word: props.search.debounced.disableOnWords,
+          prioritize_domain_specifc_words:
+            props.search.debounced.prioritize_domain_specifc_words,
         },
         highlight_options: {
           highlight_results: props.search.debounced.highlightResults ?? true,

frontends/search/src/components/SearchForm.tsx

@@ -1058,6 +1058,7 @@ const SearchForm = (props: {
                                   twoTypoWordRangeMax: null,
                                   disableOnWords: [],
                                   typoTolerance: false,
+                                  prioritize_domain_specifc_words: true,
                                   highlightResults: true,
                                   highlightDelimiters: ["?", ".", "!"],
                                   highlightMaxLength: 8,
@@ -1219,6 +1220,26 @@ const SearchForm = (props: {
                             }}
                           />
                         </div>
+                        <div class="flex items-center justify-between space-x-2 p-1">
+                          <label>Auto-require Domain Keywords</label>
+                          <input
+                            class="h-4 w-4"
+                            type="checkbox"
+                            checked={
+                              tempSearchValues()
+                                .prioritize_domain_specifc_words ?? false
+                            }
+                            onChange={(e) => {
+                              setTempSearchValues((prev) => {
+                                return {
+                                  ...prev,
+                                  prioritize_domain_specifc_words:
+                                    e.target.checked,
+                                };
+                              });
+                            }}
+                          />
+                        </div>
                         <div class="items flex justify-between space-x-2 p-1">
                           <label>One typo min word length:</label>
                           <input

frontends/search/src/hooks/useSearch.ts

@@ -65,6 +65,7 @@ export interface SearchOptions {
   oneTypoWordRangeMax: number | null;
   twoTypoWordRangeMin: number;
   twoTypoWordRangeMax: number | null;
+  prioritize_domain_specifc_words: boolean | null;
   disableOnWords: string[];
   sort_by: SortByField | SortBySearchType;
   pageSize: number;
@@ -102,6 +103,7 @@ const initalState: SearchOptions = {
   oneTypoWordRangeMax: 8,
   twoTypoWordRangeMin: 8,
   twoTypoWordRangeMax: null,
+  prioritize_domain_specifc_words: true,
   disableOnWords: [],
   highlightResults: true,
   highlightStrategy: "exactmatch",
@@ -140,6 +142,8 @@ const fromStateToParams = (state: SearchOptions): Params => {
     oneTypoWordRangeMax: state.oneTypoWordRangeMax?.toString() ?? "8",
     twoTypoWordRangeMin: state.twoTypoWordRangeMin.toString(),
     twoTypoWordRangeMax: state.twoTypoWordRangeMax?.toString() ?? "",
+    prioritize_domain_specifc_words:
+      state.prioritize_domain_specifc_words?.toString() ?? "",
     disableOnWords: state.disableOnWords.join(","),
     highlightStrategy: state.highlightStrategy,
     highlightResults: state.highlightResults.toString(),
@@ -184,6 +188,8 @@ const fromParamsToState = (
     oneTypoWordRangeMax: parseIntOrNull(params.oneTypoWordRangeMax),
     twoTypoWordRangeMin: parseInt(params.oneTypoWordRangeMin ?? "8"),
     twoTypoWordRangeMax: parseIntOrNull(params.twoTypoWordRangeMax),
+    prioritize_domain_specifc_words:
+      (params.prioritize_domain_specifc_words ?? "true") === "true",
     disableOnWords: params.disableOnWords?.split(",") ?? [],
     highlightResults: (params.highlightResults ?? "true") === "true",
     highlightStrategy: isHighlightStrategy(params.highlightStrategy)

server/Cargo.toml

@@ -162,12 +162,10 @@ murmur3 = "0.5.2"
 tantivy = "0.22.0"
 strsim = "0.11.1"
 levenshtein_automata = "0.2.1"
-bktree = "1.0.1"
 flate2 = "1.0.31"
 bincode = "1.3"
 rayon = "1.10.0"
 crossbeam = "0.8.4"
-bloomfilter = "1.0.14"
 dashmap = "6.0.1"
 
 

server/src/data/models.rs

@@ -5296,6 +5296,8 @@ pub struct TypoOptions {
     pub two_typo_word_range: Option<TypoRange>,
     /// Words that should not be corrected. If not specified, this defaults to an empty list.
     pub disable_on_word: Option<Vec<String>>,
+    /// Auto-require non-english words present in the dataset to exist in each results chunk_html text. If not specified, this defaults to true.
+    pub prioritize_domain_specifc_words: Option<bool>,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, ToSchema, Default)]