#40 add fuzzy search

server/src/handlers/search.rs

@@ -2,7 +2,7 @@ use crate::{appstate::AppState, errors::BetterResult};
 use actix_web::{web, HttpResponse};
 use atomic_lib::{urls, Resource, Storelike};
 use serde::Deserialize;
-use std::{ops::RangeBounds, sync::Mutex};
+use std::sync::Mutex;
 use tantivy::{collector::TopDocs, query::QueryParser};
 
 #[derive(Deserialize, Debug)]
@@ -16,8 +16,7 @@ pub struct SearchQuery {
 /// Parses a search query and responds with a list of resources
 pub async fn search_query(
     data: web::Data<Mutex<AppState>>,
-    query: web::Query<SearchQuery>,
-    body: String,
+    params: web::Query<SearchQuery>,
     req: actix_web::HttpRequest,
 ) -> BetterResult<HttpResponse> {
     let context = data
@@ -28,24 +27,41 @@ pub async fn search_query(
     let searcher = context.search_reader.searcher();
     let fields = crate::search::get_schema_fields(&context);
 
-    // construct the query
-    let query_parser = QueryParser::for_index(
-        &context.search_index,
-        vec![
-            fields.subject,
-            // I don't think we need to search in the property
-            // fields.property,
-            fields.value,
-        ],
-    );
-    let tantivy_query = query_parser
-        .parse_query(&query.q)
-        .map_err(|e| format!("Error parsing query {}", e))?;
+    let mut should_fuzzy = true;
+    let query = params.q.clone();
+    // If any of these substrings appear, the user wants an exact / advanced search
+    let dont_fuzz_strings = vec!["*", "AND", "OR", "[", "\""];
+    for dont_fuzz in dont_fuzz_strings {
+        if query.contains(dont_fuzz) {
+            should_fuzzy = false
+        }
+    }
+
+    let query: Box<dyn tantivy::query::Query> = if should_fuzzy {
+        let term = tantivy::Term::from_field_text(fields.value, &params.q);
+        let query = tantivy::query::FuzzyTermQuery::new_prefix(term, 3, true);
+        Box::new(query)
+    } else {
+        // construct the query
+        let query_parser = QueryParser::for_index(
+            &context.search_index,
+            vec![
+                fields.subject,
+                // I don't think we need to search in the property
+                // fields.property,
+                fields.value,
+            ],
+        );
+        let tantivy_query = query_parser
+            .parse_query(&params.q)
+            .map_err(|e| format!("Error parsing query {}", e))?;
+        tantivy_query
+    };
 
     // execute the query
     let top_docs = searcher
-        .search(&tantivy_query, &TopDocs::with_limit(10))
-        .map_err(|e| "Error with creating docs for search")?;
+        .search(&query, &TopDocs::with_limit(10))
+        .map_err(|_e| "Error with creating docs for search")?;
     let mut subjects: Vec<String> = Vec::new();
 
     // convert found documents to resources
@@ -80,6 +96,7 @@ pub async fn search_query(
     // Create a valid atomic data resource
     let mut results_resource = Resource::new(subject);
     results_resource.set_propval(urls::IS_A.into(), vec![urls::ENDPOINT].into(), store)?;
+    results_resource.set_propval(urls::DESCRIPTION.into(), atomic_lib::Value::Markdown("Full text-search endpoint. You can use the keyword `AND` and `OR`, or use `\"` for advanced searches. ".into()), store)?;
     results_resource.set_propval(urls::ENDPOINT_RESULTS.into(), resources.into(), store)?;
     results_resource.set_propval(
         urls::ENDPOINT_PARAMETERS.into(),

server/src/search.rs

@@ -1,6 +1,9 @@
+//! Full-text search, powered by Tantivy.
+//! A folder for the index is stored in the config.
+//! You can see the Endpoint on `http://localhost/search`
+
 use atomic_lib::Resource;
 use atomic_lib::Storelike;
-/// Full-text search, powered by Tantivy.
 use tantivy::schema::*;
 use tantivy::Index;
 use tantivy::IndexWriter;
@@ -10,6 +13,8 @@ use crate::appstate::AppState;
 use crate::config::Config;
 use crate::errors::BetterResult;
 
+/// The actual Schema used for search.
+/// It mimics a single Atom (or Triple).
 pub struct Fields {
     pub subject: Field,
     pub property: Field,
@@ -27,6 +32,7 @@ pub fn build_schema() -> BetterResult<tantivy::schema::Schema> {
     Ok(schema)
 }
 
+/// Creates or reads the index from the `search_index_path` and allocates some heap size.
 pub fn get_index(config: &Config) -> BetterResult<(IndexWriter, Index)> {
     let schema = build_schema()?;
     std::fs::create_dir_all(&config.search_index_path).unwrap();
@@ -39,6 +45,7 @@ pub fn get_index(config: &Config) -> BetterResult<(IndexWriter, Index)> {
     Ok((index_writer, index))
 }
 
+/// Returns the schema for the search index.
 pub fn get_schema_fields(appstate: &AppState) -> Fields {
     let subject = appstate.search_schema.get_field("subject").unwrap();
     let property = appstate.search_schema.get_field("property").unwrap();
@@ -51,6 +58,8 @@ pub fn get_schema_fields(appstate: &AppState) -> Fields {
     }
 }
 
+/// Indexes all resources from the store to search.
+/// At this moment does not remove existing index.
 pub fn add_all_resources(appstate: &AppState) -> BetterResult<()> {
     log::info!("Building search index...");
     for resource in appstate.store.all_resources(true) {
@@ -82,6 +91,6 @@ pub fn get_reader(index: &tantivy::Index) -> BetterResult<tantivy::IndexReader>
         .reader_builder()
         .reload_policy(ReloadPolicy::OnCommit)
         .try_into()
-        .map_err(|e| "Failed getting reader")
+        .map_err(|_e| "Failed getting search reader")
         .unwrap())
 }