made a few things a bit more flexible

.env

@@ -16,3 +16,6 @@ MEILI_MASTER_KEY=CHANGE_ME
 
 # general
 # LOG_LEVEL = debug # can be uaed to overide the log level
+
+MEILI_OLLAMA_URL=http://131.159.2.46:11434/api/embeddings
+MEILI_HYBRID_RATIO=0.1

docker-compose.local.yml

@@ -48,6 +48,8 @@ services:
       CONNECTUM_OAUTH_CLIENT_SECRET: ${CONNECTUM_OAUTH_CLIENT_SECRET}
       GITHUB_TOKEN: ${GITHUB_TOKEN}
       JWT_KEY: ${JWT_KEY}
+      MEILI_OLLAMA_URL: ${MEILI_OLLAMA_URL}
+      MEILI_HYBRID_RATIO: ${MEILI_HYBRID_RATIO}
     depends_on:
       meilisearch:
         condition: service_healthy

docker-compose.yml

@@ -147,6 +147,8 @@ services:
       CONNECTUM_OAUTH_CLIENT_SECRET: ${CONNECTUM_OAUTH_CLIENT_SECRET}
       GITHUB_TOKEN: ${GITHUB_TOKEN}
       JWT_KEY: ${JWT_KEY}
+      MEILI_OLLAMA_URL: ${MEILI_OLLAMA_URL}
+      MEILI_HYBRID_RATIO: ${MEILI_HYBRID_RATIO}
     depends_on:
       meilisearch:
         condition: service_healthy

server/Cargo.toml

@@ -48,7 +48,7 @@ chrono = { version = "0.4.38", default-features = false, features = ["serde"] }
 
 # search
 # IFF (open for 4 Months now) meilisearch merges my PR, change back to the main release
-meilisearch = { git = "https://github.com/commanderstorm/meilisearch-rust.git", branch = "vector-search-embedder" }
+meilisearch-sdk = { git = "https://github.com/commanderstorm/meilisearch-rust.git", branch = "vector-search-embedder" }
 logos = "0.14.1"
 regex = "1.10.6"
 

server/src/search/search_executor/query.rs

@@ -1,13 +1,13 @@
+use crate::search::search_executor::parser::{Filter, ParsedQuery, TextToken};
+use crate::search::{Highlighting, Limits};
+use actix_web::dev::ResourcePath;
 use meilisearch_sdk::client::Client;
 use meilisearch_sdk::errors::Error;
 use meilisearch_sdk::indexes::Index;
 use meilisearch_sdk::search::{MultiSearchResponse, SearchQuery, Selectors};
 use serde::Deserialize;
 use std::fmt::{Debug, Formatter};
 
-use crate::search::search_executor::parser::{Filter, ParsedQuery, TextToken};
-use crate::search::{Highlighting, Limits};
-
 #[derive(Deserialize, Default, Clone)]
 #[allow(dead_code)]
 pub(super) struct MSHit {
@@ -178,12 +178,20 @@ impl GeoEntryQuery {
         &'b self,
         entries: &'a Index,
     ) -> SearchQuery<'a, meilisearch_sdk::DefaultHttpClient> {
+        let semantic_ratio = match std::env::var("MEILI_HYBRID_RATIO").ok() {
+            None => None,
+            Some(s) if s.trim().is_empty() => None,
+            Some(s) => Some(s),
+        }
+        .unwrap_or("0.1".into())
+        .parse()
+        .unwrap_or(0.1);
         SearchQuery::new(entries)
             .with_facets(Selectors::Some(&["facet"]))
             .with_highlight_pre_tag(&self.highlighting.pre)
             .with_highlight_post_tag(&self.highlighting.post)
             .with_attributes_to_highlight(Selectors::Some(&["name"]))
-            .with_hybrid("default", 0.1)
+            .with_hybrid("default", semantic_ratio)
             .build()
     }
 
@@ -208,11 +216,19 @@ impl GeoEntryQuery {
         entries: &'a Index,
         query: &'a str,
     ) -> SearchQuery<'a, meilisearch_sdk::DefaultHttpClient> {
+        let semantic_ratio = match std::env::var("MEILI_HYBRID_RATIO").ok() {
+            None => None,
+            Some(s) if s.trim().is_empty() => None,
+            Some(s) => Some(s),
+        }
+        .unwrap_or("0.1".into())
+        .parse()
+        .unwrap_or(0.1);
         self.common_query(entries)
             .with_query(query)
             .with_limit(2 * self.limits.buildings_count) // we might do reordering later
             .with_filter(&self.filters.buildings)
-            .with_hybrid("default", 0.1)
+            .with_hybrid("default", semantic_ratio)
             .build()
     }
 

server/src/setup/meilisearch.rs

@@ -1,5 +1,5 @@
 use meilisearch_sdk::client::Client;
-use meilisearch_sdk::settings::{Embedder, HuggingFaceEmbedderSettings, Settings};
+use meilisearch_sdk::settings::{Embedder, OllamaEmbedderSettings, Settings};
 use meilisearch_sdk::tasks::Task;
 use serde_json::Value;
 use std::collections::HashMap;
@@ -66,15 +66,11 @@ pub async fn setup(client: &Client, vector_search: bool) -> anyhow::Result<()> {
         .wait_for_completion(client, POLLING_RATE, TIMEOUT)
         .await?;
     let entries = client.index("entries");
-    let en_embedder =Embedder::HuggingFace(HuggingFaceEmbedderSettings{
-        model: Some("BAAI/bge-base-en-v1.5".to_string()),
+    let en_embedder = Embedder::Ollama(OllamaEmbedderSettings{
+        api_key: None,
+        url: match std::env::var("MEILI_OLLAMA_URL").ok(){ None=>None,Some(s) if s.trim().is_empty()=> None,Some(s)=>Some(s)},
+        model: "mxbai-embed-large".to_string(),
         document_template: Some("A room titled '{{doc.name}}' with type '{{doc.type_common_name}}' used as '{{doc.usage}}'".to_string()),
-        ..Default::default()
-    });
-    let _de_embedder=Embedder::HuggingFace(HuggingFaceEmbedderSettings{
-        model: Some("google-bert/bert-base-german-cased".to_string()),
-        document_template: Some("Ein Raum '{{doc.name}}' vom typ '{{doc.type_common_name}}' benutzt als '{{doc.usage}}'".to_string()),
-        ..Default::default()
     });
 
     let mut settings = Settings::new()