feat: Add GeoJSON Analyzer (#87)

* Initial support for GeoJson Analyzer (3.8.*)

* Add test

* Fix instructions

Co-authored-by: Evan Chan <evan@urbanlogiq.com>

README.md

@@ -314,6 +314,10 @@ println!("{:?}", resp);
 
 Contributions and feed back are welcome following Github workflow.
 
+Setup instructions:
+1. Install a local ArangoDB, version 3.8 or above, port 8529.  Use `docker compose`.
+2. Run `tests/init_db.sh`
+
 ### License
 
 `arangors` is provided under the MIT license. See [LICENSE](./LICENSE).

src/analyzer.rs

@@ -24,6 +24,14 @@ pub enum NgramStreamType {
     Utf8,
 }
 
+#[derive(Debug, Serialize, Deserialize, PartialEq)]
+#[serde(rename_all = "lowercase")]
+pub enum GeoJsonType {
+    Shape,
+    Centroid,
+    Point,
+}
+
 #[derive(Debug, Serialize, Deserialize, TypedBuilder, PartialEq)]
 #[builder(doc)]
 pub struct DelimiterAnalyzerProperties {
@@ -114,6 +122,17 @@ pub struct TextAnalyzerProperties {
     pub stemming: Option<bool>,
 }
 
+#[derive(Debug, Serialize, Deserialize, TypedBuilder, PartialEq)]
+#[builder(doc)]
+pub struct GeoJsonAnalyzerProperties {
+    /// Whether to index all GeoJSON geometry types, just the centroid, or just points
+    #[serde(skip_serializing_if = "Option::is_none")]
+    #[builder(default, setter(strip_option))]
+    pub r#type: Option<GeoJsonType>,
+
+    // Skip the options as they "generally should remain unchanged"
+}
+
 #[derive(Debug, Serialize, Deserialize, PartialEq)]
 #[serde(rename_all = "camelCase", tag = "type")]
 pub enum AnalyzerInfo {
@@ -173,6 +192,16 @@ pub enum AnalyzerInfo {
         #[serde(skip_serializing_if = "Option::is_none")]
         properties: Option<TextAnalyzerProperties>,
     },
+
+    Geojson {
+        name: String,
+
+        #[serde(skip_serializing_if = "Option::is_none")]
+        features: Option<Vec<AnalyzerFeature>>,
+
+        #[serde(skip_serializing_if = "Option::is_none")]
+        properties: Option<GeoJsonAnalyzerProperties>,
+    },
 }
 
 #[derive(Debug, Serialize, Deserialize)]

tests/analyzer.rs

@@ -10,7 +10,7 @@ use uclient::ClientExt;
 
 use arangors::analyzer::{
     AnalyzerCase, AnalyzerFeature, AnalyzerInfo, NgramAnalyzerProperties, NgramStreamType,
-    NormAnalyzerProperties,
+    NormAnalyzerProperties, GeoJsonAnalyzerProperties, GeoJsonType,
 };
 use arangors::{
     collection::{
@@ -65,6 +65,24 @@ async fn create_ngram_analyzer<C: ClientExt>(
     database.create_analyzer(info).await
 }
 
+#[maybe_async]
+async fn create_geo_analyzer<C: ClientExt>(
+    database: &Database<C>,
+    analyzer_name: String,
+) -> Result<AnalyzerInfo, ClientError> {
+    let info = AnalyzerInfo::Geojson {
+        name: analyzer_name,
+        features: Some(vec![AnalyzerFeature::Frequency, AnalyzerFeature::Norm]),
+        properties: Some(
+            GeoJsonAnalyzerProperties::builder()
+                .r#type(GeoJsonType::Centroid)
+                .build(),
+        ),
+    };
+
+    database.create_analyzer(info).await
+}
+
 #[maybe_async::test(
     any(feature = "reqwest_blocking"),
     async(any(feature = "reqwest_async"), tokio::test),
@@ -109,6 +127,28 @@ async fn test_create_and_drop_ngram_analyzer() {
     assert_eq!(result.is_err(), false);
 }
 
+#[maybe_async::test(
+    any(feature = "reqwest_blocking"),
+    async(any(feature = "reqwest_async"), tokio::test),
+    async(any(feature = "surf_async"), async_std::test)
+)]
+async fn test_create_and_drop_geo_analyzer() {
+    test_setup();
+    let analyzer_name = "test_analyzer_geo_create".to_string();
+    let conn = connection().await;
+    let database = conn.db("test_db").await.unwrap();
+
+    let analyzer = create_geo_analyzer(&database, analyzer_name.clone()).await;
+
+    trace!("{:?}", analyzer);
+
+    assert_eq!(analyzer.is_err(), false);
+
+    let result = database.drop_analyzer(&analyzer_name).await;
+
+    assert_eq!(result.is_err(), false);
+}
+
 #[maybe_async::test(
     any(feature = "reqwest_blocking"),
     async(any(feature = "reqwest_async"), tokio::test),