Search for news based on ranking (#35)

graphql/api/src/helpers/handler.rs

@@ -11,6 +11,7 @@ use chrono::Utc;
 use crawler::RssNews;
 use search::Search;
 use std::sync::Arc;
+use tokio::sync::RwLock;
 use url::Url;
 
 #[derive(Debug, Default)]
@@ -42,7 +43,7 @@ fn find_media(from_url: &str) -> Media {
 /// Handling incoming messages from MPSC channel.
 pub async fn process_article(
     article: RssNews,
-    searcher: &Arc<Search>,
+    searcher: &Arc<RwLock<Search>>,
     ranker: &mut Ranker,
 ) -> Result<(), Box<dyn std::error::Error>> {
     let summary = summary::get_summary(&article.content).await?;
@@ -94,7 +95,12 @@ pub async fn process_article(
 
         ranker.add_entry(&news.title).await?;
 
-        searcher.add_entry(news).await.map_err(|e| e.into())
+        searcher
+            .read()
+            .await
+            .add_entry(news)
+            .await
+            .map_err(|e| e.into())
     } else {
         Err("No media found with this URL".into())
     }

graphql/api/src/helpers/ranking.rs

@@ -31,12 +31,12 @@ impl Ranker {
     }
 
     /// Get a single ranking from multiple sources.
-    pub async fn get_rank(&self) -> Result<Vec<String>, Error> {
+    pub async fn get_rank(&self, length: u32) -> Result<Vec<String>, Error> {
         let mut result = Vec::new();
 
         if let Some(squid) = &self.squid {
             result.append(
-                &mut squid.write().await.leaderboard().await.map_err(
+                &mut squid.write().await.leaderboard(length).await.map_err(
                     |error| {
                         Error::new(
                             Unspecified,

graphql/api/src/main.rs

@@ -15,6 +15,7 @@ use search::{Attributes, Search};
 use std::{sync::Arc, time::Duration};
 use strum::IntoEnumIterator;
 use tokio::sync::mpsc;
+use tokio::sync::RwLock;
 use tracing::{debug, error, info, Level};
 use tracing_subscriber::fmt;
 use url::Url;
@@ -120,15 +121,25 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
     crawler.crawl()?;
 
     // Create meilisearch client.
-    let searcher = Arc::new(
+    let searcher = Arc::new(RwLock::new(
         Search::new(
             std::env::var("MEILISEARCH_URL")
                 .unwrap_or("http://localhost:7700".into()),
             std::env::var("MEILISEARCH_URL").ok(),
         )?
         .index("news".into())
         .await,
-    );
+    ));
+
+    // Add country field as a filterable attribute.
+    searcher
+        .write()
+        .await
+        .index
+        .as_ref()
+        .unwrap()
+        .set_filterable_attributes(&["source.country"])
+        .await?;
 
     // Create ranking platform.
     let ranker = Ranker::new().await?;

graphql/api/src/schema/mod.rs

@@ -8,14 +8,15 @@ use mcq::QuestionQuery;
 use news::NewsQuery;
 use search::Search;
 use std::sync::Arc;
+use tokio::sync::RwLock;
 
 pub type Date = DateTime<Utc>;
 
 /// Define the context for the GraphQL schema.
 #[derive(Clone, Debug)]
 pub struct Context {
     /// [`search::Search`] meilisearch client.
-    pub meilisearch: Arc<Search>,
+    pub meilisearch: Arc<RwLock<Search>>,
     /// Custom [`Ranker`] supporting multiple sources.
     pub ranker: Ranker,
 }

graphql/api/src/schema/news.rs

@@ -1,7 +1,7 @@
 use crate::models::{image::Image, news::News, source::Media};
 use crate::schema::Date;
 use crate::Context;
-use juniper::{graphql_object, FieldResult};
+use juniper::{graphql_object, graphql_value, FieldError, FieldResult};
 
 /// Implement GraphQL on News structure.
 #[graphql_object(context = Context, description = "A media article.")]
@@ -49,18 +49,59 @@ pub struct NewsQuery;
 /// Implement the GraphQL object for the news query.
 #[graphql_object(context = Context)]
 impl NewsQuery {
-    /// Get the most relevant news of the day.
+    /// Get the *three* most relevant news of the day.
+    async fn get_top_news(
+        ctx: &Context,
+        #[graphql(description = "ISO 3166-1 alpha-2 country code.")]
+        country: String,
+        #[graphql(description = "Maximum number of articles sent.")] limit: i32,
+    ) -> FieldResult<Vec<News>> {
+        let rank = ctx.ranker.get_rank(limit.try_into()?).await?;
+        let mut most_revelant_news = Vec::new();
+
+        for word in rank {
+            let news = ctx
+                .meilisearch
+                .write()
+                .await
+                .index
+                .as_ref()
+                .ok_or_else(|| FieldError::new("Index not found", graphql_value!({ "internal_error": "Meilisearch index is not selected" })))?
+                .search()
+                .with_query(&word)
+                .with_limit(1)
+                .with_filter(&format!("source.country={:?}", country))
+                .with_attributes_to_search_on(&["title"])
+                .execute::<News>()
+                .await?;
+
+            most_revelant_news.push(news.hits[0].result.clone())
+        }
+
+        Ok(most_revelant_news)
+    }
+
+    /// Get news of the day.
     async fn get_news(
         ctx: &Context,
         #[graphql(description = "ISO 3166-1 alpha-2 country code.")]
-        _country: String,
-        #[graphql(description = "Maximum number of articles sent.")]
-        _limit: i32,
+        country: String,
+        #[graphql(description = "Maximum number of articles sent.")] limit: i32,
     ) -> FieldResult<Vec<News>> {
-        let _rank = ctx.ranker.get_rank().await?;
+        let news = ctx
+            .meilisearch
+            .write()
+            .await
+            .index
+            .as_ref()
+            .ok_or_else(|| FieldError::new("Index not found", graphql_value!({ "internal_error": "Meilisearch index is not selected" })))?
+            .search()
+            .with_query("*")
+            .with_limit(limit.try_into()?)
+            .with_filter(&format!("source.country={:?}", country))
+            .execute::<News>()
+            .await?;
 
-        Ok(vec![News {
-            ..Default::default()
-        }])
+        Ok(news.hits.iter().map(|r| r.result.clone()).collect())
     }
 }

graphql/rank/src/squid.rs

@@ -38,8 +38,11 @@ impl Squid {
     }
 
     /// Get top 10 most used words in news articles.
-    pub async fn leaderboard(&mut self) -> Result<Vec<String>, Status> {
-        let request = tonic::Request::new(LeaderboardRequest { length: 10 });
+    pub async fn leaderboard(
+        &mut self,
+        length: u32,
+    ) -> Result<Vec<String>, Status> {
+        let request = tonic::Request::new(LeaderboardRequest { length });
 
         let leaderboard = self
             .client

graphql/search/src/lib.rs

@@ -22,7 +22,8 @@ pub trait Attributes {
 #[derive(Clone, Debug)]
 pub struct Search {
     client: Client,
-    index: Option<Index>,
+    /// Meilisearch index.
+    pub index: Option<Index>,
     ttl_row: Option<String>,
 }