Crawl for number of online and total users

Cargo.lock

@@ -418,7 +418,6 @@ dependencies = [
  "reqwest",
  "serde",
  "tokio",
- "url",
 ]
 
 [[package]]

Cargo.toml

@@ -6,7 +6,6 @@ edition = "2018"
 
 [dependencies]
 reqwest = { version = "0.10.10", features = ["json"] }
-url = "2.2.1"
 serde = { version = "1.0.123", features = ["derive"] }
 anyhow = "1.0.38"
 tokio = { version = "0.2.25", features = ["full"] }

src/federated_instances.rs

@@ -1,14 +1,14 @@
 use serde::Deserialize;
 
-#[derive(Deserialize, Debug)]
+#[derive(Deserialize, Debug, Clone)]
 pub struct GetSiteResponse {
-  pub online: usize,
+    pub online: usize,
-  pub federated_instances: Option<FederatedInstances>,
+    pub federated_instances: Option<FederatedInstances>,
 }
 
-#[derive(Deserialize, Debug)]
+#[derive(Deserialize, Debug, Clone)]
 pub struct FederatedInstances {
-  pub linked: Vec<String>,
+    pub linked: Vec<String>,
-  pub allowed: Option<Vec<String>>,
+    pub allowed: Option<Vec<String>>,
-  pub blocked: Option<Vec<String>>,
+    pub blocked: Option<Vec<String>>,
-}
+}

src/lib.rs

@@ -1,2 +1,2 @@
+pub mod federated_instances;
 pub mod node_info;
-pub mod federated_instances;

src/main.rs

@@ -1,18 +1,65 @@
 use anyhow::Error;
-use url::Url;
-use lemmy_stats_crawler::node_info::NodeInfo;
 use lemmy_stats_crawler::federated_instances::GetSiteResponse;
+use lemmy_stats_crawler::node_info::NodeInfo;
+
+#[derive(Default, Debug)]
+struct TotalStats {
+    users: i64,
+    online_users: i32,
+}
 
 #[tokio::main]
 pub async fn main() -> Result<(), Error> {
-    let url = Url::parse("https://lemmy.ml/nodeinfo/2.0.json")?;
+    let mut pending_instances = vec!["lemmy.ml".to_string()];
-    let node_info: NodeInfo = reqwest::get(url).await?.json().await?;
+    let mut crawled_instances = vec![];
+    let mut total_stats = TotalStats::default();
+    while let Some(pi) = pending_instances.iter().next() {
+        crawled_instances.push(pi.to_owned());
+        let instance_details = fetch_instance_details(&pi).await.ok();
+        pending_instances = pending_instances
+            .iter()
+            .filter(|i| i != &pi)
+            .map(|i| i.to_owned())
+            .collect();
 
-    dbg!(node_info);
+        if let Some(details) = instance_details {
-
+            total_stats.online_users += details.online_users;
-    let url = Url::parse("https://lemmy.ml/api/v2/site")?;
+            total_stats.users += details.total_users;
-    let site_info: GetSiteResponse = reqwest::get(url).await?.json().await?;
+            // remove all which are in crawled_instances
-
+            for ci in details.linked_instances {
-    dbg!(site_info);
+                if !crawled_instances.contains(&ci) {
+                    pending_instances.push(ci);
+                }
+            }
+        }
+    }
+    dbg!(total_stats);
     Ok(())
-}
+}
+
+struct InstanceDetails {
+    domain: String,
+    online_users: i32,
+    total_users: i64,
+    linked_instances: Vec<String>,
+}
+
+async fn fetch_instance_details(domain: &str) -> Result<InstanceDetails, Error> {
+    dbg!(domain);
+
+    let node_info_url = format!("https://{}/nodeinfo/2.0.json", domain);
+    let node_info: NodeInfo = reqwest::get(&node_info_url).await?.json().await?;
+
+    let site_info_url = format!("https://{}/api/v2/site", domain);
+    let site_info: GetSiteResponse = reqwest::get(&site_info_url).await?.json().await?;
+
+    Ok(InstanceDetails {
+        domain: domain.to_owned(),
+        online_users: site_info.online as i32,
+        total_users: node_info.usage.users.total,
+        linked_instances: site_info
+            .federated_instances
+            .map(|f| f.linked)
+            .unwrap_or(vec![]),
+    })
+}

src/node_info.rs

@@ -3,31 +3,31 @@ use serde::Deserialize;
 #[derive(Deserialize, Debug)]
 #[serde(rename_all = "camelCase")]
 pub struct NodeInfo {
-  pub version: String,
+    pub version: String,
-  pub software: NodeInfoSoftware,
+    pub software: NodeInfoSoftware,
-  pub protocols: Vec<String>,
+    pub protocols: Vec<String>,
-  pub usage: NodeInfoUsage,
+    pub usage: NodeInfoUsage,
-  pub open_registrations: bool,
+    pub open_registrations: bool,
 }
 
 #[derive(Deserialize, Debug)]
 pub struct NodeInfoSoftware {
-  pub name: String,
+    pub name: String,
-  pub version: String,
+    pub version: String,
 }
 
 #[derive(Deserialize, Debug)]
 #[serde(rename_all = "camelCase")]
 pub struct NodeInfoUsage {
-  pub users: NodeInfoUsers,
+    pub users: NodeInfoUsers,
-  pub local_posts: i64,
+    pub local_posts: i64,
-  pub local_comments: i64,
+    pub local_comments: i64,
 }
 
 #[derive(Deserialize, Debug)]
 #[serde(rename_all = "camelCase")]
 pub struct NodeInfoUsers {
-  pub total: i64,
+    pub total: i64,
-  pub active_halfyear: i64,
+    pub active_halfyear: i64,
-  pub active_month: i64,
+    pub active_month: i64,
-}
+}