Cargo.toml

@@ -5,18 +5,13 @@ authors = ["Felix Ableitner"]
 edition = "2018"
 
 [dependencies]
-reqwest = { version = "0.11.13", default-features = false, features = ["json", "rustls-tls"] }
-serde = { version = "1.0.149", features = ["derive"] }
-anyhow = "1.0.66"
-tokio = { version = "1.22.0", features = ["macros", "rt-multi-thread"] }
-futures = "0.3.25"
-serde_json = "1.0.89"
-semver = "1.0.14"
-once_cell = "1.16.0"
-lemmy_api_common = "=0.16.0"
-lemmy_db_schema = "=0.16.0"
-async-recursion = "1.0.0"
-log = "0.4.17"
-derive-new = "0.5.9"
-stderrlog = "0.5.4"
-structopt = "0.3.26"
+reqwest = { version = "0.11.10", default-features = false, features = ["json", "rustls-tls"] }
+serde = { version = "1.0.137", features = ["derive"] }
+anyhow = "1.0.57"
+tokio = { version = "1.18.1", features = ["macros", "rt-multi-thread"] }
+futures = "0.3.21"
+serde_json = "1.0.81"
+clap = "3.1.15"
+semver = "1.0.9"
+once_cell = "1.10.0"
+lemmy_api_common = "0.16.0"

src/crawl.rs

@@ -1,135 +1,121 @@
-use crate::node_info::{NodeInfo, NodeInfoWellKnown};
-use crate::CLIENT;
-use anyhow::{anyhow, Error};
-use async_recursion::async_recursion;
-use futures::future::join_all;
+use crate::REQUEST_TIMEOUT;
+use anyhow::anyhow;
+use anyhow::Error;
 use lemmy_api_common::site::GetSiteResponse;
-use log::debug;
-use reqwest::Url;
+use once_cell::sync::Lazy;
+use reqwest::Client;
 use semver::Version;
-use std::collections::HashSet;
-use std::ops::Deref;
-use std::sync::Arc;
-use tokio::sync::Mutex;
+use serde::Serialize;
+use std::collections::VecDeque;
 
-#[derive(new)]
-pub struct CrawlJob {
-    domain: String,
-    current_distance: i32,
-    params: Arc<CrawlParams>,
-}
+static CLIENT: Lazy<Client> = Lazy::new(Client::default);
 
-#[derive(new)]
-pub struct CrawlParams {
-    min_lemmy_version: Version,
-    exclude_domains: Vec<String>,
+pub async fn crawl(
+    start_instances: Vec<String>,
+    exclude: Vec<String>,
     max_depth: i32,
-    crawled_instances: Arc<Mutex<HashSet<String>>>,
+) -> Result<(Vec<InstanceDetails>, i32), Error> {
+    let mut pending_instances: VecDeque<CrawlInstance> = start_instances
+        .iter()
+        .map(|s| CrawlInstance::new(s.to_string(), 0))
+        .collect();
+    let min_lemmy_version = min_lemmy_version().await?;
+    let mut crawled_instances = vec![];
+    let mut instance_details = vec![];
+    let mut failed_instances = 0;
+    while let Some(current_instance) = pending_instances.pop_back() {
+        crawled_instances.push(current_instance.domain.clone());
+        if current_instance.depth > max_depth || exclude.contains(&current_instance.domain) {
+            continue;
+        }
+        match fetch_instance_details(&current_instance.domain, &min_lemmy_version).await {
+            Ok(details) => {
+                if let Some(federated) = &details.site_info.federated_instances.as_ref() {
+                    for i in &federated.linked {
+                        let is_in_crawled = crawled_instances.contains(i);
+                        let is_in_pending = pending_instances.iter().any(|p| &p.domain == i);
+                        if !is_in_crawled && !is_in_pending {
+                            let ci = CrawlInstance::new(i.clone(), current_instance.depth + 1);
+                            pending_instances.push_back(ci);
+                        }
+                    }
+                }
+                instance_details.push(details);
+            }
+            Err(e) => {
+                failed_instances += 1;
+                eprintln!("Failed to crawl {}: {}", current_instance.domain, e)
+            }
+        }
+    }
+
+    // Sort by active monthly users descending
+    instance_details.sort_by_key(|i| {
+        i.site_info
+            .site_view
+            .as_ref()
+            .map(|s| s.counts.users_active_month)
+            .unwrap_or(0)
+    });
+    instance_details.reverse();
+
+    Ok((instance_details, failed_instances))
 }
 
-#[derive(Debug)]
-pub struct CrawlResult {
+#[derive(Serialize, Debug)]
+pub struct InstanceDetails {
     pub domain: String,
-    pub node_info: NodeInfo,
-    pub site_info: Option<GetSiteResponse>,
+    pub site_info: GetSiteResponse,
 }
 
-impl CrawlJob {
-    #[async_recursion]
-    pub async fn crawl(self) -> Vec<Result<CrawlResult, Error>> {
-        // need to acquire and release mutex before recursing, otherwise it will deadlock
-        {
-            let mut crawled_instances = self.params.crawled_instances.deref().lock().await;
-            if crawled_instances.contains(&self.domain) {
-                return vec![];
-            } else {
-                crawled_instances.insert(self.domain.clone());
-            }
-        }
+struct CrawlInstance {
+    domain: String,
+    depth: i32,
+}
 
-        if self.current_distance > self.params.max_depth
-            || self.params.exclude_domains.contains(&self.domain)
-        {
-            return vec![];
+impl CrawlInstance {
+    pub fn new(domain: String, depth: i32) -> CrawlInstance {
+        CrawlInstance { domain, depth }
     }
+}
 
-        debug!(
-            "Starting crawl for {}, distance {}",
-            &self.domain, &self.current_distance
-        );
-        let (node_info, site_info) = match self.fetch_instance_details().await {
-            Ok(o) => o,
-            Err(e) => return vec![Err(e)],
-        };
-        let mut crawl_result = CrawlResult {
-            domain: self.domain.clone(),
-            node_info,
-            site_info: None,
-        };
+async fn fetch_instance_details(
+    domain: &str,
+    min_lemmy_version: &Version,
+) -> Result<InstanceDetails, Error> {
+    let client = Client::default();
 
-        if let Some(site_info) = site_info {
-            match Version::parse(&site_info.version) {
-                Ok(version) => {
-                    if version < self.params.min_lemmy_version {
-                        return vec![Ok(crawl_result)];
-                    }
-                }
-                Err(e) => return vec![Err(e.into())],
-            }
-
-            let mut result = vec![];
-            if let Some(federated) = &site_info.federated_instances {
-                for domain in federated.linked.iter() {
-                    let crawl_job = CrawlJob::new(
-                        domain.clone(),
-                        self.current_distance + 1,
-                        self.params.clone(),
-                    );
-                    result.push(crawl_job.crawl());
-                }
-            }
-
-            let mut result2: Vec<Result<CrawlResult, Error>> =
-                join_all(result).await.into_iter().flatten().collect();
-            debug!("Successfully finished crawl for {}", &self.domain);
-            crawl_result.site_info = Some(site_info);
-            result2.push(Ok(crawl_result));
-
-            result2
-        } else {
-            vec![Ok(crawl_result)]
-        }
-    }
-
-    async fn fetch_instance_details(&self) -> Result<(NodeInfo, Option<GetSiteResponse>), Error> {
-        let rel_node_info: Url = Url::parse("http://nodeinfo.diaspora.software/ns/schema/2.0")
-            .expect("parse nodeinfo relation url");
-        let node_info_well_known = CLIENT
-            .get(&format!("https://{}/.well-known/nodeinfo", &self.domain))
-            .send()
-            .await?
-            .json::<NodeInfoWellKnown>()
-            .await?;
-        let node_info_url = node_info_well_known
-            .links
-            .into_iter()
-            .find(|l| l.rel == rel_node_info)
-            .ok_or_else(|| anyhow!("failed to find nodeinfo link for {}", &self.domain))?
-            .href;
-        let node_info = CLIENT
-            .get(node_info_url)
-            .send()
-            .await?
-            .json::<NodeInfo>()
-            .await?;
-
-        let site_info = CLIENT
-            .get(&format!("https://{}/api/v3/site", &self.domain))
+    let site_info_url = format!("https://{}/api/v3/site", domain);
+    let site_info = client
+        .get(&site_info_url)
+        .timeout(REQUEST_TIMEOUT)
         .send()
         .await?
         .json::<GetSiteResponse>()
-            .await
-            .ok();
-        Ok((node_info, site_info))
+        .await?;
+
+    let version = Version::parse(&site_info.version)?;
+    if &version < min_lemmy_version {
+        return Err(anyhow!("lemmy version is too old ({})", version));
     }
+
+    Ok(InstanceDetails {
+        domain: domain.to_owned(),
+        site_info,
+    })
+}
+
+/// calculate minimum allowed lemmy version based on current version. in case of current version
+/// 0.16.3, the minimum from this function is 0.15.3. this is to avoid rejecting all instances on
+/// the previous version when a major lemmy release is published.
+async fn min_lemmy_version() -> Result<Version, Error> {
+    let lemmy_version_url = "https://raw.githubusercontent.com/LemmyNet/lemmy-ansible/main/VERSION";
+    let req = CLIENT
+        .get(lemmy_version_url)
+        .timeout(REQUEST_TIMEOUT)
+        .send()
+        .await?;
+    let mut version = Version::parse(req.text().await?.trim())?;
+    version.minor -= 1;
+    Ok(version)
 }

src/lib.rs

@@ -1,138 +1,9 @@
-#[macro_use]
-extern crate derive_new;
-
-use crate::crawl::{CrawlJob, CrawlParams, CrawlResult};
-use crate::node_info::{NodeInfo, NodeInfoUsage, NodeInfoUsers};
-use anyhow::Error;
-use futures::future::join_all;
-use lemmy_api_common::site::GetSiteResponse;
-use log::warn;
-use once_cell::sync::Lazy;
-use reqwest::{Client, ClientBuilder};
-use semver::Version;
-use serde::Serialize;
-use std::collections::HashSet;
-use std::sync::Arc;
 use std::time::Duration;
-use tokio::sync::Mutex;
 
 pub mod crawl;
-mod node_info;
 
-const REQUEST_TIMEOUT: Duration = Duration::from_secs(10);
-
-static CLIENT: Lazy<Client> = Lazy::new(|| {
-    ClientBuilder::new()
-        .timeout(REQUEST_TIMEOUT)
-        .user_agent("lemmy-stats-crawler")
-        .build()
-        .expect("build reqwest client")
-});
-
-#[derive(Serialize, Debug)]
-pub struct CrawlResult2 {
-    pub domain: String,
-    pub site_info: GetSiteResponse,
-    pub federated_counts: Option<NodeInfoUsage>,
-}
-
-pub async fn start_crawl(
-    start_instances: Vec<String>,
-    exclude_domains: Vec<String>,
-    max_distance: i32,
-) -> Result<Vec<CrawlResult2>, Error> {
-    let params = Arc::new(CrawlParams::new(
-        min_lemmy_version().await?,
-        exclude_domains,
-        max_distance,
-        Arc::new(Mutex::new(HashSet::new())),
-    ));
-    let mut jobs = vec![];
-    for domain in start_instances.into_iter() {
-        let job = CrawlJob::new(domain, 0, params.clone());
-        jobs.push(job.crawl());
-    }
-
-    let crawl_results: Vec<CrawlResult> = join_all(jobs)
-        .await
-        .into_iter()
-        .flatten()
-        .inspect(|r| {
-            if let Err(e) = r {
-                warn!("{}", e)
-            }
-        })
-        .filter_map(Result::ok)
-        .collect();
-    let mut crawl_results = calculate_federated_site_aggregates(crawl_results)?;
-
-    // Sort by active monthly users descending
-    crawl_results.sort_unstable_by_key(|i| {
-        i.site_info
-            .site_view
-            .as_ref()
-            .map(|s| s.counts.users_active_month)
-            .unwrap_or(0)
-    });
-    crawl_results.reverse();
-    Ok(crawl_results)
-}
-
-/// calculate minimum allowed lemmy version based on current version. in case of current version
-/// 0.16.3, the minimum from this function is 0.15.3. this is to avoid rejecting all instances on
-/// the previous version when a major lemmy release is published.
-async fn min_lemmy_version() -> Result<Version, Error> {
-    let lemmy_version_url = "https://raw.githubusercontent.com/LemmyNet/lemmy-ansible/main/VERSION";
-    let req = CLIENT
-        .get(lemmy_version_url)
-        .timeout(REQUEST_TIMEOUT)
-        .send()
-        .await?;
-    let mut version = Version::parse(req.text().await?.trim())?;
-    version.minor -= 1;
-    Ok(version)
-}
-
-fn calculate_federated_site_aggregates(
-    crawl_results: Vec<CrawlResult>,
-) -> Result<Vec<CrawlResult2>, Error> {
-    let node_info: Vec<(String, NodeInfo)> = crawl_results
-        .iter()
-        .map(|c| (c.domain.clone(), c.node_info.clone()))
-        .collect();
-    let lemmy_instances: Vec<(String, GetSiteResponse)> = crawl_results
-        .into_iter()
-        .filter_map(|c| {
-            let domain = c.domain;
-            c.site_info.map(|c2| (domain, c2))
-        })
-        .collect();
-    let mut ret = vec![];
-    for instance in &lemmy_instances {
-        let federated_counts = if let Some(federated_instances) = &instance.1.federated_instances {
-            node_info
-                .iter()
-                .filter(|i| federated_instances.linked.contains(&i.0) || i.0 == instance.0)
-                .map(|i| i.1.usage.clone())
-                .reduce(|a, b| NodeInfoUsage {
-                    users: NodeInfoUsers {
-                        total: a.users.total + b.users.total,
-                        active_halfyear: a.users.active_halfyear + b.users.active_halfyear,
-                        active_month: a.users.active_month + b.users.active_month,
-                    },
-                    posts: a.posts + b.posts,
-                    comments: a.comments + b.comments,
-                })
-        } else {
-            None
-        };
-        // TODO: workaround because GetSiteResponse doesnt implement clone
-        let site_info = serde_json::from_str(&serde_json::to_string(&instance.1)?)?;
-        ret.push(CrawlResult2 {
-            domain: instance.0.clone(),
-            site_info,
-            federated_counts,
-        });
-    }
-    Ok(ret)
-}
+pub const REQUEST_TIMEOUT: Duration = Duration::from_secs(10);
+pub const DEFAULT_START_INSTANCES: &str = "lemmy.ml";
+pub const DEFAULT_MAX_CRAWL_DEPTH: &str = "20";
+pub const EXCLUDE_INSTANCES: &str =
+    "ds9.lemmy.ml, enterprise.lemmy.ml, voyager.lemmy.ml, test.lemmy.ml";

src/main.rs

@@ -1,65 +1,64 @@
 use anyhow::Error;
-use lemmy_stats_crawler::{start_crawl, CrawlResult2};
+use clap::{Arg, Command};
+use lemmy_stats_crawler::crawl::{crawl, InstanceDetails};
+use lemmy_stats_crawler::{DEFAULT_MAX_CRAWL_DEPTH, DEFAULT_START_INSTANCES, EXCLUDE_INSTANCES};
 use serde::Serialize;
-use structopt::StructOpt;
-
-#[derive(StructOpt, Debug)]
-#[structopt()]
-struct Parameters {
-    #[structopt(short, long, use_delimiter = true, default_value = "lemmy.ml")]
-    start_instances: Vec<String>,
-    #[structopt(
-        short,
-        long,
-        use_delimiter = true,
-        default_value = "ds9.lemmy.ml,enterprise.lemmy.ml,voyager.lemmy.ml,test.lemmy.ml"
-    )]
-    exclude_instances: Vec<String>,
-    #[structopt(short, long, default_value = "20")]
-    max_crawl_distance: i32,
-    /// Silence all output
-    #[structopt(short, long)]
-    quiet: bool,
-}
 
 #[tokio::main]
 pub async fn main() -> Result<(), Error> {
-    let params = Parameters::from_args();
-
-    stderrlog::new()
-        .module(module_path!())
-        .quiet(params.quiet)
-        .verbosity(1)
-        .init()?;
+    let matches = Command::new("Lemmy Stats Crawler")
+        .arg(
+            Arg::new("start-instances")
+                .long("start-instances")
+                .takes_value(true),
+        )
+        .arg(Arg::new("exclude").long("exclude").takes_value(true))
+        .arg(
+            Arg::new("max-crawl-depth")
+                .long("max-crawl-depth")
+                .takes_value(true),
+        )
+        .get_matches();
+    let start_instances: Vec<String> = matches
+        .value_of("start-instances")
+        .unwrap_or(DEFAULT_START_INSTANCES)
+        .split(',')
+        .map(|s| s.trim().to_string())
+        .collect();
+    let exclude: Vec<String> = matches
+        .value_of("exclude")
+        .unwrap_or(EXCLUDE_INSTANCES)
+        .split(',')
+        .map(|s| s.trim().to_string())
+        .collect();
+    let max_crawl_depth: i32 = matches
+        .value_of("max-crawl-depth")
+        .unwrap_or(DEFAULT_MAX_CRAWL_DEPTH)
+        .parse()?;
 
     eprintln!("Crawling...");
-    let instance_details = start_crawl(
-        params.start_instances,
-        params.exclude_instances,
-        params.max_crawl_distance,
-    )
-    .await?;
-    let total_stats = aggregate(instance_details);
+    let (instance_details, failed_instances) =
+        crawl(start_instances, exclude, max_crawl_depth).await?;
+    let total_stats = aggregate(instance_details, failed_instances);
 
     println!("{}", serde_json::to_string_pretty(&total_stats)?);
     Ok(())
 }
 
-// TODO: lemmy stores these numbers in SiteAggregates, would be good to simply use that as a member
-//       (to avoid many members). but SiteAggregates also has id, site_id fields
 #[derive(Serialize)]
 struct TotalStats {
     crawled_instances: i32,
+    failed_instances: i32,
     online_users: usize,
     total_users: i64,
     users_active_day: i64,
     users_active_week: i64,
     users_active_month: i64,
     users_active_halfyear: i64,
-    instance_details: Vec<CrawlResult2>,
+    instance_details: Vec<InstanceDetails>,
 }
 
-fn aggregate(instance_details: Vec<CrawlResult2>) -> TotalStats {
+fn aggregate(instance_details: Vec<InstanceDetails>, failed_instances: i32) -> TotalStats {
     let mut online_users = 0;
     let mut total_users = 0;
     let mut users_active_day = 0;
@@ -80,6 +79,7 @@ fn aggregate(instance_details: Vec<CrawlResult2>) -> TotalStats {
     }
     TotalStats {
         crawled_instances,
+        failed_instances,
         online_users,
         total_users,
         users_active_day,

src/node_info.rs

@@ -1,47 +0,0 @@
-use reqwest::Url;
-use serde::{Deserialize, Serialize};
-
-#[derive(Deserialize, Debug)]
-pub struct NodeInfoWellKnown {
-    pub links: Vec<NodeInfoWellKnownLinks>,
-}
-
-#[derive(Deserialize, Debug)]
-pub struct NodeInfoWellKnownLinks {
-    pub rel: Url,
-    pub href: Url,
-}
-
-#[derive(Deserialize, Debug, Clone)]
-#[serde(rename_all = "camelCase")]
-pub struct NodeInfo {
-    pub version: String,
-    pub software: NodeInfoSoftware,
-    pub protocols: Vec<String>,
-    pub usage: NodeInfoUsage,
-    pub open_registrations: bool,
-}
-
-#[derive(Deserialize, Debug, Clone)]
-pub struct NodeInfoSoftware {
-    pub name: String,
-    pub version: String,
-}
-
-#[derive(Deserialize, Serialize, Debug, Clone, Default)]
-#[serde(rename_all = "camelCase", default)]
-pub struct NodeInfoUsage {
-    pub users: NodeInfoUsers,
-    #[serde(rename(deserialize = "localPosts"))]
-    pub posts: i64,
-    #[serde(rename(deserialize = "localComments"))]
-    pub comments: i64,
-}
-
-#[derive(Deserialize, Serialize, Debug, Clone, Default)]
-#[serde(rename_all = "camelCase", default)]
-pub struct NodeInfoUsers {
-    pub total: i64,
-    pub active_halfyear: i64,
-    pub active_month: i64,
-}