wip: try to implement parallel crawl #8

Closed
nutomic wants to merge 2 commits from parallel-crawl into main

View file

@ -3,52 +3,75 @@ use crate::node_info::NodeInfo;
use crate::REQUEST_TIMEOUT; use crate::REQUEST_TIMEOUT;
use anyhow::anyhow; use anyhow::anyhow;
use anyhow::Error; use anyhow::Error;
use futures::try_join; use futures::executor::block_on_stream;
use futures::future::join_all;
use futures::stream::FuturesUnordered;
use futures::{future, stream, try_join, StreamExt, TryStreamExt};
use reqwest::Client; use reqwest::Client;
use serde::Serialize; use serde::Serialize;
use std::cmp::max;
use std::collections::VecDeque; use std::collections::VecDeque;
use std::future::Future;
use std::sync::{Arc, Mutex};
pub async fn crawl( pub async fn crawl(
start_instances: Vec<String>, start_instances: Vec<String>,
exclude: Vec<String>, exclude: Vec<String>,
max_depth: i32, max_depth: i32,
) -> Result<(Vec<InstanceDetails>, i32), Error> { ) -> Result<(Vec<InstanceDetails>, i32), Error> {
let mut pending_instances: VecDeque<CrawlInstance> = start_instances let exclude = Arc::new(exclude);
let mut pending_instances: VecDeque<CrawlInstanceTask> = start_instances
.iter() .iter()
.map(|s| CrawlInstance::new(s.to_string(), 0)) .map(|s| CrawlInstanceTask::new(s.to_string(), 0, exclude.clone()))
.collect(); .collect();
let mut crawled_instances = vec![]; let mut crawled_instances = Mutex::new(vec![]);
let mut instance_details = vec![]; //let mut instance_details = vec![];
let mut failed_instances = 0; //let mut failed_instances = 0;
while let Some(current_instance) = pending_instances.pop_back() {
crawled_instances.push(current_instance.domain.clone()); let stream = Box::pin(
if current_instance.depth > max_depth || exclude.contains(&current_instance.domain) { stream::iter(pending_instances)
continue; .then(|task: CrawlInstanceTask| async {
} crawled_instances.lock().unwrap().push(task.domain.clone());
match fetch_instance_details(&current_instance.domain).await { crawl_instance(task, max_depth).await.unwrap()
Ok(details) => { })
instance_details.push(details.to_owned()); .flat_map(|(instance_details, task)| {
for i in details.linked_instances { let futures = instance_details.linked_instances.iter().map(|i| {
let is_in_crawled = crawled_instances.contains(&i); crawled_instances.lock().unwrap().push(i.clone());
let is_in_pending = pending_instances.iter().any(|p| p.domain == i); crawl_instance(
if !is_in_crawled && !is_in_pending { CrawlInstanceTask::new(i.clone(), task.depth + 1, task.exclude.clone()),
let ci = CrawlInstance::new(i, current_instance.depth + 1); max_depth,
pending_instances.push_back(ci); )
} });
}
} stream::iter(futures)
Err(e) => { }),
failed_instances += 1; );
eprintln!("Failed to crawl {}: {}", current_instance.domain, e)
} let crawl_result: Vec<Result<InstanceDetails, Error>> = stream
} .buffer_unordered(10)
} .map_ok(|(details, _)| details)
.collect()
.await;
todo!()
/*
// Sort by active monthly users descending // Sort by active monthly users descending
instance_details.sort_by_key(|i| i.users_active_month); crawl_result.sort_by_key(|i| i.users_active_month);
instance_details.reverse(); crawl_result.reverse();
Ok((instance_details, failed_instances)) Ok((crawl_result, failed_instances))
*/
}
async fn crawl_instance(
task: CrawlInstanceTask,
max_depth: i32,
) -> Result<(InstanceDetails, CrawlInstanceTask), Error> {
if task.depth > max_depth || task.exclude.contains(&task.domain) {
return Err(anyhow!("max depth reached"));
}
Ok((fetch_instance_details(&task.domain).await?, task))
} }
#[derive(Serialize, Clone)] #[derive(Serialize, Clone)]
@ -70,14 +93,19 @@ pub struct InstanceDetails {
pub linked_instances: Vec<String>, pub linked_instances: Vec<String>,
} }
struct CrawlInstance { struct CrawlInstanceTask {
domain: String, domain: String,
depth: i32, depth: i32,
exclude: Arc<Vec<String>>,
} }
impl CrawlInstance { impl CrawlInstanceTask {
pub fn new(domain: String, depth: i32) -> CrawlInstance { pub fn new(domain: String, depth: i32, exclude: Arc<Vec<String>>) -> CrawlInstanceTask {
CrawlInstance { domain, depth } CrawlInstanceTask {
domain,
depth,
exclude,
}
} }
} }