internet_mapper/src/main.rs

79 lines
2.2 KiB
Rust
Raw Normal View History

2024-08-23 11:22:49 +00:00
use std::cell::LazyCell;
use std::sync::{Arc, RwLock};
use serde::{Deserialize, Serialize};
use spider::hashbrown::HashMap;
use surrealdb::engine::remote::ws::{Ws, Client};
use surrealdb::opt::auth::Root;
use surrealdb::sql::Thing;
use surrealdb::Surreal;
use tokio::spawn;
use tokio::sync::broadcast::{self, Sender, Receiver};
#[derive(Debug, Deserialize)]
struct Record {
#[allow(dead_code)]
id: Thing,
}
#[derive(Debug, Serialize)]
struct Website<'a> {
url: &'a str,
crawled: bool
}
#[tokio::main]
async fn main() -> surrealdb::Result<()> {
let db_handle = tokio::spawn(async move {
let db = Surreal::new::<Ws>("127.0.0.1:8000").await.unwrap();
db.signin(Root {
username: "root",
password: "root",
})
.await.unwrap();
db.use_ns("test").use_db("test").await.unwrap();
let mut sub = PIPE.subscribe();
loop {
match sub.recv().await {
Ok((from, to)) => {
// let f: Vec<Record> = db.create("website").content(Website {url: &from, crawled: true}).await.unwrap();
// let t: Vec<Record> = db.create("website").content(Website {url: &to, crawled: false}).await.unwrap();
// println!("{:?} {:?}", f , t);
},
Err(e) => {
eprintln!("Error: {}", e);
},
}
}
});
let mut site = spider::website::Website::new("https://oliveratkinson.net")
.with_budget(Some(HashMap::from([
("*", 25),
])))
.with_tld(true)
.with_on_link_find_callback(Some(|from, to| {
let from = from.as_ref().to_string();
let to = to.as_ref().to_string();
match PIPE.send((from.clone(), to.clone())) {
Ok(_) => {},
Err(e) => {
eprintln!("{:?}", e);
},
};
println!("{from} -> {to}");
}))
.build()
.unwrap();
site.crawl().await;
let _ = db_handle.await;
Ok(())
}
const PIPE: LazyCell<Sender<(String, String)>> = LazyCell::new(|| {
broadcast::channel(100).0
});