unifed settings for testing
This commit is contained in:
parent
0f8a3d7215
commit
c1c8cf07bb
18
src/main.rs
18
src/main.rs
@ -27,28 +27,30 @@ struct Config<'a> {
|
||||
|
||||
#[tokio::main]
|
||||
async fn main() {
|
||||
let total_runtime = Timer::start("Completed");
|
||||
|
||||
tracing_subscriber::fmt()
|
||||
.with_env_filter(EnvFilter::from_default_env())
|
||||
.with_line_number(true)
|
||||
.without_time()
|
||||
// .without_time()
|
||||
.init();
|
||||
debug!("Starting...");
|
||||
|
||||
let config = Config {
|
||||
surreal_ns: "test",
|
||||
surreal_db: "v1.7",
|
||||
surreal_url: "localhost:8000",
|
||||
surreal_username: "root",
|
||||
surreal_password: "root",
|
||||
surreal_ns: "test",
|
||||
surreal_db: "custom-engine-v2",
|
||||
s3_bucket: "custom-engine-v2",
|
||||
s3_url: "http://localhost:9000",
|
||||
s3_bucket: "v1.7",
|
||||
s3_access_key: "8tUJn7e1paMFZQr0PKIT",
|
||||
s3_secret_key: "uSMvYxNOeCejCUgXVqgTfYlUEcmiZY0xcZ91M9E0",
|
||||
s3_access_key: "0zv7GbLQsw4ZI8TclMps",
|
||||
s3_secret_key: "5dB7QkGFw7fYbUJ5LpHk2GbWR7Bl710HlRz4NbzB",
|
||||
};
|
||||
|
||||
// Would probably take these in as parameters from a cli
|
||||
let starting_url = "https://oliveratkinson.net/";
|
||||
let budget = 15;
|
||||
let budget = 5;
|
||||
let mut crawled = 0;
|
||||
|
||||
let s3 = S3::connect(&config).await.expect("Failed to connect to minio, aborting.");
|
||||
@ -94,6 +96,7 @@ async fn main() {
|
||||
drop(span);
|
||||
|
||||
info!("Done");
|
||||
drop(total_runtime);
|
||||
}
|
||||
|
||||
#[instrument(skip_all)]
|
||||
@ -117,6 +120,7 @@ async fn get(
|
||||
// Parse document and store relationships
|
||||
parser::parse(db, site, data).await;
|
||||
*count += 1;
|
||||
return;
|
||||
}
|
||||
trace!("Failed to get: {}", site.to_string());
|
||||
}
|
||||
|
@ -7,8 +7,10 @@ use html5ever::tokenizer::{Token, TokenSink, TokenSinkResult, Tokenizer, Tokeniz
|
||||
use html5ever::{local_name, tendril::*};
|
||||
use surrealdb::engine::remote::ws::Client;
|
||||
use surrealdb::Surreal;
|
||||
use tracing::instrument;
|
||||
|
||||
use crate::db::Website;
|
||||
use crate::Timer;
|
||||
|
||||
#[derive(Clone)]
|
||||
struct LinkParser<'a> {
|
||||
@ -67,6 +69,7 @@ impl TokenSink for LinkParser<'_> {
|
||||
}
|
||||
}
|
||||
|
||||
#[instrument(skip_all)]
|
||||
pub async fn parse(db: &Surreal<Client>, site: &mut Website, data: String) {
|
||||
|
||||
site.set_crawled();
|
||||
@ -79,6 +82,7 @@ pub async fn parse(db: &Surreal<Client>, site: &mut Website, data: String) {
|
||||
|
||||
let token = Tokenizer::new(sink.clone(), TokenizerOpts::default());
|
||||
|
||||
let t = Timer::start("Stored pages");
|
||||
let mut links_to = Vec::new();
|
||||
while !input.is_empty() {
|
||||
if let TokenizerResult::Script(s) = token.feed(&mut input) {
|
||||
@ -89,6 +93,7 @@ pub async fn parse(db: &Surreal<Client>, site: &mut Website, data: String) {
|
||||
}
|
||||
}
|
||||
}
|
||||
drop(t);
|
||||
sink.site.links_to(links_to, db).await;
|
||||
assert!(input.is_empty());
|
||||
token.end();
|
||||
|
Loading…
Reference in New Issue
Block a user