checkpoint

.gitignore

@@ -4,4 +4,5 @@
 perf.data
 flamegraph.svg
 perf.data.old
-/docker/logs/*
+/docker/logs/*
+/downloaded

Cargo.toml

@@ -1,5 +1,5 @@
 [package]
-name = "surreal_spider"
+name = "internet_mapper"
 version = "0.1.0"
 edition = "2021"
 
@@ -11,8 +11,8 @@ metrics-exporter-prometheus = { version = "0.16.2", features=["http-listener"]}
 # minio = "0.1.0"
 minio = {git="https://github.com/minio/minio-rs.git", rev = "c28f576"}
 reqwest = { version = "0.12", features = ["gzip", "default", "rustls-tls"] }
+rusqlite = { version = "0.34.0", features = ["bundled"] }
 serde = { version = "1.0", features = ["derive"] }
-surrealdb = "2.2"
 tokio = { version="1.41.0", features = ["full"] }
 toml = "0.8.20"
 tracing = "0.1"

Crawler.toml

@@ -3,13 +3,7 @@ surreal_url = "localhost:8000"
 surreal_username = "root"
 surreal_password = "root"
 surreal_ns = "test"
-surreal_db = "v1.15.4"
+surreal_db = "v1.19.2"
-
-# Minio config
-s3_bucket = "v1.15.4"
-s3_url = "http://localhost:9000"
-s3_access_key = "3ptjsHhRHCHlpCmgFy9n"
-s3_secret_key = "68CmV07YExeCxb8kJhosSauEizj5CAE7PINZIfQz"
 
 # Crawler config
 crawl_filter = "en.wikipedia.com" 

README.md

@@ -2,11 +2,24 @@
 
 Crawls sites saving all the found links to a surrealdb database. It then proceeds to take batches of 100 uncrawled links untill the crawl budget is reached. It saves the data of each site in a minio database.
 
-
 ### TODO
 
 - [ ] Domain filtering - prevent the crawler from going on alternate versions of wikipedia.
 - [ ] Conditionally save content - based on filename or file contents
-- [ ] GUI / TUI ?
+- [x] GUI / TUI ? - Graphana
-- [ ] Better asynchronous getting of the sites. Currently it all happens serially.3/19/25: Took 20min to crawl 100 pages
+- [x] Better asynchronous getting of the sites. Currently it all happens serially.
-This ment we stored 100 pages, 142,997 urls, and 1,425,798 links between the two.
+- [ ] Allow for storing asynchronously
+
+3/17/25: Took >1hr to crawl 100 pages
+
+3/19/25: Took 20min to crawl 1000 pages
+This ment we stored 1000 pages, 142,997 urls, and 1,425,798 links between the two.
+
+3/20/25: Took 5min to crawl 1000 pages
+
+3/21/25: Took 3min to crawl 1000 pages
+
+# About
+
+![Screenshot](/pngs/graphana.png)
+

docker/compose.yml

@@ -14,22 +14,6 @@ services:
       - --pass
       - root
       - rocksdb:/mydata/database.db
-  minio:
-    image: quay.io/minio/minio
-    ports:
-      - 9000:9000
-      - 9001:9001
-    environment:
-      - MINIO_ROOT_USER=root
-      - MINIO_ROOT_PASSWORD=an8charpassword
-      - MINIO_PROMETHEUS_AUTH_TYPE=public
-    volumes:
-      - minio_storage:/data
-    command:
-      - server
-      - /data
-      - --console-address
-      - ":9001"
 
   alloy:
     image: grafana/alloy:latest

docker/prometheus.yaml

@@ -8,13 +8,10 @@ scrape_configs:
     # change this your machine's ip, localhost won't work
     # because localhost refers to the docker container.
       - targets: ['172.20.239.48:2500']
+        #- targets: ['192.168.8.209:2500']
   - job_name: loki
     static_configs:
       - targets: ['loki:3100']
   - job_name: prometheus
     static_configs:
       - targets: ['localhost:9090']
-  - job_name: minio
-    metrics_path: /minio/v2/metrics/cluster
-    static_configs:
-      - targets: ['minio:9000']

src/db.rs

@@ -1,35 +1,15 @@
-use base64::{
-    alphabet,
-    engine::{self, general_purpose},
-    Engine,
-};
 use metrics::counter;
-use serde::{ser::SerializeStruct, Deserialize, Serialize};
+use rusqlite::Connection;
-use std::{fmt::Debug, sync::LazyLock, time::Instant};
+use std::fmt::Debug;
-use surrealdb::{
+use serde::{Deserialize, Serialize};
-    engine::remote::ws::{Client, Ws},
+use tracing::{error, instrument, trace};
-    opt::auth::Root,
-    sql::Thing,
-    Error::Api,
-    Response, Surreal,
-};
-use tokio::sync::Mutex;
-use tracing::{error, instrument, trace, warn};
 use url::Url;
 
-use crate::{Config, Timer};
+use crate::Config;
 
-// static LOCK: LazyLock<Arc<Mutex<bool>>> = LazyLock::new(|| Arc::new(Mutex::new(true)));
+const STORE: &str = "surql_store_calls";
-static LOCK: LazyLock<Mutex<bool>> = LazyLock::new(|| Mutex::new(true));
 
-const CUSTOM_ENGINE: engine::GeneralPurpose =
+#[derive(Serialize, Deserialize, Clone, Eq, PartialEq, Hash)]
-    engine::GeneralPurpose::new(&alphabet::URL_SAFE, general_purpose::NO_PAD);
-
-const TIME_SPENT_ON_LOCK: &'static str = "surql_lock_waiting_ms";
-const STORE: &'static str = "surql_store_calls";
-const LINK: &'static str = "surql_link_calls";
-
-#[derive(Deserialize, Clone)]
 pub struct Website {
     /// The url that this data is found at
     pub site: Url,
@@ -37,18 +17,6 @@ pub struct Website {
     pub crawled: bool,
 }
 
-impl Serialize for Website {
-    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
-    where
-        S: serde::Serializer {
-            let mut state = serializer.serialize_struct("Website", 2)?;
-            state.serialize_field("crawled", &self.crawled)?;
-            // to_string() calls the correct naming of site
-            state.serialize_field("site", &self.site.to_string())?;
-            state.end()
-    }
-}
-
 // manual impl to make tracing look nicer
 impl Debug for Website {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
@@ -74,84 +42,17 @@ impl Website {
         self.crawled = true
     }
 
-    pub fn get_url_as_string(site: &Url) -> String {
-        let domain = match site.domain() {
-            Some(s) => s.to_string(),
-            None => {
-                warn!("Failed to get domain of URL: {}, falling back to 'localhost'", site.to_string());
-                "localhost".to_string()
-            }
-        };
-        let path = site.path();
-
-        domain + path
-    }
-    pub fn get_url_as_b64_path(site: &Url) -> String {
-        let domain = site.domain().unwrap_or("DOMAIN").to_string();
-        let path = &CUSTOM_ENGINE.encode(site.path());
-
-        domain + path
-    }
-
-    #[instrument(skip_all)]
-    pub async fn links_to(&self, other: Vec<Thing>, db: &Surreal<Client>) {
-        let len = other.len();
-        if len == 0 {
-            return;
-        }
-        
-        let from = &self.site;
-
-        // let to = other.site.to_string();
-        trace!("Linking {} pages to {from}", other.len());
-        let msg = format!("Linked {len} pages to {from}");
-        let timer = Timer::start(&msg);
-        // prevent the timer from being dropped instantly.
-        let _ = timer;
-        counter!(LINK).increment(1);
-        match db
-            .query("COUNT(RELATE (SELECT id FROM website WHERE site = $in) -> links_to -> $out)")
-            .bind(("in", from.clone()))
-            .bind(("out", other))
-            .await
-        {
-            Ok(mut e) => {
-                // The relate could technically "fail" (not relate anything), this just means that
-                // the query was ok.
-                let _: Response = e;
-                if let Ok(vec) = e.take(0) {
-                    let _: Vec<usize> = vec;
-                    if let Some(num) = vec.get(0) {
-                        if *num == len {
-                            trace!("Link for {from} OK - {num}/{len}");
-                            return;
-                        } else {
-                            error!("Didn't link all the records. {num}/{len}. Surreal response: {:?}", e);
-                            return;
-                        }
-                    }
-                }
-                error!("Linking request succeeded but couldn't verify the results.");
-            }
-            Err(e) => {
-                error!("{}", e.to_string());
-            }
-        }
-    }
-
     // Insert ever item in the vec into surreal, crawled state will be preserved as TRUE
     // if already in the database as such or incoming data is TRUE.
-    pub async fn store_all(all: Vec<Self>, db: &Surreal<Client>) -> Vec<Thing> {
+    pub async fn store_all(all: Vec<Self>, db: &Connection) {
         counter!(STORE).increment(1);
         let mut things = Vec::with_capacity(all.len());
 
-        // TODO this only allows for one thread to be in the database at a time.
+        rusqlite::ParamsFromIter;
-        // This is currently required since otherwise we get write errors.
+
-        // If the default `crawled` is set to false, we might not need to write any more
+        db.execute("",
-        // than just the name. `accessed_at` is fun but not needed.
+            params![]
-        let now = Instant::now();
+        );
-        let lock = LOCK.lock().await;
-        counter!(TIME_SPENT_ON_LOCK).increment(now.elapsed().as_millis() as u64);
 
         match db
             .query(
@@ -167,19 +68,12 @@ impl Website {
         {
             Ok(mut id) => match id.take::<Vec<Thing>>(0) {
                 Ok(mut x) => things.append(&mut x),
-                Err(err) => match err {
+                Err(err) => error!("{:?}", err),
-                    Api(error) => {
-                        eprintln!("{:?}", error);
-                        error!("{:?}", error);
-                    }
-                    _ => error!("{:?}", err),
-                },
             },
             Err(err) => {
                 error!("{:?}", err);
             }
         }
-        drop(lock);
         things
     }
 }
@@ -196,31 +90,10 @@ pub struct Record {
     pub id: Thing,
 }
 
-#[instrument(skip_all, name = "SurrealDB")]
+#[instrument(skip_all, name = "sqlite_connect")]
-pub async fn connect(config: &Config) -> surrealdb::Result<Surreal<Client>> {
+pub async fn connect(config: &Config) -> Result<Connection, rusqlite::Error> {
-    trace!("Establishing connection to surreal...");
+    trace!("Establishing connection to sqlite...");
     // Connect to the server
-    let db = Surreal::new::<Ws>(&config.surreal_url).await?;
+    Connection::open("./squeelite.db")
-
-    trace!("Logging in...");
-    // Signin as a namespace, database, or root user
-    db.signin(Root {
-        username: &config.surreal_username,
-        password: &config.surreal_password,
-    })
-    .await?;
-
-    // Select a specific namespace / database
-    db.use_ns(&config.surreal_ns)
-        .use_db(&config.surreal_db)
-        .await?;
-
-    let setup = include_bytes!("setup.surql");
-    let file = setup.iter().map(|c| *c as char).collect::<String>();
-
-    db.query(file)
-        .await
-        .expect("Failed to setup surreal tables.");
-
-    Ok(db)
 }
+

src/filesystem.rs

@@ -0,0 +1,70 @@
+use std::{ffi::OsStr, path::PathBuf};
+
+use tokio::fs;
+use tracing::{debug, error, instrument, trace, warn};
+use url::Url;
+
+#[instrument(skip(data))]
+pub async fn store(data: &str, url: &Url) {
+    // extract data from url to save it accurately
+    let url_path = PathBuf::from("./downloaded/".to_string() + url.domain().unwrap_or("UnknownDomain") + url.path());
+
+    // if it's a file
+    let (basepath, filename) = if url_path.extension().filter(valid_file_extension).is_some() {
+        // get everything up till the file
+        let basepath = url_path.ancestors().skip(1).take(1).collect::<PathBuf>();
+        // get the file name
+        let filename = url_path.file_name().expect("This should exist").to_string_lossy();
+        trace!("Save path: {:?} and base path: {:?}", &url_path, &basepath);
+        (basepath, filename.to_string())
+    } else {
+        (url_path.clone(), "index.html".into())
+    };
+
+    debug!("Writing at: {:?} {:?}", basepath, filename);
+
+    // create the folders
+    if let Err(err) = fs::create_dir_all(&basepath).await {
+        error!("Dir creation: {err} {:?}", basepath);
+    } else {
+        // FIXME I don't think this handles index.html files well...
+        // TODO this should probably append .html to non-described files
+        // create the file if that was successful
+        if let Err(err) = fs::write(&basepath.join(filename), data).await {
+            error!("File creation: {err} {:?}", url_path);
+        }
+    }
+}
+
+fn valid_file_extension(take: &&OsStr) -> bool {
+    let los = take.to_string_lossy();
+    let all = los.split('.');
+    match all.last() {
+        Some(s) => {
+            match s.to_lowercase().as_str() {
+                "html" => true,
+                "css" => true,
+                "js" => true,
+                "ts" => true,
+                "otf" => true, // font
+
+                "png" => true,
+                "svg" => true,
+                "jpg" => true,
+                "jpeg" => true,
+                "mp4" => true,
+                "mp3" => true,
+                "webp" => true,
+
+                "pdf" => true,
+                "json" => true,
+                "xml" => true,
+                _ => {
+                    warn!("Might be forgetting a file extension: {s}");
+                    false
+                }
+            }
+        },
+        None => false,
+    }
+}

src/main.rs

@@ -3,30 +3,26 @@
 extern crate html5ever;
 
 use std::{
-    fs::File,
+    collections::HashSet, fs::File, io::Read, net::{IpAddr, Ipv4Addr}
-    io::Read,
-    net::{IpAddr, Ipv4Addr},
-    time::Instant,
 };
 
 use db::{connect, Website};
 use metrics::{counter, gauge};
 use metrics_exporter_prometheus::PrometheusBuilder;
-use s3::S3;
 use serde::Deserialize;
 use surrealdb::{engine::remote::ws::Client, Surreal};
 use tokio::task::JoinSet;
-use tracing::{debug, error, info, instrument, trace, trace_span, warn};
+use tracing::{debug, error, info, instrument, level_filters::LevelFilter, trace, trace_span};
 use tracing_subscriber::{fmt, layer::SubscriberExt, EnvFilter, Layer, Registry};
 
 mod db;
 mod parser;
-mod s3;
+mod filesystem;
 
-const GET_METRIC: &'static str = "total_gets";
+const GET_METRIC: &str = "total_gets";
-const GET_IN_FLIGHT: &'static str = "gets_in_flight";
+const GET_IN_FLIGHT: &str = "gets_in_flight";
-const SITES_CRAWLED: &'static str = "pages_crawled";
+const SITES_CRAWLED: &str = "pages_crawled";
-const BEING_PROCESSED: &'static str = "pages_being_processed";
+const BEING_PROCESSED: &str = "pages_being_processed";
 
 #[derive(Deserialize)]
 struct Config {
@@ -36,26 +32,21 @@ struct Config {
     surreal_username: String,
     surreal_password: String,
 
-    s3_url: String,
-    s3_bucket: String,
-    s3_access_key: String,
-    s3_secret_key: String,
-
     crawl_filter: String,
     budget: usize,
 }
 
 #[tokio::main]
 async fn main() {
-    let total_runtime = Timer::start("Completed");
-
     let writer = std::fs::OpenOptions::new()
         .append(true)
         .create(true)
         .open("./docker/logs/tracing.log")
         .expect("Couldn't make log file!");
 
-    let filter = EnvFilter::from_default_env();
+    let filter = EnvFilter::builder()
+        .with_default_directive(LevelFilter::DEBUG.into())
+        .from_env_lossy();
 
     let registry = Registry::default().with(
         fmt::layer()
@@ -78,7 +69,7 @@ async fn main() {
         .install()
         .expect("failed to install recorder/exporter");
 
-    debug!("Starting...");
+    info!("Starting...");
     // Would probably take these in as parameters from a cli
     let starting_url = "https://en.wikipedia.org/";
     // When getting uncrawled pages, name must contain this variable. "" will effectively get ignored.
@@ -95,9 +86,6 @@ async fn main() {
     let db = connect(&config)
         .await
         .expect("Failed to connect to surreal, aborting.");
-    let s3 = S3::connect(&config)
-        .await
-        .expect("Failed to connect to minio, aborting.\n\nThis probably means you need to login to the minio console and get a new access key!\n\n(Probably here) http://localhost:9001/access-keys/new-account\n\n");
 
     let reqwest = reqwest::Client::builder()
         // .use_rustls_tls()
@@ -110,8 +98,8 @@ async fn main() {
     let span = trace_span!("Pre-Loop");
     let pre_loop_span = span.enter();
     // Download the site
-    let site = Website::new(&starting_url, false);
+    let site = Website::new(starting_url, false);
-    process(site, db.clone(), reqwest.clone(), s3.clone()).await;
+    process(site, db.clone(), reqwest.clone()).await;
 
     drop(pre_loop_span);
 
@@ -125,28 +113,23 @@ async fn main() {
         };
 
         let uncrawled = get_uncrawled_links(&db, get_num, config.crawl_filter.clone()).await;
-        if uncrawled.len() == 0 {
+        if uncrawled.is_empty() {
             info!("Had more budget but finished crawling everything.");
             return;
         }
-        debug!("Crawling {} pages...", uncrawled.len());
-
-        let span = trace_span!("Crawling");
-        let _ = span.enter();
 
         {
             let mut futures = JoinSet::new();
             for site in uncrawled {
                 gauge!(BEING_PROCESSED).increment(1);
-                futures.spawn(process(site, db.clone(), reqwest.clone(), s3.clone()));
+                futures.spawn(process(site, db.clone(), reqwest.clone()));
                 // let percent = format!("{:.2}%", (crawled as f32 / budget as f32) * 100f32);
                 // info!("Crawled {crawled} out of {budget} pages. ({percent})");
             }
-            debug!("Joining {} futures...", futures.len());
 
             let c = counter!(SITES_CRAWLED);
             // As futures complete runs code in while block
-            while let Some(_) = futures.join_next().await {
+            while futures.join_next().await.is_some() {
                 c.increment(1);
                 gauge!(BEING_PROCESSED).decrement(1);
                 crawled += 1;
@@ -155,33 +138,27 @@ async fn main() {
     }
     drop(span);
 
-    debug!("Done");
+    info!("Done");
-    drop(total_runtime);
 }
 
-#[instrument(skip(db, s3, reqwest))]
+#[instrument(skip(db, reqwest))]
 /// Downloads and crawls and stores a webpage.
 /// It is acceptable to clone `db`, `reqwest`, and `s3` because they all use `Arc`s internally. - Noted by Oliver
-async fn process(mut site: Website, db: Surreal<Client>, reqwest: reqwest::Client, s3: S3) {
+async fn process(mut site: Website, db: Surreal<Client>, reqwest: reqwest::Client) {
     
     // METRICS
     trace!("Process: {}", &site.site);
-    let timer = Timer::start("Built request");
     // Build the request
-    let request_builder = reqwest.get(&site.site.to_string());
+    let request_builder = reqwest.get(site.site.to_string());
-    // METRICS
-    timer.stop();
 
     // METRICS
     let g = gauge!(GET_IN_FLIGHT);
     g.increment(1);
-    let timer = Timer::start("Got page");
 
     // Send the http request (get)
     if let Ok(response) = request_builder.send().await {
 
         // METRICS
-        timer.stop();
         g.decrement(1);
         counter!(GET_METRIC).increment(1);
 
@@ -190,22 +167,30 @@ async fn process(mut site: Website, db: Surreal<Client>, reqwest: reqwest::Clien
             .text()
             .await
             .expect("Failed to read http response's body!");
+
         // Store document
-        s3.store(&data, &site.site).await;
+        filesystem::store(&data, &site.site).await;
 
         // Parse document and get relationships
         let sites = parser::parse(&site, &data).await;
         
         // update self in db
         site.set_crawled();
-        Website::store_all(vec![site.clone()], &db).await;
+        Website::store_all(vec![site], &db).await;
+
+        // De-duplicate this list
+        let prev_len = sites.len();
+        let set = sites.into_iter().fold(HashSet::new(), |mut set,item| {
+            set.insert(item);
+            set
+        });
+        let de_dupe_sites: Vec<Website> = set.into_iter().collect();
+        let diff = prev_len - de_dupe_sites.len();
+        trace!("Saved {diff} from being entered into the db by de-duping");
 
         // Store all the other sites so that we can link to them.
-        // let mut links_to = Vec::new();
+        let _ = Website::store_all(de_dupe_sites, &db).await;
-        let others = Website::store_all(sites, &db).await;
 
-        // Make the database's links reflect the html links between sites
-        site.links_to(others, &db).await;
     } else {
         error!("Failed to get: {}", &site.site);
     }
@@ -234,33 +219,3 @@ async fn get_uncrawled_links(
         .expect("Returned websites couldn't be parsed")
 }
 
-pub struct Timer<'a> {
-    start: Instant,
-    msg: &'a str,
-}
-
-impl<'a> Timer<'a> {
-    #[inline]
-    pub fn start(msg: &'a str) -> Self {
-        Self {
-            start: Instant::now(),
-            msg,
-        }
-    }
-    pub fn stop(&self) -> f64 {
-        let dif = self.start.elapsed().as_micros();
-        let ms = dif as f64 / 1000.;
-
-        if ms > 200. {
-            warn!("{}", format!("{} in {:.3}ms", self.msg, ms));
-        }
-
-        ms
-    }
-}
-
-impl Drop for Timer<'_> {
-    fn drop(&mut self) {
-        self.stop();
-    }
-}

src/parser.rs

@@ -5,19 +5,21 @@ use html5ever::tokenizer::{BufferQueue, TokenizerResult};
 use html5ever::tokenizer::{StartTag, TagToken};
 use html5ever::tokenizer::{Token, TokenSink, TokenSinkResult, Tokenizer, TokenizerOpts};
 use html5ever::{local_name, tendril::*};
-use tracing::instrument;
+use tracing::{debug, error, instrument, trace, warn};
+use url::Url;
 
 use crate::db::Website;
-use crate::Timer;
 
 impl TokenSink for Website {
     type Handle = Vec<Website>;
 
+    #[instrument(skip(token, _line_number))]
     fn process_token(&self, token: Token, _line_number: u64) -> TokenSinkResult<Self::Handle> {
         match token {
             TagToken(tag) => {
                 if tag.kind == StartTag {
                     match tag.name {
+                        // this should be all the html elements that have links
                         local_name!("a")
                         | local_name!("audio")
                         | local_name!("area")
@@ -32,23 +34,18 @@ impl TokenSink for Website {
                                 let attr_name = attr.name.local.to_string();
                                 if attr_name == "src" || attr_name == "href" || attr_name == "data"
                                 {
-                                    // Get clone of the current site object
+                                    trace!("Found `{}` in html `{}` tag", &attr.value, tag.name);
-                                    let mut web = self.clone();
+                                    let url = try_get_url(&self.site, &attr.value);
 
-                                    // Set url
+                                    if let Some(mut parsed) = url {
-                                    let mut url = web.site;
+                                        parsed.set_query(None);
-                                    url.set_fragment(None); // removes #xyz
+                                        parsed.set_fragment(None);
-                                    let joined = url
+                                        debug!("Final cleaned URL: `{}`", parsed.to_string());
-                                        .join(&attr.value)
+                                        let web = Website::new(&parsed.to_string(), false);
-                                        .expect("Failed to join url during parsing!");
+                                        links.push(web);
-                                    web.site = joined;
+                                    }
-
-                                    web.crawled = false;
-
-                                    links.push(web);
                                 }
                             }
-
                             return TokenSinkResult::Script(links);
                         }
                         local_name!("button") | local_name!("meta") | local_name!("iframe") => {
@@ -69,22 +66,21 @@ impl TokenSink for Website {
 pub async fn parse(site: &Website, data: &str) -> Vec<Website> {
     // prep work
     let mut other_sites: Vec<Website> = Vec::new();
-    let _t = Timer::start("Parsed page");
 
     // change data into something that can be tokenized
-    let chunk = Tendril::from_str(&data).expect("Failed to parse string into Tendril!");
+    let chunk = Tendril::from_str(data).expect("Failed to parse string into Tendril!");
     // create buffer of tokens and push our input into it
-    let mut token_buffer = BufferQueue::default();
+    let token_buffer = BufferQueue::default();
     token_buffer.push_back(
         chunk
             .try_reinterpret::<fmt::UTF8>()
-            .expect("Failed to reinterprt chunk!"),
+            .expect("Failed to reinterpret chunk!"),
     );
     // create the tokenizer
     let tokenizer = Tokenizer::new(site.clone(), TokenizerOpts::default());
 
     // go thru buffer
-    while let TokenizerResult::Script(mut sites) = tokenizer.feed(&mut token_buffer) {
+    while let TokenizerResult::Script(mut sites) = tokenizer.feed(&token_buffer) {
         other_sites.append(&mut sites);
         // other_sites.push(sites);
     }
@@ -93,3 +89,57 @@ pub async fn parse(site: &Website, data: &str) -> Vec<Website> {
 
     other_sites
 }
+
+#[instrument]
+fn try_get_url(parent: &Url, link: &str) -> Option<Url> {
+    match Url::parse(link) {
+        Ok(ok) => Some(ok),
+        Err(e) => {
+            if link.starts_with('#') {
+                trace!("Rejecting # url");
+                None
+            } else if link.starts_with("//") {
+                // if a url starts with "//" is assumed that it will adopt
+                // the same scheme as it's parent
+                // https://stackoverflow.com/questions/9646407/two-forward-slashes-in-a-url-src-href-attribute
+                let scheme = parent.scheme();
+
+                match Url::parse(&format!("{scheme}://{}", link)) {
+                    Ok(url) => Some(url),
+                    Err(err) => {
+                        error!("Failed parsing realative scheme url: {}", err);
+                        None
+                    }
+                }
+            } else {
+                // # This is some sort of realative url, gonna try patching it up into an absolute
+                // url
+                match e {
+                    url::ParseError::RelativeUrlWithoutBase => {
+                        // Is: scheme://host:port
+                        let origin = parent.origin().ascii_serialization();
+                        let url = origin.clone() + link;
+
+                        trace!("Built `{url}` from `{origin} + {}`", link.to_string());
+
+                        if let Ok(url) = Url::parse(&url) {
+                            trace!("Saved relative url `{}` AS: `{}`", link, url);
+                            Some(url)
+                        } else {
+                            error!(
+                                "Failed to reconstruct a url from relative url: `{}` on site: `{}`",
+                                link,
+                                parent.to_string()
+                            );
+                            None
+                        }
+                    }
+                    _ => {
+                        error!("MISC error: {:?} {:?}", e, link);
+                        None
+                    }
+                }
+            }
+        }
+    }
+}

src/s3.rs

@@ -1,101 +0,0 @@
-use metrics::counter;
-use minio::s3::{
-    args::{BucketExistsArgs, MakeBucketArgs},
-    client::ClientBuilder,
-    creds::StaticProvider,
-    error::Error,
-    http::BaseUrl,
-    Client,
-};
-use tracing::{instrument, trace, warn};
-use url::Url;
-
-use crate::{db::Website, Config, Timer};
-
-const S3_ROUND_TRIP_METRIC: &'static str = "s3_trips";
-
-#[derive(Clone)]
-pub struct S3 {
-    bucket_name: String,
-    client: Client,
-}
-
-impl S3 {
-    #[instrument(skip_all, name = "S3")]
-    pub async fn connect(config: &Config) -> Result<Self, Error> {
-        let base_url = config
-            .s3_url
-            .parse::<BaseUrl>()
-            .expect("Failed to parse url into BaseUrl");
-
-        let static_provider =
-            StaticProvider::new(&config.s3_access_key, &config.s3_secret_key, None);
-
-        let client = ClientBuilder::new(base_url)
-            .provider(Some(Box::new(static_provider)))
-            .build()?;
-
-        trace!("Checking bucket...");
-        let exists = client
-            .bucket_exists(
-                &BucketExistsArgs::new(&config.s3_bucket)
-                    .expect("Failed to check if bucket exists"),
-            )
-            .await?;
-        counter!(S3_ROUND_TRIP_METRIC).increment(1);
-
-        if !exists {
-            trace!("Creating bucket...");
-            client
-                .make_bucket(
-                    &MakeBucketArgs::new(&config.s3_bucket).expect("Failed to create bucket!"),
-                )
-                .await?;
-        }
-        counter!(S3_ROUND_TRIP_METRIC).increment(1);
-
-        trace!("Connection successful");
-
-        Ok(Self {
-            bucket_name: config.s3_bucket.to_owned(),
-            client,
-        })
-    }
-
-    #[instrument(name = "s3_store", skip_all)]
-    pub async fn store(&self, data: &str, url: &Url) {
-        let counter = counter!(S3_ROUND_TRIP_METRIC);
-        let t = Timer::start("Stored page");
-        let _ = t; // prevent compiler drop
-
-        let filename = Website::get_url_as_string(url);
-        trace!("Storing {} as {filename}", url.to_string());
-
-        counter.increment(1);
-        let _ = match &self
-            .client
-            .put_object_content(&self.bucket_name, &filename, data.to_owned())
-            .send()
-            .await
-        {
-            Ok(_) => {}
-            Err(err) => match err {
-                Error::InvalidObjectName(_) => {
-                    // This code will really only run if the url has non-english chars
-                    warn!("Tried storing invalid object name, retrying with Base64 encoding. Last try.");
-
-                    let filename: String = Website::get_url_as_b64_path(url);
-
-                    counter.increment(1);
-                    let _ = &self
-                        .client
-                        .put_object_content(&self.bucket_name, &filename, data.to_owned())
-                        .send()
-                        .await
-                        .unwrap();
-                }
-                _ => {}
-            },
-        };
-    }
-}