changes

Reviewed-on: #6

.gitignore

@@ -5,3 +5,4 @@ perf.data
 flamegraph.svg
 perf.data.old
 /docker/logs/*
+/downloaded

.vscode/launch.json

@@ -7,18 +7,15 @@
         {
             "type": "lldb",
             "request": "launch",
-            "name": "Debug executable 'surreal_spider'",
-            "env": {
-                "RUST_LOG": "surreal_spider=trace,reqwest=info",
-            },
+            "name": "Debug executable 'internet_mapper'",
             "cargo": {
                 "args": [
                     "build",
-                    "--bin=surreal_spider",
-                    "--package=surreal_spider"
+                    "--bin=internet_mapper",
+                    "--package=internet_mapper"
                 ],
                 "filter": {
-                    "name": "surreal_spider",
+                    "name": "internet_mapper",
                     "kind": "bin"
                 }
             },
@@ -28,16 +25,16 @@
         {
             "type": "lldb",
             "request": "launch",
-            "name": "Debug unit tests in executable 'surreal_spider'",
+            "name": "Debug unit tests in executable 'internet_mapper'",
             "cargo": {
                 "args": [
                     "test",
                     "--no-run",
-                    "--bin=surreal_spider",
-                    "--package=surreal_spider"
+                    "--bin=internet_mapper",
+                    "--package=internet_mapper"
                 ],
                 "filter": {
-                    "name": "surreal_spider",
+                    "name": "internet_mapper",
                     "kind": "bin"
                 }
             },

.vscode/settings.json

@@ -3,6 +3,6 @@
         "creds",
         "reqwest",
         "rustls",
-        "surql"
+        "surql",
     ]
 }

Cargo.lock

@@ -1966,6 +1966,7 @@ name = "internet_mapper"
 version = "0.1.0"
 dependencies = [
  "base64 0.22.1",
+ "futures-util",
  "html5ever 0.29.1",
  "metrics",
  "metrics-exporter-prometheus",

Cargo.toml

@@ -5,12 +5,13 @@ edition = "2021"
 
 [dependencies]
 base64 = "0.22.1"
+futures-util = "0.3.31"
 html5ever = "0.29"
 metrics = "0.24.1"
 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"] }
+reqwest = { version = "0.12", features = ["gzip", "default", "rustls-tls", "stream"] }
 serde = { version = "1.0", features = ["derive"] }
 surrealdb = "2.2"
 tokio = { version="1.41.0", features = ["full"] }

Crawler.toml

@@ -3,14 +3,10 @@ surreal_url = "localhost:8000"
 surreal_username = "root"
 surreal_password = "root"
 surreal_ns = "test"
-surreal_db = "v1.17"
-
-# Minio config
-s3_bucket = "v1.17"
-s3_url = "http://localhost:9000"
-s3_access_key = "Ok6s9uQEvKrqRoGZdacm"
-s3_secret_key = "qubeSkP787c7QZu4TvtnuwPTGIAq6ETPupCxvv6K"
+surreal_db = "v1.21.1"
 
 # Crawler config
-crawl_filter = "en.wikipedia.com" 
-budget = 1000
+crawl_filter = "https://ftpgeoinfo.msl.mt.gov/Data/Spatial/MSDI" 
+start_url = "https://ftpgeoinfo.msl.mt.gov/Data/Spatial/MSDI"
+budget = 10000
+batch_size = 50

README.md

@@ -2,13 +2,43 @@
 
 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.
 
+## How to use
+
+1. Clone the repo and `cd` into it.
+2. Build the repo with `cargo build -r`
+3. Start the docker conatiners
+	1. cd into the docker folder `cd docker`
+	2. Bring up the docker containers `docker compose up -d`
+4. From the project's root, edit the `Crawler.toml` file to your liking.
+5. Run with `./target/release/internet_mapper`
+
+You can view stats of the project at `http://<your-ip>:3000/dashboards`
+
+```bash
+# Untested script but probably works
+git clone https://git.oliveratkinson.net/Oliver/internet_mapper.git
+cd internet_mapper
+
+cargo build -r
+
+cd docker
+docker compose up -d
+cd ..
+
+$EDITOR Crawler.toml
+
+./target/release/internet_mapper
+
+```
+
 ### TODO
 
-- [ ] Domain filtering - prevent the crawler from going on alternate versions of wikipedia.
+- [x] Domain filtering - prevent the crawler from going on alternate versions of wikipedia.
 - [ ] Conditionally save content - based on filename or file contents
 - [x] GUI / TUI ? - Graphana
 - [x] Better asynchronous getting of the sites. Currently it all happens serially.
-- [ ] Allow for storing asynchronously
+- [x] Allow for storing asynchronously - dropping the "links to" logic fixes this need
+- [x] Control crawler via config file (no recompliation needed)
 
 3/17/25: Took >1hr to crawl 100 pages
 
@@ -17,6 +47,8 @@ This ment we stored 1000 pages, 142,997 urls, and 1,425,798 links between the tw
 
 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
@@ -82,4 +66,3 @@ volumes:
   grafana_storage:
   alloy_storage:
   surrealdb_storage:
-  minio_storage:

docker/prometheus.yaml

@@ -7,14 +7,11 @@ scrape_configs:
     static_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: ['192.168.1.200:2500']
+        #- 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

@@ -1,57 +1,37 @@
-use base64::{
-    alphabet,
-    engine::{self, general_purpose},
-    Engine,
-};
 use metrics::counter;
-use serde::{ser::SerializeStruct, Deserialize, Serialize};
-use std::{fmt::Debug, sync::LazyLock, time::Instant};
+use std::fmt::Debug;
+use serde::{Deserialize, Serialize};
 use surrealdb::{
     engine::remote::ws::{Client, Ws},
     opt::auth::Root,
     sql::Thing,
-    Response, Surreal,
+    Surreal,
 };
-use tokio::sync::Mutex;
-use tracing::{error, instrument, trace, warn};
+use tracing::{error, instrument, trace};
 use url::Url;
 
 use crate::Config;
 
-// static LOCK: LazyLock<Arc<Mutex<bool>>> = LazyLock::new(|| Arc::new(Mutex::new(true)));
-static LOCK: LazyLock<Mutex<bool>> = LazyLock::new(|| Mutex::new(true));
-
-const CUSTOM_ENGINE: engine::GeneralPurpose =
-    engine::GeneralPurpose::new(&alphabet::URL_SAFE, general_purpose::NO_PAD);
-
-const TIME_SPENT_ON_LOCK: &str = "surql_lock_waiting_ms";
 const STORE: &str = "surql_store_calls";
-const LINK: &str = "surql_link_calls";
 
-#[derive(Deserialize, Clone)]
+#[derive(Serialize, Deserialize, Clone, Eq, PartialEq, Hash)]
 pub struct Website {
     /// The url that this data is found at
     pub site: Url,
     /// Wether or not this link has been crawled yet
     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()
-    }
+    /// 200, 404, etc
+    pub status_code: u16,
 }
 
 // manual impl to make tracing look nicer
 impl Debug for Website {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        f.debug_struct("Website").field("site", &self.site).finish()
+        f.debug_struct("Website")
+            .field("host", &self.site.host())
+            .field("path", &self.site.path())
+            .field("status_code", &self.status_code)
+            .finish()
     }
 }
 
@@ -64,96 +44,25 @@ impl Website {
         };
         Self {
             crawled,
-            site
-        }
-    }
-
-    pub fn set_crawled(&mut self) {
-        trace!("Set crawled to true");
-        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());
-        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.first() {
-                        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());
-            }
+            site,
+            status_code: 0,
         }
     }
 
     // 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.
+    #[instrument(skip(db))]
     pub async fn store_all(all: Vec<Self>, db: &Surreal<Client>) -> Vec<Thing> {
         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.
-        // 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
-        // than just the name. `accessed_at` is fun but not needed.
-        let now = Instant::now();
-        let lock = LOCK.lock().await;
-        counter!(TIME_SPENT_ON_LOCK).increment(now.elapsed().as_millis() as u64);
-
+        // FIXME failes *sometimes* because "Resource Busy"
         match db
             .query(
                 "INSERT INTO website $array
                     ON DUPLICATE KEY UPDATE
                         accessed_at = time::now(),
+                        status_code = $input.status_code,
                         crawled = crawled OR $input.crawled
                     RETURN VALUE id;
                  ",
@@ -169,7 +78,6 @@ impl Website {
                 error!("{:?}", err);
             }
         }
-        drop(lock);
         things
     }
 }

src/filesystem.rs

@@ -0,0 +1,66 @@
+use std::{io::ErrorKind, path::PathBuf};
+
+use reqwest::header::HeaderValue;
+use tokio::fs;
+use tracing::{error, trace, warn};
+use url::Url;
+
+pub fn as_path(url: &Url, content_type: &HeaderValue) -> PathBuf {
+    // extract data from url to save it accurately
+    let mut url_path = PathBuf::from("./downloaded/".to_string() + url.domain().unwrap_or("UnknownDomain") + url.path());
+
+    if let Ok(header) = content_type.to_str() {
+        // text/html; charset=UTF-8; option=value
+        let ttype = if let Some((t, _)) = header.split_once(';') {
+            t
+        } else {
+            header
+        };
+
+        if let Some((ttype, subtype)) = ttype.split_once('/') {
+            trace!("Found Content-Type to be: {ttype}/{subtype} for {}", url.to_string());
+            // If the Content-Type header is "*/html" (most likely "text/html") and the path's
+            // extension is anything but html:
+            if subtype=="html" && !url_path.extension().is_some_and(|f| f=="html" || f=="htm" ) {
+                // time to slap a index.html to the end of that path there!
+                url_path = url_path.join("index.html");
+            }
+        }
+    } else {
+        warn!("Header: {:?} couldn't be parsed into a string!", content_type);
+    }
+    trace!("Final path for {} is: {:?}", url, url_path);
+
+    url_path
+}
+
+pub async fn init(filename: &PathBuf) -> Option<fs::File> {
+    let file = async || tokio::fs::OpenOptions::new()
+        .append(true)
+        .create(true)
+        .open(&filename).await;
+
+    match file().await {
+        Ok(ok) => Some(ok),
+        Err(err) => {
+            // the file/folder isn't found
+            if err.kind() == ErrorKind::NotFound {
+                if let Some(parent ) = &filename.parent() {
+                    // create the folders
+                    if let Err(err) = fs::create_dir_all(&parent).await {
+                        error!("Dir creation: {err} {:?}", filename);
+                        eprintln!("{}", err)
+                    } else if let Ok(ok) = file().await {
+                        return Some(ok);
+                    }
+                } else {
+                    error!("Couldn't get file's parents: {:?}", &filename);
+                }
+            } else {
+                error!("File open error: {err} {:?}", filename);
+            }
+            // we don't care about other errors, we can't/shouldn't fix them
+            None
+        }
+    }
+}

src/main.rs

@@ -1,8 +1,13 @@
 #![feature(ip_from)]
+#![warn(clippy::expect_used)]
+#![deny(clippy::unwrap_used)]
 
 extern crate html5ever;
 
+use futures_util::StreamExt;
+
 use std::{
+    collections::HashSet,
     fs::File,
     io::Read,
     net::{IpAddr, Ipv4Addr},
@@ -11,16 +16,15 @@ use std::{
 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};
+use tokio::{io::{AsyncWriteExt, BufWriter}, task::JoinSet};
+use tracing::{debug, debug_span, error, info, instrument, level_filters::LevelFilter, trace, trace_span, warn};
 use tracing_subscriber::{fmt, layer::SubscriberExt, EnvFilter, Layer, Registry};
 
 mod db;
+mod filesystem;
 mod parser;
-mod s3;
 
 const GET_METRIC: &str = "total_gets";
 const GET_IN_FLIGHT: &str = "gets_in_flight";
@@ -35,24 +39,25 @@ 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,
+    start_url: String,
     budget: usize,
+    batch_size: usize,
 }
 
 #[tokio::main]
 async fn main() {
+    println!("Logs and metrics are provided to the Grafana dashboard");
+
     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()
@@ -75,9 +80,8 @@ async fn main() {
         .install()
         .expect("failed to install recorder/exporter");
 
-    debug!("Starting...");
-    // Would probably take these in as parameters from a cli
-    let starting_url = "https://en.wikipedia.org/";
+    info!("Starting...");
+
     // When getting uncrawled pages, name must contain this variable. "" will effectively get ignored.
     // let crawl_filter = "en.wikipedia.org/";
     // let budget = 50;
@@ -88,13 +92,11 @@ async fn main() {
     let _ = file.read_to_string(&mut buf);
 
     let config: Config = toml::from_str(&buf).expect("Failed to parse Crawler.toml");
+    let starting_url = &config.start_url;
 
     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()
@@ -108,38 +110,28 @@ async fn main() {
     let pre_loop_span = span.enter();
     // Download the site
     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);
 
     let span = trace_span!("Loop");
     let span = span.enter();
     while crawled < config.budget {
-        let get_num = if config.budget - crawled < 100 {
-            config.budget - crawled
-        } else {
-            100
-        };
-
-        let uncrawled = get_uncrawled_links(&db, get_num, config.crawl_filter.clone()).await;
+        let uncrawled =
+            get_uncrawled_links(&db, config.budget - crawled, config.crawl_filter.clone(), &config).await;
         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
@@ -152,14 +144,23 @@ async fn main() {
     }
     drop(span);
 
-    debug!("Done");
+    if let Ok(mut ok) = db
+        .query("count(select id from website where crawled = true)")
+        .await
+    {
+        let res = ok.take::<Option<usize>>(0);
+        if let Ok(Some(n)) = res {
+            info!("Total crawled pages now equals {n}");
+        }
+    }
+
+    info!("Done");
 }
 
-#[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);
     // Build the request
@@ -171,48 +172,110 @@ async fn process(mut site: Website, db: Surreal<Client>, reqwest: reqwest::Clien
 
     // Send the http request (get)
     if let Ok(response) = request_builder.send().await {
+        let headers = response.headers();
+        let code = response.status();
+
+        #[allow(non_snake_case)]
+        let CT = headers.get("Content-Type");
+        let ct = headers.get("content-type");
+
+        let ct = match (CT,ct) {
+            (None, None) => {
+                warn!("Server did not respond with Content-Type header. Url: {} Headers: ({:?})", site.site.to_string(), headers);
+                return
+            },
+            (None, Some(a)) => a,
+            (Some(a), None) => a,
+            (Some(a), Some(_)) => a,
+        };
+
+        // create filepath (handles / -> /index.html)
+        let path = filesystem::as_path(&site.site, ct);
+
+        // make sure that the file is good to go
+        if let Some(file) = filesystem::init(&path).await {
+            // Get body from response
+            // stream the response onto the disk
+            let mut stream = response.bytes_stream();
+
+            let should_parse = path.to_string_lossy().ends_with(".html");
+            let mut writer = BufWriter::new(file);
+            let mut buf: Vec<u8> = Vec::new();
+
+            // Write file to disk
+            info!("Writing at: {:?}", path);
+            while let Some(data) = stream.next().await {
+                match data {
+                    Ok(data) => {
+                        let _ = writer.write_all(&data).await;
+                        // If we are going to parse this file later, we will save it
+                        // into memory as well as the disk.
+                        // We do this because the data here might be incomplete
+                        if should_parse {
+                            data.iter().for_each(|f| buf.push(*f));
+                        }
+                    },
+                    Err(err) => {
+                        eprintln!("{}", err)
+                    },
+                }
+            }
+            let _ = writer.flush();
+
+
+            // (If needed) Parse the file
+            if should_parse {
+                let span = debug_span!("Should Parse");
+                let enter = span.enter();
+
+                // Parse document and get relationships
+                let sites = parser::parse(&site, &buf).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 _ = Website::store_all(de_dupe_sites, &db).await;
+
+                drop(enter);
+            }
 
             // METRICS
             g.decrement(1);
             counter!(GET_METRIC).increment(1);
 
-        // Get body from response
-        let data = response
-            .text()
-            .await
-            .expect("Failed to read http response's body!");
-        // Store document
-        s3.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], &db).await;
+            site.crawled = true;
+            site.status_code = code.as_u16();
+            Website::store_all(vec![site.clone()], &db).await;
+        } else {
+            error!("File failed to cooperate: {:?}", path);
+        }
 
-        // Store all the other sites so that we can link to them.
-        // let mut links_to = Vec::new();
-        let _ = Website::store_all(sites, &db).await;
-
-        // Make the database's links reflect the html links between sites
-        // site.links_to(others, &db).await;
+        trace!("Done processing: {}", &site.site);
     } else {
         error!("Failed to get: {}", &site.site);
     }
 }
 
 /// Returns uncrawled links
-#[instrument(skip(db))]
+#[instrument(skip(db, config))]
 async fn get_uncrawled_links(
     db: &Surreal<Client>,
     mut count: usize,
     filter: String,
+    config: &Config,
 ) -> Vec<Website> {
-    if count > 100 {
-        count = 100
+    if count > config.batch_size {
+        count = config.batch_size;
     }
-    debug!("Getting uncrawled links");
+
+    debug!("Getting {} uncrawled links", count);
 
     let mut response = db
         .query("SELECT * FROM website WHERE crawled = false AND site ~ type::string($format) LIMIT $count;")
@@ -224,4 +287,3 @@ async fn get_uncrawled_links(
         .take(0)
         .expect("Returned websites couldn't be parsed")
 }
-

src/parser.rs

@@ -1,22 +1,24 @@
 use std::default::Default;
-use std::str::FromStr;
 
 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;
 
 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")
@@ -31,23 +33,18 @@ 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
-                                    let mut web = self.clone();
-
-                                    // Set url
-                                    let mut url = web.site;
-                                    url.set_fragment(None); // removes #xyz
-                                    let joined = url
-                                        .join(&attr.value)
-                                        .expect("Failed to join url during parsing!");
-                                    web.site = joined;
-
-                                    web.crawled = false;
+                                    trace!("Found `{}` in html `{}` tag", &attr.value, tag.name);
+                                    let url = try_get_url(&self.site, &attr.value);
 
+                                    if let Some(mut parsed) = url {
+                                        parsed.set_query(None);
+                                        parsed.set_fragment(None);
+                                        trace!("Final cleaned URL: `{}`", parsed.to_string());
+                                        let web = Website::new(&parsed.to_string(), false);
                                         links.push(web);
                                     }
                                 }
-
+                            }
                             return TokenSinkResult::Script(links);
                         }
                         local_name!("button") | local_name!("meta") | local_name!("iframe") => {
@@ -65,18 +62,20 @@ impl TokenSink for Website {
 
 #[instrument(skip_all)]
 /// Parses the passed site and returns all the sites it links to.
-pub async fn parse(site: &Website, data: &str) -> Vec<Website> {
+pub async fn parse(site: &Website, data: &[u8]) -> Vec<Website> {
+    debug!("Parsing {}", site.site.to_string());
     // prep work
     let mut other_sites: Vec<Website> = Vec::new();
 
     // change data into something that can be tokenized
-    let chunk = Tendril::from_str(data).expect("Failed to parse string into Tendril!");
+    let s: Result<Tendril<fmt::UTF8>, ()> = Tendril::try_from_byte_slice(data);
+    if let Ok(chunk) = s {
         // create buffer of tokens and push our input into it
         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());
@@ -88,7 +87,65 @@ pub async fn parse(site: &Website, data: &str) -> Vec<Website> {
         }
         assert!(token_buffer.is_empty());
         tokenizer.end();
+    } else {
+        warn!("Tendril failed to parse on: {}", site.site.to_string());
+    }
 
     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 relative 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 mut origin = parent.origin().ascii_serialization();
+                        if !origin.ends_with('/') && !link.starts_with('/') {
+                            origin += "/";
+                        }
+                        let url = origin.clone() + link;
+
+                        if let Ok(url) = Url::parse(&url) {
+                            trace!("Built `{url}` from `{origin} + `{}`", link.to_string());
+                            Some(url)
+                        } else {
+                            error!(
+                                "Failed to reconstruct a url from relative url: `{}` on site: `{}`. Failed url was: {}",
+                                link,
+                                parent.to_string(),
+                                url
+                            );
+                            None
+                        }
+                    }
+                    _ => {
+                        error!("MISC error: {:?} {:?}", e, link);
+                        None
+                    }
+                }
+            }
+        }
+    }
+}

src/s3.rs

@@ -1,100 +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};
-
-const S3_ROUND_TRIP_METRIC: &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 filename = Website::get_url_as_string(url);
-        trace!("Storing {} as {filename}", url.to_string());
-
-        counter.increment(1);
-        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();
-                }
-                _ => {}
-            },
-        };
-    }
-}
-