Merge pull request 'foss_storage' (#3) from foss_storage into main

Reviewed-on: #3

Crawler.toml

@@ -3,8 +3,9 @@ surreal_url = "localhost:8000"
 surreal_username = "root"
 surreal_password = "root"
 surreal_ns = "test"
-surreal_db = "v1.19.2"
+surreal_db = "v1.19.5"
 
 # Crawler config
-crawl_filter = "en.wikipedia.com" 
+crawl_filter = "en.wikipedia.org" 
-budget = 1000
+start_url = "https://en.wikipedia.org"
+budget = 100

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
 

docker/compose.yml

@@ -66,4 +66,3 @@ volumes:
   grafana_storage:
   alloy_storage:
   surrealdb_storage:
-  minio_storage:

src/db.rs

@@ -49,6 +49,7 @@ impl Website {
 
     // 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());

src/filesystem.rs

@@ -5,7 +5,10 @@ use tracing::{debug, error, instrument, trace, warn};
 use url::Url;
 
 #[instrument(skip(data))]
-pub async fn store(data: &str, url: &Url) {
+/// Returns whether or not the saved file should be parsed.
+/// If the file is just data, like an image, it doesn't need to be parsed.
+/// If it's html, then it does need to be parsed.
+pub async fn store(data: &str, url: &Url) -> bool {
     // extract data from url to save it accurately
     let url_path = PathBuf::from("./downloaded/".to_string() + url.domain().unwrap_or("UnknownDomain") + url.path());
 
@@ -21,19 +24,20 @@ pub async fn store(data: &str, url: &Url) {
         (url_path.clone(), "index.html".into())
     };
 
+    let should_parse = filename.ends_with(".html");
+
     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);
         }
     }
+
+    should_parse
 }
 
 fn valid_file_extension(take: &&OsStr) -> bool {
@@ -41,28 +45,14 @@ fn valid_file_extension(take: &&OsStr) -> bool {
     let all = los.split('.');
     match all.last() {
         Some(s) => {
-            match s.to_lowercase().as_str() {
+            // FIXME it's worth noting that the dumb tlds like .zip are in here,
-                "html" => true,
+            // which could cause problems
-                "css" => true,
+            let all_domains = include_str!("tlds-alpha-by-domain.txt");
-                "js" => true,
-                "ts" => true,
-                "otf" => true, // font
 
-                "png" => true,
+            // check if it is a domain
-                "svg" => true,
+            match all_domains.lines().map(str::to_lowercase).find(|x| x==s.to_lowercase().as_str()) {
-                "jpg" => true,
+                Some(_) => false,
-                "jpeg" => true,
+                None => 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

@@ -24,6 +24,8 @@ const GET_IN_FLIGHT: &str = "gets_in_flight";
 const SITES_CRAWLED: &str = "pages_crawled";
 const BEING_PROCESSED: &str = "pages_being_processed";
 
+const BATCH_SIZE: usize = 2;
+
 #[derive(Deserialize)]
 struct Config {
     surreal_ns: String,
@@ -33,11 +35,14 @@ struct Config {
     surreal_password: String,
 
     crawl_filter: String,
+    start_url: String,
     budget: 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)
@@ -70,8 +75,7 @@ async fn main() {
         .expect("failed to install recorder/exporter");
 
     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.
     // let crawl_filter = "en.wikipedia.org/";
     // let budget = 50;
@@ -82,6 +86,7 @@ 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
@@ -106,13 +111,7 @@ async fn main() {
     let span = trace_span!("Loop");
     let span = span.enter();
     while crawled < config.budget {
-        let get_num = if config.budget - crawled < 100 {
+        let uncrawled = get_uncrawled_links(&db, config.budget - crawled, config.crawl_filter.clone()).await;
-            config.budget - crawled
-        } else {
-            100
-        };
-
-        let uncrawled = get_uncrawled_links(&db, get_num, config.crawl_filter.clone()).await;
         if uncrawled.is_empty() {
             info!("Had more budget but finished crawling everything.");
             return;
@@ -138,6 +137,15 @@ async fn main() {
     }
     drop(span);
 
+    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(i) = res {
+            if let Some(n) = i {
+                info!("Total crawled pages now equals {n}");
+            }
+        }
+    }
+
     info!("Done");
 }
 
@@ -158,26 +166,26 @@ 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 {
 
-        // METRICS
+        // TODO if this will fail if the object we are downloading is
-        g.decrement(1);
+        // larger than the memory of the device it's running on.
-        counter!(GET_METRIC).increment(1);
+        // We should store it *as* we download it then parse it in-place.
-
         // Get body from response
         let data = response
             .text()
             .await
             .expect("Failed to read http response's body!");
 
-        // Store document
+        // METRICS
-        filesystem::store(&data, &site.site).await;
+        g.decrement(1);
+        counter!(GET_METRIC).increment(1);
 
+        // Store document
+        let should_parse = filesystem::store(&data, &site.site).await;
+
+        if should_parse {
             // 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;
-
             // De-duplicate this list
             let prev_len = sites.len();
             let set = sites.into_iter().fold(HashSet::new(), |mut set,item| {
@@ -190,6 +198,11 @@ async fn process(mut site: Website, db: Surreal<Client>, reqwest: reqwest::Clien
 
             // Store all the other sites so that we can link to them.
             let _ = Website::store_all(de_dupe_sites, &db).await;
+        }
+
+        // update self in db
+        site.set_crawled();
+        Website::store_all(vec![site], &db).await;
 
     } else {
         error!("Failed to get: {}", &site.site);
@@ -203,9 +216,11 @@ async fn get_uncrawled_links(
     mut count: usize,
     filter: String,
 ) -> Vec<Website> {
-    if count > 100 {
+
-        count = 100
+    if count > BATCH_SIZE {
+        count = BATCH_SIZE;
     }
+
     debug!("Getting uncrawled links");
 
     let mut response = db