cleanup and more accuratly use metrics

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 {
@@ -41,35 +45,14 @@ fn valid_file_extension(take: &&OsStr) -> bool {
     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
+            // FIXME it's worth noting that the dumb tlds like .zip are in here,
+            // which could cause problems
+            let all_domains = include_str!("tlds-alpha-by-domain.txt");
 
-                "png" => true,
-                "svg" => true,
-                "jpg" => true,
-                "jpeg" => true,
-                "mp4" => true,
-                "mp3" => true,
-                "webp" => true,
-
-                "pdf" => true,
-                "json" => true,
-                "xml" => true,
-
-                // IGNORE
-                // TODO Should this be a list of all domains?
-                "org" => false,
-                "com" => false,
-                "net" => false,
-
-                _ => {
-                    warn!("Might be forgetting a file extension: {s}");
-                    false
-                }
+            // check if it is a domain
+            match all_domains.lines().map(str::to_lowercase).find(|x| x==s.to_lowercase().as_str()) {
+                Some(_) => false,
+                None => true
             }
         },
         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,
@@ -109,13 +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 {
-            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()).await;
         if uncrawled.is_empty() {
             info!("Had more budget but finished crawling everything.");
             return;
@@ -170,26 +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
-        g.decrement(1);
-        counter!(GET_METRIC).increment(1);
-
+        // TODO if this will fail if the object we are downloading is
+        // larger than the memory of the device it's running on.
+        // 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
-        filesystem::store(&data, &site.site).await;
+        // METRICS
+        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| {
@@ -202,6 +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);
@@ -215,9 +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