Merge pull request #672 from epi052/661-fix-double-dir-scan

661 fix double dir scan

src/scan_manager/scan.rs

@@ -32,6 +32,9 @@ pub struct FeroxScan {
     /// The URL that to be scanned
     pub(super) url: String,
 
+    /// A url used solely for comparison to other URLs
+    pub(super) normalized_url: String,
+
     /// The type of scan
     pub scan_type: ScanType,
 
@@ -79,6 +82,7 @@ impl Default for FeroxScan {
             num_requests: 0,
             scan_order: ScanOrder::Latest,
             url: String::new(),
+            normalized_url: String::new(),
             progress_bar: Mutex::new(None),
             scan_type: ScanType::File,
             output_level: Default::default(),
@@ -191,6 +195,7 @@ impl FeroxScan {
     ) -> Arc<Self> {
         Arc::new(Self {
             url: url.to_string(),
+            normalized_url: format!("{}/", url.trim_end_matches('/')),
             scan_type,
             scan_order,
             num_requests,
@@ -332,10 +337,11 @@ impl Serialize for FeroxScan {
     where
         S: Serializer,
     {
-        let mut state = serializer.serialize_struct("FeroxScan", 4)?;
+        let mut state = serializer.serialize_struct("FeroxScan", 6)?;
 
         state.serialize_field("id", &self.id)?;
         state.serialize_field("url", &self.url)?;
+        state.serialize_field("normalized_url", &self.normalized_url)?;
         state.serialize_field("scan_type", &self.scan_type)?;
         state.serialize_field("status", &self.status)?;
         state.serialize_field("num_requests", &self.num_requests)?;
@@ -387,6 +393,11 @@ impl<'de> Deserialize<'de> for FeroxScan {
                         scan.url = url.to_string();
                     }
                 }
+                "normalized_url" => {
+                    if let Some(normalized_url) = value.as_str() {
+                        scan.normalized_url = normalized_url.to_string();
+                    }
+                }
                 "num_requests" => {
                     if let Some(num_requests) = value.as_u64() {
                         scan.num_requests = num_requests;
@@ -480,6 +491,7 @@ mod tests {
         let scan = FeroxScan {
             id: "".to_string(),
             url: "".to_string(),
+            normalized_url: String::from("/"),
             scan_type: ScanType::Directory,
             scan_order: ScanOrder::Initial,
             num_requests: 0,

src/scan_manager/scan_container.rs

@@ -213,8 +213,10 @@ impl FeroxScans {
     /// on the given URL
     pub fn contains(&self, url: &str) -> bool {
         if let Ok(scans) = self.scans.read() {
+            let normalized = format!("{}/", url.trim_end_matches('/'));
+
             for scan in scans.iter() {
-                if scan.url == url {
+                if scan.normalized_url == normalized {
                     return true;
                 }
             }
@@ -225,8 +227,10 @@ impl FeroxScans {
     /// Find and return a `FeroxScan` based on the given URL
     pub fn get_scan_by_url(&self, url: &str) -> Option<Arc<FeroxScan>> {
         if let Ok(guard) = self.scans.read() {
+            let normalized = format!("{}/", url.trim_end_matches('/'));
+
             for scan in guard.iter() {
-                if scan.url == url {
+                if scan.normalized_url == normalized {
                     return Some(scan.clone());
                 }
             }
@@ -589,7 +593,8 @@ impl FeroxScans {
     ///
     /// Also return a reference to the new `FeroxScan`
     pub fn add_directory_scan(&self, url: &str, scan_order: ScanOrder) -> (bool, Arc<FeroxScan>) {
-        self.add_scan(url, ScanType::Directory, scan_order)
+        let normalized = format!("{}/", url.trim_end_matches('/'));
+        self.add_scan(&normalized, ScanType::Directory, scan_order)
     }
 
     /// Given a url, create a new `FeroxScan` and add it to `FeroxScans` as a File Scan

src/scan_manager/tests.rs

@@ -277,7 +277,7 @@ fn ferox_scan_serialize() {
         None,
     );
     let fs_json = format!(
-        r#"{{"id":"{}","url":"https://spiritanimal.com","scan_type":"Directory","status":"NotStarted","num_requests":0}}"#,
+        r#"{{"id":"{}","url":"https://spiritanimal.com","normalized_url":"https://spiritanimal.com/","scan_type":"Directory","status":"NotStarted","num_requests":0}}"#,
         fs.id
     );
     assert_eq!(fs_json, serde_json::to_string(&*fs).unwrap());
@@ -296,7 +296,7 @@ fn ferox_scans_serialize() {
     );
     let ferox_scans = FeroxScans::default();
     let ferox_scans_json = format!(
-        r#"[{{"id":"{}","url":"https://spiritanimal.com","scan_type":"Directory","status":"NotStarted","num_requests":0}}]"#,
+        r#"[{{"id":"{}","url":"https://spiritanimal.com","normalized_url":"https://spiritanimal.com/","scan_type":"Directory","status":"NotStarted","num_requests":0}}]"#,
         ferox_scan.id
     );
     ferox_scans.scans.write().unwrap().push(ferox_scan);
@@ -556,6 +556,7 @@ fn feroxscan_display() {
     let scan = FeroxScan {
         id: "".to_string(),
         url: String::from("http://localhost"),
+        normalized_url: String::from("http://localhost/"),
         scan_order: ScanOrder::Latest,
         scan_type: Default::default(),
         num_requests: 0,
@@ -600,6 +601,7 @@ async fn ferox_scan_abort() {
     let scan = FeroxScan {
         id: "".to_string(),
         url: String::from("http://localhost"),
+        normalized_url: String::from("http://localhost/"),
         scan_order: ScanOrder::Latest,
         scan_type: Default::default(),
         num_requests: 0,

tests/test_scan_manager.rs

@@ -20,11 +20,13 @@ fn resume_scan_works() {
     // localhost:PORT/ <- complete
     // localhost:PORT/js <- will get scanned with /css and /stuff
     let complete_scan = format!(
-        r#"{{"id":"057016a14769414aac9a7a62707598cb","url":"{}","scan_type":"Directory","status":"Complete"}}"#,
-        srv.url("/")
+        r#"{{"id":"057016a14769414aac9a7a62707598cb","url":"{}","normalized_url":"{}","scan_type":"Directory","status":"Complete"}}"#,
+        srv.url("/"),
+        srv.url("/"),
     );
     let incomplete_scan = format!(
-        r#"{{"id":"400b2323a16f43468a04ffcbbeba34c6","url":"{}","scan_type":"Directory","status":"NotStarted"}}"#,
+        r#"{{"id":"400b2323a16f43468a04ffcbbeba34c6","url":"{}","normalized_url":"{}/","scan_type":"Directory","status":"NotStarted"}}"#,
+        srv.url("/js"),
         srv.url("/js")
     );
     let scans = format!(r#""scans":[{},{}]"#, complete_scan, incomplete_scan);