Fixes

casr/src/bin/casr-cluster.rs

@@ -61,7 +61,7 @@ fn make_clusters(
     }
 
     // Get casreps with stacktraces and crashlines
-    let (casreps, stacktraces, crashlines, badreports) = util::reports_from_dirs(casreps, jobs);
+    let (casreps, stacktraces, crashlines, badreports) = util::reports_from_paths(casreps, jobs);
 
     if !badreports.is_empty() {
         fs::create_dir_all(format!("{}/clerr", &outpath.display()))?;
@@ -336,7 +336,7 @@ fn update_clusters(
 ) -> Result<(usize, usize, usize, usize, usize, usize)> {
     // Get new casreps
     let casreps = util::get_reports(newpath)?;
-    let (casreps, stacktraces, crashlines, _) = util::reports_from_dirs(casreps, jobs);
+    let (casreps, stacktraces, crashlines, _) = util::reports_from_paths(casreps, jobs);
     let casreps = casreps
         .iter()
         .zip(stacktraces.iter().zip(crashlines.iter()));
@@ -346,12 +346,8 @@ fn update_clusters(
         .unwrap()
         .map(|path| path.unwrap().path())
         .filter(|path| {
-            path.clone()
-                .file_name()
-                .unwrap()
-                .to_str()
-                .unwrap()
-                .starts_with("cl")
+            let name = path.file_name().unwrap().to_str().unwrap();
+            name.starts_with("cl") && !name.starts_with("clerr")
         })
         .collect();
     cluster_dirs.sort();
@@ -363,15 +359,13 @@ fn update_clusters(
     // Get casreps from each existing cluster
     for cluster in &cluster_dirs {
         // Get cluster number
-        let Ok(i) = cluster.clone().file_name().unwrap().to_str().unwrap()[2..]
+        let i = cluster.clone().file_name().unwrap().to_str().unwrap()[2..]
             .to_string()
             .parse::<usize>()
-        else {
-            continue;
-        };
+            .unwrap();
         // Get casreps from cluster
         let casreps = util::get_reports(cluster)?;
-        let (_, stacktraces, crashlines, _) = util::reports_from_dirs(casreps, jobs);
+        let (_, stacktraces, crashlines, _) = util::reports_from_paths(casreps, jobs);
         // Fill cluster info structures
         clusters.push(Cluster::new(i, stacktraces));
         if dedup {
@@ -496,12 +490,8 @@ fn avg_sil(dir: &Path, jobs: usize) -> Result<f64> {
         .unwrap()
         .map(|path| path.unwrap().path())
         .filter(|path| {
-            path.clone()
-                .file_name()
-                .unwrap()
-                .to_str()
-                .unwrap()
-                .starts_with("cl")
+            let name = path.file_name().unwrap().to_str().unwrap();
+            name.starts_with("cl") && !name.starts_with("clerr")
         })
         .collect();
     dirs.sort();
@@ -519,12 +509,15 @@ fn avg_sil(dir: &Path, jobs: usize) -> Result<f64> {
         // Get casreps from cluster
         let casreps = util::get_reports(dir)?;
         // Get stacktraces from cluster
-        let (_, stacktraces, _, _) = util::reports_from_dirs(casreps, jobs);
+        let (_, stacktraces, _, _) = util::reports_from_paths(casreps, jobs);
         // Update size
         size += stacktraces.len();
         // Add stacktraces
         clusters.push(stacktraces);
     }
+    if size == 0 {
+        bail!("{} valid reports, nothing to calculate...", size);
+    }
     // Init sil sum
     let mut sum = 0f64;
     // Calculate silhouette coefficient for each casrep

casr/src/util.rs

@@ -445,7 +445,7 @@ pub fn get_reports(dir: &Path) -> Result<Vec<PathBuf>> {
 /// * A vector of reports stacktraces
 /// * A vector of reports crashlines
 /// * A vector of bad reports
-pub fn reports_from_dirs(
+pub fn reports_from_paths(
     casreps: Vec<PathBuf>,
     jobs: usize,
 ) -> (Vec<PathBuf>, Vec<Stacktrace>, Vec<String>, Vec<PathBuf>) {

libcasr/src/stacktrace.rs

@@ -445,11 +445,11 @@ fn diam(stacktraces: &[Stacktrace]) -> f64 {
 /// "a" subcoefficient silhouette coefficient
 fn sil_subcoef_a(num: usize, stacktraces: &[Stacktrace]) -> f64 {
     let mut sum = 0f64;
-    for i in 0..stacktraces.len() {
+    for (i, stacktrace) in stacktraces.iter().enumerate() {
         if i == num {
             continue;
         }
-        sum += 1.0 - similarity(&stacktraces[num], &stacktraces[i]);
+        sum += 1.0 - similarity(&stacktraces[num], stacktrace);
     }
     sum / (stacktraces.len() - 1) as f64
 }
@@ -461,24 +461,24 @@ fn sil_subcoef_a(num: usize, stacktraces: &[Stacktrace]) -> f64 {
 ///
 /// * `num` - given stacktrace number
 ///
-/// * `cl` - cluster number of given stacktrace
+/// * `i` - cluster number of given stacktrace
 ///
 /// * `clusters` - a vector of clusters represented as slice of `Stacktrace` structures
 ///
 /// # Return value
 ///
 /// "b" subcoefficient silhouette coefficient
-fn sil_subcoef_b(num: usize, cl: usize, clusters: &[Vec<Stacktrace>]) -> f64 {
+fn sil_subcoef_b(num: usize, i: usize, clusters: &[Vec<Stacktrace>]) -> f64 {
     let mut min = MAX;
-    for j in 0..clusters.len() {
-        if j == cl {
+    for (j, cluster) in clusters.iter().enumerate() {
+        if j == i {
             continue;
         }
         let mut sum = 0f64;
-        for i in 0..clusters[j].len() {
-            sum += 1.0 - similarity(&clusters[cl][num], &clusters[j][i]);
+        for stacktrace in cluster {
+            sum += 1.0 - similarity(&clusters[i][num], stacktrace);
         }
-        let res = sum / clusters[j].len() as f64;
+        let res = sum / cluster.len() as f64;
         if res < min {
             min = res;
         }
@@ -499,7 +499,7 @@ fn sil_subcoef_b(num: usize, cl: usize, clusters: &[Vec<Stacktrace>]) -> f64 {
 ///
 /// # Return value
 ///
-/// "b" subcoefficient silhouette coefficient
+/// Silhouette coefficient
 pub fn sil_coef(num: usize, i: usize, clusters: &[Vec<Stacktrace>]) -> f64 {
     if clusters[i].len() != 1 {
         let a = sil_subcoef_a(num, &clusters[i]);