Fixes

casr/src/bin/casr-cluster.rs

@@ -85,7 +85,6 @@ fn make_clusters(inpath: &Path, outpath: Option<&Path>, jobs: usize, dedup: bool
                     traces.write().unwrap().push(trace);
                     filtered_casreps.write().unwrap().push(casreps[i].clone());
                 }
-                // TODO: Try to avoid of checking same cond
                 if dedup {
                     crashlines.write().unwrap().push(report.crashline);
                 }
@@ -410,7 +409,6 @@ fn main() -> Result<()> {
         )
         .get_matches();
 
-    // Init log
     init_ignored_frames!("cpp", "rust", "python", "go", "java");
 
     // Get number of threads

casr/tests/tests.rs

@@ -2416,6 +2416,7 @@ fn test_casr_cluster_c() {
     // Check crashline deduplication:
     // 2.casrep and 20.caserp without crashlines => no dedup
     // 3.casrep and 30.caserp with crashlines => dedup
+    // Thus, cluster with 2.casrep has 2 casreps and others have 1 casrep
     for i in 1..clusters_cnt + 1 {
         let cluster_path = paths[1].to_owned() + "/cl" + &i.to_string();
         let size = std::fs::read_dir(cluster_path.clone()).unwrap().count();

docs/usage.md

@@ -243,7 +243,7 @@ Tool for clustering CASR reports
 
 If you set `CASR_CLUSTER_UNIQUE_CRASHLINE` environment variable as "1"
 casr-cluster will deduplicate reports by crashline for each cluster and won't if
-you set ant other value or do not set anything.
+you set any other value or do not set anything.
 
 Report deduplication and clustering is based on stack trace comparison from
 [gdb-command](https://github.com/anfedotoff/gdb-command). The idea is to run

libcasr/src/stacktrace.rs

@@ -259,11 +259,9 @@ pub fn cluster_stacktraces(stacktraces: &[Stacktrace]) -> Result<Vec<usize>> {
 /// * 'clusters' - An vector of the same length as `crashlines`.
 /// Vec\[i\] is the flat cluster number to which original casrep i belongs.
 ///
-/// * cluster_num - number of cluster
-///
 /// # Return value
 ///
-/// An vector of the same length as `crashlines`.
+/// An vector of the same length as `clusters`.
 /// Vec\[i\] is the flat cluster number to which original casrep i belongs,
 /// if Vec\[i\] is 0, casrep crashline is duplicate of some other
 pub fn dedup_crashlines(crashlines: &[String], clusters: Vec<usize>) -> Result<Vec<usize>> {
@@ -275,8 +273,7 @@ pub fn dedup_crashlines(crashlines: &[String], clusters: Vec<usize>) -> Result<V
     let mut unique_crashlines: Vec<HashSet<String>> = vec![HashSet::new(); cluster_num];
 
     // Dedup reports by crashlibe
-    for i in 0..crashlines.len() {
-        let crashline = &crashlines[i];
+    for (i, crashline) in crashlines.iter().enumerate() {
         // Leave report in the cluster if crashline is absent
         if crashline.is_empty() || unique_crashlines[clusters[i] - 1].insert(crashline.to_string())
         {