Optimize function that is causing OOM errors

Update .gitignore

.gitignore

@@ -14,6 +14,7 @@ cobertura.ser
 *~
 build.properties
 /index/
-
+/indexed/
+collector-02.torproject.org
 /.idea/
 .iml

build @ e1152533

-Subproject commit 467f5e132c71b9dbbcdc2513927a742f6b1e43a1
+Subproject commit e11525332d5d2f0b50f17815ba06de7afaec5db8

src/main/java/org/torproject/metrics/collector/bridgepools/BridgePoolAssignmentsProcessor.java

@@ -249,12 +249,6 @@ public class BridgePoolAssignmentsProcessor extends CollecTorMain {
             LocalDateTime bridgePoolAssignmentTime = LocalDateTime.parse(
                 line.substring("bridge-pool-assignment ".length()),
                 this.assignmentFormat);
-            if (readBridgePoolAssignments.containsKey(
-                bridgePoolAssignmentTime)) {
-              logger.warn("Input file {} contains duplicate line: {}. "
-                  + "Discarding previously read line and subsequent assignment "
-                  + "lines.", assignmentFile, line);
-            }
             currentAssignments = new TreeMap<>();
             readBridgePoolAssignments.put(bridgePoolAssignmentTime,
                 currentAssignments);
@@ -275,10 +269,6 @@ public class BridgePoolAssignmentsProcessor extends CollecTorMain {
             logger.warn("Unrecognized line '{}'. Aborting.", line);
             break;
           }
-          if (currentAssignments.containsKey(parts[0])) {
-            logger.warn("Input file {} contains duplicate line: {}. "
-                + "Discarding previously read line.", assignmentFile, line);
-          }
           currentAssignments.put(parts[0], parts[1]);
         }
       }

src/main/java/org/torproject/metrics/collector/relaydescs/ReferenceChecker.java

@@ -290,29 +290,32 @@ public class ReferenceChecker {
 
   private void checkReferences() {
     Set<String> knownDescriptors = new HashSet<>();
+    double totalMissingDescriptorsWeight = 0.0;
+
+    // First pass: collect known descriptors
     for (Reference reference : this.references) {
       knownDescriptors.add(reference.referencing);
     }
-    double totalMissingDescriptorsWeight = 0.0;
-    Set<String> missingDescriptors = new TreeSet<>();
-    StringBuilder sb = new StringBuilder("Missing referenced "
-        + "descriptors:");
+
+    // StringBuilder to accumulate the log message
+    StringBuilder sb = new StringBuilder("Missing referenced descriptors:");
+
+    // Second pass: check for missing descriptors
     for (Reference reference : this.references) {
-      if (reference.referenced.length() > 0
-          && !knownDescriptors.contains(reference.referenced)) {
-        if (!missingDescriptors.contains(reference.referenced)) {
-          totalMissingDescriptorsWeight += reference.weight;
-        }
-        missingDescriptors.add(reference.referenced);
+      if (reference.referenced.length() > 0 && !knownDescriptors.contains(reference.referenced)) {
+        totalMissingDescriptorsWeight += reference.weight;
         sb.append(String.format("%n%s -> %s (%.4f -> %.4f)",
-            reference.referencing, reference.referenced, reference.weight,
-            totalMissingDescriptorsWeight));
+                reference.referencing, reference.referenced, reference.weight,
+                totalMissingDescriptorsWeight));
       }
     }
+
+    // Log the accumulated string
     logger.info(sb.toString());
+
+    // Warn if the total missing descriptors weight exceeds the threshold
     if (totalMissingDescriptorsWeight > 0.999) {
-      logger.warn("Missing too many referenced descriptors ({}).",
-          totalMissingDescriptorsWeight);
+      logger.warn("Missing too many referenced descriptors ({}).", totalMissingDescriptorsWeight);
     }
   }