Merge branch 'master' of github.com:internetarchive/heritrix3

commons/src/main/java/org/archive/io/WriterPool.java

@@ -92,7 +92,7 @@ public abstract class WriterPool {
             final int poolMaximumActive, final int poolMaximumWait) {
         logger.info("Initial configuration:" +
                 " prefix=" + settings.getPrefix() +
-                ", suffix=" + settings.getTemplate() +
+                ", template=" + settings.getTemplate() +
                 ", compress=" + settings.getCompress() +
                 ", maxSize=" + settings.getMaxFileSizeBytes() +
                 ", maxActive=" + poolMaximumActive +

commons/src/main/java/org/archive/io/arc/ARCReaderFactory.java

@@ -285,7 +285,11 @@ implements ARCConstants {
         			compressedARCFile = true;
         		}
         	}
+        } else {
+        	// Some old arcs don't have an extra header at all, but they're still compressed
+        	compressedARCFile = true;
         }
+        
         return compressedARCFile;
     }
 

engine/src/main/java/org/archive/crawler/framework/ToePool.java

@@ -283,30 +283,32 @@ public class ToePool extends ThreadGroup implements Reporter {
         w.print(map.get("toeCount"));
         w.print(" threads: ");
         
-        TreeSet<Map.Entry<Object,Long>> sortedSteps = (TreeSet<Entry<Object, Long>>) map.get("steps");
-        if(sortedSteps.size()==0) {
-            return;
+        LinkedList<String> sortedSteps = (LinkedList<String>)map.get("steps");
+        {
+        	Iterator<String> iter = sortedSteps.iterator();
+        	if (!iter.hasNext()) {
+        		return;
+        	}
+        	w.print(iter.next());
+        	if (iter.hasNext()) {
+        		w.print(", ");
+        		w.print(iter.next());
+        		if (iter.hasNext()) {
+        			w.print(", etc...");
+        		}
+        	}
+        	w.print("; ");
         }
-        w.print(Histotable.entryString(sortedSteps.first()));
-        if(sortedSteps.size()>1) {
-            Iterator<Map.Entry<Object,Long>> iter = sortedSteps.iterator();
-            iter.next();
-            w.print(", ");
-            w.print(Histotable.entryString(iter.next()));
-        }
-        if(sortedSteps.size()>2) {
-            w.print(", etc...");
-        }
-        w.print("; ");
-        TreeSet<Map.Entry<Object,Long>> sortedProcessors = (TreeSet<Entry<Object, Long>>) map.get("processors");
-        w.print(Histotable.entryString(sortedProcessors.first()));
-        if(sortedProcessors.size()>1) {
-            Iterator<Map.Entry<Object,Long>> iter = sortedProcessors.iterator();
-            iter.next();
-            while(iter.hasNext()) {
-                w.print(", ");
-                w.print(Histotable.entryString(iter.next()));
-            }
+        LinkedList<String> sortedProcesses = (LinkedList<String>)map.get("processors");
+        {
+        	Iterator<String> iter = sortedProcesses.iterator();
+        	if (iter.hasNext()) {
+        		w.print(iter.next());
+        		while (iter.hasNext()) {
+        			w.print(", ");
+        			w.print(iter.next());
+        		}
+        	}
         }
 
     }

engine/src/main/java/org/archive/crawler/reporting/CrawlerLoggerModule.java

@@ -235,7 +235,7 @@ public class CrawlerLoggerModule
         closeLogFiles();
     }
     
-    private void setupLogs() throws IOException {
+    protected void setupLogs() throws IOException {
         String logsPath = getPath().getFile().getAbsolutePath() + File.separatorChar;
         uriProcessing = Logger.getLogger(LOGNAME_CRAWL + "." + logsPath);
         runtimeErrors = Logger.getLogger(LOGNAME_RUNTIME_ERRORS + "." +

engine/src/main/resources/org/archive/crawler/restlet/Job.ftl

@@ -164,7 +164,7 @@ ${line?html}
 	</#if>
 
 	<h2>Files</h2>
-	<h3>Browser <a href='jobdir'>Job Directory</a></h3>
+	<h3>Browse <a href='jobdir'>Job Directory</a></h3>
 	<h3>Configuration-referenced Paths</h3>
 	<#assign configRefPaths=job.configFiles! />
 	<#if !configRefPaths?has_content >