Merge pull request #215 from ukwa/use-latest-checkpoint

Extend API to simplify using the latest checkpoint

Author: nlevitt

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

@@ -393,6 +393,22 @@ public synchronized void setRecoveryCheckpointByName(String selectedCheckpoint)
         if(isRunning) {
             throw new RuntimeException("may not set recovery Checkpoint after launch");
         }
+        // If the selectedCheckpoint is 'latest', pick up the latest checkpoint
+        // and use that:
+        if ("latest".equalsIgnoreCase(selectedCheckpoint)) {
+            List<File> cps = this.findAvailableCheckpointDirectories();
+            if (cps == null || cps.size() == 0) {
+                LOGGER.warning(
+                        "Cannot find any checkpoints so cannot choose the latest one! Assuming we should launch a new crawl.");
+                return;
+            }
+            // As per the API above, the most recent checkpoint is the first in
+            // the list:
+            File latestFile = cps.get(0);
+            // For the checkpoint we use the folder name:
+            selectedCheckpoint = latestFile.getName();
+        }
+        // Now setup the checkpoint:
         Checkpoint recoveryCheckpoint = new Checkpoint();
         recoveryCheckpoint.getCheckpointDir().setBase(getCheckpointsDir());
         recoveryCheckpoint.getCheckpointDir().setPath(selectedCheckpoint);