Fix AWS Glacier/Deep Archive loading of restored objects (#759)

* Naive headObject method wrapper for GLACIER_ARCHIVE restore. Observed header combinations while debugging do not seem to align docs, more investigation is needed.

* The AWS SDK Java v2 HeadObject.restore() is not reliable as of version 2.8.5, see aws/aws-sdk-java-v2#670 (comment)

* Other than fixing the Glacier/Archive tier error handling, also implemented support for File->Load From URL if the user knows the S3 URL, they can paste it there directly. Also aws-java-sdk-v2 version bump from Maven

* This field gets populated on newer versions of the aws-java-sdk-v2, fortunately

* Apply suggestions/code review from @reisingerf. Rollback aws-sdk version bump since it affects error messaging and "germane-ity" of this PR. Killing Triple/Tuple in next commit.

Author: brainstorm

src/main/java/org/broad/igv/aws/S3LoadDialog.java

@@ -48,9 +48,10 @@
 import java.util.Collections;
 import java.util.List;
 
-import software.amazon.awssdk.services.s3.model.S3Error;
 import software.amazon.awssdk.services.s3.model.S3Exception;
 
+import static org.broad.igv.util.AmazonUtils.isObjectAccessible;
+
 
 public class S3LoadDialog extends JDialog {
 
@@ -102,7 +103,8 @@ private void loadButtonActionPerformed(ActionEvent e) {
                 if (isFilePath(path)) {
                     Triple<String, String, String> bucketKeyTier = getBucketKeyTierFromTreePath(path);
 
-                    if(!isObjectAccessible(bucketKeyTier)) return;
+                    AmazonUtils.s3ObjectAccessResult res = isObjectAccessible(bucketKeyTier.getLeft(), bucketKeyTier.getMiddle());
+                    if(!res.getObjAvailable()) { MessageUtils.showErrorMessage(res.getErrorReason(), null); return; }
 
                     preLocatorPaths.add(bucketKeyTier);
                 }
@@ -200,23 +202,6 @@ private void updateModel(DefaultMutableTreeNode parent) {
         }
     }
 
-    // Determines whether the object is immediately available.
-    // On AWS this means present in STANDARD, STANDARD_IA, INTELLIGENT_TIERING object access tiers.
-    // Tiers GLACIER and DEEP_ARCHIVE are not immediately retrievable without action.
-    private boolean isObjectAccessible(Triple S3Obj) {
-        String S3ObjectBucket = S3Obj.getLeft().toString();
-        String S3ObjectKey = S3Obj.getMiddle().toString();
-        String S3ObjectStorageClass = S3Obj.getRight().toString();
-
-        if (S3ObjectStorageClass.contains("DEEP_ARCHIVE") ||
-                S3ObjectStorageClass.contains("GLACIER")) {
-            MessageUtils.showErrorMessage("Amazon S3 object is in " + S3ObjectStorageClass + " storage tier, not accessible at this moment. " +
-                    "Please contact your local system administrator about object: s3://" + S3ObjectBucket +  "/" + S3ObjectKey, null);
-            return false;
-        }
-
-        return true;
-    }
 
     private void initComponents() {
         // JFormDesigner - Component initialization - DO NOT MODIFY  //GEN-BEGIN:initComponents
@@ -256,7 +241,8 @@ public void mousePressed(MouseEvent e) {
                         if (isFilePath(selPath)) {
                             Triple<String, String, String> bucketKeyTier = getBucketKeyTierFromTreePath(selPath);
 
-                            if(!isObjectAccessible(bucketKeyTier)) return;
+                            AmazonUtils.s3ObjectAccessResult res = isObjectAccessible(bucketKeyTier.getLeft(), bucketKeyTier.getMiddle());
+                            if(!res.getObjAvailable()) { MessageUtils.showErrorMessage(res.getErrorReason(), null); return;}
 
                             ResourceLocator loc = getResourceLocatorFromBucketKey(bucketKeyTier);
                             IGV.getInstance().loadTracks(Collections.singletonList(loc));

src/main/java/org/broad/igv/ui/action/LoadFromURLMenuAction.java

@@ -33,13 +33,13 @@
 import org.broad.igv.exceptions.HttpResponseException;
 import org.broad.igv.feature.genome.GenomeManager;
 import org.broad.igv.google.GoogleUtils;
-import org.broad.igv.google.OAuthUtils;
 import org.broad.igv.prefs.Constants;
 import org.broad.igv.prefs.PreferencesManager;
 import org.broad.igv.ui.IGV;
 import org.broad.igv.ui.IGVMenuBar;
 import org.broad.igv.ui.util.LoadFromURLDialog;
 import org.broad.igv.ui.util.MessageUtils;
+import org.broad.igv.util.AmazonUtils;
 import org.broad.igv.util.HttpUtils;
 import org.broad.igv.util.ResourceLocator;
 
@@ -52,6 +52,8 @@
 import java.util.HashMap;
 import java.util.Map;
 
+import static org.broad.igv.util.AmazonUtils.isObjectAccessible;
+
 /**
  * @author jrobinso
  */
@@ -96,6 +98,20 @@ public void actionPerformed(ActionEvent e) {
                             MessageUtils.showMessage("Error loading url: " + url + " (" + ex.toString() + ")");
                         }
                     } else {
+                        try {
+                            // If AWS support is active, check if objects are in accessible tiers via Load URL menu...
+                            if (AmazonUtils.isAwsS3Path(url)) {
+                                String bucket = AmazonUtils.getBucketFromS3URL(url);
+                                String key = AmazonUtils.getKeyFromS3URL(url);
+
+                                AmazonUtils.s3ObjectAccessResult res = isObjectAccessible(bucket, key);
+                                if (!res.getObjAvailable()) { MessageUtils.showErrorMessage(res.getErrorReason(), null); return; }
+                            }
+                        } catch (NullPointerException npe) {
+                            // User has not yet done Amazon->Login sequence
+                            AmazonUtils.checkLogin();
+                        }
+
                         ResourceLocator rl = new ResourceLocator(url.trim());
 
                         if (dlg.getIndexURL() != null) {

src/main/java/org/broad/igv/util/AmazonUtils.java

@@ -2,6 +2,7 @@
 
 import com.google.gson.JsonObject;
 import htsjdk.samtools.util.Tuple;
+import org.apache.commons.lang3.tuple.Triple;
 import org.apache.log4j.Logger;
 import org.broad.igv.Globals;
 import org.broad.igv.aws.IGVS3Object;
@@ -203,6 +204,100 @@ public static List<String> ListBucketsForUser() {
         return bucketsFinalList;
     }
 
+    public static HeadObjectResponse getObjectMetadata(String bucket, String key) {
+        HeadObjectRequest HeadObjReq = HeadObjectRequest.builder()
+                                                        .bucket(bucket)
+                                                        .key(key).build();
+        HeadObjectResponse HeadObjRes = s3Client.headObject(HeadObjReq);
+        log.debug("getObjectMetadata(): "+HeadObjRes.toString());
+        return HeadObjRes;
+    }
+
+
+    // Holds whether a S3 object is accessible or not and reason/error msg in case it's not.
+    public static class s3ObjectAccessResult {
+        private boolean objAvailable;
+        private String errorReason;
+
+        public boolean getObjAvailable() {
+            return objAvailable;
+        }
+
+        public void setObjAvailable(boolean objAvailable) {
+            this.objAvailable = objAvailable;
+        }
+
+        public String getErrorReason() {
+            return errorReason;
+        }
+
+        public void setErrorReason(String errorReason) {
+            this.errorReason = errorReason;
+        }
+    }
+
+    // Determines whether the object is immediately available.
+    // On AWS this means present in STANDARD, STANDARD_IA, INTELLIGENT_TIERING object access tiers.
+    // Tiers GLACIER and DEEP_ARCHIVE are not immediately retrievable without action.
+    public static s3ObjectAccessResult isObjectAccessible(String bucket, String key) {
+        s3ObjectAccessResult res = new s3ObjectAccessResult();
+        HeadObjectResponse s3Meta;
+
+        String s3ObjectStorageStatus = null;
+        String s3ObjectStorageClass;
+
+        s3Meta = AmazonUtils.getObjectMetadata(bucket, key);
+        s3ObjectStorageClass = s3Meta.storageClass().toString();
+
+        // Determine in which state this object really is:
+        // 1. Archived.
+        // 2. In the process of being restored.
+        // 3. Restored
+        //
+        // This is important because after restoration the object mantains the Tier (DEEP_ARCHIVE) instead of
+        // transitioning that attribute to STANDARD, we must look at head_object response for the "Restore"
+        // attribute.
+        //
+        // Possible error reason messages for the users are:
+
+        String archived = "Amazon S3 object is in " + s3ObjectStorageClass + " storage tier, not accessible at this moment. " +
+                "Please contact your local system administrator about object: s3://" + bucket + "/" + key;
+        String restoreInProgress = "Amazon S3 object is in " + s3ObjectStorageClass + " and being restored right now, please be patient, this can take up to 48h. " +
+                "For further enquiries about this dataset, please use the following path when communicating with your system administrator: s3://" + bucket + "/" + key;
+
+        if (s3ObjectStorageClass.contains("DEEP_ARCHIVE") ||
+            s3ObjectStorageClass.contains("GLACIER")) {
+            try {
+                s3ObjectStorageStatus = s3Meta.sdkHttpResponse().headers().get("x-amz-restore").toString();
+                //S3ObjectStorageStatus = S3Meta.restore();
+            } catch(NullPointerException npe) {
+                res.setObjAvailable(false);
+                res.setErrorReason(archived);
+                return res;
+            }
+
+            if(s3ObjectStorageStatus.contains("ongoing-request=\"true\"")) {
+                res.setObjAvailable(false);
+                res.setErrorReason(restoreInProgress);
+
+            // "If an archive copy is already restored, the header value indicates when Amazon S3 is scheduled to delete the object copy"
+            } else if(s3ObjectStorageStatus.contains("ongoing-request=\"false\"") && s3ObjectStorageStatus.contains("expiry-date=")) {
+                res.setObjAvailable(true);
+            } else {
+            // The object has never been restored?
+                res.setObjAvailable(false);
+                res.setErrorReason(archived);
+            }
+        } else {
+            // The object must be either in STANDARD, INFREQUENT_ACCESS, INTELLIGENT_TIERING or
+            // any other "immediately available" tier...
+            res.setErrorReason("Object is in an accessible tier, no errors are expected");
+            res.setObjAvailable(true);
+        }
+
+        return res;
+    }
+
     private static List<String> getReadableBuckets(List<String> buckets) {
         List<CompletableFuture<String>> futures =
                 buckets.stream()
@@ -289,18 +384,19 @@ public static ArrayList<IGVS3Object> ListBucketObjects(String bucketName, String
         return objects;
     }
 
-    public static Tuple<String, String> bucketAndKey(String S3urlString) {
-        AmazonS3URI s3URI = new AmazonS3URI(S3urlString);
-        String bucket = s3URI.getBucket();
-        String key = s3URI.getKey();
+    public static String getBucketFromS3URL(String s3URL) {
+        AmazonS3URI s3URI = new AmazonS3URI(s3URL);
+        return s3URI.getBucket();
 
-        log.debug("bucketAndKey(): " + bucket + " , " + key);
-        return new Tuple(bucket, key);
+    }
+
+    public static String getKeyFromS3URL(String s3URL) {
+        AmazonS3URI s3URI = new AmazonS3URI(s3URL);
+        return s3URI.getKey();
     }
 
     // Amazon S3 Presign URLs
     // Also keeps an internal mapping between ResourceLocator and active/valid signed URLs.
-
     private static String createPresignedURL(String s3Path) throws IOException {
         // Make sure access token are valid (refreshes token internally)
         OAuthProvider provider = OAuthUtils.getInstance().getProvider("Amazon");
@@ -318,11 +414,10 @@ private static String createPresignedURL(String s3Path) throws IOException {
                 .region(getAWSREGION())
                 .build();
 
-        Tuple<String, String> bandk = bucketAndKey(s3Path);
-        String bucket = bandk.a;
-        String filename = bandk.b;
+        String bucket = getBucketFromS3URL(s3Path);
+        String key = getKeyFromS3URL(s3Path);
 
-        URI presigned = s3Presigner.presignS3DownloadLink(bucket, filename);
+        URI presigned = s3Presigner.presignS3DownloadLink(bucket, key);
         log.debug("AWS presigned URL from translateAmazonCloudURL is: " + presigned);
         return presigned.toString();
     }