Determine Azure directories more accurately using metadata and blob properties

plugins/nf-azure/src/main/nextflow/cloud/azure/nio/AzFileAttributes.groovy

@@ -58,23 +58,42 @@ class AzFileAttributes implements BasicFileAttributes {
         objectId = "/${client.containerName}/${client.blobName}"
         creationTime = time(props.getCreationTime())
         updateTime = time(props.getLastModified())
-        directory = client.blobName.endsWith('/')
-        size = props.getBlobSize()
-
-        // Support for Azure Data Lake Storage Gen2 with hierarchical namespace enabled
+
+        // Determine if this is a directory using metadata only (most reliable):
         final meta = props.getMetadata()
-        if( meta.containsKey("hdi_isfolder") && size == 0 ){
-            directory = meta.get("hdi_isfolder")
+        if( meta != null && meta.containsKey("hdi_isfolder") && meta.get("hdi_isfolder") == "true" ){
+            directory = true
+            size = 0
+        }
+        else {
+            // Without metadata, default to treating as file
+            // This aligns with Azure SDK's approach where explicit directory markers are required
+            directory = false
+            size = props.getBlobSize()
         }
     }
 
     AzFileAttributes(String containerName, BlobItem item) {
         objectId = "/${containerName}/${item.name}"
-        directory = item.name.endsWith('/')
-        if( !directory ) {
-            creationTime = time(item.properties.getCreationTime())
-            updateTime = time(item.properties.getLastModified())
-            size = item.properties.getContentLength()
+
+        // Determine if this is a directory using reliable methods only:
+        // 1. Check if it's marked as a prefix (virtual directory) - Most reliable
+        if( item.isPrefix() != null && item.isPrefix() ) {
+            directory = true
+            // Virtual directories don't have properties like creation time
+            size = 0
+        }
+        // 2. Check metadata for hierarchical namespace (ADLS Gen2)
+        else if( item.getMetadata() != null && item.getMetadata().containsKey("hdi_isfolder") && item.getMetadata().get("hdi_isfolder") == "true" ) {
+            directory = true
+            size = 0
+        }
+        // 3. Default: treat as file
+        else {
+            directory = false
+            creationTime = time(item.getProperties().getCreationTime())
+            updateTime = time(item.getProperties().getLastModified())
+            size = item.getProperties().getContentLength()
         }
     }
 

plugins/nf-azure/src/test/nextflow/cloud/azure/nio/AzFileAttributesTest.groovy

@@ -0,0 +1 @@
+

plugins/nf-azure/src/test/nextflow/cloud/azure/nio/AzNioTest.groovy

@@ -928,4 +928,51 @@ class AzNioTest extends Specification implements AzBaseSpec {
         deleteBucket(bucket1)
     }
 
+    def 'should detect directory with hdi_isfolder metadata' () {
+        given:
+        def bucketName = createBucket()
+        def dirPath = "$bucketName/test-dir"
+
+        when:
+        // Create a directory marker with hdi_isfolder metadata
+        def containerClient = storageClient.getBlobContainerClient(bucketName)
+        def blobClient = containerClient.getBlobClient("test-dir/")
+        blobClient.upload(new ByteArrayInputStream(new byte[0]), 0)
+        blobClient.setMetadata(['hdi_isfolder': 'true'])
+
+        and:
+        def path = Paths.get(new URI("az://$dirPath/"))
+        def attrs = Files.readAttributes(path, BasicFileAttributes)
+
+        then:
+        attrs.isDirectory()
+        !attrs.isRegularFile()
+
+        cleanup:
+        deleteBucket(bucketName)
+    }
+
+    def 'should not treat file with trailing slash as directory without metadata' () {
+        given:
+        def bucketName = createBucket()
+
+        when:
+        // Create a file with trailing slash but no directory metadata
+        def containerClient = storageClient.getBlobContainerClient(bucketName)
+        def blobClient = containerClient.getBlobClient("test-file/")
+        blobClient.upload(new ByteArrayInputStream("content".bytes), "content".length())
+
+        and:
+        def path = Paths.get(new URI("az://$bucketName/test-file/"))
+        def attrs = Files.readAttributes(path, BasicFileAttributes)
+
+        then:
+        // Without metadata or isPrefix, it should be treated as a file
+        attrs.isRegularFile()
+        !attrs.isDirectory()
+
+        cleanup:
+        deleteBucket(bucketName)
+    }
+
 }