From cedc59fef6709e81c2cdc700d7eaad7023863119 Mon Sep 17 00:00:00 2001
From: Jan van Mansum <janvanmansum@gmail.com>
Date: Wed, 14 Feb 2024 13:17:42 +0100
Subject: [PATCH] Fixed + improved dv-dataset-destroy-migration-placeholder
 (#54)

* Modified dv-dataset-destroy-migration-placeholder so that it also recognizes easy-migration.zip
as a placeholder file, and not only and easy-migration folder.

* - Implemented missing dry-run option for dv-dataset-destroy-migration-placeholder
- Refactored filtering on datasets with only 'easy-migration' files + also allowing easy-migration.zip as 'easy-migration' file (formerly, easy-migration
was expected to be a directory containing up to 4 files).
---
 pyproject.toml                                |  2 +-
 .../dataverse/destroy_placeholder_dataset.py  |  9 ++++----
 ...v_dataset_destroy_migration_placeholder.py |  3 ++-
 src/tests/test_destroy_placeholder_dataset.py | 22 +++++++++++++++++++
 4 files changed, 30 insertions(+), 6 deletions(-)
 create mode 100644 src/tests/test_destroy_placeholder_dataset.py

diff --git a/pyproject.toml b/pyproject.toml
index 89e54c0..c910aea 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dans-datastation-tools"
-version = "0.38.0"
+version = "0.39.0"
 description = "Command line utilities for Data Station application management"
 authors = ["DANS-KNAW"]
 packages = [
diff --git a/src/datastation/dataverse/destroy_placeholder_dataset.py b/src/datastation/dataverse/destroy_placeholder_dataset.py
index 94f3f6e..60d4174 100644
--- a/src/datastation/dataverse/destroy_placeholder_dataset.py
+++ b/src/datastation/dataverse/destroy_placeholder_dataset.py
@@ -13,8 +13,9 @@ def matches(description):
     return matches
 
 
-def has_directory_label_different_from(file_metadata, dir_label):
-    return 'directoryLabel' not in file_metadata or file_metadata['directoryLabel'] != dir_label
+def is_migration_file(file_metadata):
+    return ('directoryLabel' in file_metadata and file_metadata['directoryLabel'] == 'easy-migration') or \
+        ('directoryLabel' not in file_metadata and file_metadata['label'] == 'easy-migration.zip')
 
 
 def destroy_placeholder_dataset(dataset_api: DatasetApi, description_text_pattern, csv_report: CsvReport,
@@ -45,12 +46,12 @@ def destroy_placeholder_dataset(dataset_api: DatasetApi, description_text_patter
             messages.append(f"Found {len(files)} files <= 4: OK")
 
         non_easy_migration_files = list(
-            filter(lambda m: has_directory_label_different_from(m, 'easy-migration'), files))
+            filter(lambda m: not is_migration_file(m), files))
         logging.debug(f"Non easy-migration files: {non_easy_migration_files}")
 
         if len(non_easy_migration_files) > 0:
             blocker = True
-            messages.append(f"Files other than 'easy-migration' found: {len(non_easy_migration_files)}: BLOCKER")
+            messages.append(f"Files other than 'easy-migration/*' or 'easy-migration.zip' found: {len(non_easy_migration_files)}: BLOCKER")
         else:
             messages.append("Only found easy-migration files: OK")
 
diff --git a/src/datastation/dv_dataset_destroy_migration_placeholder.py b/src/datastation/dv_dataset_destroy_migration_placeholder.py
index 1121e6b..3a7a76e 100644
--- a/src/datastation/dv_dataset_destroy_migration_placeholder.py
+++ b/src/datastation/dv_dataset_destroy_migration_placeholder.py
@@ -28,7 +28,8 @@ def main():
     batch_processor.process_pids(pids,
                                  callback=lambda pid, csv_report: destroy_placeholder_dataset(dataverse.dataset(pid),
                                                                                               description_text_pattern,
-                                                                                              csv_report))
+                                                                                              csv_report,
+                                                                                              dry_run=args.dry_run))
 
 
 if __name__ == '__main__':
diff --git a/src/tests/test_destroy_placeholder_dataset.py b/src/tests/test_destroy_placeholder_dataset.py
new file mode 100644
index 0000000..d321d49
--- /dev/null
+++ b/src/tests/test_destroy_placeholder_dataset.py
@@ -0,0 +1,22 @@
+import datastation.dataverse.destroy_placeholder_dataset
+
+
+def test_is_migration_file_returns_true_for_file_with_label_easy_migration_dot_zip():
+    file_metadata = {'label': 'easy-migration.zip'}
+    assert datastation.dataverse.destroy_placeholder_dataset.is_migration_file(file_metadata) == True
+
+
+def test_is_migration_file_returns_true_for_file_with_directory_label_easy_migration():
+    file_metadata = {'directoryLabel': 'easy-migration'}
+    assert datastation.dataverse.destroy_placeholder_dataset.is_migration_file(file_metadata) == True
+
+
+def test_is_migration_file_returns_false_for_file_with_label_not_easy_migration_dot_zip():
+    file_metadata = {'label': 'not-easy-migration.zip'}
+    assert datastation.dataverse.destroy_placeholder_dataset.is_migration_file(file_metadata) == False
+
+
+def test_is_migration_file_returns_false_for_file_with_directory_label_not_easy_migration():
+    file_metadata = {'directoryLabel': 'not-easy-migration'}
+    assert datastation.dataverse.destroy_placeholder_dataset.is_migration_file(file_metadata) == False
+