weko#24923 - Handle check consistence between file_path and filename when import

Author: weko3-dev040

modules/weko-search-ui/weko_search_ui/translations/ja/LC_MESSAGES/messages.mo

@@ -747,3 +747,6 @@ msgstr "（{}）に指定したファイルが存在しません。"
 
 msgid "The file specified in ({}) does not exist.<br/>The file will not be updated. Update only the metadata with tsv contents."
 msgstr "（{}）に指定したファイルが存在しません。<br/>ファイルの更新はしません。tsv内容でメタデータのみ更新します。"
+
+msgid "The file name specified in {} and {} do not match."
+msgstr "{}に指定されたファイル名と{}が一致しません。"

modules/weko-search-ui/weko_search_ui/translations/ja/LC_MESSAGES/messages.po

@@ -702,3 +702,6 @@ msgstr ""
 
 msgid "The file specified in ({}) does not exist.<br/>The file will not be updated. Update only the metadata with tsv contents."
 msgstr ""
+
+msgid "The file name specified in {} and {} do not match."
+msgstr ""

modules/weko-search-ui/weko_search_ui/translations/messages.pot

@@ -412,7 +412,8 @@ def convert_data(pro, path=None):
         if key in item_path_not_existed:
             continue
         key_path = handle_generate_key_path(key)
-        if value or key_path[0] in ['file_path', 'thumbnail_path']:
+        if value or key_path[0] in ['file_path', 'thumbnail_path'] \
+                or key_path[-1] == 'filename':
             set_nested_item(result, key_path, value)
 
     convert_data(result)
@@ -2734,10 +2735,14 @@ def handle_check_file_content(record, data_path):
     errors = []
     warnings = []
 
+    file_paths = record.get('file_path', [])
+    # check consistence between file_path and filename
+    filenames = get_filenames_from_metadata(record['metadata'])
+    errors.extend(handle_check_filename_consistence(file_paths, filenames))
+
+    # check if file_path exists
     error, warning = handle_check_file_path(
-        record.get('file_path', []),
-        data_path,
-        record['status'] == 'new')
+        file_paths, data_path, record['status'] == 'new')
     if error:
         errors.append(error)
     if warning:
@@ -2778,3 +2783,53 @@ def handle_check_thumbnail(record, data_path):
         warnings.append(warning)
 
     return errors, warnings
+
+
+def get_filenames_from_metadata(metadata):
+    """Check thumbnails metadata.
+
+    :argument
+        metadata -- {dict} record metadata.
+    :return
+        filenames -- {list} List filename data.
+    """
+    filenames = []
+    file_meta_ids = []
+    for key, val in metadata.items():
+        if isinstance(val, list):
+            for item in val:
+                if isinstance(item, dict) and 'filename' in item:
+                    file_meta_ids.append(key)
+                    break
+
+    count = 0
+    for _id in file_meta_ids:
+        for file in metadata[_id]:
+            data = {
+                'id': '.metadata.{}[{}].filename'.format(_id, count),
+                'filename': file.get('filename', '')
+            }
+            filenames.append(data)
+            count += 1
+
+    return filenames
+
+
+def handle_check_filename_consistence(file_paths, meta_filenames):
+    """Check thumbnails metadata.
+
+    :argument
+        file_paths -- {list} List file_path.
+        meta_filenames   -- {list} List filename from metadata.
+    :return
+        errors -- {list} List errors.
+    """
+    errors = []
+    msg = _('The file name specified in {} and {} do not match.')
+    for idx, path in enumerate(file_paths):
+        meta_filename = meta_filenames[idx]
+        if path and path.split('/')[-1] != meta_filename['filename']:
+            errors.append(msg.format(
+                'file_path[{}]'.format(idx), meta_filename['id']))
+
+    return errors