feat: UI support for zip content previewing

* New entrypoint for zip enhanced previewer
* Two new routes for preview and download container items
* Zip enhanced previewer
* Decorator to pass container item to views
* view function to preview specific container item and download it
* JS and HTML files

Co-authored-by: Ondřej Ruml <Ondrej.Ruml@cesnet.cz>
Co-authored-by: Arsenii Pogodin <pogodars@cvut.cz>
Co-authored-by: Martin Čorovčák <Martin.Corovcak@cesnet.cz>

Author: 4 people

invenio_app_rdm/config.py

@@ -805,7 +805,9 @@ def files_rest_permission_factory(obj, action):
     "record_detail": "/records/<pid_value>",
     "record_export": "/records/<pid_value>/export/<export_format>",
     "record_file_preview": "/records/<pid_value>/preview/<path:filename>",
+    "record_container_item_preview": "/records/<pid_value>/preview/<path:filename>/container/<path:path>",
     "record_file_download": "/records/<pid_value>/files/<path:filename>",
+    "record_container_item_download": "/records/<pid_value>/files/<path:filename>/container/<path:path>",
     "record_thumbnail": "/records/<pid_value>/thumb<int:size>",
     "record_media_file_download": "/records/<pid_value>/media-files/<path:filename>",
     "record_from_pid": "/<any({schemes}):pid_scheme>/<path:pid_value>",
@@ -1168,14 +1170,28 @@ def github_link_render(record):
     "video_videojs",
     "audio_videojs",
     "ipynb",
-    "zip",
+    "previewable_zip",
     "txt",
 ]
 """Preferred previewers."""
 
 PREVIEWER_ABSTRACT_TEMPLATE = "invenio_previewer/rdm_abstract_previewer.html"
 """Override the abstract template with an RDM-specific one."""
 
+CONTAINER_ITEM_PREVIEWER_PREFERENCE = [
+    "csv_papaparsejs",
+    "pdfjs",
+    "simple_image",
+    "json_prismjs",
+    "xml_prismjs",
+    "mistune",
+    "video_videojs",
+    "audio_videojs",
+    "ipynb",
+    "zip",
+    "txt",
+]
+
 RECORDS_RESOURCES_IMAGE_FORMATS = ["." + ext for ext in IIIF_FORMATS.keys()]
 """RECORDS_RESOURCES_IMAGE_FORMATS must contain all possible IIIF formats to ensure their metadata is extracted."""
 

invenio_app_rdm/records_ui/previewer/previewable_zip.py

@@ -0,0 +1,142 @@
+# -*- coding: utf-8 -*-
+#
+# This file is part of Invenio.
+# Copyright (C) 2025 CESNET i.a.l.e.
+#
+# Invenio is free software; you can redistribute it and/or modify it
+# under the terms of the MIT License; see LICENSE file for more details.
+
+"""Simple ZIP archive previewer."""
+
+import sys
+
+from flask import render_template
+from invenio_access.permissions import system_identity
+from invenio_base import invenio_url_for
+from invenio_previewer.proxies import current_previewer
+from invenio_previewer.views import is_container_item_previewable
+
+from ..views.records import PreviewContainerItem
+
+previewable_extensions = ["zip"]
+
+
+def create_container_item_preview_link(record_id, container_filename, item_path):
+    """Create preview link for a container item."""
+    values = {
+        "pid_value": record_id,
+        "filename": container_filename,  # specific .zip file
+        "path": item_path,
+    }
+    return invenio_url_for(
+        "invenio_app_rdm_records.record_container_item_preview", **values
+    )
+
+
+def convert_zip_list_container(entries, folders, record_id, container_filename):
+    """Convert structure returned by files.list_container(...).to_dict()."""
+    counter = iter(range(sys.maxsize))
+
+    def create_parent_hierarchy(node, path):
+
+        for path_item in path:
+            items = node["children"]
+            for child_item in items:
+                if child_item["name"] == path_item:
+                    node = child_item
+                    break
+            else:
+                node = {
+                    "type": "folder",
+                    "name": path_item,
+                    "id": f"folder{next(counter)}",
+                    "children": [],
+                }
+                items.append(node)
+        return node
+
+    def convert_file_entry(key, node):
+        """Convert one node (file or folder)."""
+        converted = {
+            "name": key,
+            "type": "item",
+            "id": f"item{next(counter)}",
+        }
+
+        # Copy metadata fields if they exist
+        for field in ("size", "compressed_size", "mime_type", "crc", "links"):
+            if field in node:
+                converted[field] = node[field]
+
+        # create preview link
+        container_item_extension = key.split(".")[-1].lower()
+        if is_container_item_previewable(container_item_extension):
+            converted["links"].update(
+                {
+                    "preview": create_container_item_preview_link(
+                        record_id, container_filename, node["key"]
+                    )
+                }
+            )
+        return converted
+
+    def convert_folder(key, node):
+        """Convert one node (file or folder)."""
+        converted = {
+            "name": key,
+            "type": "folder",
+            "id": f"folder{next(counter)}",
+            "children": [],
+            "links": node["links"],
+        }
+        return converted
+
+    # Root folder
+    root = {"type": "folder", "id": -1, "children": []}
+
+    # Convert items of root
+    for folder in sorted(folders, key=lambda x: x["key"]):
+        folder_key = folder["key"].split("/")
+        converted = convert_folder(folder_key[-1], folder)
+        hierarchy_position = create_parent_hierarchy(root, folder_key[:-1])
+        hierarchy_position["children"].append(converted)
+
+    for entry in sorted(entries, key=lambda x: x["key"]):
+        entry_key = entry["key"].split("/")
+        converted = convert_file_entry(entry_key[-1], entry)
+        hierarchy_position = create_parent_hierarchy(root, entry_key[:-1])
+        hierarchy_position["children"].append(converted)
+
+    return root
+
+
+def can_preview(file):
+    """Return True if filetype can be previewed."""
+    return (
+        file.is_local()
+        and file.has_extensions(".zip")
+        and not isinstance(file, PreviewContainerItem)  # we are top level file
+    )
+
+
+def preview(file):
+    """Return the appropriate template and pass the file and an embed flag."""
+    from invenio_rdm_records.proxies import current_rdm_records_service
+
+    tree_raw = current_rdm_records_service.files.list_container(
+        system_identity, file.record["id"], file.filename
+    ).to_dict()
+
+    converted_tree = convert_zip_list_container(
+        tree_raw["entries"], tree_raw["folders"], file.record["id"], file.filename
+    )
+    tree_list = converted_tree["children"]
+    return render_template(
+        "invenio_previewer/previewable_zip.html",
+        file=file,
+        tree=tree_list,
+        limit_reached=False,
+        error=None,
+        js_bundles=current_previewer.js_bundles + ["previewable-zip.js"],
+        css_bundles=current_previewer.css_bundles + ["zip_css.css"],
+    )

invenio_app_rdm/records_ui/views/__init__.py

@@ -48,6 +48,8 @@
 from .records import (
     draft_not_found_error,
     not_found_error,
+    record_container_item_download,
+    record_container_item_preview,
     record_detail,
     record_export,
     record_file_download,
@@ -114,13 +116,25 @@ def create_blueprint(app):
             default_view_func=record_file_preview,
         )
     )
+    blueprint.add_url_rule(
+        **create_url_rule(
+            routes["record_container_item_preview"],
+            default_view_func=record_container_item_preview,
+        )
+    )
 
     blueprint.add_url_rule(
         **create_url_rule(
             routes["record_file_download"],
             default_view_func=record_file_download,
         )
     )
+    blueprint.add_url_rule(
+        **create_url_rule(
+            routes["record_container_item_download"],
+            default_view_func=record_container_item_download,
+        )
+    )
     blueprint.add_url_rule(
         **create_url_rule(
             routes["record_thumbnail"],

invenio_app_rdm/records_ui/views/decorators.py

@@ -3,6 +3,7 @@
 # Copyright (C) 2019-2025 CERN.
 # Copyright (C) 2019-2025 Northwestern University.
 # Copyright (C)      2021 TU Wien.
+# Copyright (C) 2025 CESNET i.a.l.e.
 #
 # Invenio App RDM is free software; you can redistribute it and/or modify it
 # under the terms of the MIT License; see LICENSE file for more details.
@@ -260,6 +261,48 @@ def view(**kwargs):
     return decorator
 
 
+def pass_container_item():
+    """Decorator to pass a extracted file from container (e.g. zip)."""
+
+    def decorator(f):
+        @wraps(f)
+        def view(**kwargs):
+            pid_value = kwargs.get("pid_value")
+            file_key = kwargs.get("filename")
+            path = kwargs.get("path")
+            extract_kwargs = {
+                "id_": pid_value,
+                "file_key": file_key,
+                "identity": g.identity,
+                "path": path,
+            }
+
+            from invenio_records_resources.proxies import current_service_registry
+
+            file_service = current_service_registry.get("files")
+
+            try:
+                item = file_service.extract_container_item(**extract_kwargs)
+
+                kwargs["container_item"] = item
+                return f(**kwargs)
+
+            except RecordDeletedException:
+                # Redirect to the record page which has proper tombstone handling
+                return redirect(
+                    url_for(
+                        "invenio_app_rdm_records.record_detail",
+                        pid_value=pid_value,
+                    ),
+                    # Use 302 (temporary) instead of 301 since records can be restored
+                    code=302,
+                )
+
+        return view
+
+    return decorator
+
+
 def pass_file_metadata(f):
     """Decorate a view to pass a file's metadata using the files service."""