Merge branch 'master' into develop

Author: MarkCalvert

ckanext/datavic_odp_theme/config.py

@@ -6,7 +6,7 @@
 
 
 CONFIG_BASE_MAP = "ckanext.datavicmain.dtv.base_map_id"
-DEFAULT_BASE_MAP = "basemap-positron"
+DEFAULT_BASE_MAP = "basemap-vic-topographic"
 
 CONFIG_BASE_URL = "ckanext.datavicmain.odp.public_url"
 

ckanext/datavic_odp_theme/plugin.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import logging
 from typing import Any, Optional
 
 from flask import Response, session
@@ -15,6 +16,8 @@
 from ckanext.datavic_odp_theme.views import get_blueprints
 from ckanext.datavic_odp_theme.helpers import get_helpers, group_list
 
+log = logging.getLogger(__name__)
+
 
 class DatavicODPTheme(p.SingletonPlugin):
     p.implements(p.IConfigurer)
@@ -198,24 +201,103 @@ def after_dataset_create(self, context, pkg_dict):
                 group.add_package_by_name(pkg_dict.get('name', None))
 
     def after_dataset_update(self, context, pkg_dict):
-        self._trigger_after_resource_create(pkg_dict)
+        self._submit_new_resources_only(pkg_dict)
+
         group_id = pkg_dict.get('category', None)
         if group_id:
             group = model.Group.get(group_id)
             groups = context.get('package').get_groups('group')
             if group not in groups:
                 group.add_package_by_name(pkg_dict.get('name'))
 
+    def _submit_new_resources_only(self, pkg_dict):
+        """Submit only newly added resources to xloader during dataset update.
+
+        Compares current resource IDs against the previous activity snapshot
+        to detect new resources. URL changes for existing resources are
+        handled by the parent xloaderPlugin via ``notify()``.
+
+        Falls back to submitting all resources if no activity data is
+        available (e.g. activity plugin disabled, data migration).
+        """
+        current_resources = pkg_dict.get("resources", [])
+        current_res_ids = {
+            r.get("id") for r in current_resources if r.get("id")
+        }
+
+        previous_res_ids = self._get_previous_resource_ids(pkg_dict.get("id"))
+
+        if previous_res_ids is None:
+            log.info(
+                "No previous activity for package %s — "
+                "submitting all %d resources",
+                pkg_dict.get("id"),
+                len(current_resources),
+            )
+            for resource in current_resources:
+                self._infer_format_and_submit(resource)
+            return
+
+        new_res_ids = current_res_ids - previous_res_ids
+
+        if not new_res_ids:
+            return
+
+        log.info(
+            "Detected %d new resource(s) for package %s: %s",
+            len(new_res_ids),
+            pkg_dict.get("id"),
+            new_res_ids,
+        )
+
+        for resource in current_resources:
+            if resource.get("id") in new_res_ids:
+                self._infer_format_and_submit(resource)
+
+    def _get_previous_resource_ids(self, pkg_id):
+        """Return resource IDs from the most recent activity, or ``None``
+        if unavailable.
+        """
+        if not pkg_id or not p.plugin_loaded("activity"):
+            return None
+
+        try:
+            activities = tk.get_action("package_activity_list")(
+                {"ignore_auth": True},
+                {
+                    "id": pkg_id,
+                    "limit": 1,
+                    "include_hidden_activity": True,
+                },
+            )
+        except Exception:
+            return None
+
+        if not activities:
+            return None
+
+        prev_pkg = activities[0].get("data", {}).get("package", {})
+        prev_resources = prev_pkg.get("resources", [])
+        return {r.get("id") for r in prev_resources if r.get("id")}
+
+    def _infer_format_and_submit(self, resource):
+        """Infer the resource format from its URL if missing, then submit."""
+        if resource and not resource.get("format"):
+            if not resource.get("url_type"):
+                url_without_params = resource.get("url", "").split("?")[0]
+                resource["format"] = (
+                    url_without_params.split(".")[-1].lower()
+                )
+        self._submit_to_xloader(resource)
+
     def _trigger_after_resource_create(self, pkg_dict):
-        """Dataset syndication doesn't trigger the `after_resource_create` method.
-        So here we want to run submit for each resource after dataset creation.
+        """Submit all resources after dataset creation.
+
+        Syndication via ``package_create`` does not trigger
+        ``after_resource_create``, so we handle it here.
         """
         for resource in pkg_dict.get("resources", []):
-            if resource and not resource.get("format"):
-                if not resource["url_type"]:
-                    url_without_params = resource["url"].split('?')[0]
-                    resource["format"] = url_without_params.split('.')[-1].lower()
-            self._submit_to_datapusher(resource)
+            self._infer_format_and_submit(resource)
 
     def _submit_to_datapusher(self, resource_dict):
         """The original method doesn't check if `url_type` is here. Seems like