Merge pull request #282 from GSA/collection

Collection

ckanext/geodatagov/helpers.py

@@ -1,9 +1,10 @@
 import json
 import logging
 
+from ckan import model
 from ckan import plugins as p
 from ckanext.harvest.model import HarvestSource
-from ckan.logic import NotFound, NotAuthorized
+from ckan.logic import NotFound, NotAuthorized, get_action
 
 log = logging.getLogger(__name__)
 
@@ -59,12 +60,45 @@ def get_harvest_source_config(harvester_id):
     return source_config
 
 
-def get_collection_package(collection_package_id):
-    try:
-        package = p.toolkit.get_action('package_show')({}, {'id': collection_package_id})
-        return package
-    except (NotFound, NotAuthorized):
-        pass
+def count_collection_package(source_id, identifier):
+    context = {'model': model, 'session': model.Session}
+    package_search = get_action('package_search')
+    search_params = {
+        'fq': f'harvest_source_id:{source_id} isPartOf:{identifier} include_collection:true',
+        'rows': 0,
+    }
+
+    search_result = package_search(context, search_params)
+
+    return search_result['count'] if search_result['count'] else 0
+
+
+def get_collection_package(source_id, identifier):
+    context = {'model': model, 'session': model.Session}
+
+    package_search = get_action('package_search')
+    search_params = {
+        'fq': f'harvest_source_id:{source_id} identifier:{identifier}',
+        'rows': 1,
+    }
+
+    search_result = package_search(context, search_params)
+
+    ret = None
+
+    if search_result['results']:
+        collection_package_id = search_result['results'][0]['id']
+
+        try:
+            package = p.toolkit.get_action('package_show')(
+                context,
+                {'id': collection_package_id}
+            )
+            ret = package
+        except (NotFound, NotAuthorized):
+            pass
+
+    return ret
 
 
 def string(value):

ckanext/geodatagov/plugin.py

@@ -499,11 +499,22 @@ def before_dataset_search(self, search_params):
             pattern = r'collection_info:"([^"]+?) ([^"]+)"'
             fq = re.sub(pattern, r'harvest_source_id:"\1" isPartOf:"\2"', fq)
             log.info('FQ changed for collection_info')
-        elif 'bulk_process' not in path:
+        elif 'bulk_process' not in path and 'include_collection' not in fq:
             # hide collection's children datasets from regular search
             fq += ' -isPartOf:["" TO *]'
             log.info('Added FQ to hide collection')
 
+        # fq comes in as a string such as '(a:1 b:"2" c:["" to *])'
+        # remove string include_collection=true from fq, if found.
+        # Other values of include_collection will end up with a search term that return no results
+        pattern = r'include_collection:"?([^",\s)]+)"?'
+        match = re.search(pattern, fq, re.IGNORECASE)
+        if match and match.group(1).lower() == 'true':
+            fq = re.sub(pattern, '', fq, flags=re.IGNORECASE).strip()
+            # if include_collection=true is the only fq, we could end up with a set of parentheses.
+            if fq == "()":
+                fq = ""
+
         search_params['fq'] = fq
         return search_params
 
@@ -536,6 +547,7 @@ def get_helpers(self):
             'get_harvest_source_type': geodatagov_helpers.get_harvest_source_type,
             'get_harvest_source_config': geodatagov_helpers.get_harvest_source_config,
             'get_collection_package': geodatagov_helpers.get_collection_package,
+            'count_collection_package': geodatagov_helpers.count_collection_package,
         }
 
     # IActions

ckanext/geodatagov/templates/package/read.html

@@ -3,25 +3,42 @@
 {% set pkg_dict = c.pkg_dict %}
 
 {% block collection_resources %}
-{% set collection_package_id = h.get_pkg_dict_extra(c.pkg_dict, 'collection_package_id', '') %}
-{% if h.get_pkg_dict_extra(c.pkg_dict, 'collection_metadata', '') %}
-<section class="module-content">
-    <h3>{{ _('Collection') }}</h3>
-    <p>{{ _('This dataset is a collection of other datasets.') }}</p>
-    <p><a href="{{ h.url_for('search', collection_package_id=pkg_dict.id) }}" class="btn-collection">{{ _('Search datasets within this collection') }}</a></p>
-</section>
-{% elif collection_package_id %}
-{% set collection_package = h.get_collection_package(collection_package_id) %}
+
+{% set identifier = h.get_pkg_dict_extra(c.pkg_dict, 'Identifier', '') %}
+{% set collection_sourceid = h.get_pkg_dict_extra(c.pkg_dict, 'harvest_source_id', '') %}
+{% set collection_ispartof = h.get_pkg_dict_extra(c.pkg_dict, 'isPartOf', '') %}
+
+{#
+ 1.  we check each dataset to see if it is a collection parent using its identifier.
+ if count_collection returns a value other than 0, then we know that this dataset is a collection parent.
+
+ 2.  we use collection_ispartof to tell if this dataset is a collection child. 
+ if collection_package returns a dataset, then we know its collection parent is found.
+#}
+
+{% set count_collection = h.count_collection_package(collection_sourceid, identifier) %}
+
+{% if count_collection or collection_ispartof %}
 <section class="module-content">
     <h3>{{ _('Collection') }}</h3>
+    {% if count_collection %}
+    {% set collection_info =  collection_sourceid ~ ' ' ~  identifier %}
+    <p>This dataset is a collection of {{ count_collection }} other datasets.</p>
+    <p><a href="{{ h.url_for('search', collection_info=collection_info) }}" class="btn-collection">{{ _('Search datasets within this collection') }}</a></p>
+    {% endif %}    
+
+    {% if collection_ispartof %}
+    {% set collection_info =  collection_sourceid ~ ' ' ~  collection_ispartof %}
+    {% set collection_package = h.get_collection_package(collection_sourceid, collection_ispartof) %}
     {% if collection_package %}
         <p>{{ _('This dataset is part of the following collection:') }}</p>
         <ul class="dataset-list unstyled">
             {% snippet "snippets/package_item.html", package=collection_package, truncate=75 %}
         </ul>
     {% else %}
         <p>{{ _('This dataset is part of a deleted collection.') }}</p>
-        <p><a href="{{ h.url_for('search', collection_package_id=collection_package_id) }}" class="btn-collection">{{ _('Search other datasets within the same collection') }}</a></p>
+        <p><a href="{{ h.url_for('search', collection_info=collection_info) }}" class="btn-collection">{{ _('Search other datasets within the same collection') }}</a></p>
+    {% endif %}
     {% endif %}
 </section>
 {% endif %}