[FEATURE] Creating a route to GET duplicate documents

https://forge.iscpif.fr/issues/1360

gargantext_web/api.py

@@ -136,9 +136,8 @@ _ngrams_order_columns = {
 
 from rest_framework.views import APIView
 from rest_framework.response import Response
-from rest_framework.exceptions import APIException
+from rest_framework.exceptions import APIException as _APIException
 
-_APIException = APIException
 class APIException(_APIException):
     def __init__(self, message, code=500):
         self.status_code = code
@@ -167,6 +166,53 @@ def Root(request, format=None):
     })
 
 
+class NodesChildrenDuplicates(APIView):
+
+    def get(self, request, node_id):
+        # input validation
+        if 'keys' not in request.GET:
+            raise APIException('Missing GET parameter: "keys"', 400)
+        keys = request.GET['keys'].split(',')
+        # metadata retrieval
+        metadata_query = (Metadata
+            .query(Metadata)
+            .filter(Metadata.name.in_(keys))
+        )
+        # build query elements
+        columns = []
+        aliases = []
+        for metadata in metadata_query:
+            # aliases
+            _Metadata = aliased(Metadata)
+            _Node_Metadata = aliased(Node_Metadata)
+            aliases.append(_Node_Metadata)
+            # what shall we retrieve?
+            columns.append(
+                getattr(_Node_Metadata, 'value_' + metadata.type)
+            )
+        # build the query!
+        groups = list(columns)
+        duplicates_query = (get_session()
+            .query(*( [func.count()] + columns ))
+            .select_from(Node)
+        )
+        for _Node_Metadata, metadata in zip(aliases, metadata_query):
+            duplicates_query = duplicates_query.outerjoin(_Node_Metadata, _Node_Metadata.node_id == Node.id)
+            duplicates_query = duplicates_query.filter(_Node_Metadata.metadata_id == metadata.id)
+        duplicates_query = duplicates_query.filter(Node.parent_id == node_id)
+        duplicates_query = duplicates_query.group_by(*columns)
+        duplicates_query = duplicates_query.order_by(func.count().desc())
+        duplicates_query = duplicates_query.having(func.count() > 1)
+        # return results
+        return JsonHttpResponse([
+            {
+                'count': duplicate[0],
+                'values': duplicate[1:],
+            }
+            for duplicate in duplicates_query
+        ])
+
+
 class NodesChildrenMetatadata(APIView):
 
     def get(self, request, node_id):

gargantext_web/urls.py

@@ -48,6 +48,7 @@ urlpatterns = patterns('',
     url(r'^api$', gargantext_web.api.Root),
     url(r'^api/nodes/(\d+)/children/metadata$', gargantext_web.api.NodesChildrenMetatadata.as_view()),
     url(r'^api/nodes/(\d+)/children/queries$', gargantext_web.api.NodesChildrenQueries.as_view()),
+    url(r'^api/nodes/(\d+)/children/duplicates$', gargantext_web.api.NodesChildrenDuplicates.as_view()),
     url(r'^api/nodes/(\d+)$', gargantext_web.api.Nodes.as_view()),
     url(r'^api/nodes$', gargantext_web.api.NodesList.as_view()),